Vous êtes sur la page 1sur 872

1/11111/11111111111111111111111

PB97-103964

Volume II
Sessions 1 through 4

December, 1994

Sponsor: U.S. Federal Highway Administration (FHWA)


Co-Sponsors: ADSC: International Association of Foundation Drilling
Deep Foundation Institute (DFI)
Transportation Research Board (TRB)
American Association of State Highway and Transportation
Officials (AASHTO)

REPRODUCED BY: ~
U.s. Depanment of Commerce
NatiDnal Technicallnformalion Service
Springfield, Virginia 22161
Technical Report Documentation Page
1. Report No. P897-1eJ3964 3. Recipient's Catalog No.
FHWA-SA':96-085 1111111111111111111111111111111
4. Title and Subtitle 4. Report Date
Proceedings: International Conference on Design and December 6, 1994
Construction of Deep Foundations. VOLUME 2
6. Performing Organization Code:

7. Author(s) 8. Performing Organization Report No.

9. Performing Organization Name and Address 10. Work Unit No.(TRAIS)


John Hou All American Soils, Inc. 711 South Bristol St.,
Santa Ana, CA 92703
11. Contract or Grant No.
DTFH61-92-Z-00114
12. Sponsoring Agency Name and Address 13 Type of Report and Period Covered
Conference Proceedings
Office of Technology Application
Office ,of Engineering/Bridge Division
Federal Highway Administration
400 Seventh Street, S.W.
Washington, D.C. 20590
14. Sponsoring Agency Code

HTA-20
15. Supplementary Notes
Federal Highway Administration (FHWA) Project Manager: Chien-Tan Chang
Conference Chairperson and Technical Consultant: Jerry DiMaggio
16. Abstract
, The U.S. Federal Highway Administration (FHWA), the International Association of Foundation
Drilling (ADSC), Deep Foundation Institute (DFI), Transportation Research board (TRB) and American
Association of State Highway and Transportation Officials (AASHTO) co-sponsored an International
Conference on Design and Construction of Deep Foundations, from December 6, to 8, 1994 in Orlando,
Florida.
The principal objective of the conference was to improve the cost-effectiveness of deep
foundation systems for transportation-related projects by documentation and exchange of innovative
practices and improved techniques. The conference provided an opportunity to share technological
advances in deep foundation from a global perspective. The conference subjects ranged from theoretical
modeling of pile foundation systems to driven and drilled pile transportation-related case studies.
Volume 1 of the proceedings consists of papers by eleven internationally known keynote speakers
invited to provide state-of-the-art knowledge and perspectives on Deep Foundations. Volumes 2 and 3 of
the proceedings consist of 115 papers selected from 191 abstracts in response to the call for papers in
1992. The papers were w~tt~n by more th~~_30~ authors and co-authors representing 23 countries.
17. KeyWords Drilled Shafts, Auger Case Piles, Pile 18. Distribution Statement
Testing, Axial and Lateral Loads, Stanamic No restrictions. This document is available to the
Testing, Materials and Equipment, Driven Piles, public from the National Technical Information
Drilled Piles, Integrity Testing, Axial Capacity,
physical Modeling, Mumerical Techniques, Soil Service, Springfield, Virginia 22161.
Structure Interaction, Driveability. Dynamic
Capacity Prediction.
19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No. of 22. Price
Unclassified Unclassified Pages
855
Form COT F 1700.7 (8-72) Reproduction of completed page authorized
Preface

In 1992, the United States Federal Highway Administration in its continuing efforts to
improve quality in the design and construction of highway and roadway systems, initiated this
international conference on the Design and Construction of Deep Foundations, with Orlando,
Florida selected as the conference site. The purpose of the conference was to provide an
international forum concerning design, analysis, and construction of deep foundation systems.
Accordingly, the following conference topics were selected:

Numerical Techniques
Physical Modeling
Integrity & Capacity Testing of Load Bearing Elements
New & Innovative Driven & Drilled Pile Types
Load Transfer Behavior (Single Elements & Groups)
Deep Foundation Experiences
State of Practice
Specifications and Contracting Documents
Compression, Tension, and Lateral Loads
Special Design Events (i.e., Seismic, Ship Impact and Scour)
Geologic and Subsurface Data (Interpretation & Application)
Economic Considerations and Selection Methods
Innovations in Material and Equipment
Soil-Structure Interaction between Foundations & Superstructures
Load & Resistance Factor Design (LRFD) vs Working Stress Design

In cognizance of these conference themes, eleven Keynote Speakers were invited to


provide State-of-the-Art knowledge and perspectives; these are:

1. Dr. George G. Goble, GRL & Assoc., Boulder, Colorado "Pile Driving - An International State-of-the-Art"

2. Dr. Gary M. Norris, University of Nevada, Reno, Nevada "Seismic Bridge Pile Foundation Behavior"

3. Manuel A. Fine, Berminghammer Corp. Ltd., Hamilton, Ontario, "Innovation in Equipment and Materials
and Their Impact on Design and Construction of Driven Pile Foundations"

4. Dr, Michael w. O'Neill, University of Houston, Houston, Texas "Drilled Shafts: Effect of Construction on Performance
and Design Criteria"

5. Dr. John H. Schrnertmann, Schrnertmann & Crapps, Inc. Gainesville, Florida "Past, Present, and Future Practice in
Deep Foundations, with Florida Emphasis"

6. Lawrence H. Roth, CH2M Hill, Oakland, California "Current Practice Issues for Consulting Engineers and DOTs"

7. Dr. Za-Chieh Moh, Moh & Assoc, Inc. Taiwan, Rep. of China, "Current Deep Foundation Practice in Taiwan and
Southeast Asia"

8. Dr. Ken Fleming, Cementation Piling & Foundation, Ltd. Hertfordshire, England, "Current Understanding and Control
of Continuous Flight Auger Piling"

9. Clyde N. Baker, STS Consultants, North Brook, Illinois "Current' U.S. Design and Construction Practices for Drilled
Piers"

10. Dr. Roger Frank, Centre D'Enseignement et de Recherche en Mecanique des Sols, Paris, France "The New Eurocode
and the New French Code for the Design of Deep Foundations"

11. Herb Minatre, Bay Shore Systems, Inc. Benicia, California "Tools and Equipment Used in the Drilled Shaft Industry"
The call for papers it 1992 drew a response of 191 abstracts. After the review process,
115 papers were selected and published in the Proceedings. The papers were written by more
than 300 authors and co-authors representing 23 countries. Every paper in this Proceeding has
undergone a rigorous review process, to those reviewers, the Steering Committee is most grateful.
At the conference, up to 100 Exhibitors displayed and demonstrated their products.

We, the steering committee, trust that this conference and Proceedings provided the
participants and readers a useful forum and infonnation source for deep foundation design and
construction, to the benefit of public users of highway systems. We wish to thank the Federal
Highway Administration, ADSC: The International Association of Foundation Drilling, the Deep
Foundations Institute (DFI), the Transportation Research Board (TRB), and American Association
of State Highway and Transportation Officials (AASHTO) for their sponsorship of this
conference, the authors and co-authors, invited keynote Speakers, and exhibitors who have
assisted and supported the "International Conference on Design and Construction of Deep
Foundations"

December 1994.

Steering Committee:

Jerry DiMaggio (Chairperson) Federal Highway Administration (FHWA)


John Hooks Federal Highway Administration (FHWA)
AI DiMiIlio Federal Highway Administration (FHWA)
Chien-Tan Chang Federal Highway Administration (FHWA)
AI Kilian Federal Highway Administration (FHWA) (formerly Washington DOT)
John Ledbetter North Carolina DOT
George Goble GRL & Associates
Scot Litke ADSC: The International Association of Foundation Drilling (ADSC)
Robert Compton Deep Foundations Institute (DFI)
Joe Caliendo Utah State University
Frank C. Townsend University of Florida
John P. Hou All American Soils, Inc. (AASn

Moderators

The Steering Committee would like to thank the following persons who served as
moderators, and thus greatly assisted in guiding the technical sessions of the conference:

Ted Ferra gut Alan 1. Luttenegger Dan Bruce


Joe Caliendo Mike O'Neill AI DiMillio
John Walkinshaw John Ledbetter Dan Brown
Clyde Baker Jim Cahill Manuel Fine
AI Kilian Mary Ellen Hynes Roger Frank
Frank Townsend Ilan Juran Ed Keane
Wes Moody Fred Kulhawy Mike McVay
Barry Berkovitz Pricilla Nelson Mark Morvant
George Goble Bengt Fellenius
Manuscript Reviewers

The Steering Committee would like to thank the following persons who reviewed
manuscripts, and thus greatly assisted in the quality and technical standards of the papers in these
Proceedings:
ABAR, NARlMAN GRABNER, WILLIAM J. PIRES, JOSE A.
ADAMS, GREGG B. GRAHAM, JAMES S. PRYSOCK,ROD
ALAMPALLI, SRIEENUAS HANNIGAN, PATRICK PUTCHA, SASTRY
ALLEN, TONY HASAN, SHAFl RA THFON, SCOTT
ALPERSTEIN, ROBERT HOLDER, SAM RAUSCHE, FRANK
ALW AHAB, RIYAD M. HOLEYMAN, ALAIN E. REESE, LYMON C.
ARGO, RICHARD H. HOOVER, K.R. RIAZ, MUHAMMED
ARMSTRONG, STAN HORVATH, JOHN S. RODRIGUEZ-PEREZ, CARLOS E.
ARONOWITZ, ARNOLD HOUGHTON, ROBERT C. ROLLINS, KYLE M.
AVASARALA. SWAMY K.V. HOVRANI, NABIL ROSCHKE, PAUL N.
BAILEY, WARREN HUSSEIN, MOHAMAD RUPIPER, STAN
BAKER, CLYDE N JR. ISENHOWER, WILLIAM M. SAMARA, EMILE
BEIKAE, MOHSEN ISKANDER, MAGUED SAMMAN, MAC.
BENAMAR, AHMED JAY SCOTT DE NATALE SA YED, SA YED M.
BENDA, CHRIS JIMENEZ, PEDRO SCHAEFFER, VERNON R.
BENVIE, DONALD JOLLY, JOSEPH P. SCHNORE, AUSTARS R.
BERKOVITZ, BARRY JORY, BRIAN W. SHEAHAN, JAMES
BHARIL, RAJNEESH KEANE, EDWARD SHEFFIELD, RJCHARD
BOGHRAT, ALIREA KILIAN, ALAN SIEL, BARRY
BONOMO, RONALD J. KNIGHT, BUBBA SIZEMORE, JEFF
BOOTH, ANDREW T. KOON, MENG CHUA SMITH, ALEC D.
BORDEN, ROY H. KULHAWY, FRED H. SMITH, TREVOR D.
BRENNAN, JIM LAI, PETER W. SPEER, DANIEL
BRENNER, BRIAN, LAIER, JAMES E. SVINKIN, MARK
BRJAUD, JEAN-LOUIS LAMBRECHTS, JIM TAWFlG, KAMAL S.
BROWN, DAN A. LEDBEITER, JOHN F. JR. TOMMEN, GLEN
BRUCE, DONALD A. LENS, JOHN E. TOWNSEND, FRANK C.
BURCH, SCOTT LEVINE, MIKE WALKINSHAW, JOHN
BUTLER, BERI'-.1ARD E. LEW, MARSHALL WARRINGTON, DON C.
BYRNE, JOSEPH H. LIANG, ROBERT Y. WEATHERBY, DAVID E.
CALIENDO, JOSEPH A. LIKINS, GARLAND WEBSTER, SCOTT
CANNER,RON LONG, JAMES H. WHITAKER, SCOTT S.
CANNON, RAI\'DY LONG, RICHARD P. WITHIAM, JAMES L.
CARVILLE, CHESTER A. LUKAS, ROBERT G. WONG, DANIEL
CHASSIE, RONALD G. LUTENEGGER, ALAN J. WOO, EDWIN P.
CHENEY, RICHARD MAMOON, SAIF M. YANG,CHANGZHONG
CLARK, G.N. MATillS, HENRY, JR YEH, SHAN-TAl
DASH, UMAKANT MAYNE, PAUL W. YOKLEY, BUDDY
DA VlDSON, JOHN L. McCASKIE, STEPHEN L. YORK, DONALD L.
DEMIR, SlDDIKA McGILLIVRAY, ROSS T. ZANDI, FIROOZ
DIMILLIO, ALBERT F. MCVAY, MICHAEL
DIRKS, KERMIT MECKLIN, PAUL R.
DRUMM, ERIC C. MENSAH, FRANCIS D.
DUMAS, CHRISTOPHER MIDDENDORP, PETER
DUNN, PHILIP MOODY, WESLEY
EALY, CARL D. MORGAN, MELVIN W.
ELIAS, VICTOR NAGLE, GALEN S.
ENGEL, RICHARD L. NEELY, WILLIAM J.
ENGLE, DAVE NEFF, THOMAS
ERIKSON, CHRIS M. NEGUSSEY, DAWIT
ESNARD, 1. B. NORRIS, GARY
FENNESSEY, TOM O'NEILL, MICHAEL W.
FERREGUT, CARLOS O'ROURKE, PAT
FLICK, LOREN D. OTOOLE, DONALD
FONG, MITCHELL L. ODEM, GEORGE
FOSHEE, JON OSBORN, PETER
FRITZ, MIKE OSTERBERG, JORJ O.
GEARY, MALOl'.'E PASSE, PAUL, D.
GOBLE, GEORGE G. PEARLMAN, SETH L.
GOEITLE, RICHARD J. III PE1RASSIC, KERRY
For any additional reviewers whose names were inadvertently omitted, we offer our
sincere apology for this oversight. the quality of these proceedings reflect the efforts of these
reviewers.
TABLE OF CONTENT

(Volume I)
Keynote Papers

"Pile Driving - An International State-of-the-Art" . . .. .... .... . ... 1


George G. Goble (26 pgs)

"Seismic Bridge Pile Foundation Behavior" . . .. .... .... .... '" 27


Gary M. Norris (110 pgs)

"Innovation in Equipment and Materials and Their Impact on Design and Construction of
Driven Pile Foundations"
Manuel A. Fine (Presentation Only)

"Drilled Shafts: Effects of Construction on Performance and Design Criteria". '" 137
Michael W. O/Neill and Khaled M. Hassen (51 pgs)

"Past, Present and Future Practice in Deep Foundations, with Florida Emphasis" . . . 188
John H. Schmertmann and David K. Crapps (21 pgs)

"Current Practice Issues for Consulting Engineers and DOTs" .. .... .... '" 209
Lawrence H. Roth (27 pgs)

"Current Deep Foundation Practice in Taiwan and Southeast Asia" ... .... '" 236
Za-Chieh Moh (24 pgs)

"Current Understanding and Control of Continuous Flight Auger Piling" .... '" 260
W.G. Ken Fleming (19 pgs)

"The New Eurocode and the New French Code for


the Design of Deep Foundations" . . .. .... .... .... '" 279
Roger Frank (26 pgs)

"Current U.S. Design and Construction Practices for Drilled Piers" '" .... . .. 305
Clyde N. Baker, Jr. (19 pgs)

"Tools and Equipment Used in the Drilled Shaft Industry"


Herb Minatre (Presentation Only)
(Volume m
Session lA
Drilled Shaft Case Histories

"Some Experiences on Bored Cast-In-Situ Reinforced Concrete Piles


in Bangladesh - A Case Study" . . . .. .... . 325
!vID.N. Amin, M.F. Karim, and A. Uz Zaman (12 pgs)

"Compression Bored Piles in Singapore Old Alluvium: Performance and Design" .. 337
C.E. Ho and e.G. Tan (15 pgs)

"Axial compression Behavior of Two Drilled Shafts in Piedmont Residual Soils" .. 352
D.E. Harris and P. Mayne (16 pgs)

Session IB
Auger Cast Piles

"Auger-Cast Piles in Clays" . . .. .... .... . .. 368


D. O. Wong (17 pgs)

"Utilization and Quality Control of Augercast Piles" ... .... .... .... . .. 385
A.T. Booth, and K.A. McIntosh (16 pgs)

"Auger-Cast Piles". . ... . . .. .... .... .... . .. 401


J.J. Brennan (14 pgs)

Session lC
Special Design Events

"Limit States Design for Deep Foundations" . . .. .... . .. 415


B.H. Fellenius (12 pgs)

"Soil Resistance Factors for LRFD of Driven Piles" . . .. .... .... .... . .. 427
J. Berger and G.G. Goble (12 pgs)

"Seismic Retrofit of Foundations for a Double-Deck Viaduct" ., .... .... . .. 439


M.E. Fowler, R.E. Johnston, G.S. Nagle (15 pgs)

Session ID
Expert Systems
"Diagnostic Expert System for Drilled Shaft Foundation Construction" 454
S. Demir, DJ. Fisher, and M.W. O'Neill (15 pgs)
"A Prototype Expert System for Foundation Design" . .. .... .... .... '" 469
P.N. Roschke, J. Briaud, and E.G. Funegard (15 pgs)

"A Numerical Solution for the Dynamic Pile Driving Problem". .... .... '" 484
S.M. Mamoon (9 pgs)

Session 2A
Static Axial Pile Testing

"Pile Load Test is a Proof Test, But?" .. . '" 493


F. Abdrabbo and R. El-Hansy (20 pgs)

"Optimum Specification for Static Load Testing" .. 513


M. England and W.G.K. Fleming (12 pgs)

"Effects of Plugging on Piles Installed in an Overconsolidated Clay" .. .... '" 525


G.A. Miller and AJ. Lutenegger (16 pgs)

Session 2B
Design Methods for Axial and Lateral Loads

"Drilled Shaft Load Test Database and an Evaluation of the Program SHAFTUF" . . 541
J.L. Davidson, L.D. Spears, and P.W. Lai (15 pgs)

"A New Approach to the Prediction of Drilled Pier Performance in Rock" . .. .., 556
J.P. Seidel and C.M. Haberfield (15 pgs)

"Improved Methods for Evaluation of Bending Stiffness of Deep Foundations" ... 571
W.M. Isenhower (15 pgs)

Session 2C
Statnamic Testing

"Statnamic Tests on Steel Pipe Piles Driven in a Soft Rock" . .. .... .... '" 586
T. Matsumoto and M. Tsuzuki (15 pgs)

"Analytical Study of Statnamic Test of a Cast-In-Place Concrete Pile". .... '" 601
K. Yamashita, Y. Tsubakihara, M. Kakurai, and T. Fukuhara (15 pgs)

"A Comparison of Statnamic and Static Field Tests at Seven FHWA Sites" ... 616
P. Bermingham, C.D. Ealy, and J.K. White (15 pgs)
Session 3A
Innovations in Materials and Equipment

"A Promising Method for Improving Drilled Pier Performance in Rock" .... ... 631
C.M. Haberfield, S. Baycan, and TD. Chamberlain (15 pgs)

"Design Guidelines for Screw Anchors" . . ." ... 646


C.A. Carville and RW. Walton (10 pgs)

"Hardware Solutions for Quality Control of Deep Foundations-Overview" ... . .. 656


G. Likins, F. Rausche, and D. Peterman (14 pgs)

"The Development of a New Pile Load Testing System" .. 670


T. Fujioka and K Yamada (15 pgs)

"Polyethylene Coating for Downdrag Mitigation on Abutment Piles" .. .... ... 685
KS. Tawfiq (14 pgs)

"Correlations Between the Standard Penetration Test (SPT) and


the Measured Shear Strength of Florida Natural Rock" . .... .... .... . .. 699
H.R Ramos, J.A. Antorena, and G.T McDaniel (13 pgs)

Session 3B
Design Methods for Driven Piles

"An Evaluation of Predicted Ultimate Capacity of


Single Piles From SPILE and UNIPILE Programs" . . .. .... .... .... . .. 712
S. Kumar, V. Avasarala, J.L. Davidson, M.C. McVay (12 pgs)

"Static Pile Capacity Predictions with SPT91" .. .. 724


J.A. Caliendo, M. Bartholomew, P.W. Lai, F.C. Townsend, and M.e. McVay (14 pgs)

"Neural Network Predictions of Load- Deflection Curves for Concrete


Piles in Florida" .. .... . . .. .... . . .. . .. 738
G.G. Goble, J.F. Kreider, P. Curtis and J. Berger (21 pgs)

"Pile Load Test Database and an Evaluation of the Program SPT91" .. .... . .. 759
J.L. Davidson, F.C. Townsend, P.F. Ruesta, and lA. Caliendo (15 pgs)

"Design Parameters for Steel Pipe Piles Driven in a Soft Rock" ... 774
Y. Michi, M. Tsuzuki, and T Matsumoto (15 pgs)

"Aspects of Pile Design for Cyclic and Dynamic Loading


with Reference to API Conditions for Offshore Structures" ... .... .... ... 789
lP. Sully, M. Paga, R Bea, E. Gajardo, R Gonzalez, and A.F. Fernandez (14 pgs)
Session 3C
Numerical Method

"Integrity Testing of Drilled Shafts: A Computer Vision Approach" .. .... '" 803
M.M. Samman and M. Biswas (14 pgs)

"Bearing Capacity of Expanded-Base Piles in Till" .... . .. 0 0 0 ' " 817


W. J. Neely (12 pgs)

"Elasto-Plastic Analysis of Laterally Loaded Piles" o. . o. .... .... .


0 0 0 829 0

S. Tanaka, Y. Sawano, Fo Okumura, A. Nishimura, and T. Watanabe (13 pgs)

"Analysis of Piles Under Lateral Loading with Nonlinear Flexural Rigidity" o. '" 842
L.C. Reese and S.T. Wang (15 pgs)

"Detennination of P-Y Curves in Fractured Rock Using Inclinometer Data" .. '" 857
D.A. Brown and S. Zhang (16 pgs)

"Modelling of the Shaft Capacity of Grouted Driven Piles in Calcareous Soils" '" 873
H.A. Joer and M.F. Randolph (15 pgs)

Session 4A
New and Innovative Driven and Drilled Pile Types

"A Novel Foundation Piling System - The Spear Pile" 0 888


R.T. McGillivray and M.B. Hussein (15 pgs)

"Effect of Cantilever Plate of a Foundation Pile on


Pile Deflection Under Lateral Load" .. 0 0 0 0 o. . 0 903
K. Trojnar (10 pgs)

"The GEWI-PILE, A Micropile for Retrofitting,


Seismic Upgrading and Difficult Installation" .. . .. 0 0 0 0 0 913
T.F. Herbst (18 pgs)

"The Drill Pile Method - New Low-Noise, Low Vibration Piling Method" . .. '" 931
M. Hashimoto, O. Hashimoto, So Nishizawa, K. Ishihara, and Y. Sakurai (15 pgs)

"Researches Into the Behavior of High Capacity Pin Piles'" ... 0... 0... '" 946
D.A. Bruce, loR. Wolosick and A.L. Rechenmacher (23 pgs)

"Testing of Geojet Units Under Lateral Loading" . 0 . 969


D. Spear, L.C. Reese, GT. Reavis, and So Wang (11 pgs)

"Helical Plate Bearing Members, A Practical Solution to Deep Foundations" . '" 980
S. Rupiper (12 pgs)
Session 4B
Load Transfer Behavior & Soil Structure Interaction

"Recent Developments in the Design of Piles Loaded by Lateral Soil Movements" . 992
D.P. Stewart, M.P. Randolph, and R.I. Jewell (15 pgs)

"Deep Timber Pile Foundations Using Geotextile Reinforced Embankments" . .. 1007


J.H. Byrne (15 pgs)

"Design of Deep Foundations for Cut-and-Cover Tunnels" . 0 0 0... 1022


B. Brenner, C. Gagnon, and c.K. Shah (15 pgs)

"The Computational Method of Settlement and Loads of Pile Group" 0 0 1037


L. Jie, Q.S. Yin, and Z.G. Ming (8 pgs)

"Group Load Test: 9-Timber Piles" o 0" 0 0... 1045


R. Alperstein and C. Dobryn (9 pgs)

"Footings with Settlement-Reducing Piles in Non-Cohesive Soil" .... .... . 0 1054


P. D. Long (15 pgs)

"The Mechanics of Piled Embankment" 0 0 1069


B.B.K. Huat, W.H. Craig, and P. Ali (14 pgs)

Session 4C
Driveability and Dynamic Capacity Prediction

"Dynamic Testing of an Instrumented Drilled Shaft" '" .. o' .... .. 1083


FoC. Townsend, J.F. Theos, D. Bloomquist, and M. Husse~n (12 pgs)

"Validity of Predicting Pile Capacity by Pile Driving Analyzer" .. 1095


P. Lai and C.L. Kuo (8 pgs)

"Dynamic and Static Tests on Driven and Cast-In-Place Piles" . .... .... .. 1103
M.H. Hussein and W.M. Camp (15 pgs)

"A Rational and Usable Wave Equation Soil Model


Based on Field Test Correlation" . . . . . . . . . . . . . . 1118
F.Rausche, G. Likins, and G. Goble (15 pgs)

"Examination of the Energy Approach for Capacity Evaluation of Driven Piles" .. 1133
S.G. Paikowsky and V.A. LaBelle (17 pgs)

"Influence of Pile Parameters on Pile Driveability" .... .... .... .... .. 1150
M.R. Svinkin (15 pgs)
"Soil Modeling for Pile Vibratory Driving" . . .. 1165
A.E. Holeyman and C. Legrand (14 pgs)

(Volume ITn
Session SA
State of Practice

"Specifications: Constructible as Well as Thorough" . . .. .... .... " 1179


E.C. Yokley, Jr. (13 pgs)

"Drilled Shaft Bridge Foundations in North Carolina" .. .... .... .... " 1192
IF. Ledbetter, Jr. (11 pgs)

"The Practice of Deep Foundations in Highly Degradable Soils" .... .... " 1203
E. P. Corona (13 pgs)

"A Failure Case Study of Island Method Excavation in Soft Clay" ... .... .. 1216
Y.K. Chu (15 pgs)

Session 5B
Integrity Testing

"Quality Assurance of Drilled Shaft Foundations with Nondestructive Testing" " 1231
L.D. Olson, M. Lew, G.C. Phelps, K.N. Murthy, and B.M. Ghadiali (13 pgs)

"Pile Integrity Testing in North Carolina" ................ " 1244


N. Abar (18 pgs)

"Load and Integrity Testing of Auger Cast Piles for a Multi-Level Building" . .. 1262
C. Mirza and M. Montgomery (10 pgs)

"A Comparison of the Efficiency of Drilled Shaft Down-Hole Integrity Tests" " 1272
A.G. Davis and B.H. Hertlein (15 pgs)

Session 5C
Interpretation, Application, and Use of
Geologic and Subsurface Investigation Data

"Prediction and Performance of Pile Using In Situ Test Data" .. . . .. " 1287
A. Tanaka (9 pgs)
"Response of Drilled Shaft Foundations in Karst During Construction Loading" .. 1296
Q. Tang, E.C. Drumm, and RM. Bennett (14 pgs)

"Site Variability in Karst Fonnations" . . . . . . . . . . . . . . 1310


P.D. Passe (11 pgs)

"Dynamic Probing Test to Detennine Driven Pile Capacity" . .. .... .... .. 1321
REngel, M. Riaz, V. Dalal, D. Hanhilammi, A.I. Husein, and R Y. Liang (16 pgs)

Session 6A
New and Innovative Driven and Drilled Shafts

"Newly Developed Toe-Grouted Bored Pile in


Soft Bangkok Clay: Performance and Behavior" 1337
W. Teparaksa (lS pgs)

"The Design of Starsol Piles" . .. 1352


L.J. Whitworth and C.M. Gilbert (14 pgs)

"Spin-Fin Pile Perfonnance" .... . . .. .... .... .... .. 1366


D. Nottingham (11 pgs)

Session 6B
Numerical Techniques

"Static Lateral and Moment Behavior of Rigid Drilled Shafts in Sand" 1377
S."W. Agaiby and F.H. Kulhawy (13 pgs)

"A Model of Shaft Pile-Soil Interaction During Driving. Numerical Simulation" .. 1390
A. Benamar (13 pgs)

"Analysis and Design of Piles Through Embankments" . .... .... .... .. 1403
H.G. Poulos (19 pgs)

Session 6C
Case Histories - Driven Piles
"Experience in High Capacity Pile Driving Monitored
Using Pile Dynamic Analyzer" . . . . .. .... .. 1422
J.P. Jolly, S.L. Rathfon, and A. Donofrio (19 pgs)

"Dynamic Pile Testing for Five Mile Bridge and Tunnel Crossing" ... .... .. 1441
S.D. Webster, PJ. Hannigan, and D.A. Lawler (14 pgs)
"Design and Construction of Deep Pile Foundations, Kings Bay, Georgia" . .. .. 1455
A.D. Smith and E.B. Kinner (16 pgs)

Session 7A
Physical Modeling

"Ohio DOT Research on the Development of In-Situ Testing Techniques for Wave
Equation Soil Parameters" . . . .. .... .... .... 1471
REngel, M. Riaz, V. Dalal, D. Hanhilammi, and RY. Liang (14 pgs)

"Drilled Shaft Friction Evaluation via Pullout Tests" ... .... . ... .. 1485
F.C. Townsend, C.E. Dunkelberger, and D. Bloomquist (14 pgs)

"Effect of Surcharge Loading on Pile Capacity" .... " 1499


G. Achari and RC. Joshi (12 pgs)

"Experimental Program to Monitor the Behavior of Piers in Expansive Soils". " 1511
J. Behar, C. Ferregut, and M. Picornell (9 pgs)

"Laboratory Modelling of Drilled Piers in Soft Rock" ., .... .... .... .. 1520
C.M. Haberfield, J.P. Seidel, and LW. Johnston (15 pgs)

"Response of a Flexible Cyclic Loaded Pile in Sand by Centrifuge Modelling" .. 1535


D. Levacher and F. Schoefs (17 pgs)

Session 7B
Case Histories - Driven Piles

"Lateral Load Test at Thorofare Bay, North Carolina" . . .. .... .. 1552


D. T. O'Toole (16 pgs)

"A Case History of the Foundation Design Procedures for the Palm Valley Bridge" 1568
S. Burch (14 pgs)

"Comparative Results of Static and Dynamic Tests of High Capacity Piles" .. .. 1582
E.P. Woo, T.W. Pelnik III, F. Rausche, and S.R Weaver (15 pgs)

"The Use of the PDA in the Construction of the Manati River Bridge" 1597
P. Jimenez (15 pgs)

"Installation of High Capacity Driven Piles in Coastal Northern Florida" .... " 1612
M. Sharp, B. Jory, W. Knight, and M. Hussein (15 pgs)

"Shaft Resistance on H-Piles in Cohesionless Soils" .... . . .. .... " 1627


P.F. Bailey, S.L. Borg, and A.R Schnore (15 pgs)
"Driven Steel Pile in Bouldery Soils" . . . . .. 0 0 0 0 0 0 0 0 0 0 0 1642
SoT Yeh, R.D. Andrew and T.D. Bowen (15 pgs)

Session 7C
Axial Capacity

"Performance of Long, Slender Piles Installed by Jacking" .... .. 0 0 0 0 o' . 0 1657


LHo Wong (9 pgs)

"Testing, Performance and Reliability of Small Diameter Pipe Piles" o. . 0 0 0 0 0 1666


CoA. Carville and J.So Pack (10 pgs)

"Pile Load Test Data" 0" 0 1676


J. Vrymoed (15 pgs)

"Tension and Compression Testing of Various Piles in San Francisco Bay Mud" . 1691
J.L. Walkinshaw and So Healow (26 pgs)

"Uplift Load Tests on Drilled Shafts in Gravels to Evaluate Side Friction" . .. .. 1717
KoM. Rollins, R.C. Mikesell, R.J. Clayton, and E. Keane (15 pgs)

"Long Term Relaxation Test on a Pile Foundation" . . . . .. 1732


L. Vulliet, H. Fleischer, and S. Terentieff (12 pgs)

Papers Not Presented In The Conference

"Site Geology and Selection of Optimum Foundation System" . . .. .. 1747


S.K. Jain and K. Kishore (14 pgs)

"Behaviour of Bored Piles of A Cooling Tower and Chimney:


Group and FEM Analysis. . . .. .... .... .... .. 1761
M. Chiorboli, 1. Limido, M. Nesti, F.A. Uliana (15 pgs)

"Modeling of Swift Delta Bridge Piles From


Soil Nailed Wall Construction" . . . .. .... .... ., 1776
ToD. Smith, R. Kimmerling, R. Barrows (13 pgs)

"Utility of Drilled Shaft Load Test Results" '" 0 . . 0... .... .. 1789
M.H. Wysockey and J.H. Long (15 pgs)

"Predicting the Ultimate Compressive Capacity of a


Long 12-H-74 Steel Pile" . . . .. .... . .. 0 1804
I.E. Laier (15 pgs)

"Lug Behavior For Model Steel Piles In Frozen Sand - Theory And Experiment"
R. M. Al Wahab (7 pgs) 0 . . . .. . . . . . . . . 0... . . 1819
SOME EXPERIENCES ON BORED CAST-IN-SITU REINFORCED
CONCRETE PILES IN BANGLADESH - A CASE STUDY

ABSTRACT

Geologically Bangladesh is composed of young fluvio-deltaic


deposits, except some older deposi ts of Mio-Pleistocene age.
Most of this landmass is inundated during monsoon. Due to
intricate river system and flooding, bridges and drainage
structures are integral components of the highways. Until
the 1970's the bridges were normally constructed on masonry
caisson foundations. Presently instead of caissons, a bored
cast-in-situ reinforced concrete pile is widely adopted in
bridge foundation, because it is found cost-effective, it
saves appreciable construction time, and the equipment and
technology are readily available.

The varied and difficult sub-soil condition and uncontrolled


construction procedure, frequently cause defects in the
piles. This paper identifies the causes of defects and
suggests quality control measures for bored cast-in-situ
piles.

INTRODUCTION

Since the independence (1971) development activities in


Bangladesh increased manifold. Obviously the transportation
sector got the top priority, which involved construction of
bridges and drainage structures. These structures are
generally founded on deep foundation. The existence of weak
upper stratum (Amin et al., 1987) and scouring of bed and
bank of the rivers are the primary reasons for adopting deep
foundations.

Previously in Bangladesh the bridges were normally


established on open caisson of masonry or reinforced
concrete construction where excavation were carried out by
grabbing. Presently the caisson foundations are replaced by

MD. N. Amin, M. F. Karim and A. Uz Zaman.


325
bored cast-in-situ piles. The bored cast-in-situ pile
foundations are widely constructed for readily available
technology and equipment.

Construction of such piles involves excavation of an uncased


or unlined pile hole and under water tremie concreting.
Drilling of pile hole is done by percussion method i.e. by
means of a chopping bit attached at the end of drill rods.
The bore hole is filled with bentonite-slurry or sometime
locally available clay-slurry.

The slurry circulates through drill rods and flow up along


the sides of the bore holes with cuttings to the surface and
are separated from the slurry by decantation for
recirculation.

The bentonite-slurry stabilizes the sides of boreholes. And


in addi tion the hydrostatic head due to higher speci fie
gravity of fluid does not allow the borehole to collapse.
Back flow of water into the borehole is avoided. The
thixotropic action of the bentoni te mud on the wall of
boreholes acts as further protection against caving in.

In earlier times, pile capacity and its integri ty were


determined by static load test. Recently pile integri ty
testing by non-destructive dynamic response method has been
introduced to determine structural soundness of the piles.

To identify the type and causes of defects in the pile


shafts, a number of pile integrity tests are performed on
the piles constructed by different agencies in several
geological formations. The pile integrity tests were
performed by using low strain integrity tester (GRL, 1987).
Often defects in the pile shafts (like decrease or increase
in diameters, vo ids, separat ion, incl us ion, neck ing and
shortening in length) are identified. These defects
orig inate either for di ff icul t subso i 1 condition or for
faul ty construction procedure. Some of these defects and
their remedial measures are discussed.

BORED CAST-IN-SITU PILE IN DIFFERENT GEOLOGICAL ZONES

Bangladesh is a large fluvio-del taic plain (Alam et al.,


1990). It is composed of young unconsolidated sediments
deposited in a prograding deltaic environment.

The engineering characteristics of soils are varied and


dependent on active geological processes. The vertical and
spatial extent are distinguished by local physiographical
aspects. For the purpose of geotechnical engineering, the
geological deposits of Bangladesh can be broadly divided

326
into following five zones ( Figure 1 I.

Zone A: Alluvial deposits


Zone B: Old deltaic deposits
Zone C: Young deltaic and coastal deposits
Zone D: Residual soil
Zone E: Mio-Pliocene hills

Except the zone E all zones are considered in this paper for
observation of bored cast-in-si tu piles construction and
testing (Figure 3) by the pile integrity tester (GRL,
1987)

o 100 200Km.
! ! I

zaN E - A
IN 0 I A .~ A II \J II i a \-
. deposits
~"-~.'-'.~...JI".~~ .........
........ ZaN E ... 8
) ~ aI d . d.e I t a i c
deposits
ZaNE-C
~ Young deltaic
Q-nd coastal
deposits
('.~ zaN E- 0
~
). .
ill]] Residual
\ \.
,,,j
( So i I

IN 0 I A za N E - E

o Mio.,.Pliacene
Hill s

BENGAL

Figure 1. Map of Bangladesh showing the geological zones of


different engineering soil.
(1 - Dhaka City, 2 - Parbatipur Railway Workshop,
3 - Chittagong, 4 - Location of Bridge Structure
- 3, Cox's Bazar-Dohazari Road).

327
GRAIN SIZE GRAIN SIZE
S P T Su kN/m! DIST. ~ S P T Su kN/m!
DIST. "
25 75 10 30 20 60 25 75 10 30 20 60

2 t-'~ 2 -

)
.7~
E~
4 ;::t.l: 4 -

6
~~
~~~
6 -
s>!
8 ~~E 8 -
=~o.r-.

};.;:
10- "~'f:
. r-~::
10-
~2; ,::
1 2- 12-

14- 14-

16-
;1 "
16-

18- 1~
.
20 2b

~- GRAIN SIZE ~- GRAIN SIZE


s
l1, DIST." S P T SII kN/m! eJB DIST.% P T Su kN/m!
!~_ 25 75 10 30 20 60 Q _ 25 75 30 20 60
1..---1----1'_ ---L---J---L.-...

2 -

6
/ 1

6
- ------\

8 8 -

10-

12-
,,-L

14-

16- '_-;
"

18- 18-
1 .-~

2c 2d
D SAND I,;; I SILT I_I CLAY

Figure 2. Soil profiles of different geological zones


(2a - Zone A, 2b - ZoneB, 2c Zone C and
2d - Zone D; S.P.T. Standard Penetration
Test blows count; Su - Undrained Shear Strength)

328
),...---<1 1.50 M&

r
V.l-Avg-Ampd

, -r~T- -~iT'--ri--!I,6-11--~IO 1F!4--,,-----~.r;


J
~
'TT-T' " ! "I I I i
1.--'\, ! ---~----~--~ I ~r tM
t_ .L!
._ I.i __ . _J
!
.L. _j __ .1: __ i_.
-.L ij_
_.__ L I
- --L

~50.0

cc======:::._:-=_::-...:.: :-.: -_: _-: ._: :._:.:=:-:::=~=====~~== ====:J 24.00


4.00
m
m/m.
(A) Separation of concrete at depth 19.54m.

IJel-Avg-AMpd
~ ,i\', 8 n

\
..-
'
.'-

2.il
ill.eil M

(8) Increase and decrease in diameter of pile shaft


-----4 1. 50 Ms ,. .e
Vel-Avg-Ampd

I
1;\ 4 8
j2 6

--rn ,
4 m

--
1I '

i I . .J----- t--. I
I r-- 1'------ ;~
UJ
I' ~

I
=:==1
~16.00
10 0
.
m

(e) Shortening in length of pile (Drilled 16m, achieved 11.25m).


1.50 Hs
V~]-Avg-Ampd

c ,

I
~ :8 ~:2 ~6
i ,i I I !
i ~O
: I
I
~~ I i ~ r---... ! t-, ~ I
i
I i
- -~
I
I
""T-~
r, {\ r i

1\
,~

I V
I VI \1)vf\ ,-'I_I
I
I
I
i
!
I I :
i
I
Ii i
I J
I
I
I I

C ---d
===========::::::=:===-:::'==-===
100.0

j~.

~
DC

.co
I,

0.1"5

(D) Large necking (36.4% decrease) in pile shaft at 9m length.

Figure 3. Results of pile integrity tests.

329
Zone A - Alluvial deposits

The zone A is formed of alluvial fan deposits and floodplain


deposi ts. These deposits are composed of coarse to fine
silty sand with lenses of channel sand (at places pebbly)
and inter-stream silts. To the lower regime the floodplain
deposits are relatively finer and composed of sandy silt and
silt with occasional thin clay patches.

A generalized soil profile from this zone is shown in Figure


2a. The subsoil upto a depth of about 15 m is composed of
medium to loose silty sand. The underlying strata are mostly
composed of dense monotonous medium to coarse grained sand.
Severe caving occurs during excavation in this sand.

In this zone, significant variation in pile diameter


throughout the shaft is observed by pile integri ty test
(Fig. 3). The wiggles, waists and necks are both for poor
constructional methods and presence of alternate sand and
silt of inter-stream deposits.

Zone B - Old deltaic deposits

The zone B is formed of old deltaic deposits laid down by


the Ganges, Brahmaputra and Meghna river system. The soils
of this zone are composed of sand and silt with patches of
marshy clay and peat. A generalized soil profile of this
zone is shown in Figure 2b. The upper stratum up to a depth
of about 8m is composed of soft clayey silt. Below this the
subsoil is composed of dense sand and silt which overlies
dense silty sand.

The observed bored cast-in-si tu piles of this zone show


al ternate increase and decrease in diameter wi thin the depth
of 6 to 9m. These are due to the presence of al ternate
deltaic deposits of silt and fine sand. The holes remain
moderately stable at depth.

Zone C - Young deltaic and coastal deposits

The vast plain land in the southern part of Bangladesh is


covered with young deltaic and coastal (tidal) deposits
(Amin et. al 1987). In Figure-2c a generalized soil profile
of the zone is shown. The sub- so il is composed of so ft
clayey and sandy silt. The sub-soil towards the coast are
often dominated by woody organic rich mangrove deposits.

A significant increase in pile diameter throughout the shaft


is observed in this zone. However, in some cases severe
reduction and increase in pile diameter is observed near the
pile top due to alternate layers of soft sandy to clayey
silt.
330
Zone D - Residual soil

The upper layer in the zone D is composed of highly plastic,


stiff to very stiff sandy clay - silt underlain by dense to
very dense silty sand. It is considered to be over
consolidated. A generalized soil profile from this zone is
shown in Figure-2d.

Most of the piles constructed in this zone are structurally


sound. However, defective piles are also observed which may
be attributed to poor construction procedure. Some of the
piles brought under observation are found shorter in length
than the desired length. The shortening of the piles
occurred due to inadequate cleaning and bore hole collapse.
Also the expansive red soil causes necking in the pile
shaft.

CASE OBSERVATIONS

In the process of constructing bored cast-in-si tu reinforced


cement concrete (R.C.C. ) piles, detail study and
observations were made on locally developed equipment,
slurry fluid characteristics, sub-soil parameters, drilling
techniques, cleaning of boreholes, installation of re-bar
cage and tremie pouring of concrete. These reveal various
shortcomings in the prevailing practices, procedures and
techniques of bored cast-in-si tu R. C. C. pile construction in
Bangladesh. It was found that adoption of certain
inexpens i ve and simple qual i ty control measures to the
existing procedure can some times produce good results. The
study undertaken in the major geological regions of the
country are summarized in the following paragraphs.

Excavation and stabilization of pile holes

Structurally sound bored cast-in-si tu piles are achieved


where the construction process, from the beginning of
drilling or excavation to the completion of concrete
pouring, is uninterrupted and continuous.

It is observed that the choice of right personnel is very


important for the construction of piles. Present
construction equipment comprising of tripod set, power
winch, mud pump, drill rod, circular chopping bits and
tremie concrete pouring operations are effectively used to
construct sound vertical piles. This however, needs services
of qualified and experienced drilling foreman~ conversant
with relevant geological formations.

The stabil i ty of pile holes is pre-requisite for


constructing structurally sound piles. The hydraulic drive

331
rotary drill rigs wi th direct mud circulation method are
found to excavate more stable boreholes. Even unlined stable
pile holes are effectively drilled by rotary drill rigs for
the construction of raker piles (1:5; Horizontal to
Vertical). The reverse circulation rotary rig which produces
suction and conventional direct circulation percussion rig
where excavation is made by chopping bit, often destabilize
the pile holes.

Drilling rate is found as an important aspect for stability


of bore holes. The satisfactory drilling rate in zone B for
1200 mm diam pile hole is 4m/hr and in zone C for 600 mm
pile hole it is 5m/hr.

It is observed that the unlined pile holes in sub-soil with


8PT 0 to 1 (zone C) are unstable and require permanent steel
lining and temporary casings are required in sub-soil where
8PT is at least greater than one.

In all the geological regions the pile holes were found to


be stable when drilling fluid level within the temporary
casing, maintained around two meters above adjacent ground
water table. The suitable characteristics of drilling mud
for respective geological zones are shown in Table - 1. It
has been observed that the higher density slurry is required
in poorer soils.

Table - 1. Characteristics of drilling fluids (ICE, 1978).

~ Viscosity with I Density II


Zone Marsh Funnel (gm/cm 3 )
Viscometer
Zone A 30 - 90s 1.05-1.10
Zone B 30 - 90s 1. 04-1. 07
Zone C 30 - 90s 1.03-1.05
Zone D 30 - 90s 1. 03-1. 05

The design values of viscosity and density of drilling


fluid, vis-a-vis stability of pile holes are effectively
determined from a trial pile hole.

The pile holes for 924 number of piles in Parbatipur Railway


Workshop Project (Zone D) were stabilized efficiently. The
typical quality control practice was as follows:

The mud characteristics and drilling technique are pre-


determined by a trial pile hole.

332
The drilling mud is then checked for stability of borehole
to twice the construction period of a pile.

Cleaning of pile holes

A continuous supply of fresh slurry from a reserve tank for


removal of contaminated slurry and cuttings from the base
normally ensures clean pile hole which leads to firm end
bearing.

In the difficult sub-soil condition, removal of contaminated


slurry and debris from the pile hole by compressed air
injected through inverted funnel attached to tremie tip is
found effecti ve. The example of two piles in Dhaka city
located 5m apart is shown in Figure -4 where performance of
two identical piles is found to be quite different. These
piles are taken under static load test. The pile hole of
Pile No. 1 was cleaned with compressed air and the hole of
Pile No. 2 was cleaned by conventional technique i.e. by
flushing of fresh slurry. The appl ied test loads are 350
Tons and 275 tons on Pile No.1 and 2 respectively. The
established ultimate load carrying capacity of Pile No.2 is
229 tons and above 350 tons for Pile No.1 (Figure - 4).

The pressfre of slurry-fluid during drilling in excess of


1.5 kg/cm (20 psi) or jetting downward, is found to loosen
the subsoil around the pile tip and reduces the end bearing
vis a vis capacity of a pile.

Pouring of concrete

Present observation indicates that a controlled concrete


pouring operation is essential in achieving .structurally
sound piles.

The grade C 25 (compressive strength 3 25 N/mm 2 ) concrete


with minimum cement content of 400 kg/m and high slump (125
- 150 mm) with well graded shingles ensures free flow and
forms continuous monolithic concrete shaft. The tremie
diameter of 200 mm for under water concrete pouring is found
most efficient.

Satisfactory pouring and clean pile tip are ensured by using


disposable plug with tremie, lowered within the bottom of
150 mm of a clear pile hole. The first charge of concrete
shall fill the tremie length plus twice the volume required
to fill the space below the tremie tip. The tip of the
tremie shall always be placed around 2m deep wi thin the
fluid concrete. If accidently the tremie be lifted above the
level of concrete, the tremie shall completely be removed
from pile hole, cleaned and installed again with disposable
plug.
333
350T
350

135T

100

50

20 TYPE OF PILE:BORED CAST-IN-SITU


DIAMETER: 600 mm.
LENGTH: 19m.

r-
300 (J) PILE NO.2 275T
z
2500
229 I-
z
200 - 183 T
o
<t
150 0
-J

100

50

_-'----I_...J.-_.L--+_-..L_~~-&4+-:.::.28=......-3::..0~=-=3~2=-..:3~4~.-=4~6~4~
50 40 30 -20 10 2
H0 U R S

10 ~
~

20 z
30 I-
Z
I1J
40 ~
I1J

50 ~
t-
I1J
(J)

Figure 4. Load test diagrams for two identical R.C.C


Bored Cast-in-Situ Piles in Dhaka City.
Test Load of Pile No.1 - 350 tons.
Test Load of Pile No.2 - 275 tons.

334
Study on a bridge foundation

A Comparative study between bored cast-in-situ pile


foundation and caisson foundation of Bridge Structure - 3 at
Dohazari - Cox's Bazar Road section, under Road Improvement
Project (RIP) situated in zone E (Figure - 1) is summarized
in Table-2.

The analysis reflects that the cost for caisson foundation


is 68% higher than that of bored cast-in-situ pile
foundation. The time requirement for completing foundation
work of the same bridge by bored piling is only one fifth of
the time required for caissons.

TABLE - 2. Cost and time for pile and caisson foundations.


Foundation Cost Presumptive
type (US $ in million) time of
execution
Pile 0.192 60 Days
Caisson 0.320 200 Days

CONCLUSIONS

Bangladesh is covered with young geological deposits


composed of unconsol idated sand, silt and clay Deep
foundation are required for most of the bridges, culverts,
and drainage structures.

Bored cast-in-situ R.C.C. pile foundation is widely accepted


for readily available equipment and technical know-how. A
cost comparison indicates that the bored cast-in-situ R.C.C.
pile foundation is more cost effective than the caissons or
other type of deep foundation and reduces substantial
construction time.

The bored cast-in-situ piles are often found to be


structurally defective. The identified defects are increase
or decrease in diameter, voids, separation, inclusion,
necking of pile shaft and shortening in length. These
defects are formed either for poor construction procedure or
ignoring the geological aspects of the deposi ts. These
defects are successfully identified by low strain I no-
destructive pile integrity tester.

Reasonably good piles can be constructed by utilizing


existing techniques and equipment. This requires some
development in the pile hole stabilization, cleaning and
concrete pouring operations.

335
REFERENCES

Alam, M.K., Hasan, A.K.M.S., Khan, M.R. & Whitney, J.W.


(1990); Geological Map of Bangladesh. Geological
Survey of Bangladesh Publication, Dhaka.

Amin, M.N., Kabir, M.H., Saha, G.P. and Ahmed, M. (1987);


Geotechnical behavior of soils from coastal region of
Bangladesh. Ninth south east Asian Geotechnical Conference,
Bangkok, Thailand. Vol-1, pp 5-1 - 5-13.

I.C.E. (1978); Model Procedures & Specifications,


Institution of Civil Engineers, London, UK, p.79,82.

G.R.L. (1987); A Catalogue of Computed Pile Top Velocities


from Travelling Wave Reflections Nonuniform piles, Goble
Rausche Likins and Associates, Inc., Ohio, USA.

336
COMPRESSION BORED PILES IN SINGAPORE
OLD ALLUVIUM: PERFORMANCE AND DESIGN

Chu Eu HOI and Chin Gee Tan2

ABSTRACT

Bored piles have been widely used as fOWldations in the Old Alluvium in Singapore.
The boreholes are usually very stable and dry due to its partially cohesive or cemented
properties and low penneability. The design of bored piles has traditionally been based
on Meyerhofs equations for cohesionless soils. A review of results from more recent
ultimate load tests showed that in the case of the Old AlluviUlIl, the skin friction
resistance developed were very much higher than estimated The measured ultimate end
bearing resistance were however lower than estimated. The purpose of this paper is to
present data gathered from several ultimate load tests on instrumented bored piles of
various diameters constructed in the Old Alluvium. A new procedure for designing
bored piles in Old Alluvium is proposed.

1Associate, 2Geotechnical Engineer, Arup Geotechnics, Ove Amp and Partners


Singapore, 5001 Beach Road, #04-03, Golden Mile Complex, Singapore 0719.

337 Ho C E and Tan C G


INTRODUCTION

The geology of a large part of Eastern Singapore comprises mainly of the Old Alluvium
formation. Bored piles have been widely used as foundations in the Old Alluvium due
to its relative ease of construction. The borehole is usually very stable and dry.
Although much experience has been accumulated in the past in constructing bored piles
in the Old Alluvium, few piles have been instrumented and tested to failure to
investigate the shaft friction and end bearing capacities for design.

The behavior of bored piles in compression has previously been reported by a number
of investigators, for example, Yong et al (1982), Chin et al (1985), Buttling and
Robinson (1987) and Chan and Lee (1990). Current design practice is commonly
based on Meyerhofs equations for cohesionless soils (Meyerhof, 1976). However,
experience has shown that actual load carrying capacities were very much higher than
those estimated by Meyerhofs equations. Due to more recent foundation construction
in the Old Alluvium, in which the authors were involved, opportunity was sought to
carry out ultimate load tests on instrumented bored piles to study the mobilization of
skin mcrion resistance and end bearing resistance in this soil material. Based on the
results of these ultimate load tests and a review of available pile test data reported by
others, a new procedure for designing bored piles in Old Alluvium is proposed.

PROPERTIES OF SINGAPORE OLD ALLUVIUM

The engineering geology and properties of the Old Alluvium Formation in Singapore
has been reviewed by Tan et al (1980) and Poh et al (1987). The Old Alluvium can be
largely described as an overconsolidated sandy material consisting of medium dense to
very dense soil grains with deferring degrees of cementation. The characteristic soils are
poorly-graded clayey sands or sand-clay mixtures. 1bin beds of clay and silt can also be
found within the overconsolidated sandy deposits. The soil grains are well compacted
and permeability is usually very low. For the sandy soils, the internal mction angle cP
was reported to be between 30 to 45, and the cohesion c between 20 to 60 kPa. For
the clayey soils, the cohesion Cu ranges from 40 to 120 kPa.

The two sites where bored piles were installed for testing were located in the Eastern
part of Singapore Island, one at Tampines and the other at Changi. At both sites, the
existing formation levels have been derived from cutting down of hilly outcrops of Old
Alluvium in the past.

The properties of the Old Alluvimn at the test sites are given in Fig. 1. The soils were
mainly medium dense to very dense silty sands. The grain size fraction for clay and silt
particles were less than 15% and does not appear to vary significantly with depth. The
proportion of sand grains accounted for about 40 to 80% of the soil mass. The amount
of gravel size particles was more significant at the Tampines site with grain size
fractions of up to 37% being recorded. The sand fraction generally decreased with
depth whereas the gravel fraction increased with depth. The total sand and gravel

338 Ho C E and Tan C G


0 0

TP1 1
TP1
-I
BH11 BH11
-2 2

oJ
Sand -4
E E
~
5
-5 Gravel -5
Q,
Q,

c! c! -{l

-8
Si~
-9

10
0 40 BO 120 160 200
20 40 60 80
SPTValue, N ( blow/0.3m )
Grain size fraction ('lEI)

0
0
TP2
-2 TP2
BH19 -2
BH19
-4

-{l

E E -8

.:::: -10
Q. .::::
c!
a
c! -12

14

16

-18

-20 20
0 10 20 30 40 50 60 70 BO 0 40 BO 120 160 200
Grain size fraction ('lEI) SPT Value, N ( blow/0.3m )

-2
TP3
BH12
-4

-{l

-8
E E
10
.:::: .::::
Q. a -12
c! c! -14

16

18

20
-22
10 20 30 40 50 60 70 BO 0 40 BO 120 160 200
Grain size fraction ('lEI) SPT Value, N ( blow/0.3m )

Figure 1(a) Properties of Old Alluvium at Changi

339 Ho C E and Tan C G


0.-----------...,
-2
TP4
BH2

Grain Size
Fraction
Not Measured
atBH2

40 80 120 160 200


SPT Value. N ( blow/O.:Jm )

0,.....---------....,
TP5
0,.....---------...,
2 2
TPS
BH21 BH21

e e -a
10
=
Co t
c! ~ 12
14

-16

18 18

-20 20
0 10 20 :JO 40 50 60 70 80 0 40 80 120 160 200
Grain size fraction (%) SPTValue, N ( blowIO.:Jm )

0 0
TP6 TP6
-2
BH52 BH52
-4 -4

e -a -e -a
.:=
Q. -10 t -10
c! ~
-12 -12

-14 -14

-16 16

18 18
0 10 20 :JO 40 50 60 70 80 0 40 80 120 160 200
Grain size fraction (%) SPTVaJue, N ( blowIO.:Jm )

Figure 1(b) Properties of Old Alluvium at Tampines

340 Hoe E and Tan C G


fraction accounted for between 75 to 80% on average. The Standard Penetration Test
(SPT) N values showed large variations within the body of the Old Alluvium, Fig. 1.
The range of SPT N values were typically between 40 to 100 blows/O.3m. There were
indications of increase in soil strength with depth, although this was not so obvious for
the Changi site.

PILE CONSTRUCTON AND TESTING

All test piles were constructed using conventional augers and cleaning buckets. The
boreholes were generally dry. The test piles were located at the position of existing
boreholes to provide meaningful correlation and interpretation of test results. The piles
were instrumented with six levels of vibrating wire strain gauges (A to F), each with
four gauges set diametrically across. The reinforcement cages were placed to the toe of
the piles. Three 8mm diameter rod extensometers were also installed within 25mm
diameter steel pipes at 3 levels within the pile shaft to measure the displacement along
the pile. All test piles were cast with ready mixed concrete with compressive strength of
35MPa.

TABLE 1 Test Pile Details

Test pile TP1 TP2 TP3 TP4 TP5 TP6

Site Changi Changi Changi Tampines Tampines Tampines

Nominal 600 1350 1000 900 600 750


Diameter ( mm )

Actual 680 1360 1425 940-1180 9151060 645 740 - 945


Diameter ( mm )

Pile length ( m ) 6.40 13.15 13.20 21.55 18.70 14.45

Casing length no casing no casing 3.40 10.70 5.00 6.35


below ground
level (m)

Date of 3.5.93 7.6.93 24.3.93 19.3.93 13.1.93 5.11.92


installation

Date of testing 18.5.93 - 19.7.93 17.4.93 5.4.93 - 2.2.93 - 18.11.92 -


23.5.93 24.7.93 21.4.93 8.4.93 6.2.93 22.11.92

Maximum 5918 29990 16932 15137 7308 11566


test load ( KN )

341 Ho C E and Tan C G


0
A A
1

2 ~

e
H e
-3

::t: ::t:
l- I-
e. -4 e.
W
Q
w
Q 14
5 16

-18
~
20

7
0 '000 2000
LOAD
3000 4000
( KN)
5000 6000 6
LOAD
8 10
(MN)
'2 '4 16

-2

-4

e e

342 Ho C E and Tan C G


6000,..----------------... 0.030,..-----------------,
Qu = 5942KN

5000 Applied load, P 0.025


Qsu 3343KN = x

4000 0.020
/ z
::.::
x
/
/ " E
3000 " E 0.015 x
""o - -
Shaft resistance, Os Q::
x

.... 2000 <l 0.010 x


x
0.005
x
1000
TP1 TP1
O+*-...--__......~-.....,...-"""T'"-...,....-"T""""~ O.ooo+--...---.---.---.---...---..---.----l
o 20 40 60 SO 100 120 140 160 o 20 40 60 eo 100 120 140 160
Pilehead selIlement 6 (mm) Pilehead selllement6 (mm)

JO.,.....-------------""""----. 5.0.,---------------.,
x
Qu = 36933KN
4.5

25
Qsu 21876KN =
4.0

z
::::E
20

.- .- -- i
E
3.5

3.0
x
x

./ Shaft resistance, Os E 2.5 x


15
x

.3"
Q:: 2.0
10 <l 1.5
x
1.0
x
5
TP2 0.5 ~:l TP2
o..jll---.---ll;.,-.---.---.--..,....-..,.............,....----l 0.0 +--..----,...----,...----,---,---,..--,.---1
a 20 40 60 SO 100 120 140 160 o 20 40 60 eo 100 120 140 160
Pilehead seWement, 6 (mm) Pilehead settlement Co (mm)

18.--------------------, 1.6,----------------,..-..,
Qu = 20326KN
16 1.4 Qsu :::0 12784KN

1.2
/-

12 "
z / / "Shaft resistacce, Os z 1.0
::::E ,/ ::E x
10 E
E 0.8
x
8
x
Q:: 0.6 x
6 <l x
0.4 )O(X
4 ..."X
TP3 0.2
TP3
2

O.__............,...--...---"""T'"--+---,----! 0.0+--.....,...--..,....---,..---...,---,----1
JO
a 5 10 15 20 25 JO o 5 10
Pi!ehead settlement..6.. (mm)
15 20 25
Pilehead sealement ~ (mm)

Figure 3(a) Load Versus Settlement Curves and Stability Plots at Changi

343 Ho C Eand Tan C G


344 Ho C E and Tan C G
The Maintained Load Method was adopted in all ultimate load tests. Loads were
applied using hydraulic jacks and acting through calibrated load cells. Pilehead and
extensometer displacements were measured using dial gauges and survey levels read
against reference beams. Any movement of the reference beams were corrected with
respect to two remote bench marks. In all cases, loads were applied using kentledge
blocks as reaction. The base of the kentledge blocks rested on the existing ground. A
pilecap was constructed to ensure even spread of load. The loads were applied in three
cycles to ultimate failure load or a minimum of 3 times the working test load. Each
intennediate load step was observed for 1 hour. The maximum load at each cycle was
maintained for between 18 to 24 hours. Due to the excellent conditions of the Old
Alluvium when excavated, overbreak was minimal. Details of the piles tested are
swnmarized in Table 1.

Fig. 2 shows the load transfer curves for Stratum 1 (A - B), Stratum 2 (B - C), Stratum
3 (C - D), Stratum 4 (D - E) and Stratum 5 (E - F). The load transferred to the pile toe
was between 5.8 and 35.1 % of the maximum applied test load at the pilehead. The
performance of the piles in Old Alluvium were therefore largely dictated by the
characteristics of the shaft resistance. The load versus settlement response of the test
piles is depicted in Fig. 3. It was observed that deviation from the initial linear load-
settlement beha~or commenced at a pilehead settlement of about 4 to 8rnm. Maximum
pile head settlement ranged between 2.7 to 26.5% of the nominal pile diameter.

SKIN FRICTION RESISTAl~CE

The mobilization curves for skin friction resistance fs are shown in Fig. 4. It can be
seen that except for the upper portion of the pile shaft just below the pilecap (Stratum
1), the mobilized skin friction values exhibited a characteristic ductile response. Where
peak and residual values of fs were observed, the difference between them were very
small, ranging typically between 0.4 to 16.6% and averaging about 6%.

For the part of the pile shaft just beneath the pilecap (Stratum 1), there appeared to be
an absence of the ductile response. It was observed that the skin friction resistance
increased indefinitely with increasing applied loads. The skin friction values were also
very much higher than would have been expected of soils of similar strength. This
anomaly may be explained by the fact that the bearing pressures exerted on the ground
beneath the soffit of the pilecap caused an increase in lateral confining stress around the
pile shaft immediately below it. The effect of the increased confining stress resulted in
additional skin friction resistance being induced. This phenomenon was very significant
in the case of Old Alluvium since it is an overconsolidated soil with rather high
stiffness.

Fig. 5 shows the relationship between fsIN ratio and N for soil strata which were not
influenced by the pilecap. In this case, the N value is the average value along the length
of pile shaft over which fs is obtained. It is obvious that there exists a trend of
decreasing fsIN ratio with increasing N values. This implies that the reduction of fs due

Ho C Eand Tan C G
345
1000..------------------, 700,....------------------,

900 TP1 TP4


1__ Stratum 1 ~ Slrawm 2 ... Stlalum 3 600
800 _ S1raIum 04 __ S1raIum 5

700

600

500

400
- - Stratum 1 -e- Stratum 2 ....... Stratum 3
- S1raIum 04 __ S1raIum 5

100

O~---r--_,_-_,-- ........-_,-__I
20 40 60 80 100 120 140 1.60 o 10 20 30 40 50 60
Shaft displacement, S~ (mm) Shaft displacement 6s (mm)

8OOT"""---------------., 300,......---------------.,
700 TP5

~200
(/J

- - S11alum 1 -e- Strarum 2 ........ Strarum 3


_ Stratum 4 -10- Slralum 5

~ 40 60 80 100 1~ 1W 2 4 6 8 10 12 14
Shaft displacement, tfs (mm) Shaft displacement, 6~ (mm)

800,....-----------------, 5OOT"""--------------..
700
TP3 450 TP6

350

- - Strmum 1 -e- Strarum 2 - . - Strarum 3


- - Slrawm 4 __ S1ralIlm 5
c
i:J 100

50

o.i----.----,-----,.--_,----I OJ-_'---'--"""'T'"-_'_-'''''---''-"""'T'"--,---I
o 5 10 15 20 25 o 5 10 15 20 25 30 J5 40 45
Shaft displacement. ~ 5 (mm) Shaft displacement, 8~ (mm)

Figure 4 Mobilization Curves for Skin Friction Resistance, fs

346 Ho C Eand Tan C G


to loss of confining stress during pile construction affects the stronger soils more than
the weaker soils. The range offslN ratio varied largely between 0.8 to 5.4 with extreme
values of 0.8 (lower bound) and 9.1 (upper bound). The fsIN ratio were generally
much larger than the value of 1.0 recommended by Meyerhof (1976) for cohesionless
soils. Chin et al (1985) reported fsIN values of between 3.8 and 6.0 for 30<N<60
blows/0.3m in slightly to very slightly cemented dense sands. Chan and Lee (1990)
suggested fsIN=llK.s(tsf) for Ks=32, i.e. fsIN=3.1(kPa) for Old Alluvium consisting
mainly of very dense, slightly cemented silty or clayey course sands with N>50
blows/0.3m. However, fsIN of 0.8 was obtained for N>100 blows/O.3m in the present
case, Fig. 5. The consistently high values of fsIN ratio is indicative of the effect of
dilatancy of the dense sands as well as the large normal stiffness of the
overconsolidated soil mass constraining the pile shaft.

END BEARING RESISTANCE

Fig. 6 shows the mobilization curves for the end bearing resistance qb. It can be seen
that the trend indicates a steeply increasing curvilinear response of qb with pile toe
displacement 6b. This implies that the end bearing resistance may be more significantly
influenced by the effects of dilatancy of the dense sand, compared with the skin friction
resistance. The reason may be due to the increase of soil strength and stiffness with
depth. The maximum measured qb ranged from 1297.3 to 5642.5 kPa , although none
were yet to be fully mobilized, even at a maximum toe displacement of 25.2% of the
pile diameter for TP1. It was observed that the initial response of the pile toe was very
stiff in all cases. The estimated initial modulus of subgrade reaction for the pile toe was
between 157 to 372 !v1N/m3 This confumed that the accumulation of debri at the pile
toe prior to concreting was minimal and no soft toes were fonned. Fig. 6 indicates an
asymptotic value of 6000kPa for qb.

The relationship between the end bearing resistance qb and the Standard Penetration
Test value Nb is depicted in Fig. 7. Based on the procedure of Bowles(1988), the
value of Nb was obtained by taking the average of the N values between 3 D below
the pile toe and 8 D above the pile toe, where D is the pile diameter. The length of piles
Ls socketted within the Old Alluvium were between 9.7 D and 31.1 D. The maximum
qblNb ratio obtained was 65.7 for TP2 (q,135Omm) for a pile toe displacement of 9.9%
of the pile diameter, although the pile toe response suggested that the ultimate value of
qb had not yet been attained. For the case of TPI (q,60Omm), which was very close to
ultimate failure, the qblNb ratio obtained was 55.0 for a pile toe displacement of 25.2%
of the pile diameter. These values were however very much lower than the qblNb ratio
of 400/3 ( or 133 ) suggested by Meyerhof (1976) for LsID equal or greater than 10.
The much lower qblNb ratio may be due to the sensitivity of the pile toe to stress relief
during pile construction in an overconsoildated soil such as Old Alluvium.

347 Ho C Eand Tan C G


PILE SETTLEMENT AND FACTOR OF SAFETY

The applied load at the pilehead P and the shaft resistance Qs is plotted against pilehead
settlement in Fig.3. It can be seen that at the initial loading stages, the load-settlement
behavior at the pilehead was dominated by the shaft resistance Qs. Beyond a pilehead
settlement of about 4 to 8mm , the load-settlement curve deviates from the initial elastic
response and end bearing resistance begins to develop more significantly as the shaft
resistance becomes fully mobilized. As the response of the skin friction resistance fs is
essentially ductile (Fig. 4), it implies that once slip occurs at the pile-soil interrace , the
settlement of the pilehead becomes very sensitive to the end bearing response. In the
case where soft debris has accumulated at the toe of the pile due to poor workmanship,
very large settlement may result. In this respect, it is suggested that allowable working
load should be limited to a value at which the ultimate shaft resistance Qsu is not
exceeded

For a limiting pilehead settlement of 12mm, the corresponding load at the pilehead PI
for the various test piles are summarized in Table 2. Qsmax is the maximum mobilized
shaft resistance. Qbmax is computed based on an ultimate qb of 6000 kPa. The
ultimate bearing capacity is given by Qu = Qsmax + Qbmax. The calculated Factor
of Safety on the shaft resistance ( Fs = QsmaxlPl ) ranged between 1.05 and 1.68.
The calculated global Factor of Safety ( FOS = QulPI) ranged between 1.35 to 2.65.
It can be seen that although the global Factor of Safety FOS was less than the
traditionally adopted value of 2.5 to 3.0, pilehead settlement of 12mm can still be
achieved, provided that adequate factor of safety Fs on the ultimate shaft resistance is
available. The base resistance Qb of the pile is therefore comparatively less important
than the shaft resistance Qs in controlling pile settlement. The base resistance of the pile
only provides additional safety against total collapse of the foundation. It is suggested

TABLE 2 Limiting Pile Load PI and Factor of Safety

Test pile Pmax PI Qsmax Qbmax Qu Fs=Qsmax/PI FOS=Qu/PI


(KN) (KN) (KN) (KN) (KN)

TPI 5918 2274 3837 2179 6016 1.687 2.65

TP2 29990 14384 21806 8716 30522 1.516 2.12

TP3 16932 14135 14871 4164 19035 1.052 1.35

TP4 15137 8864 13735 3945 17680 1.549 1.99

TP5 7308 6293 6885 1960 8845 1.094 1.40

TP6 11566 n62 10334 2580 12914 1.331 1.66

348 Ho C E and Tan C G


10

9 +
TPI600mm

8
+
.
Cl

+ TP4900mm
TP2 1350mm
TP3-'OOOmm

Cl X TP5600mm
7 TP6750mm .

6
Z x
5
~
4
. Cl
Cl

2
x
x!1
. Cl

+.
x
0~"""T'"-....-...,...--r-.....--.....,..-.,...... ........---.,r--1
o ~ ~ ~ ~ 1001~1~1~1~200
Average SPT value, N (blow/0.3m )
Figure 5 Relationship Between Skin Friction'
Resistance fs and Standard
Penetration Test N Value
6000 70
__ TPI 600mm x
.
0-
5000 ~
__ TP2 1350mm
- - TP3 'OOOmm
:.:: ..... TP4900mm
- - TP5600mm

~ 4000
50 -e- TP6 750mm
0" 0
:I
If)"
<J c! ~
i 3000 __ TP1600mm
Vi
<ii ~

! -e-TP2 1350mm z 30 Cl
CI
__ TP3 , CXlOmm :c-O"
.
c
"t:

.8
2000 - - TP4900mm
-TP56OOmm
~TP6750mm'
~

." +
c 1000
w
10

O....-,......---.---.--...,...--r--..,.....-,.......~
o
O+---.--.....--.....,..-.....--.....,..-,......-,...-,.........,...---i
20 ~ ~ ~ 100 120 1~ 160 o 10 20 30 ~ 50 ~ 70 80 90 100
Toe movement, oSb (inm) Average SPT value, Nb ( blow/0.3m )
Figure 6 Mobilization of End Bearing Figure 7 Relationship Between End Bearing
Resistance qb Resistance qb and Standard
Penetration Test Value Nb
25...----------------"71/ ~ ~,......-------------/""?'I
/
r/ ::l 35
/
Z / o /
::E 20
" 8 30 /
~
::I /
o /
/
8c 15 ./ l 25 /
]
~ /. /
<ii /
.2
S
~ /
!
"i 10 ~ ." /
. 15
/
. ./
J:.

."
-=-g
If) / /
U U 10
.
'Ei
et
5
~
/.
5 ".
/
/
//
0 1
/
O~-r----..-.....,..--.--""T"""-.,......-r--~
o 5 10 15 20 25 a 5 10 15 20 25 30 35 40
Measured shalt resistance, Osma:& ( MN ) Measured maximum applied load. Pmax (MN)
Figure 8 Measured and Predicted Shaft Figure 9 Measured and Predicted Ultimate
Resistance, Qsu after Total Pile Resistance, Qu after
Chin (1970) Chin ( 1970)

349 Ho C E and Tan C G


that the design of pile length for bored piles in Singapore Old Alluvium should be based
on a minimum Factor of Safety of 1.5 on the ultimate shaft resistance so as to achieve
tolerable pile settlement up to about 12mm. The base resistance may be ignored since
pile toe conditions are often uncertain and movement required to mobilize it can be
very large.

For piles that had not been loaded to failure, the estimation of the ultimate shaft
resistance Qsu and the ultimate total resistance Qu can be obtained from the inverse
slopes of a stability plot as proposed by Chin (1970), Fig. 3. Fig. 8 illustrates that the
estimation of Qsu using this method can be reasonably obtained. The value of Qu
obtained by Chin's method is about 20 to 25% higher than the maximum applied test
load Pmax, Fig. 9. However, since the ultimate total resistance has not been fully
mobilized at the value of Pmax, it is expected that the value predicted by Chin's method
would provide a close approximation to the fully mobilized value of Qu.

CONCLUSION

The results of several ultimate load tests on bored piles constructed in the Old Alluvium
in Singapore has been analyzed in this paper. Test data showed that the skin friction
resistance in Old Alluvium was very much larger than estimated from the traditional
Meyerhofs equation for cohesionless soils. The measured fsIN ratios ranged typically
between 0.8 and 5.4. The estimation of the end bearing resistance qb using Meyerhofs
equations however overpredicted the actual end bearing resistance. The end bearing
resistance in Old Alluvium appears to be limited to about 6000 kPa, with a qblNb ratio
of between 55.0 and 65.7.

The bearing capacity of bored piles in Old Alluvium is dominated by the shaft
resistance. It is recommended that design of bored piles in Old Alluvium be based on a
minimum Factor of Safety of 1.5 on the ultimate shaft resistance Qsu. This would
ensure that a tolerable pile settlement of less than 12mm can be obtained, with
reasonable confidence, at working load. For piles not loaded to failure, the ultimate
shaft resistance Qsu may be estimated from the inverse slopes of Chin's stability plot.

REFERENCES

1. Bowles J E (1988) "Foundation Analysis and Design" 4th Edition, Mcgraw


Hill, Singapore, pp 743

2. Buttting S and Robinson S A ( 1987) " Bored Piles - Design and Testing
"Proc. The Singapore Mass Rapid Transit Conference, Singapore, pp 155 -175

3. Chan S F and Lee PC S ( 1990) " The Design of Foundations for the Suntec
City, Singapore" Proc. Conference on Deep Foundation Practice, Singapore,
pp 27-32

350 Ho C E and Tan C G


4. Chin F K ( 1970) "Estimation of the Ultimate Load of Pile Tests Not
Carned to Failure" Proc. 2nd Southeast Asian Conference on Soil
Engineering, Singapore, pp 81-90

5. Chin Y 1<, Tan S L and Tan S B ( 1985) "Ultimate Load Tests on


Instrumented Bored Piles in Singapore Old Alluvium" Proc. 8th Southeast
Asian Geotechnical Conference, Kuala Lumpur, Malaysia, pp 2-54/2-66

6. Meyerhof G G ( 1976) ""Bearing Capacity and Settlement of Pile


FOlmdations '", J. Geot. Engng. Div. , ASCE, 102 ( GT3 ), pp 197-228

7. Poh K B, Burtling S and Hwang R ( 1981) "Some MRT Experience of the


Soils and Geology of Singapore" Proc. The Singapore Mass Rapid Transit
Conference, Singapore, pp 177-185

8. Tan S B ,Loy W C and Lee K W ( 1980) "Engineering Geology of the Old


Alluvium in Singapore" Proc. 6th Southeast Asian Conference on Soil
Engineering, Taipei,Taiwan, pp 673-683

9. Yong K Y, Cheah W B and Yap N C (1982) "Ultimate Load Test of an


Instrumented Cast Insitu Bored Pile installed in Stiff Silty Clay" Proc. 7th
Southeast Asian Geotechnical Conference, Hong Kong, pp 453-463

351 Ho C E and Tan C G


AXIAL COMPRESSION BEHAVIOR OF TWO DRILLED SHAFTS IN
PIEDMONT RESIDUAL SOILS

Dean E. Harris 1
Dr. Paul Mayne2

INTRODUCTION

Residual soils of the Piedmont Geologic Province underlie several important


cities such as Atlanta, GA, Washington, D.C., and Philadelphia, PA, as shown on
Figure 1. In the Piedmont, drilled shafts are a commonly used foundation for
heavily loaded structures (Gardner, 1987; Schwartz, 1987), however, the approach
to drilled shaft design varies considerably. The side resistance, or "skin friction"
developed in Piedmont residual soils has been a particularly controversial issue.

Design procedures for drilled shafts are complicated by the physical


characteristics of residual soils. The soils are primarily the product of the in-place
weathering of schists, gneisses, and granites and often retain the mineral
segregation and alignment, and structural defects of the parent rock (Sowers and
Richardson, 1983). These factors cause nonhomogeneity and anisotropy, with wide
variations in penetration resistance over short distances in both the horizontal and
vertical direction. The soils are difficult to characterize because they exhibit
behavioral aspects of both clay and sand and because of sample disturbance caused
by swelling of the micaceous mineral components and destructurization of fabric.

Because of the issues concerning drilled shaft design in the Piedmont,


members of the Geotechnical Committee of the Georgia Section of the American
Society of Civil Engineers (ASCE) formed ajoint committee with the Southeastern
Chapter of the International Association of Foundation Drilling (known as ADSC)
to implement a load test program to evaluate load transfer and load displacement
characteristics of drilled shafts in the Piedmont residual soils.

IGeotechnical Engineer, CH2M HILL, 825 NE Multnomah, Suite 1300, Portland,


Oregon, 97232-2146
2Associate Professor, Georgia Institute of Technology, Atlanta, Georgia, 30332

352 Harris
Figure 1. Region of the Atlantic Piedmont in the Eastern U.S.

SCOPE OF PROGRAM

The load test program was conducted on the Georgia Tech campus in Atlanta,
Georgia. The scope of the program consisted of extensive site characterization and
axial compression tests of two 76.2 cm (30 in.) diameter drilled shafts: one an end-
bearing shaft on rock with a length of 21.4 m (70.2 ft.) designated as Cl, and the
other, a 16.8 m (55.5 ft.) long "floating" shaft designated as C2. For the purposes
of the load test program, the depth to rock was defined as the refusal depth
encountered by the CME 550 auger drill rig used for standard penetration testing
during site characterization. Locations of the shafts within the site boundaries are
shown on Figure 2.

SITE CHARACTERIZATION

Site characterization included Standard Penetration Tests (SPT) , Flat Blade


Dilatometer Tests (DMT), Pressuremeter Tests (PMT), and Cone Penetrometer
Tests (CPT). Rock coring was also performed in several of the borings. Samples
from the field exploration were used in a laboratory testing program that included
grain size analysis, Atterberg Limits, triaxial shear tests, and consolidation tests.
Figure 2 shows the locations of the field tests performed. Test locations were

353 Harris
chosen to coincide with the locations of the test and reaction shafts. Total depths
of the borings, including coring, ranged from 20.0 to 27.8 m (65.5 to 91.0 ft.)
below the ground surface.

Borings at the site encountered a typical profile of residual soils, except that
fill and debris was encountered to depths ranging from 0.6 to 3.7 m (2 to 12 ft.).
Beneath this fill, residual silty sands were encountered to depths ranging from 15.8
to 19.5 m (52 to 64 ft.) where the borings encountered what is commonly termed
partially weathered rock (PWR). This less weathered soil retains much more of
the structure and hardness of the parent rock, though it is typically sampled as a
silty sand due to the hammering action of the SPT. The thickness of the partially
weathered rock varied from 0.6 to 7.3 m (2 to 24 ft.). Bedrock refusal was
encountered in the borings at depths ranging from 20.0 to 24.8 m (65.5 to 81.5 ft.)
and consisted of schistose to granitic gneiss. Recovery from the coring ranged
from 49 to 100 percent, with rock quality designations (RQDs) ranging from 29 to
47 percent. Water levels measured after 24 hours of stabilization ranged from 15.8
to 16.8 m (52.0 to 55.2 ft.).

Standard Penetration Testing

Standard penetration test resistances (N values) recorded in the soil at the test
site varied from about 8 to 31 blows per 0.3 m (1 foot). SPTs in the partially
weathered rock resulted in blow counts greater than 50, and equivalent N values

4 \
J SCALE(m)

1~~5
Line of SASW Profile

,
EllCPT.1 TSB-9
,,"Ell' "
Ell CPT-2 I

, TSB-4 TSB6
,,"Ell
C2 OMT-2

, TSB5
ElT
a: TSB-8 TSB-10 Ell
DMT-1
'W ,DMT.3

Ell TSB11 $,,";5B.2

--1';'B~7
\
" Limits of Test Area

Figure 2. Locations of Test Shafts and In-Situ Testing

354 Harris
Figure 3. Profile of Average N-Values.

were calculated by dividing the number of blows by the actual penetration of the
sampler. Differences were seen in the relative magnitude of N values from the
three different crews who performed the drilling, though data from each crew
showed the same general trend. Without actual measurements of the hammer
efficiencies from each crew, "true" N values cannot be determined, therefore all
values at each depth were averaged and assumed to be approximately equal to N60'
designating that 60% efficiency has been achieved, considered normal for U.S.
practice (Skempton, 1986).

A profile of the average N value is shown in Figure 3. N values increase with


depth, which is to be expected considering that the degree of soil weathering
decreases with depth. The average N value in the soil increases linearly from an
average of about 9 blows per 0.3 meter at a depth of 3.0 m (10 ft.) to about 22
blows per 0.3 meter at 15.2 m (50 ft.), then increases abruptly at a depth of 16.5
m (54 ft.), indicating the transitional PWR material.

Dilatometer Tests

Three dilatometer test (DMT) soundings were made at the site to depths ranging
from 10.4 m to 13.0 m (34.1 to 42.5 ft.). In the three soundings made at this site,
the DMT blade was pushed using the drill rig hydraulics, and blade resistance (qD)

355 Harris
Or----~=__-----_.. . . .. : .~ ..
: e o.l 0

.
~.

o 8 ~ +.
o..;e. II ,......-----;
I

I" .....:.. o
ca:.
8lC~ -.:..

DMT-l
DMT-2

--c..
o
5
.....a"....
~ \ :-,. DMT-3
E CD ~.

I~ -::.;,

! .......
.s=
<1>
o ...I:...I.
... :~
10
.. o~oo
00


t..


DMT-1
DMT-2

..
rI' t

co.
00
~
~~
o
it
..


DMT-3 Po PI
15 ........................................................~. . . . . . . . . . . . . . . . . . . . . . . . . . ..0...1
o 5 10 15 20 25 30 o 0.5 1.0 1.5 2.0 2.5
Blade Resistance, qD (MPa) Pressures, Po and P1 (MPa)

Figure 4. Profile of Measured Dilatometer Parameters.

measurements were made by recording the hydraulic pressure at the top of the drill
rods. Figure 4 shows profiles of the qD' PO' and PI readings for the three
soundings. Profiles of ID , KD , and ED are shown together in Figure 5. As with
the N values from the SPTs, the DMT strength and deformability parameters (qD'
PO' PI' and ED) each show a trend of increasing value with depth. The I D values
are relatively constant beneath the surface fill. The KD parameter shows a slight
decrease in the upper soil, but is relatively constant below about 9 m (30 ft.) It
should be noted that the correlations on which the DMT output are based were
likely made using databases which did not include residual soil, and their
applicability to the soils of the Piedmont has yet to be proved.

Pressuremeter Tests

A Menard-type GAM pressuremeter with a probe diameter of 70 mm (2.76 in.),


and an initial volume of 790 cc (48.21 in. 3) was used to perform five
pressuremeter tests (PMTs) in a single boring advanced using mud rotary drilling
methods. Plots showing the measured pressure-volume curves from the five tests
are shown in Figure 6. Unload-reload cycles were not included in the tests.

356 Harris
o
'~f ,....


. ...; ....

~
~,
....
., ~~

()

- 5

1:
f- I- ...
I- ':

E
1;.~ .(~

10 I-
,f
=if'" ~ ,,:I ..~"
~.

15 I
S

I I
~
I

I I I I .1
"
..,....
I

o 2 3 4 0 5 10 o 10 20 30 40
Ko ED {MPa}

DMT-' ... DMT-2 DMT3

Figure 5. Profile of Interpreted Dilatometer Parameters.

2.5
0--0 4.6 m Depth
t:.-t:. 9.1 m Depth
13.7 m Depth

- 2.0 0-0

0-0 16.8 m Depth


ctJ 'V-'V 19.8 m Depth
a..
~
......- 1.5
Q)
~

:::J
(/)
(/) 1.0
Q)
~

a..
0.5

a
a 200 400 600 800
Volume (cc)

Figure 6. Pressure-Volume Curves from PMT Testing.

357 Harris
Cone Penetration Tests

Two cone penetration tests (CPT) were performed using an electric friction cone
o
penetrometer with a 600 apex, 10 cm2 (1.55 in2) projected cone tip area, and a
150 cm 2 (23.25 in 2) friction sleeve. The CPT's were performed to depths of 19.2
m (63.0 ft.) and 9.7 m (31.8 ft.) below ground. CPT 2 was terminated early due
to an obstruction (stone), and rainy weather. The tip resistance and friction
resistance from the deeper sounding, CPT 1, are shown in Figure 7. As with the
previous in-situ tests, the tip resistance and the friction resistance increase with
depth.

Laboratory Testing

Samples obtained during field exploration were subjected to a series of


laboratory tests including grain size analyses (mechanical and hydrometer),
Atterberg limits, triaxial shear tests, and one-dimensional consolidation tests. The
scope of the laboratory testing program was determined by the ASCE load test
committee.

Soil Classification Tests. Grain size distributions were determined using


mechanical sieves and hydrometer tests. A total of 113 tests were performed on
samples obtained from standard penetration testing at the site. The results from the

qc (MPa) fs (kPa)
00 2 4 6 8 10 0 100 200 300

---E
-
.r:.

0
aQ).
10

15

CPT 1

20

Figure 7. Profile of CPT measurements

358 Harris
tests indicate extreme uniformity of particle sizes within the soils. The median
grain size is D 50 = 0.14 mm (5.5 mils) and the material technically classifies as
a silty sand (SM). The soils have an average fines content (percent passing #200
sieve) of 33% and a clay content (percent finer than 0.002 mm) averaging 8%.
Atterberg limits testing revealed that the soils are non-plastic, except for some of
the fill and near-surface residual soils.

Triaxial Shear Tests. Thirteen isotropically-consolidated undrained triaxial


compression (CIUC) tests with pore pressure measurements were performed on
saturated specimens retrieved from the tube samples obtained at boring TSB-lO.
Because of sample disturbance, in some instances, only one or two specimens were
available from a sample, and multi-stage tests were performed on these specimens.
Stress paths from the CIUC tests are shown together in Figure 8 in MIT q-p space.
It is apparent from inspection of the data that the failure envelope passes through
the origin, with no cohesion intercept (c'). A least squares regression confirmed
that c' = O. Therefore, a best fit regression (with a forced intercept of zero) was
made and has been shown along with the data. The failure envelope indicates an
effective angle of internal friction (4) ') of 36.1 degrees. Note that Figure 8
includes specimens from a range of depths, and N-Values.

Consolidation Tests. Consolidation tests were also performed on specimens


from Shelby tube samples. Most of the test results were so highly affected by
sample disturbance that a determination of a preconsolidation stress (<1 ') was
difficult. Various methods were used to interpret a yield stress from the Pest data.
The results of these interpretations varied widely. Estimated overconsolidation
ratios ranged from approximately 3 at a depth of 25 feet, to about 1.5 at a depth
of 40 feet. The virgin compression index Cc was relatively consistent from test to
test, and an average value of 0.28 was calculated. The average value of the
recompression index (C r ) was 0.04.

WAD TEST PROCEDURES

Excavation and construction required for the load tests were. performed by
members of the Southeastern Chapter of the International Association of
Foundation Drillers (ADSC). A Hughes LDH auger rig was used to excavate each
shaft, and construction began on May 11, 1992. Both test shafts (C1 and C2) were
excavated using earth augers only. Temporary casing was not used during
construction of the test shafts. Concrete was allowed to free fall into the shaft
excavations, though hand shovels were used to guide the flow, and to minimize the
amount of concrete striking the reinforcing steel or the sides of the excavation.

The hydraulic jack used in load testing of the shafts had a maximum rated
capacity of 8.9 MN (1000 tons) and had been calibrated approximately 8 months
prior to the load test. Hydraulic pressures from the jack were used to estimate the
total load on the shafts.

359 Harris
600
as' (kPa)
Best Fit Regression: 41 .. 36.1 0
500 r2 .. 0.998
+-+ 565.0

-CIS
Q..
n .. 13
0-0 496.1
284.0
;:'-l;.

-
..x:: 400 0-047.5

-...
C\I

300
l'
0-0 95.8

'J-'J 191.5
b

-
b
II
0'"
200
rl
+
o ;.-
,
o \~I
,
e-e 135.0
A-A 268.0
---536.0
..- .. 164.0
o -II lI(-lI( 81.3
b~ 48.2
100 Ciji
Q-Q

B +
()-() 79.9
0_
, ,+
-~
o '.,f-
a
a 200 400 600 800 1000
p' = (CJI' + CJ3')/2 (kPa)

Figure 8. Stress Paths from CIUC Tests.

Vibrating wire strain gauges were welded to the shaft reinforcing steel at the
top and base of each test shaft, as well as intermediate depths. Gauges were placed
at the top to allow correlation with the hydraulic jack, since the axial load and the
jack load should be equal at this point. In shaft C1, the remaining gauges were
placed at 9.1 m, 16.8 m and 21.3 m (30 ft., 55 ft., and 70 ft.) below ground. In
shaft C2, the gauges were placed at 9.1 m and 16.8 m (30 ft. and 55 ft.) below
ground. These depths were chosen in order to separate the side resistance effects
from regions with different N-value ranges, and to separate the effects of partially
weathered rock from those of the soil. Four gauges were placed at each depth for
redundancy. Figure 9 shows a generalized representation of the test shafts and
strain gauge configuration.

Four independent measurements of the displacements at the tops of each test


shaft were measured using dial gauges fixed to the top of the shaft and marked
scales fixed to the hydraulic jack. Tip displacements were measured using a tell-
tale cast within the concrete during construction.

The loading of the drilled shafts followed a quick loading procedure, similar
to the method recommended by ASTM D 1143. Using the quick load procedure,
the shaft is loaded in equal increments, usually 10 to 15 percent of the proposed
design load, or anticipated failure load, and each load is maintained for a minimal

Harris
360
period of time, usually 2.5 minutes. In this program, shaft displacement readings
were made immediately upon reaching each load level, and again after a period of
stabilization, during which time, readings from the vibrating wire gauges were
taken. The time required to read all of the vibrating wire gauges varied from about
3 to 5 minutes. After completion of all readings and the second reading of the
shaft displacement, the load on the shaft was then increased to the next load level.
An initial load increment of 223 kN (25 tons) was used to apply a seating load to
each of the two drilled shafts. After applying this seating load, and unloading the
jack, load increments of 890 kN (100 tons) were used for shaft Cl, and increments
of 445 kN (50 tons) were used for shaft C2. Two unload-reload loops were
included in each of the load tests.

WAD TEST RESULTS

Load-Displacement Behavior

Readings made immediately after application of each load increment were used
to represent the load-displacement response of the two drilled shafts. The load-
displacement response measured for the two shafts are shown in Figure 10. Four
separate measurements of displacement are shown for Shaft C 1, with maximum
differences between measurements on the order of 2.5 mm (0.1 in.). Two unload
reload loops were performed at 0.9 and 3.6 MN (100 tons and 400 tons) during
the load test. A maximum settlement of 2.56 cm (1.010 in.) was measured in the
load test of the 21.4 m (70.2 ft) long, end bearing shaft, C1. The maximum load
measured in this load test was 8.9 MN (1000 tons). Loading of the shaft was
halted at this load, since the maximum safe capacity of the jack had been reached.
The shaft was unloaded incrementally, with a permanent residual settlement of
1.288 cm (0.507 in.) remaining aftecthe load was removed. The 16.9 m (55.5 ft.)
long, floating pile, Shaft C2, was loaded to a maximum load of 4.5 MN (500
tons), where loading was halted, due to inability to maintain pressure by the
hydraulic jack. A maximum butt settlement of 16.38 cm (6.45 in.) was recorded
at this load. After unloading, a residual settlement of 15.5 cm (6.1 in.) was
measured.

Nine methods interpreting failure from pile load tests were applied to the
results of the ASCEI ADSC drilled shaft load test. For shaft C1, none of the
methods indicated that failure had occurred but the Davisson method (Davisson,
1972) indicated that failure was imminent. Interpreted failure loads for shaft C2
ranged from 2.2 to 5.1 MN (251 tons to 574 tons). The average of the nine
interpreted failure loads was 3.2 MN (360 tons). The Davisson limit is achieved
at an axial load of 2.7 MN (312 tons), just below the "average" failure load. The
ultimate capacity has been taken to be 3.1 MN (350 tons).

361 Harris
SOIL PROFILE DRILLED SHAFTS
DESCRIPTION DEPTH (m) SHAFT Cl SHAFTC2

0.0
\ \ \ ALL \ \ \ \ x x
\\\\\\\\\\ 1.8

RESIDUAL SILTV x x
SAND (SM)

16.9 x x
\ \ PARTIALLY \ \
WEATHERED
ROCK
~ -22.0 x
GRANITIC GNEISS

Both shafts 76 em diameter


X Depth of Instrumentation

Figure 9. Generalized Representation of Test Shaft


and Strain Gauge Configuration.

12 7 .-0 Gage (0-2)

10
ShaftC1
6
Shaft C2
a-a
. - Gage (0-3)
Level (Jl)
. - . Wire (J-2)

-z 8
- 5

."..-a-a-a--\a
z
-ro
~
"C 6 -
~
"C
4

0
....J
4
ro
0
....J
3

2 (1 aI
2
I
aI
o I I
0.5 1.0 1.5 2.0 2.5 3.0 o 5 10 15 20
Displacement (cm) Displacement (cm)

Figure 10. Load Displacement from Load Shaft Cl and Shaft C2.

362 Harris
Load-Transfer

Axial loads at each instrumented depth for shafts Cl and C2 are shown
Figure 11. In the load test of the deeper shaft, C 1, a majority of the load transfer
occurs along a section between a depth of 16.8 m (55 ft.), and the tip of the shaft
at 21.4 m (70.2 ft.). This corresponds to the zone where partially weathered rock
was encountered at this shaft location. In contrast load transfer in C2 is relatively
constant with depth. This shaft was constructed within the soil profile, with no
partially weathered rock. Also there were fewer instrumented depths in this shaft.

Figure 12 shows the variation of the side and tip resistance with settlement for
shafts Cl and C2. It is clear from both plots that the majority of the shaft load
was carried by side resistance. In shaft C 1, 65 percent of the total load was
carried by side resistance at the final load increment. In Shaft C2, approximately
71 percent of the total load was carried by side resistance at the last stage of the
load test.

Using the load transfer data, and failure loads determined from the load-
displacement measurements, ultimate values of the side and base resistance
components can be computed for Shaft C2. Since shaft C 1 could not be loaded to
failure, ultimate values of the side and base resistance could not be determined, but
maximum recorded values can be used to approximate the ultimate side resistance.

In shaft Cl, at the maximum total load of 8.9 MN (1000 tons), approximately
2.9 MN (328 tons) was supported by side resistance in the residual soil, and 2.6
MN (292 tons) was supported by side resistance in the partially weathered rock.
Using the shaft length of 21.4 m (70.2 ft.) and the shaft diameter of 0.76 m (2.5
ft.) average unit side resistances of 73 kPa (0.8 tsf) and 234 kPa (2.4 tsf) are
calculated for the residual soil and partially weathered rock, respectively. In the
load test of C2, after the shaft load reached a value of 2.5 to 2.7 MN (280 to 300
tons), relatively little additional load is carried by side resistance. An ultimate side
resistance capacity of 2.7 MN (300 tons) was estimated. By dividing the total side
resistance capacity by the total surface area of the shaft, an average unit side
resistance of 66 kPa (0.7 tsf) was calculated. The tip resistance versus butt
displacement curve for shaft C2, shown in Figure 12, shows an inflection at a tip
load of 0.5 MN (50 tons), indicating an ultimate base resistance of 975 kPa (10 tsf)
for this shaft.

DISCUSSION

Results from the laboratory test program and in-situ testing were used to
estimate the axial compression capacity of Test shaft C2. Separate calculations
were made using a total stress analysis and effective stress methods. Calculation
of the axial compression capacity of drilled shaft foundations is discussed by
Kulhawy, 1991.

363 Harris
0 0

-
.s
.t=.
10
-.s
.t=.
5

0.. 0..
CD Q)
10
Cl Cl
15

20 15
Shaft C2

0 2 4 6 8 10 0 2 3 4 5
Axial Load (MN) Axial Load (MN)

Figure 11. Average Axial Load at Instrumented Depths, Shaft Cl and C2.

12 6
Shaft Cl Shaft C2
10 5
Total Total

-
Z
8 4

-
~
"0
l'Cl
6 3
0
....J
4 2

0.5 1.0 1.5 2.0 2.5 3.0 5 10 15 20


Displacement (cm) Displacement (cm)

Figure 12. Variation of Side and Base Resistance with


Settlement of Shafts Cl and C2.

364 Harris
Calculation of the shaft resistance and base resistance by the total stress
method requires an estimate of the soil undrained shear strength. Estimates of
undrained shear strength were made using the CIUC triaxial shear tests and using
numerous empirical and theoretical correlations between undrained shear strength
and PMT, CPT, and DMT results (Mayne and Kulhawy, 1990). Shaft resistance
was estimated based on empirical correlations (the "alpha" method). Base
resistance was estimated using bearing capacity formulas.

Effective stress calculations of axial compression capacity rely primarily on


estimates of the soil angle of internal friction, cP I , and coefficient of lateral stress,
Ko. Estimates of cP I were made using CIUC test results and correlations with
SPT, DMT, and CPT data (Mayne and Kulhawy, 1990). Empirical correlations
with in-situ testing were also used to estimate ~ In addition, Ko was estimated
based on the relationship Ko = (1 - sincP I )OCR s cP (Mayne and Kulhawy, 1990),
using results from the CIUC testing and one dimensional consolidation tests.

Estimates of the axial compression capacity are shown in Table 1. Based on


a review of the results, it appears that the effective stress method was more
accurate in estimating side resistance, though slightly non-conservative. This can
be partially explained by the fact that the effective stress calculations were made
by assuming the angle of interface friction between the soil and concrete, 0, equal
to the soil angle of internal friction, cP I . In reality, 0 is probably less than cP I , due
to the effects of construction. In addition, the effective horizontal stress was
assumed to be equal to the in-situ horizontal stress existing before construction.
The stress release occurring during shaft excavation probably acts to reduce the
horizontal stress acting on the shaft. The total stress estimates of the base
resistance appear to be closer to the interpreted values from the load test.

Side resistance determinations using effective stress methods (fs = Bavo ')
have been successfully applied to both coh~sionless and cohesive soils in the past.
One hypothesis is that the shaft interface acts as a path of drainage during loading.
Also, another possible explanation is that the shear strains along the shaft produce
relatively small volume changes, resulting in little development of pore pressure.
For undrained loading of piles in clay, available experimental measurements of
pore water pressures along the shaft show essentially no development of excess ~u
(Konrad and Roy, 1987; Coop and Wroth, 1989). In contrast, for piles in soft and
stiff clays, ~u measurements beneath the foundation base show strong development
of positive pore water pressures during axial compression loading, since this is a
zone of high compression. Differences in the shaft resistance and tip resistance
behavior may also be a result of strain rate effects, particularly since the two load
tests reported herein were conducted using the quick load test procedure.

365 Harris
Table 1. Estimated Axial Compression Capacity of Shaft C2.

Axial Compression
Capacity (leN)
Method Input Qshaft 9tJase Qtotal
Total Stress Analysis CIUC 1761 827 2588
PMT 2553 871 3424
CPT 3763 1548 5311
DMT 1575 365 1939
Effective Stress Analysis PMT/CIUC 2900 1112 4012
CPT 3300 1023 4323
DMT 2936 1334 4270
I-D Consol 2847 1112 3959
ICIUC
Interpreted from Load 2669 444 3114
Test C2

Additional support for these arguments may be seen in piezocone soundings


recently advanced in Piedmont soils. Piezocones with pore pressure measurements
at the tip generally show positive pressure being developed, while piezocones with
pore pressure elements behind the tip show negative or relatively small positive
pressures.

SUMMARY AND CONCLUSIONS

The results of the load test program indicate that drilled shaft foundations in
residual soil develop their axial compression capacity primarily from side friction
resistance, and that this component should not be ignored. These findings are
consistent with common procedures for estimating the axial compression capacity
of deep foundations (Kulhawy, 1991; Fellenius 1991). Also; it appears that
effective stress methods are more appropriate for estimating the side resistance of
drilled shaft foundations. Based on the results of a single test carried to failure,
it appears that tip resistance of drilled shaft foundations in piedmont soils may be
controlled by an undrained failure mode, and that the tip resistance can be more
accurately estimated using total stress methods or undrained shear strength.

366 Harris
REFERENCES

Coop, M.R., and Wroth, C.P., "Field Studies of Instrumented Model Pile in
Clay", Geotechnigue, Vol. 39, No.4, Dec 1989, pp. 679 - 696.

Davisson, M.T., "High Capacity Piles", Proceedings, Lecture Series on


Innovations in Foundation Construction", American Society of Civil Engineers,
Illinois Section, 1972, 52 p.

Fellenius, RH., "Pile Foundations", Foundation Engineering Handbook. Second


Edition, Ed., H.Y. Fang, Van Nostrand Reinhold, New York, 1991, pp. 511 -
536. .

Gardner, W .S., "Design of Drilled Piers in the Atlantic Piedmont", Foundations


and Excavations in Decomposed Rock of the Piedmont Province, GSP 9, Ed. Ron
Smith, ASCE, New York, 1987, pp. 1-15.

Konrad I.M., and Roy, M., "Bearing Capacity of Friction Piles in Marine Clay",
Geotechnigue, Vol. 37, No.2, 1987, pp. 163 - 175.

Kulhawy, F .H., "Drilled Shaft Foundations", Foundation Engineering Handbook.


Second Edition, Ed., H.Y. Fang, Van Nostrand Reinhold, New York, 1991, pp.
537 - 552.

Mayne, P.W., and Kulhawy, F.H., "Manual on Estimating Soil Properties for
Foundation Design", Report EL-6800, Electric Power Research Institute, Palo
Alto, August 1990.

Schwartz, S.A., "Drilled Piers in the Piedmont - Minimizing Contractor - Engineer


- Owner Conflicts", Foundations and Excavations in Decomposed Rock of the
Piedmont Province, GSP 9, Ed. Ron Smith, ASCE, New York, 1987, pp. 80 -90.

Skempton, A.W., "Standard Penetration Test Procedures and the Effects on Sands
of Overburden Pressure, Relative Density, Particle Size, Aging, and
Overconsolidation", Geotechnigue, Vol. 36, No.3, Sept. 1986, pp. 425-447.

Sowers, G.F., and Richardson, T.L., "Residual Soils of the Piedmont and Blue
Ridge", Transportation Research Record 919, Washington, D.C., 1983, pp. 10 -
16.

367 Harris
Auger-Cast Piles in Clays
Daniel O. Wong

ABSTRACT

An auger-cast pile is a special deep foundation element that has many advantages
and limitations. However, the lack of knowledge about the installation details and the

understanding of the load bearing behavior have sometimes precluded its use even under

potentially favorable conditions. This paper presents a general overview of the installation

process with emphasis on several important details that may affect the pile behavior. The
static load test results of two auger-cast piles in predominately clay environment are also

presented. Reconstruction of the measured load-settlement data using a computer

solution, APILE, allows the backfiguring of the ultimate unit skin friction, f u ' profile. A

lower bound and upperbound f u profiles are backfigured to be about 0.3 and 0.4 times the

effective overburden pressures of soils. Load transfer factor a of 1 and f3 of 0.3 to 0.4 are
estimated based on the total and effective stress methods. The parameters presented are
site specific and should not be applied directly for design without field verification of other

specific site conditions.

INTRODUCTION
Installation of auger-cast piles has been practiced by the foundation industry for

more than half a century. The special technique, suitable under certain favorable
conditions, has not gained its share of popularity among some design engineers and

contractors. There are many advantages of using auger-cast piles such as relatively less-

induced vibration, no casing or slurry requirements for borehole stability, and the flexibility
of penetration depth. The disadvantages, on the other hand, consist of uncertainty in the

Principal, Tolunay-Wong Engineers, Inc., 1706 W. Sam Houston Pky N., Houston, Texas 77043

368 Daniel O. Wong


integrity of the final product, inability to install instrumentation in the foundation element,

relatively inferior lateral capacity and impracticality of installing at a larger diameter than
the typical range of 12 in. to 24 in. The uncertainty and controversy about its use are

mostly due to the lack of knowledge about the installation process and the understanding

of the load bearing behavior. As a matter of fact, there are as many names of this

foundation type as the opinions about its use. The foundation element has been called in

the literature as Auger-Grout-Injected Pile, Auger-Grout Pile, Auger-Injected Pile, Auger-

Placed Grout Piles, Auger-Cast-In-Place Pile and Auger-Cast Pile (1-3). Because of the

terminology used for the study of this specific project, auger-cast pile is adopted
throughout this paper.

This paper presents a general overview on auger-cast pile installation and the static
load test results of two auger-cast piles in clays. An attempt is made to further analyze

the load test data in order to gain some insights into the bearing capacity and load transfer

characteristics of auger-cast piles. This paper by no means provides the understanding of

the fundamental pile-soil interaction of auger-cast piles in clays. However, it serves as a

general reference for the installation process and a case history that provides some
understanding of auger-cast piles in a cohesive environment.

OVERVIEW OF INSTALLATION PROCESS


Installation of auger-cast piles requires a hydraulically powered, continuous flight

hollow shaft auger, typically as shown in Figure 1. The auger is advanced to the designed
depth or to refusal. Once the final depth is achieved, the auger is raised 0.15 m (0.5 tt)
to 0.61 m (2 ftl and the grout is pumped through the hollow shaft into the openhole. The

intent of this procedure is to build up sufficient pressure in the hole to offset the lateral
earth pressure and the hydrostatic pressure, if applicable. The auger is then lowered into

the grout to the original depth. The grout is again pumped in the hole with a minimum

head of about 2 m (6.6 ft) above the point of injection, while the auger is continually
withdrawn in a steady rate without being lifted above the level of the grout in the hole.
This presumably assures that the grout is always pumped into grout and the loose
materials can be pushed up instead of mixed with the grout.

As the auger tip reaches the ground surface, the grout is allowed to overflow from
the auger hole so that no soil cuttings may be trapped in the pile. It is a good practice to
provide a minimum 457mm (18 in.) long metal sleeve around the pile top at the ground

369 Daniel O. Wong


surface to prevent contamination by surficial materials. Reinforcement can be introduced
into the pile after the withdrawal of the auger and before the grout is set.

The grout should be a cement-based non-shrink mortar which can achieve an

ultimate compressive strength comparable to regular concrete mixture. The grout typically
consists of Portland cement (ASTM C-150l, fluidifier, fine aggregate and water so

proportioned that it is capable of suspending the solids and can be pumped without
difficulty. A minimum of 4 grout samples are recommended for each pile for strength tests

at 7 days and 28 days in accordance with ASTM C-109.

The pump should be a positive displacement piston-type equipment with a pressure

gauge and flow gauge. The gauges should be continually monitored throughout the
installation process to determine the pumping pressure and grout volume. It should be

emphasized that monitoring of the instrumentation can be a good quality control measure
as an experienced operator can coordinate the extraction rate with respect to the grout

flow and pressure.

It should also be noted that the grout volume and the grout pressure are factors

that affect the behavior of auger-cast piles. The necessity of maintaining a positive grout

pressure during installation to avoid borehole instability can positively affect the pile-soil

load transfer characteristics. It has been speculated that the grout pressure imposed
horizontally on the soil increases the frictional resistance at the pile-soil interface for auger-
cast piles. The degree of influence of that effect can not be fully assessed without

detailed measurements of the horizontal soil pressures along the pile. Another factor
influencing the behavior of auger-cast pile is the grout volume( 1). This factor is usually

expressed as the grout factor (GFl, which is defined as the ratio of actual to theoretical

grout volumes. An acceptable grout factor is greater than 1 to account for the following
reasons: (a) larger than the theoretical auger hole, (b) grout being pressurized into the

secondary features of soils, such as fissures or cracks, and (c) extra grout being pushed
out from the borehole upon completion. A typical top view of an auger-cast pile is shown

in Figure 2.

CASE HISTORY
Closely adhering to the installation details discussed previously, a group of eight

auger-cast piles was installed at a site near the Texas Gulf Coast. All piles were drilled

Daniela. Wong
370
using a 406 mm (16 in.) diameter auger. Two of the eight piles, including a 21.3 m (70
ft) and a 25.9 m (85 ft) long auger-cast piles installed about 4.9 m (16 ft) apart, were used

as test piles. The remaining six 21.3 m (70 ft) long auger cast piles were installed at an

approximately 2.6 m (8.5 ft) spacing diagonally from the test piles to serve as reaction

piles for the load tests. Each load test utilized four reaction piles and the layout of these
six reaction piles were arranged in such a way that two reaction piles were used in both
of the load tests. During pile installation, two grout samples were taken during the

construction from each of the eight auger-cast piles and tested for compressive strength

by others. All test results exceeded an ultimate compressive strength of 3000 psi at 7

days and 28 days. The load tests were performed about a week after pile installation.

The subsurface stratigraphy at the project site consisted of about 6.1 m (20 ft) to

7.3 m (24 ft) thick hydraulic fill soils, underlain by firm to very stiff consistency clays, silty
clays and sandy clays to the explored depth of about 36.6 m (120 ft). The fill soils were

typically comprised of soft to firm consistency clays, silty clays and sandy clays with

interbedding layers of very loose silty sands and sands. Thin layers of silty sands and

clayey silts were occasionally observed within the underlying natural cohesive strata.

Groundwater level measured in the open boreholes was found to be at about the 4.6 m (15

ft depth).

A plot of the moisture content profile of the cohesive soils are presented in Figure
3. Liquid limits and plastic limits measured on selected soil samples are also shown in

Figure 3. More than 95% of the moisture content data lie between 20% and 40%. The
plasticity indices (PI) of tested soil samples range from 20 to 66, with a median value of

40.

An undrained shear strength profile of the cohesive soils, based on limited number
of unconfined compression tests and unconsolidated undrained triaxial tests, is shown in
Figure 4. Total unit weights of cohesive soils range from 16.4 KN/m 3 to 20.8 KN/m 3 (104
pcf to 132 pct), with an average value of 19.3 KN/m 3 (123 pcf), based on the
measurements of 17 soil samples. Using a total unit weight of 19.3 KNfm 3 above the 15-

ft depth and a submerged unit weight of 9.5 KNfm 3 below the 15-ft depth, lines of cfp of

0.3 and 0.4 were generated (i.e. c = 0.3 or OAp) and are also plotted in Figure 4. c/p is
known as normalized undrained shear strength ratio and is a ratio of undrained cohesion
(c) and the effective overburden pressure (p). It appears that most measured cohesion

Daniela. Wong
371
data lie close or within the range of the generated c/p lines except the upper 10 to 15 ft.
The two offset data points recorded at the 13.1-m (43-ft) depth were obtained from
samples with large amounts of shell fragments and organics.

LOAD TESTS AND RESULTS


The load test arrangements, as presented in Figure 5, included three (890-KN) 100-
ton jacks and load cells, an hydraulic pump and dial gauges. Three dial gauges were
placed equally-spaced around the test pile (Figure 6) to measure the pile head movement
and possible rotation. One dial gauge was attached to each reaction pile to monitor
vertical movement.

The compression load tests were conducted in general accordance with the "Slow
Test" procedures outlined in ASTM D 1143. Test A was performed on the 21.3-m (70-ft)
long pile (Pile A) on June 11 and 12, 1992 to complete a loading-unloading cycle up to a
compressive load of 979 KN (110 tons). Pile A was reloaded on June 12, 1992 to a load
of 1691 KN (190 tons) and the test was terminated because of the failure of a steel rebar

connecting the reaction beam and one of the reaction piles. Pile A was loaded again to a
plunging load of about 2136 KN (240 tons) on June 15, 1992. Load-movement curves
for Test A are presented in Figure 7. Compression Test B was performed on the 25.9-m
(85-ft) long pile (Pile B) on June 17 and 18, 1992. Test B was loaded to a maximum load

of 2136 KN (240 tons), when plunging occurred. The load movement curve for Test B is
presented in Figure 8.

EVALUATION OF LOAD TEST DATA


Since there was no instrumentation installed along the pile below the ground
surface, the load transfer characteristics could not be readily understood. An attempt was
made to further analyze the load settlement data in a logical way in order to shed some
insights into the soil-pile behavior.

It appears from Test A that a stiffer load settlement response was obtained each
time after the pile was reloaded. This may be due to the stressing of the surrounding soils
at the pile interface and base from the previous loading cycle resulting in stiffer soils. In
order to perform an evaluation of the load test data, only virgin curves of Tests A and B
were used and the following steps were followed:

372 Daniel O. Wong


a. Reconstruction of the load settlement curve for Test B was attempted
using a computer-based numerical solution APILE (4) since Test B was a
complete virgin test.

b. Maximum end bearing of 9c was assumed in the analysis where c is the


cohesion at the depth of the pile tip.

c. Vijayvergiya's load transfer criteria (5) were used assuming a movement


of 7.6 mm (0.3 in.) to reach maximum skin friction and a movement of
25.4 mm (1 in.) to reach maximum end bearing.

d. A trial and error procedure was used to select a profile of ultimate unit skin
friction that provided the best match of the measured and reconstructed
load settlement curve.

e. Upper bound and lower bound ultimate unit skin friction profiles were
backfigured based on the nominal auger diameter and the corrected
diameter of pile, which will be further explained in the subsequent
paragraph.

f. The upper bound and lower bound ultimate skin friction profiles were used
to reconstruct the virgin load settlement curve of Test A for validation.

As mentioned previously, grout factor, GF, is an important parameter that may


impact the behavior of auger-cast piles. Based on the field documentation, GF was
estimated to be 1.52 for Pile A and 1.36 for Pile B. It should be reiterated that both Piles
A and B were drilled with a 406 mm (16 in.) diameter auger. In a clay environment, the
additional grout needed is possibly due to largely a larger hole than the theoretical auger
size. The corrected average diameters, based on this assumption, were found to be about
500 mm (19.7 in.) for Pile A and 472 mm (18.6 in.) for Pile B. One should note that the
actual average diameter of the pile is likely to be in between the nominal auger diameter
and the corrected diameter. During the trial and error process to reconstruct the measured
load settlement curve for test B, the selected ultimate unit skin friction profile that provided
the best match of the measured and reconstructed curve was defined as the upper bound

373 Daniel O. Wong


profile if the nominal auger diameter was used. Similarly, the selected ultimate skin friction
profile was defined as lower bound if the corrected diameter was used.

Following steps a through e presented above, respective upperbound and


lowerbound ultimate skin friction profiles, that were used to generate a best-matched load
settlement curve for Test 8, were obtained based on the nominal auger diameter and the
corrected diameter of piles, as represented by the following equations.

0.3 yz =:; f u =:; 0.4 yz (1)

Where f u is the ultimate unit skin friction, z is the depth and y is the effective unit
weight. Figure 9 presents the measured load settlement curves for Test 8 and the
reconstructed curves based on the upper and lower bound profiles.

Relating Equation 1 to Figure 3, the measured undrained shear strength data


approximately lie within the upper and lower bound ultimate skin friction profiles, resulting
in the following deduction:

a. for the subsurface soils in this study

0.3 =:; c/p < 0.4 (2)

or

O.3yz < c < O.4yz

b. as f u is found to be lying approximately within the same range where c is

lying,

Daniel O. Wong
374
f u =c (3)

Using the total stress method and effective stress method to estimate f u (6, 7, 8),

a and p values can be obtained as:

a. fu = ac, where a= 1 based on Eq (3)


b. fu = Pvz, where 0.3 :::;; P :::;; 0.4 based on Eq (2)

The upper and lower bound ultimate skin friction profiles were then used to

reconstruct the virgin load settlement curve for Test A. Figure 10 presents the measured
and reconstructed curves, which appears to compare well.

SUMMARY AND CONCLUSIONS


A general discussion on the installation of auger-cast piles is presented with

emphasis on the significance of certain procedures. A case history of load testing two
auger-cast piles in clays is discussed. Based on the load test results and a further analysis
of the data, the following observations may be made.

A) Reloading of the pile, based on Test A, appeared to provide stiffer load

settlement behavior. The better load bearing response may be due to the
stressing of the surrounding soils at the pile interface and base from the
previous loading sequence.

B) Backfiguring of the ultimate skin friction profiles based on the matching

technique of the measured and reconstructed load settlement curves was


performed. The exercise yielded an upper and lower bound values of 0.4

and 0.3 times the effective overburden pressures of the soils for this study.

C) Relating the upper and lower bound ultimate unit skin friction profiles to
the total stress and effective stress parameters, a appears to be
approximately 1 and P ranges from 0.3 to 0.4.

375 Daniel O. Wong


It should be emphasized that the conclusions only apply to the site-specific
conditions of the described project and the parameters and values presented should not
be used directly for design without field verification of other site conditions.

REFERENCES

1. W.J. Neely. Bearing Capacity of Auger-Cast Piles in Sands. Journal of


Geotechnical Engineering, ASCE, Vol. 117, No.2, 1991, pp. 331-345.

2. F.M. Fuller. Engineering of Pile Installations. McGraw-Hili Book Company, 1983,


pp.19-23.

3. S. Prakash and H.D. Sharma. Pile Foundations in Engineering Practice, John


Whiley & Sons, Inc., 1990, pp. 64-65.

4. L.M. Tucker, APILE, Civil Engineering Department, Texas A&M University, August
1986.

5. V.N. Vijayvergiya. Load-Movement Characteristics of Piles, 4th Annual Symposium


of the Waterway, Port, Coastal and Ocean Division of ASCE, Long Beach, March,
1977.

6. M.J. Tomlinson. The Adhesion of Piles Driven in Clay Soils. Proc. 4th
International Conference of Soil Mechanics and Foundation Engineering, Vol. 2,
1957, pp. 66-71.

7. J.B. Burland. Shaft Friction of Piles in Clays - A Simple Fundamental Approach.


Ground Engineering, Vol. 6, No.3, May 1973, pp. 30-42.

8. G.G. Meyerhof. Bearing Capacity and Settlement of Pile Foundations. Journal of


Geotechnical Engineering, ASCE, Vol. 102, No. GT3, 1976, pp. 197-228.

376 Daniel O. Wong


Figure 1. Hollow shaft auger

Figure 2. Top view of an auger-cast pile

377 Daniel O. Wong


C (kpa)

25 50 75 100 125
...-----,-----.,--------'-,-------,------.------..,0

10

20

30

10

e 0
m
::I: 40 "'IJ
~ -I
c.. :r
l.IJ
0 Ti
,j
0
15 2% 50

60

20

70

UMCONFlOIED COMPRESSION

UU TRIAXIAL

2"/. FAILURE STRAIN


80
25
OPN SYIIBOL DENOTES
SLICKENSiDED FALUAE

l.- ----l ----L ----I. -'---'- --'-''--- --' 90


o 0.5 1.5 2 2.5 3
C (ksf)

Figure 3. Undrained shear strength profile

378 Daniel O. Wong


379 Daniel O. Wong
. ':*h

Figure 5. Load test arrangement

Figure 6. Dial guages, jacks and load cells

Daniel O. Wong
380
LOAD (KN)

0'==- .:..::r=- 1"-"0T'0'-"0'--- -'1-"'-5r=-0-"'-0 2000


--=..:;=-;=____--,


0.1
::::::--- ~
5 '-::::::::.- 0.2
~-. .......
~ .......... ........
Virgin Test on June 11 & 12 "'" ......... 0.3
Reloaded Test on June 12
"-
E 10 \ 0.4
E - - - Reloaded Test on June 15 \ ......
.5
I-
Z
\
NOTE: MAXIMUM SETTLEMENT WAS ACHIEVED 1 HOUR 0.5 I-
LU
\ z
~ AFTER MAXIMUM LOAD WAS APPLIED FOR THE LU
LU ~
...J
l-
I-
15 JUNE 15 TEST. \ 0.6 ~
l-
LU
(I)
......... \ I-
LU

--
....... (I)
\ 0.7
.......
\
20
.......
- 0.8

.......
----- \
0.9

25
- --- 1.0

LOAD (ton)

Figure 7. Load movement curves for Test A

381 Daniel 0. Wong


LOAD (KN)

0 1000 1500 2000


0 0

0.1

5 0.2

0.3
--.:
E
--
E 10 0.4 --
l-
Z
I- W
Z
w :=:
w
:=:
W
NOTE: 1. Test on June 17 & 18 - 0.5 ...J
I-
...J 2. Maximum Settlement of 44 mm (1.74 in) l-
I- W
I- 15 was achieved 7 hours after maximum load (/)
w 0.6
(/) was applied.

0.7

20
0.8

0.9

25
1.0

1.1
0 20 40 60 80 100 120 140 160 180 200 220 240

LOAD (ton)

Figure 8. Load movement curve for Test B

Daniel O. Wong
382
LOAD (KN)

00 500 1000 1500 2000


..... 0
.....
0.1

-
::::-
. "'-
~- .....
5 ~ ......... 0.2
.........
~ .........
'-....
~ ...........
0.3

10
~. " :\. c
0.4
E Measured ~~ I-
Z
E
"\
UJ
"- Reconstructed using c/p = 0.3 ; dia = 472 mm 0.5 ~
I- UJ
Z Reconstructed using c/p = 0.4 ; dia = 406 mm ...I
I-
UJ
~
UJ
...I
15 '\ I-
UJ
en

~
l-
I-
UJ
en 0.7

20 \
0.8
I
I 0.9
I
25
I 1.0

l...----l.._ _..L.-_-1.._--::,.l"-_--..1.~__:~-__....,~-~-~~__,,.~-~=__~1.1
I
o 20 40 100 120 140 160 180 200 220 240

LOAD (ton)

Figure 9. Measured and reconstructed curves for Test B

383 Daniel O. Wong


LOAD (KN)

oo;:..". ---:~~-----1.:.;:0'_T0'-=0'-------1;.;:5;_:0...::0-----=::;.==--.....,
2000
0

0.1

5 0.2

0.3

E 10 0.4
E .....
.E
I-
Z 0.5 I-
w z
::E w
w ::E
..J w
l- 15
I- 0.6 ~
W I-
CI) W
CI)
Measured (Virgin) 0.7
Reconstructed using c/p = 0.3; die = 500 mm
20
- . - Reconstructed using c/p = 0.4; die = 406 mm 0.8

0.9

25
1.0

o 20 40 60 80 100 120 140 160 180 200 220

LOAD (ton)

Figure 10. Measured and reconstructed curves for Test A

Daniel O. Wong
384
UTILIZATION AND QUALITY CONTROL OF AUGERCAST
PILES

by Andrew T. Booth,P.E. 1 and Kirk A. McIntosh, P.E. 2

ABSTRACT: Augered, cast-in-place (augercast) piles have been used in the foundation
industry since the early 1950's. Thousands of projects have been completed in the
United States utilizing augercast piles. The pile type offers an advantage over some
driven pile types in its adaptability to low headroom and confined work space
conditions, variable length installation conditions and where construction noise and
vibrations are to be minimized. Although augercast piles have proven to be
successful and economical under some conditions, some government agencies do not
readily accept this foundation type over the more established alternatives, such as
driven piles or drilled shafts. Considering the fact that drilled shafts are widely
accepted, it is the opinion of the authors that the exclusion of augercast piles must lie
in the quality control/quality assurance aspect of cast-in-place pile installation. The
primary focus of this paper will be to discuss recent advances in the augercast pile
industry with regard to model specifications for pile installation and field inspection
techniques/quality assurance. Important specification input items and field quality
control aspects during pile installation are discussed. It is the intent of this paper to
show that augercast piles should be accepted as a viable pile foundation alternative
and will perform favorably when installed under proper guidelines.

'Senior Geotechnical Engineer, Law Engineering, Inc., 3901 Carmichael


Avenue, Jacksonville, Florida, 32207

2Principal Geotechnical Engineer, Inc. Law Engineering, Inc., 3901 Carmichael


Avenue, Jacksonville, Florida 32207

385
Booth/McIntosh
INTRODUCTION
Augercast piles have been utilized in the United States for a wide range of
projects since the 1950's. Due to the cast-in-place nature of the installation method
(and lack of a driving criteria); however, augercast piles have not been readily
accepted by some practicing geotechnical and structural engineers. Since augercast
piles were initially developed, the installation equipment and techniques have been
improved and refined and the number of experienced contractors has significantly
increased. Over the past 10 to 15 years, the pile type has experienced significant
growth in the United States (particularly in Florida) because engineers have become
more knowledgeable and comfortable with its use. The Miami area, in particular,
has experienced considerable growth in the utilization of augercast piles primarily due
to their usage on the $1 billion Miami Rapid Transit system in the early 1980's. The
economic advantages over other deep foundation types, coupled with some unique
installation advantages, have convinced engineers that augercast piles can be a viable
foundation alternative on many commercial projects. It is the opinion of the authors
that the augercast pile industry is currently at a similar stage of acceptance that the
drilled shaft industry experienced about 15 to 20 years ago.
Even though augercast piles have gained respect in the private sector, the pile
type has generally not been readily accepted by some government agencies. It is the
opinion of the authors that the reluctance to readily accept this foundation type stems
primarily from the quality assurance aspect of cast-in-place pile construction.
Additionally, there has not been much research activity on this foundation type,
technical papers, formal associations, or a concerted effort to standardize the
industry. The following sections of this paper will focus on the history of the pile,
augercast pile installation procedures, design considerations, quality control measures,
integrity testing, and new developments in the industry with respect to
standardization.

HISTORICAL BACKGROUND AND OVERVIEW


Augercast piles were developed in the United States in the late 1940's by
Intrusion-Prepakt, Inc. In 1956, a patent on the installation method using a hollow
stem auger was granted to Raymond Patterson. Around that time, the Lee Turzillo
Contracting Company and Berkel & Company Contractors began installing augercast
piles under license agreements. The patent expired in 1973, which opened up
opportunities for prospective contractors and resulted in increased use of the pile in
the United States and other parts of the world. The technology and procedures for
augercast pile construction have changed significantly since the 1940's with today's
contractors capable of constructing piles up to 30 inches in diameter and 130 feet in
depth in the United States. Augercast piles 30 and 36 inches in diameter have been
installed in Europe and the Orient. Although uncommon, augercast piles have been
installed over water. Low headroom piles up to 24 inches in diameter and 100 feet
in length have also been installed. The need for full-length reinforcement, however,
generally restricts the length of low headroom piles to 50 feet.
Currently, there are two national contractors and about 20 regional contractors
that specialize in augercast pile installation in the United States. Typical sizes for
augercast piles vary from 12 to 16 inches in diameter; however, 18 to 24-inch

386 Booth/McIntosh
diameter piles and even 30-inch diameter piles are not uncommon. There have been
thousands of augercast pile projects completed in the United States, including the
support of multistory office buildings, bridges, tanks, hotels, hospitals, railroad
bridges, parking garages, industrial structures, pipelines, and rapid transit structures.
Other applications of augercast piles include cast-in-place earth retaining walls,
excavation support, tiebacks, and slope reinforcement.
The cost of the pile is a function of grout cost (which can range from less than
$40/yd 3 to over $110/yd 3 in the U.S.), labor cost, pile diameter and length, the total
number of piles, and pile spacing. The cost of crane-installed piles can vary from
$7 to $25 per lineal foot. The unit cost of low headroom piles can range from $35
to $50 per foot. In many parts of the U. S., the installed cost of an augercast pile is
roughly equivalent to the material cost of a precast concrete or steel H-pile.

PAST UTILIZATION ON TRANSPORTATION PROJECTS IN THE UNITED STATES


At least three major transportation-related projects have been constructed in
Florida utilizing augercast piles--the Automated Skyway Express in downtown
Jacksonville, Florida and the Downtown People Mover and Rapid Transit Systems
in Miami, Florida. The senior author inspected hundreds of piles and evaluated
dozens of successful static load tests for the latter project. To the knowledge of the
authors, there have been no major bridges constructed in Florida supported on
augercast piles. Augercast piles have been used for bridge support in New Mexico,
Illinois, Kansas, Minnesota, Maryland, New Jersey, Nebraska, Pennsylvania, New
York, Ohio, Michigan, Washington State, Virginia, and Kentucky. Other types of
transportation-related projects where augercast piles have been successfully utilized
include railroads and port facilities.

TYPICAL INSTALLATION PROCEDURE


In general, augercast piles are installed by rotating a continuous hollow-stem
flight auger into the ground until a specified termination criteria is achieved and then
pumping a sand-cement grout with admixtures through the auger stem under pressure
as the auger is slowly withdrawn to fill the drilled hole.
The grout used in augercast pile construction is usually rich in cement content
and typically has the consistency of a thick milkshake. During the augering phase,
a cork stopper is used to plug the hole located at the hollow-stem auger tip. This
cork serves as a tremie seal to prevent soil or rock cuttings from entering the hollow
stem auger. Grout is generally placed from ready-mix trucks directly into a portable
concrete pump. A pressure gauge is installed either at the pump or in line to
measure grout pressures during installation. The entire augering and grouting
procedure generally takes 3 to 5 minutes for a 50-foot long pile. A typical set-up for
an augercast pile rig is shown in Figure 1.
In the case of low headroom piles similar installation procedures are utilized
with the exception that the piles are augered in sections, typically 3 to 12 feet each.
During grouting, each section is incrementally removed, and the auger tip dropped
2 to 3 feet or so prior to resuming withdrawal to maintain an adequate grout pressure
head. The installation of low headroom piles is considerably slower than

387 Booth/McIntosh
conventional piles, as would be expected. A typical set-up for low headroom pile
installation is illustrated in Figure 2.

SWIVEL

GEAR BOX

PILE LEADS

CONTINUOUS FLIGHT HOLLOW


STEM AUGER

GROUT HOSE

0000000000000

GROUT PUMP CRANE


COUNTERWEIGHT

FIGURE _1 TYPICAL AUGERCAST PILE RIG SET-UP (Courtesy of DFI>

LINE TO HYDRAULIC POWER UNIT

ADDITIONAL AUGER SECTIONS

HOLLOW STEM AUGER

FIGURE -.L . TYPICAL LOW-HEADROOM AUGERCAST PILE RIG SET-UP

388 Booth/McIntosh
DESIGN CONSIDERATIONS
It is not the intent of this paper to address pile capacity detennination.
Methods of estimating pile capacity have been addressed by others (Neely, 1991 and
McVay, Annaghani, and Casper, 1994). In some geographic locations, the only
limitation on allowable pile compression capacity is the building code allowable
design stress requirement (0.25 f' c in the case of the Standard Building Code). Many
building codes also restrict maximum pile lengths as a function of pile diameter
(typically 30 times the diameter). This restriction can generally be waived with
successful load testing and the recommendation of a registered professional engineer.
Because there are several factors which influence the load-carrying capabilities of the
piles, these factors will be briefly discussed in the following paragraphs.
Because augercast piles are nonnally installed to one of two primary criteria
(either auger "refusal" or to an embedment criteria), the Geotechnical Engineer
should clearly state which criteria is applicable to the specific. project. If auger
"refusal" is specified, the equipment requirements assumed (auger drive head or gear
box weight and torque) to create the refusal condition should be clearly stated. In
general, the torque of available power units range from 10,000 to 50,000 ft-lbs., with
typical values of 12,000 to 35,000 ft-lbs. A power unit having a rated torque of
12,000 ft-lbs may not be able to penetrate a weakly cemented or soft limestone;
however, a unit with a torque of 25,000-30,000 ft-lbs may be able to penetrate
several feet. The type or design of the cutting head is also important. Although
actual equipment selection should be left to the pile contractor, failure to adequately
discuss the subsurface soil and rock conditions and specify unusual equipment
requirements can result in project delays and the potential for claims. As would be
expected, higher torque is generally desirable when drilling into stiff or hard cohesive
soils, dense granular soils, or when embedment into weakly-cemented rock is
desirable.
Augercast piles can be successfully installed in almost any type of subsurface
conditions. Subsurface conditions dictating specialized procedures include:

Porous man-made fill layers (which can result in grout loss or


subsidence)
Soft cohesive soils (which can create bulging of the pile shaft)
Highly porous limestone or gravel layers (which can result in grout
loss or subsidence problems)
Organic soils (which can inhibit hydration of cement if mixing of the
organics and grout occurs)

In the above conditions, a pennanent casing may be required during pile installation.
The writers have witnessed successful pile installation in the above conditions. The
key to successful installation is coordinating the rate of auger withdrawal with
adequate grout head and pressure so that any voids are completely filled with grout.
The grout mix used in augercast piles typically consists of cement, sand,
water, flyash and a fluidifier. Mixes containing 7 to 10 sacks of cement per cubic
yard, with and without flyash, are commonly used throughout the United States.
Coarse aggregate is not used in the United States but is used in Europe. The grout

389 Booth/McIntosh
mix has a soupy consistency and cannot be tested with a slump cone; instead, a
flowcone having an orifice diameter of 3/4-inch is used. A flowcone rate of 10 to
25 seconds is commonly specified. Flow rates less than 10 seconds may be indicative
of an excessively high water-cement ratio and low grout strength. A flow rate in
excess of 25 seconds may indicate the grout is too thick to pump properly. A large
variation in flow rates with the same mix is usually indicative of a variable sand
stockpile moisture content or inadequate mixing.
The number of grout specimens for unconfined compressive strength testing
should be clearly specified, with typical frequencies of 2 sets of 6 specimens per day.
Either 2-inch square cubes or 2 by 4-inch cylinders can be used. The grout used to
make specimens should preferably be obtained from the pile itself or from the bottom
of the auger. Typical specified grout unconfined compressive strengths are in the
3, 000 to 5, 000 psi range at 28 days.
Because of the cast-in-place nature of the pile, pile caps should not be
constructed until grout strength tests indicate that the piles will reach the design 28-
day strength. To reduce the delay in cap construction as much as possible, several
intermediate tests (3, 7 and/or 14 days) are recommended to determine if a potential
strength problem exists.
Factors influencing grout mix design include sand content (which may affect
reinforcing steel embedment), use of a retarder in hot weather to control initial set,
and cement content (higher cement contents improve pumpability).
Augercast piles are often reinforced after grouting. Cages are used for lateral
load resistance and full-length reinforcing bars or threaded reinforcing bars for
tension resistance. In some parts of the U. S. , full-length reinforcing bars are inserted
into the hollow stem of the auger prior to grouting to assure full embedment and
centering of the bar.
Other factors to be considered in pile design include the following:

Minimum headroom available (minimum 9 feet desirable) and its effect


on pile reinforcing and embedment length
Prolonged drilling on a batter (which can result in large volumes of
soil being removed)
Drilling near existing shallow foundation-supported structures
(potential for inducing settlement if delays occur during grouting
phase) (Esrig, Leznicki, and Gaibrois, 1994)

The number and type of probe piles and pile load tests should be specified.
Since augercast piles are installed without a penetration resistance criteria, a more
comprehensive load test program may be warranted than with driven piles. Probe
piles (piles installed prior to production piles) are a good idea and permit both the
geotechnical engineer and contractor to develop knowledge of the drilling and
grouting characteristics of the subsurface soils and rocks prior to production. Probe
piles are generally not installed in production pile locations. As with any pile type,
it is desirable to load test piles to failure to confirm design assumptions. When
possible, test piles should be instrumented internally with multiple vibrating wire
strain gages and a full-length telltale rod to determine load distribution with depth and

390 Booth/McIntosh
load transfer. The instrumentation can be attached to a full-length steel reinforcing
bar or threadbar fairly easily and inserted into the pile after grouting. Grout cubes
or cylinders made from the test pile should be tested to determine the grout
unconfined compressive strength on the day of the load test.
Some relatively recent methods of high strain dynamic load testing of
augercast piles have been developed. One method utilizes the Pile Driving Analyzer
(PDA) and wave equation analysis. The test is performed similarly to that of re-
striking a driven pile to confirm capacity. Another test method utilizes a "statnamic"
device. Statnamic testing involves an explosion chamber located between the pile and
a reaction mass. The explosion acts to propel the mass upward and force the pile
downward. These dynamic tests can usually be performed quicker and more
economical than conventional static load tests, particularly where multiple load tests
are to be performed.

QUALITY ASSURANCE MEASURES


It is widely recognized that the success of an augercast pile project is highly
dependent on the skill and experience of the contractor. Even a properly designed
augercast pile foundation and well-planned project can result in failure if the
contractor is not willing or able to follow acceptable procedures. Since the quality
of augercast piles is so dependent on the contractor, full-time inspection during
construction is imperative. Ideally, a geotechnical engineer familiar with the site and
subsurface conditions, the design assumptions, and augercast pile installation practice
should monitor the installation. In some cases, an experienced engineering technician
working under the supervision of a geotechnical engineer would be acceptable.
Although the grouting phase is by far the most important, quality assurance measures
for both drilling and grouting phases will be briefly discussed.
Drilling Phase - Before drilling actually begins, the inspector should confirm
that the auger is initially centered over the desired pile location and that the leads are
checked for plumbness using a carpenter's level. The alignment of battered piles can
be checked using a magnetic protractor/inclinometer or a prefabricated jig and level.
It is important that the pile inspector have copies of boring logs on hand to
help confirm that the subsurface stratification at the pile location is as anticipated.
The relative degree of difficulty of drilling (easy, moderate, or hard drilling, for
example) should be noted and compared to boring logs. The relative degree of
difficulty of drilling is a function of auger drive head torque, horsepower, weight and
type of cutting head and the type of soil or rock being penetrated. Observed changes
in drilling difficulty should be documented.
If auger refusal is the applicable pile termination criterion, it is important that
the full auger drive head assembly weight is bearing on the auger stem when making
the termination observation. This can sometimes be confirmed by observing slack
in the cable from which the drive head assembly is suspended. In some subsurface
conditions, the cable cannot be allowed to go slack since the auger may "freeze" in
the hole. Piles installed to a specific length or embedment criteria should also be
carefully documented.
One aspect of the drilling phase involves prolonged drilling at one depth or
on a batter in saturated sands which can result in a loss of ground significantly

391 Booth/McIntosh
greater than the theoretical volume of the drilled hole. When drilling is interrupted
near existing structures, this loss of ground can induce settlement of soil-supported
structures and buried pipelines (Esrig, Leznicki, and Gaibrois, 1994). In this case,
the speed of rotation of the auger should be slowed down to reduce the volume of soil
removed from the ground, and every effort made to drill and completely grout each
pile in one continuous, uninterrupted operation.
Grouting Phase - The most important quality assurance aspect during
grouting is the maintenance of an adequate pressure head of grout (generally at least
10 feet) on the auger flights above the auger tip. It is generally required that this
pressure head be built up within the bottom half of the pile's length. Maintenance
of a grout head above the auger tip is important since it helps prevent caving of the
borehole sidewalls and permits some lateral flow of grout into porous soil, rock or
fill zones to occur as the auger is pulled. If an inadequate head of grout exists above
the auger tip and a soft soil layer or porous zone is encountered with the auger pulled
at a consistent rate, the auger may be pulled out of the grout column creating suction
and causing the walls of the hole to collapse. In subsurface conditions which include
a shallow water table, observation of a continuous fluid return out the top of the hole
(initially soil mixed with groundwater, then grout) is the best indication that the
maximum pressure head possible is being achieved.
Another very important aspect of the grouting procedure is confirming that at
least 115 percent of the theoretical pile volume is pumped per 5-foot increment. This
is accomplished by documenting the cumulative number of pump strokes per
increment starting at the base of the pile. If the pump calibration factor is known
(volume pumped per stroke), the minimum number of strokes necessary per
increment can be determined prior to pile installation. Any pile constructed with an
incremental stroke count less than the target would then be questioned.
Calibration of the grout pump should be performed at the start of each project
and following major maintenance on the pump. This is usually performed using a
large capacity trash can or drum as a container. The authors also recommend that
a record of the total number of pump strokes per grout truck be maintained and the
pump calibration factor checked periodically by comparing the computed grout yield
to the batched volume.
The overall grout factor (total grout volume pumped divided by theoretical
volume of the drilled hole) is often used as a means of confirming pile quality. The
authors have seen overall grout factors in the range of 1.15 to over 3 for acceptably-
installed piles. While this factor is useful when compared to other piles,
interpretation of it must be performed carefully. In general, low grout factors (1.3
or less) would be expected in stiff or hard cohesive soils or dense sands since
"bulging" of the pile shaft would not be expected. An overall grout factor less than
about 1. 10 may indicate a problem and would be cause for evaluation of the pile.
High grout factors (1.7 or greater) might be expected in porous limestone, soft soils,
gravels, or very loose sands. Piles installed with large pressure heads (a desirable
condition) will obviously result in a significant volume of grout being wasted out the
top of the pile. It is important for the pile inspector to document the reason for a
high overall grout factor when this condition occurs, especially if the high factor is
markedly different from other piles. A more useful grout factor can be calculated by

392 Booth/McIntosh
documenting the total number of pump strokes (actual grout volume) at the time grout
is first observed at the ground surface (termed the "grout return depth"). This grout
factor represents the actual volume of grout necessary to completely fill the drilled
hole (less the auger volume remaining in the hole) compared to the theoretical
volume. The remaining grout volume pumped into the pile is only necessary to
replace the volume lost as the last portion of the auger is extracted. In most cases,
the auger is extracted at the same rate as the bottom portion of the pile and much
grout is wasted at the surface. It can be seen that the grout factor at the grout return
depth is more representative of the actual grout volume used to construct the pile and
can help detect when differing subsurface conditions are encountered at a site.
A lot of controversy exists concerning the use of p'ressure gauges. In the
authors' opinion, it is not possible for a pile inspector to continuously watch a
pressure gauge and observe all the other important facets of pile installation at the
same time. At best, pressures can be periodically monitored during grouting. It is
probably more useful for the crane operator to have a gauge in his cab that can be
more continuously monitored than by the inspector. Grout pressures are highly
variable and are dependent on the location of the gauge, the height of the auger, the
length and diameter of the hose, and stiffness of the grout mix. One benefit of a
pressure gauge is that non-uniformity of grout pressure from one pile to another or
drastic changes in pressure within a single pile may be an indication of a defect or
a subsurface anomaly which should then be further evaluated.
During grouting, it is generally preferred that the auger be withdrawn at a
slow, continuous rate or in uniform 6 to 12-inch increments. This is normally
governed by whether the drilling rig is equipped with a torque converter or not. The
auger should be rotated in a clockwise manner during extraction. Rotation of the
auger stabilizes the walls of the hole and keeps spoil material moving upward and out
the top of the hole. Rotation is sometimes stopped when the grout pressure head
reaches the top of the hole. Counter-clockwise rotation of the auger can mix spoil
material with grout and should not be permitted.
Delays during grouting should be documented by the pile inspector. The
depth of the auger tip, reason for the delay, and duration should be recorded.
Depending on the time of delay, it is generally preferred that the auger tip be lowered
a specified distance (typically 5 to 10 feet) below the depth where the delay occurred
and the pile re-grouted. This procedure is normally performed as a safety measure
against the fluid grout column settling into porous soil or rock strata below the auger
tip and an inclusion or neckdown created when auger withdrawal is resumed. It is
preferred that all piles be drilled and grouted during one continuous operation.
The quality and age of the grout mix should be carefully monitored during
installation. Most specifications allow a maximum age of 60 to 90 minutes from the
time of batching. Because some grout mixtures include a retarder to promote
workability, this time restriction may be unfair in some cases. In the authors'
opinion, a logical alternative for the estimation of "initial set" is to sample each batch
of grout directly from the truck with a small disposable cup when the grout truck
arrives on site. This "control" sample is then set aside and periodically checked by
the pile inspector to determine when initial set has occurred. Initial set has occurred
when the grout assumes the shape of the container when tilted from the vertical.

393 Booth/McIntosh
ASTM C 109 addresses the making and testing of grout cubes. Extra sets of cubes/
cylinders should be made on grout batches which have reached initial set or have
been diluted with an excessive amount of water at the site. As a general rule-of-
thumb, the addition of one gallon of water per cubic yard of concrete (in excess of
the design mix amount) has the effect of reducing the 28-day compressive strength
by about 200 psi.
Another tool for the evaluation of grout acceptability is the flow cone.
Because the grout used in pile construction cannot be tested with a slump cone, a
flow cone with a 3/4-inch diameter orifice is used. Applicable standards for testing
include ASTM C 939 and Corps of Engineers CRD C 79. The flow cone is useful
in calibrating the inspector as to what the "correct grout mix" should look like. Once
calibrated, the frequency of testing can decrease, or testing only performed on
especially thick or watery batches. The authors recommend making an extra set of
cubes or small cylinders on any batch of grout which is used to construct piles and
does not achieve a flow rate within the specified range.
When grouting is complete, reinforcing steel may be added to the pile. This
should be observed and documented. Full-length reinforcing bars should easily
penetrate the grout column under their own weight. Full length rebars or threadbars
have been installed in piles 70 feet in length following grouting. Generally these bars
must be at least one-inch in diameter and possess adequate weight to penetrate the
grout column to the bottom of the pile. Lack of penetration to the bottom of the pile
can be an indication of an inclusion (defect in the pile), premature setting of grout,
or the steel striking the side wall of the pile. To facilitate installation, the authors
recommend that centralizers or spacers be used on all full-length rebars (one at the
tip and every 20 to 30 feet thereafter to the top of the pile). The minimum diameter
of the centralizers can be determined based on the specified grout compressive
strength, and maximum acceptable axial stress considerations. If the pile will be
deriving a significant portion of its capacity in end-bearing and full-length reinforcing
does not penetrate to the bottom of the grout column, the steel should be removed
and the pile re-drilled and re-grouted.
Reinforcing cages can also be installed in piles after grouting. The difficulty
in the insertion of these cages increases for cage lengths in excess of about 20 to 25
feet; however, cages have been successfully installed to a depth of 60 feet in vertical
piles. In the case of battered piles, cages are even more difficult to install beyond
a length of about 20 to 25 feet. Rollers are generally required on reinforcement
cages for battered piles. Because cage insertion can be difficult even in properly
grouted piles, most augercast pile contractors prefer that the cage length be limited
to about 25 feet, and that single full-length reinforcing bars or high strength steel
threaded bars be added for tensile reinforcement. Reinforcing steel installation is
often difficult in sands above the water table since the dry soil can "extract" water
from the fluid grout mix, causing it to stiffen rapidly. Reinforcing steel installation
may also be difficult in low headroom piles due to the period of time necessary for
pile installation and the restricted headroom.
The installation requirements of adjacent piles should be carefully specified.
There are two factors that must be considered when determining acceptable lateral
distances: (1) lateral communication between piles due to porous substrata, and (2)

394 Booth/McIntosh
vibration of fresh grout in an adjacent pile during drilling of another. With the
exception of porous limestone formations and possibly rubble fill or gravel layers,
the first factor is generally not an issue. In the case of vibrations, since the
construction of augercast piles does not generate high vibration levels, it is generally
not a problem either. The Standard Building Code requires that augercast piles not
be installed within 6 pile diameters (center-to-center) of each other until the grout
reaches an age of 24 hours. This requirement can be waived by the Building
Official. In the authors' opinion, the normally specified three pile diameter (edge to
edge) restriction placed on contractors in the construction of adjacent piles is a
reasonable starting point. A wait period of at least 12 hours for piles installed within
three diameters is also recommended. In any event, the pile inspector should monitor
grout levels in completed piles since subsidence (or rise) or disturbance can occur
even in piles installed several feet from each other.
Subsidence immediately following construction is common in augercast piles,
especially in porous limestone and soft cohesive soils. The subsidence of the piles
is generally due to the fluid weight of the grout column "pushing" the grout laterally
into pore space in the rock or expanding the shaft laterally into soft soils. Since it
is believed that the grout column will only settle as a unit while it is still in a fluid
condition, it is acceptable to top up the grout level if done prior to initial set. If the
grout level settles several feet, the grout hose can be used as a tremie and new grout
pumped into the pile to cut-off level. The important consideration here is that fresh
grout should not be added to the pile unless it can be done without mixing with soil.
Low Headroom Piles - Because low headroom piles are augered and grouted
in sections (typically 3 to 12 feet), the procedure is much slower than normal;
however, piles of virtually any length can be installed. Drilling and grouting of the
piles is similar to conventional augercast pile installation with the exception that
reinforcing steel insertion can be more difficult due to the age of the grout at the base
of the pile. Full-length reinforcement may require the use of steel threadbars spliced
with couplers. The structural design of the pile and the grout mix itself should be
carefully considered in the case of restricted headroom piles, since steel insertion can
be particularly difficult.

ADVANTAGES AND DISADVANTAGES OF AUGERCAST PILES


Augercast piles have some unique advantages over most of the driven,
displacement-type piles. These advantages include the following:

Augercast piles are low in cost as compared to other pile types;


primarily because material costs are relatively low and installation is
relatively fast. The authors have routinely witnessed production
magnitudes on the order of 40 augercast piles acceptably installed to
depths of 50 to 60 feet per day when significant delays are not
experienced.

Vibration and noise levels are relatively low, which serves to reduce
litigation and damage that could arise when installing piles near
existing residences, businesses, office buildings, hospitals, etc. In

395 Booth/McIntosh
some instances, vibrations associated with conventional pile driving
operations can damage nearby structures--usually when loose
cohesionless soils beneath a structure are densified by the vibrations
and cause the foundations to settle.

Augercast piles can be constructed in limited access conditions where


conventional pile driving equipment and premanufactured piles cannot
be utilized. Augercast piles are particularly advantageous over other
pile types when low-headroom conditions exist. The method of pile
installation can be adjusted to accommodate the low-headroom
condition.

Augercast pile lengths can be readily adjusted in the field in cases


where a termination refusal criteria is specified rather than a tip
elevation criteria. If unanticipated weaker subsurface conditions are
encountered, the piles can be lengthened without excessive extra costs
other than the unit cost for the extra pile length. This eliminates
splicing costs associated with driven piles. Additionally, if refusal
conditions are encountered at higher elevations than anticipated, there
are no wasted materials or costs associated with pile cut-offs.

Augercast piles typically generate higher skin friction capacity than


driven piles; therefore, they provide higher allowable uplift capacity
per unit of surface area. In cases where several feet of firm to dense
soils must be penetrated prior to reaching the intended pile bearing
stratum, this factor can result in significantly shorter pile lengths for
augercast piles than driven concrete piles (which generally require the
construction of a pilot hole).

Augercast piles can generally develop higher side shear capacity than
driven piles when installed into weakly-cemented rock formations.
The grout tends to interlock with rock formations as the grout flows
into fissures or solution channels in the rock. Installation of driven
displacement-type piles through weak rock formations fractures the
materials and results in lower skin friction resistance. In Miami,
Florida, for example, an allowable compression capacity of 100 to 125
tons on a 16-inch diameter augercast pile 20 to 40 feet in length
drilled into the surficial Miami Limestone formation is common.

The single, most important disadvantage of augercast piles is that the integrity
of the pile is highly dependent on the skill of the contractor's field personnel. For
this reason, only qualified, experienced contractors should be considered for projects.
Another disadvantage of augercast piles is the potential for excessive grout takes into
porous rock formations (such as limestone), porous rubble fill layers, underground
utility conduits, and "bulging" of the grout column in weak soils (such as soft clay
or peat). While acceptable piles can be constructed in these conditions, the unknowns

396 Booth/McIntosh
associated with required grout volume can result in contractor claims for additional
compensation.

NOISE AND VIBRATION CONSIDERATIONS DURING INSTALLATION


Vibration and noise levels generated by an augercast pile rig were recently
measured in Jacksonville, Florida using a series of Instantel Blastmate Series II, DS-
477 seismographs. The augercast piles were installed using a crane-operated rig with
a power unit having a variable torque of 10,000 to 35,000 ft-Ibs. The subsurface
conditions generally consisted of a very soft to loose overburden soils in the upper
30 to 40 feet underlain by a weakly-cemented clayey silty fine sand (locally termed
"marl"). The marl formation exhibits a standard penetration resistance in the range
of 35 to over 100 blows per foot. The higher level of torque was generally utilized
to penetrate this layer. The 16-inch diameter piles were installed to an embedment
of 10 feet into the marl formation, resulting in a total pile length of about 50 feet.
Figure 3 presents the results of the vibration monitoring.

1
0.9
0.8
0.7
0.6
0.5

0.4

UQ) 0.3

"'
........
c
.:.=,..- 0.2
\
\
>- E
.- E
oU"
-ltn
.
WN
> 0.1
.09
II
W .08
-lJ::
U u .07
F c
~- .06
\-
~
o..~ 05 \
::.::
\
.04

W
0..
.03 1\\
.02
\

.01
'"
10 100

DISTANCE (Feel)
(1 Fl = 0.305 meters)

FIGURE ~ I AUGERCAST PILE INSTALLATION VIBRATIONS

It was found at this site that beyond a distance of about 50 feet, vibrations generated
by the augercast pile installation could not be discerned from other distant
construction operations. The vibrations measured from the described equipment were
generally less than that of a caisson drilling operation and more than an idling crane
as shown by Wiss, 1981.

397 Booth/McIntosh
In general noise levels generated by the augercast pile rig could not be
distinguished from noise generated by distant construction equipment, such as
bulldozers and front-end loaders. The upper boundaries of the noise levels measured
generally ranged from about 80 to 90 decibels within a distance of 10 to 150 feet
from the equipment. For comparison, this level of sound is similar to that of average
street traffic (Dowding, 1985).

POST CONSTRUCTION INTEGRITY TESTING


On some augercast pile projects, questions occasionally arise concerning the
acceptability of an installed pile. When the integrity of a pile is questionable, the
preferred solution is to re-drill and re-grout the pile immediately or before the grout
in the questionable pile gains too much strength. Full scale static load tests on
questionable piles are usually too expensive and time consuming for most projects.
Some methods of non-destructive testing of augercast piles have been
developed, which include sonic logging and low strain integrity testing (DFI, 1994).
Sonic logging involves drilling multiple holes, around the pile, in which probes are
simultaneously lowered in order to collect data. The sonic logging test procedure is
relatively time consuming and expensive. Low strain integrity testing includes the
pulse echo method (PEM), transient response method (TRM) and impedance profile
analysis. Generally, low strain integrity testing involves tapping the pile head with
a hand-held hammer and measuring the pile top motion and wave reflection from the
pile tip. Non-uniformities in the pile cross-section can be determined when the data
is interpreted by an experienced engineer. Low strain integrity testing is relatively
inexpensive and numerous piles can be tested in one day. In Europe, integrity testing
is often performed on several or all production piles. Although this is not the
practice in the U. S., such testing could be performed on bridge projects at a
reasonable cost if necessary to gain acceptance of the pile type.

CONCLUSIONS/REcOMMENDATIONS
It has been the intent of this paper to show that inspection and testing
techniques have been developed which provide reasonable assurance that augercast
piles can be properly installed, given suitable subsurface conditions. The advantages
of augercast piles are too significant to preclude this pile type from consideration on
any project where they are technically and economically feasible.
In an effort to increase knowledge of this pile type amongst government
agencies, and to hopefully increase its utilization on government (and commercial)
projects, the following recommendations are offered:

A pile installation specification should be prepared for use on


government projects based on the Guideline Model Specification for
Augered Cast-In-Place Piles published by the Deep Foundations
Institute (DFI). This specification would then be modified, as
necessary, based on the site specific subsurface conditions encountered
at each site.

398 Booth/McIntosh
The Inspector's Manual for Augered Cast-In-Place Piles published by
the DFI should be used as the guide for the training of pile inspectors.
Contractors should be aware of the inspection procedures adopted by
the DFI so that they know what inspection procedures will be followed
on each specific project and accurate foundation cost estimates can be
prepared.

If necessary for acceptance, integrity testing of each production pile


could be performed on government projects.

A certification program for augercast pile contractors based on pre-


determined standards established by a committee of contractors,
academians, and engineers is recommended. The DFI is currently
evaluating the adoption of such a program. It is recommended that
geotechnical engineers employed by the government become active in
the DFI Augered Cast-In-Place Pile Committee.

Equally as important as the certification program for contractors, a


certification program for pile inspectors is highly recommended. This
program would involve classroom and field training, and knowledge
of the DFI Inspector's Manual and Guideline Specification. This
training program could be accomplished in a series of two to three-day
"workshops" instructed by trained personnel. Again, the DFI is
currently considering a certification program for augercast pile
inspectors.

An instructional video should be prepared which shows proper pile


installation and inspection procedures. Safety should also be
addressed. Some augercast pile contractors already possess videos of
pile installation; however, to the authors' knowledge, inspection
procedures have not been addressed on film.

Further academic research into the pile is necessary for it to gain


acceptance. There are currently several, well-respected consulting
engineers and academians which publish numerous papers on drilled
shafts, while very few papers are presented which deal solely with
augercast piles. The augercast pile industry could benefit from similar
university research. The funding for such research should be assisted
by grants from pile contractors and both state and federal
governments.

Government and/or contractor-sponsored test programs in selected


areas of the U.S. should be performed. In these programs, test piles
would be installed using accepted procedures and the piles statically
or dynamically load-tested. Internal instrumentation should be
included to determine load distribution and load transfer with depth.

399 Booth/McIntosh
Each test pile could also be integrity tested using one or more of the
available methods to detennine if pile defects exist. Deliberate pile
defects could be created during installation to confinn that the integrity
testing methods give reasonable results. If possible, the test piles could
be pulled following integrity testing to evaluate the accuracy of the
methods. Similar procedures were perfonned by the drilled shaft
industry 10 to 15 years ago with favorable results. Valuable
comparative information could also be gained if other deep foundation
alternatives were installed and load tested on these sites, along with
the augercast piles.

REFERENCES

DFI (1994). Augered Cast-in-Place Pile Inspector's Manual, Deep Foundations


Institute.

Dowding, C.H. (1985). Blast Vibration Monitoring and Control, Prentice-Hall, Inc.
Englewood Cliffs, N. J., 106.

Esrig, M.I., Leznicki, J.K., and Gaibrois, R.G. (1994). "Managing the Installation
of Augered Cast-in-Place Piles." 73rd Annual Meeting, Transportation
Research Board, Washington, D.C., Paper No. 940776.

McVay, M., Annaghani, B., and Casper, R. (1994). "Design and Construction of
Auger-Cast Piles in Florida." 73rd Annual Meeting, Transportation Research
Board, Washington D.C., Paper No. 940500.

Neely, W.J. (1991). "Bearing Capacity of Auger-Cast Piles in Sand." Journal of


Geotechnical Engineering, ASCE, 117(2), 331-345.

Wiss, J.F. (1981). "Construction Vibrations: State-of-the-Art." Journal of


Geotechnical Engineering, ASCE, 107(2), 167-181.

ACKNOWLEDGEMENTS
The authors wish to acknowledge the members of the Augered Cast-In-Place
Pile Committee of the DFI (particularly Mr. Joel Moskowitz of Mueser Rutledge
Consulting Engineers), Law Engineering, Inc., Berkel and Company (Mr. Mike
Jones), Richard Goettle, Inc. (Mr. Larry Rayburn), and L.G. Barcus & Sons (Mr.
Dick Hoener) -- all of which provided considerable input into this paper. The authors
are also grateful to Mr. Ralph Reese of Contract Drilling & Blasting, Inc. for
providing the equipment to record the vibration and noise level data which was
presented in this paper.
The authors also wish to thank Ms. Kathy Weaver of Law Engineering, Inc.
for typing of the manuscript, and Mr. James Patterson for drafting of the figures.

400 Booth/McIntosh
AUGER-CAST PILES

By James J. Brennan 1

Abstract
Department of Transportation Engineers have typically
been wary of utilizing auger-cast piles (augered pressure
grouted piles) for bridge structures due to the uncertainties
in the design and construction control of these foundation
elements. The Kansas Department of Transportation' s policy
for bridge foundations typically allows the use of steel
friction or end-bearing piles, drilled shafts socketed into
rock, or spread footings founded in rock. The KsDOT has used
auger-cast piles for bridge foundation elements on a limited
basis for secondary bridges in the past. Recently, a major
bridge structure was partially founded on auger-cast piles.
Auger-cast piles were elected as the foundation
element for piers spanning a distance of 146.3 m (480 ft.) of
a bridge totaling 615 m (2018 ft.) This foundation
alternative was selected since the installation of auger-cast
piles would create the least soil disturbance with a
warehouse encroaching to within 6.1 m (20 ft.) of the bridge.
The soil stratum of major concern for disturbance was a
sensitive silt which underlaid the neighboring warehouse.
This paper presents a case history of the project
detailing the engineering profile of the soils at the site,
the design considerations leading to the selection of
auger-cast piles as the foundation element, the design
procedure followed, and the load test results of the
installed pi les. The construct ion provisions developed are
presented along with the revisions that will be incorporated
after observation of the construction methods utilized.
Introduction

The Kansas Department of Transportation undertook the


design of the replacement of the U. s. 169 bridge over the
Kansas River in Kansas City, Kansas in 1988. The existing
bridge had been constructed in 1938 and consisted of a steel
truss structure 268.3 m (880.33 ft.) long spanning the river
channel and simply supported reinforced concrete deck girder
spans for a distance of 346.8 m (1137.67 ft.) The reinforced
concrete deck girder spans provided an elevated roadway above
the numerous railways adjacent to the river valley.
The original design concept envisioned constructing a
welded plate girder span over the river channel to the west
of the existing structure and rehabilitating the existing
truss. This would furnish two parallel bridges over the
river each carrying two lanes of traffic with nominal
shoulders. The existing structure spanning the railroads

1S0ilS Engineer, Kansas Department of Transportation,


2300 Van Buren, Topeka, Kansas 66611-1195 Brennan

401
would be completely demolished in stages and replaced with
mechanically stabilized embankment retaining walls (MSE) with
simply supported slabs over only the railroad tracks.
Soil Profiles
. The geotechnical site investigation performed for the
MSE and simply supported slab structures revealed two widely
differing types of soils. The soil profile for a distance of
approximately 133 m (436 ft.) north of the river channel can
be described as 0 to 1.829 m (0 to 6 ft.) of trash and rubble
fill overlying a layer of silty to medium sand .61 to 3 m (2
to 10 ft.) in thickness. This relatively minor sand layer
was further underlain by a soft silt to a depth of 10.7 m (35
ft.) before the medium to coarse sand zone is encountered.
This sand zone lies conformably on the bedrock for the area
which consists of shale at a depth of 27 to 30 m (90 to 100
ft. )
The remainder of the proposed alignment possessed
foundation soils consisting of silty sand overlying medium to
coarse sands with minor silt lenses found throughout. An
exception was found for approximately 61 m (200 ft.) of the
alignment which crossed a deep, demolition landfill
containing layers of metal, cinders, rubble, and wood
contained in an extremely soft soil matrix.
The silt encountered in the soil profile immediately
north of the river channel was described as very soft with 3
blows required to drive a Standard Penetration Test sampler
610 mm (2 feet.) In-situ vane shear tests yielded undrained
shear strengths (su) of 23 to 28.7 kPa (480 to 600 psf.)
Undisturbed samples were obtained for triaxial shear and
consolidation testing. The CIU triaxial test results
yielded: c = 40 to 89.6 kPa (835 to 1872 psf), = OOi c'
o to 4.1 kPa (86 psf), and ' = 36.6 0 to 45 0 Void ratios
in the silt ranged from 1.2021 to .7659 with the majority of
the samples exhibiting void ratios of approximately 0.9. In
addition, numerous perched water tables were found within the
silt zone creating in-situ moisture contents in the upper
30's while the liquid limit of the silt was typically in the
lower 30's.
The soil profile is shown on Figure 1 Typical
Soil Strata Profile.
Feasibility of MSE Structures
Analysis of the proposed MSE structures for these
foundation conditions indicated settlements as great as 460
mm (18 inches) could be expected using procedures found in
Lambe and Whitman (1969) with safety factors against deep
shear failures less than the minimum acceptable value.
Furthermore, the construction sequencing required the MSE
structures to be built adjacent to and under the existing
bridge while continuing to route traffic over the existing
structure. Plans from the 1938 construction indicated the
existing structure was founded on timber piles driven into
the medium to coarse sand underlying the silt. Calculations
indicated the construction of the MSE structure would
overstress the existing piles.
In light of these conditions, alternative means of
constructing the MSE structure in the zone immediately north
of the river channel were pursued. The support of the MSE
structure via stone columns was first analyzed. It was
determined a stone column supported MSE structure would still

402 Brennan
stress the existing timber piles dangerously close to their
ultimate capacity.

0.9 m (approximate) Trash and Rubble Fill

3.3 m (approximate) Silty to Medium Sand

Clayey to Sandy Silt, very soft, wet

numerous perched water tables

N = 3 blows for 610 rom

Su 23 to 28.7 kPa (vane shear)

eo = 0.9
CIU triaxial test results:

r/J = 0 0
c = 40 to 89.6 kPa

r/J' = 36.6 to 45 0

c' = 0 to 4.1 kPa


10.7 m
Medium to Coarse Sand
permanent water table at 10.7 m

N = 6 - 32
Shale at 27 - 30 m

! CD direct shear test results:

c' = 0 kPA, r/J' = 34 to 38 0

Figure 1 - Typical Soil Strata Profile

Furthermore, the construction of any bridge


replacement structure was complicated by the existence of a
warehouse which had been constructed sometime in the 1960's
or 1970' s at a distance of 6 m (20 ft.) from the proposed
bridge. Figure 2 shows the distance between the constructed
br idge and the warehouse. The warehouse was founded on
inter-piles tipping out in the medium to coarse sands
underlying the silts. The piles were loaded to 60 to 70 tons
per pile and were connected by structural grade beams in
conjunction with a structural mat foundation used for the
warehouse floor slab. Provisions had been made for the
future slab-jacking of the floor slab since deleterious
settlements were still expected to occur due to consolidation
of the silt zone coupled with the presence of a demolition
dump from the 1951 100-year flood which underlaid part of the
warehouse. The owner of the warehouse was extremely adverse
to any construction activity adjacent to his structure and
demonstrated his litigious nature early in the project
development stage.
The construction of stone column supported structures
in close proximity to the warehouse could conceivably have
initiated the settlements the warehouse owner was concerned

403 Brennan
about.
The second means of constructing the MSE structures
would have entailed staged conDtruction techniques coupled

Figure 2- Proximity of Warehouse and Bridge


with the use of wick drains. This method would also have
induced downdrag effects on the existing bridge structure and
possibly initiated consolidation of the soils underlying the
warehouse.
Isolating the warehouse structure while utilizing the
stone columns or wick drains was next considered. This
isolation was deemed appropriate since all attempts were to
be made to keep from de-watering the soils under the
warehouse and thus inducing settlement. The isolation
methods studied included the use of slurry trenches or
construction of a continuous wall of auger-cast piles around
the warehouse. These strategies would either result in
unacceptable pollution levels or were not economical. The
economics were impractical since a barrier system would have
to be constructed to allow foundation improvements which in
turn were necessary to construct an alternative to a bridge.
In a meeting held between the consultant, KsDOT Bridge
Design, and KsDOT Geotechnical Unit personnel; it was decided
to span the entire bridge length which was originally spanned
in 1938. This decision was due to the difficulties which
were being encountered in the soils immediately north of the
river channel and the 61 m (200 ft.). length-. of alignment
which crossed the demolition dump.
Bridge Foundation Alternates
The Kansas Department of Transportation s policy for
I

bridge foundations is to utilize rock-socketed drilled


shafts, spread footings in rock, or piling driven to refusal
either in end-bearing or side-friction. The use of either
rock-socketed drilled- shafts or spread footings in rock
proved impractical in the area of concern due to the

404 Brennan
excessive depth to bedrock. This left driven piling as the
acceptable foundation alternate.
Concerns were immediately raised as to the effect the
pile driving would have on the adjacent warehouse. The
determination was made that the likelihood of the medium to
coarse sands liquefying under the warehouse due to the
driving stresses imposed on the bridge piling was not of
critical concern. This position was taken since Rollins and
Seed (1990) indicate buildings have a positive effect upon
preventing liquefaction plus the largely successful history
the KsDOT possessed in driving piles in the Kansas River
valley in the vicinity of existing structures.
The development of excess pore pressures in the silt
stratum which underlaid the bridge site and the warehouse due
to pile driving and the subsequent collapse of the soil
structure was deemed to be of more pressing urgency. Using
the work of D Appolonia and Lambe as related by Poulos and
I

Davis (1980), Table 1 was derived using test results from the
in-situ vane shear, triaxial shear, and consolidation testing
and the equations as shown below:
TABLE 1 - EXCESS PORE PRESSURE SAFETY FACTORS

Area Depth m (ft. ) Safety


Factor
SW warehouse corner 4.1 m (13.5 ft. ) 0.83
SW warehouse corner 6.1 m (20.0 ft. ) 1.14
SW warehouse corner 9.1 m (30.0 ft. ) 0.85
NW warehouse corner 4.1 m (13.5 ft. ) 0.85
NW warehouse corner 6.1 m (20.0 ft. ) 1.16
NW warehouse corner 9.1 m (30.0 ft.) 0.86

~I ~l - Ko) + 2, s~ Af
0- YO crvo
~u = ~ um/(~)2
where maximum excess pore pressure
in-situ coefficient of earth
pressure at rest
Su = undrained shear strength
Af pore-pressure coefficient A at
I failure
o-vo = initial vertical effective stress
in soil

~u excess pore pressure


r = distance from the pile
R 8 times the radius of the pile for
this installation
These results and the work of Lambe and Whiteman
(1969) indicated the warehouse foundation would have to be
isolated from the bridge construction if driven piling were
to be used in the vicinity of the warehouse. Again, a slurry
trench method of isolation was proposed and discarded due to
environmental concerns over the displacement of the slurry
during backfilling operations and the applicability of the
use of a fluid as a damping medium to prevent driving-induced
ground settlement.
An auger-cast pile cut-off wall wrapping around the
west side of the warehouse was again considered to isolate

Brennan
405
the structures. The possibility of sand zones underlying the
warehouse ravelling into the auger holes plus the lack of
certainty of construction a continuous cut-off wall led this
alternative to be discarded.
The third isolation option explored the pre-drilling
of driven piles through the silt to the depth of the sand.
To ensure the bore hole would remain open, a slurry was
considered vital for this option. The necessity of using a
slurry invalidated this option again due to concerns over the
use of a fluid as a damping medium. Constructability
concerns also argued against this approach.
The decision was made to use unconventional foundation
elements through the critical areas adjacent to the
warehouse. The use of auger-cast piles with their capacities
verified by load tests was determined to be the optimum
solution. This position was taken since the auger-cast piles
avoid the construction difficulties with drilled shafts and
the detrimental effects of driven piling. Auger-cast piles
are also extremely economical and their use avoids the
complexities foreseen with the various foundation isolation
techniques previously discussed.
Auger-cast Piles
Auger-cast piles are formed by advancing hollow rotary
augers into the ground, and then pumping grout into the
boring via the hollow augers. After a sufficient head of
grout is established, the augers are withdrawn while
continuing to pump grout. The grouted pile can be reinforced
by inserting reinforcing steel into the grout after the
augers are withdrawn. Since the proximity of the warehouse
to the bridge foundation raised concerns about the ravelling
of the sand soils into the auger boring from under the
warehouse, the decision was made early in the design process
to utilize 305 nun (12 inch) diameter piles to minimize the
probability of ravelling soils.
The top 9.14 m (30 ft.) of the soil profile was
discarded in the design process as providing any frictional
support to the pile. This philosophy was adopted due to the
poor engineering characteristics of the soft silt found in
this profile coupled with the desire to pursue a conservative
approach with this type of piling. Direct shear and standard
penetration tests of the medium to coarse sand which served
as the bearing material for the piling exhibited a friction
angle of between 34 and 38 degrees. For design purposes, the
friction angle of the bearing material was taken as 30
degrees further assuring a conservative design.
At the time of the design of this bridge, auger-cast
piles were only designed to develop their bearing capacity in
frictional resistance. Since that time, development of
end-bearing capacity has been accepted by the auger-cast
industry (Berry, 1994.)
The frictional resistance to be developed by the
auger-cast piles was computed using Nordlund's Method as
related by Cheney and Chassie (1982) and by the Meyerhoff
Method as related in Bowles (1982.) Methods used to
calculate the frictional resistance for drilled shafts were
also considered. The use of these drilled shaft analysis
methods were discarded since the construction techniques used
for auger-cast piles renders the drilled shaft analysis
methods inappropriate (Drilled Shafts: Construction
Procedures and Des ign Methods.) The method adopted for
design was Nordlund's Method. This method uses the following

406 Brennan
equation to compute skin resistance when in soils of the same
effective weight and friction angle:

Qs = K
d CF Pd sinS Cd 0
where
Qs = the capacity of pile segment 0
(skin friction)
= dimensionless factor relating
normal and shear stress
correction factor for Kf when
r 1 ~ (soil friction angle)
= effective overburden pressure at
the center of depth increment d
friction angle on the surface of
sliding
Cd pile perimeter
o segment length
The correction factors used in this method can be
found in the same reference as above.
For 305 nun (12 inch) diameter auger-cast piles, the
practical bearing capacity limit is 889.6 kN (200k.) The
approach taken to compute required penetration of the medium
to coarse sand then becomes application of Nordlund's Method
to each discrete pile segment of 1.524 m (5 ft.) length until
the cumulative frictional resistance totaled 889.6 kN (200
k.) To compute the maximum allowable axial compressive load
based upon structural capacity, the allowable compressive
stress was applied to a cross-sectional area having a
diameter 50 nun (2 inches) less than the nominal diameter of
the pile as reconunended by the FHWA (Hamilton 1990.)
Finally, a safety factor of 3.6 was used to reduce the
ultimate loads to allowable. This resulted in individual
pile capacities of 244.7 kN (55k) with the safety factors
derived from partial safety factors as again provided by the
FHWA (Hamilton 1990.) These low pile capacities resulted in
groups of from 6 to 16 piles required to carry the foundation
load of the bridge. The lengths of the production piles
varied from 14.33 to 18 m (47 to 59 ft.) with the shorter
lengths placed in the more forgiving soil profile areas. The
pile tip elevations were set since the overlying silt strata
was ignored and the differences in length were derived from
the differences in ground surface elevation.
A reinforcing steel cage was required for the upper
one-third of the piling to meet seismic requirements. This
reinforcing steel proved difficult to impossible to place for
some piling and the pile had to be re-drilled.
To verify the capacities of the piles, a limited load
testing program was conducted which was designed to verify
the capacity of the piles with the use of ASTM 0 1143-81
using the loading schedules contained in Section 5.6 of the
referenced standard. These load tests were conducted on
piles which were installed using the same criteria which was
developed for the production piles. This step was taken to
familiarize the construction inspectors with the auger-cast
pile installation procedures and to acquaint all site
personnel with the problems which would be encountered.
The load tests were analyzed using techniques as put
forward in Cheney and Chassie (1982), and Butler and Hoy
(1976.) With the maximum load applied during the load tests
either 889.6 or 1067.6 kN (100 or 120 tons,) no plunging
failure load was ever attained. However, the results of the

407 Brennan
load tests did show the design loading was well within the
safe static load range of the piling. Figure 3 shows the
load-settlement curve attained for the load testing of a 305
rom (12 inch) diameter auger-cast pile 12.8 m (42 feet) in
length. This particular pile penetrated the medium to coarse
sand a distance of 3.66 m (12 feet) before stopping.

Load (kN)
356 712 1068

....c
v
E
v
=::4mml----7----+-----'---+-~----_t--~~-
v
(f)
(/)
Cf)
o
'-
C)

v
g'
'- 8mm I---...,..----+--~~"""'=::::__+---------lr+----'---
~

Figure 3 - Load Test Results for 12.8 m Pile

Construction Provisions for Auger-cast Piles

The following construction provisions were developed


for the bridge replacement project which utilized auger-cast
piles. Only the sections which dealt with the actual
construction of the piles are presented. Suggested revisions
to the construction provisions are shown in parentheses.
MIXING AND PUMPING MORTAR:
Only approved mixing and pumping equipment shall be used in
preparing and handling the mortar. All materials shall be
measured by volume or weight as they are fed into the mixer.
All oil or other rust inhibitors shall be removed from mixing
drums and mortar pumps. The time of mixing shall be such as
to produce a homogeneous mortar, whether mixed on the site or
obtained from a ready mix plant. If there is a lapse in the
operation of pumping, the mortar shall be recycled through
the pump or through the mixer drum or agitator. The minimum
and maximum mixing times, as well as the maximum recycling
time, are dependent on the contractor's mix, and will be
determined during the installation of the piles of the pile
load test.

The cement base non-shrinkage mortar defined by ASTM C1107-89


shall consist of Portland cement, a special pozzolan, a

408 Brennan
grouting agent, sand and water so proportioned and mixed as
to produce a mortar capable of maintaining the solids in
suspension without appreciable water gain; which may be
pumped without difficulty and will penetrate and fill any
open voids in the adjacent soils.
The contractor shall submit a mix-design of the cement base
mortar to the Engineer for approval prior to use in this
work. The mix design shall include the following
information:
Test results on the Fine Aggregate showing their
compliance with the specifications.
Source of the Fine Aggregate.
Weights of all materials used for one cubic yard of
fresh mixed mortar.
Brand name and type of the Portland Cement, brand name
of the grouting agent (water reducer and retarder), and
source and type of flyash (pozzolan).
Compressive strengths of test specimens made and cured
in accordance with ASTM C 192 and tested in accordance with
ASTM C 39. The materials shall be proportioned to produce a
hardened mortar with an ultimate compressive strength of
4,000 psi minimum at 28 days.

A sufficient quantity of the materials proposed for


use shall be submitted far enough in advance of use so that
the Engneer may conduct applicable tests.
The mortar flow as determined by ASTM C 939-87
(modified to a 3/4 inch cone opening) shall not be less than
17 seconds (nor more than 25 seconds). The flow of each load
shall be tested and recorded for process control. The
contractor shall provide the specified flow cone for project
inspection use.
The minimum and maximum allowable grout temperatures at time
of placement shall not be less than 500 F. and not more than
900 F.
Pressure gauges, in good condition, shall be located on the
grout pump and at the auger rig, so that the grouting
pressure may be checked by the operator and Engineer's
representative. A mechanical counter shall be used on all
pumps to monitor the quantity of grout placed. The volume of
grout displacement per piston stroke shall be verified by the
Engineer prior to grout placement.
Failure of the mixed mortar to meet compressive strength
requirements of Secion 3.3.5 will be considered grounds for
rejection of the pile. At the discretion of the Engineer, a
replacement pile shall be placed at a location determined by
the Engineer. The rejected pile will be left in place and no
payment shall be made for that pile. Additional work or
materials required to incorporate into the structure any
piles which were placed as replacements for rejected piles
shall be provided by the contractor at no additional cost to
the State. The proposed method of construction shall be
submitted to the Engineer for approval before construction of

409 Brennan
the replacement piles.
STRENGTH: During the progress of the job, standard
compression test cylinders shall be made and tested by the
Kansas Department of Transportation. A minimum of one (1)
set of three (3) cylinders shall be made for each day's work.
From each set of three (3) cylinders, one (1) shall be tested
at 7 days, one (1) at 28 days, and one (1) as directed by the
Engineer.

RECORDS:
Before Commencing Work: prior to commencing work, the
Contractor shall submit to the Engineer and obtain approval
for the following:

Sketch and/or description of the pile drilling


equipment to be utilized.
Complete description of method of installation.
Concrete mix design including preliminary mixing and
recycling times.
The proposed method for calibrating the volume of
grout displaced per piston stroke.
A dimensioned sketch of the propsed test loading
arrangement, and data on testing and measuring equipment,
including jack and gauge calibration.
During Course of Work: During the course of the work, the
Engineer shall record the following:

Load test reports, if applicable, including all test


data, and a graph of load versus settlement.

A daily pile report showing the pile number and


location, date placed, length of pile, final tip elevation
and log of boring. The daily pile report will also show
quantity of grout, reinforcing steel, mixing times, delivery
times, and unusual occurrences for each pile.

Mortar flow test results.


LOAD TESTS:
Two pile load tests shall be performed by the Contractor with
monitoring and evaluation performed by Kansas Department of
Transportation personnel. The cost of the anchor piles and
all equipment necessary to perform the pile load test shall
be paid as "Load Test". All load testing shall be performed
in accordance with ASTM D 1143-81. Loading procedures shall
be in accordance with Section 5.6 of ASTM D 1143-81 with a
loading apparatus capacity of 200 kips (400 kips). (The
loads shall be applied in increments of 10 percent of the
design load with each load held for 2-1/2 minutes. The piles
shall be loaded to failure.) The auger-cast reaction piles
will also be instrumented to determine uplift capacities
concurrently with the axial load test. (The test pile will
be instrumented with 4 dial gauges.) The test piles (and
reaction piles) shall be reinforced similar to the production
piles (and be constructed to a similar length) and shall not

410 Brennan
be loaded until the mortar has attained a minimum strength of
4,000 pounds per square inch (but not before 7 days after
pile installation.)
The test pile lengths shall be established by the Engineer,
in order to predetermine the optimum and most economical
lengths to be used.
All equipment, including load frames, jacks, dial gauges,
plates, reference beams, and all other equipment, tools, and
incidentals necessary to complete the work, will be furnished
by the contractor.
The contractor will submit for approval a dimensioned sketch
of the proposed loading arrangement, and data on testing and
measuring equipment including jack and gauge calibrations,
prior to commencing work.
CONSTRUCTION REQUIRMENTS:
The piling shall be the diameter and length as shown on the
drawings or as revised by the Engineer after evaluation of
test piles. All finished piles shall be in the location as
shown on the Plans and be within a tolerance of + 3 inches
from the center of the pile.
A continuous flight hollow shaft auger shall be drilled into
the underlying soil, to the required depth. Leaving the
auger in the hole and slowly rotating it clockwise, a cement
base non-shrinking mortar shall then be injected, under
pressure, through the hollow shaft as the auger is slowly
withdrawn. A head of mortar at least seven (7) feet above
the point of injection shall be maintained at all times
during the pumping process to remove all loose material and
retain the shape of the augered hole. The auger shall be
used to retain the shape of the hole. Since the pile may be
placed below the water table, under hydrostatic pressure,
extreme care must be exercised to prevent the lateral
pressure of both soil and water from "pinching in" and
reducing the pile diameter.
The auger shall be carefully withdrawn to preclude the
possibility of earth or mud caving into the hole. If the
auger is raised by a sudden jerk for any appreciable
distance, the hole shall be redrilled and the grouting
operation restarted.
Auger flighting shall be continuous from the auger head to
the top of the auger with no gaps or other breaks.
Augers over 40 feet in length shall contain a middle guide.
The leads must be prevented from rotating by an approved
means.
A pile shall not be installed within 4 hours of the
installation of an adjacent pile, if the adjacent pile is
within 5 feet, to preclude the possibility of the hydrostatic
head causing the mortar to break through to the hole being
drilled. The 4 hour time limit may be revised by the
Engineer, based on the set time of the mortar used in the
test piles.

411 Brennan
In the event non-augerable material is encountered, the
obstruction shall be removed and the pile completed or
another pile shall be placed in a location as directed by the
Engineer. Non-augerable material is defined as material
which causes the rate of penetration to be reduced to less
than one foot per minute, assuming an applied torque of
10,000 ft-lbs. The lineal footage of any piles which
encounter non-augerable material above the specified tip
elevation, plus the lineal footage of any replacement pile,
will be paid for at the contract unit price bid per lineal
foot for "Pressure Grouted Piles."
Pressure Grouted Piles shall be constructed to the elevations
shown on the Plans. Top of piles shall be float finished and
level.

Mortar shall be injected through the auger shaft at a


pressure between 120 and 240 psi. Mortar shall be placed in
a continuous operation from bottom to top of pile. The
minimum volume of mortar pumped into the pile will at least
equal 115 percent of the theoretical volume below elevation
720 and at least equal the theoretical volume of the augered
hole above elevation 720. The volume of mortar pumped will
be checked in 5 foot increments.
The Contractor shall coordinate the performance of the load
tests with the Engineer. The tests may be performed, with
the Engineer' s approval, prior to the general excavation,
provided the test pile is free from the top down to the
cutoff point.
Tests must be performed sequentially. The results of
previous tests shall be used to establish the depth and test
loading for the following test.
The Engineer shall make a thorough analysis of the test
results and determine the most feasible length required for
the conditions encountered.
The construction methods developed during the test pile
program shall be used for the production piles.
The reinforcing steel cage shall be completely assembled
prior to placement. The reinforcing steel cage shall be
placed following extraction of the auger and while the mortar
is still fluid. suitable centralizers shall be utilized to
insure that the specified cover is maintained.
No later than one month prior to construction pressure
grouted piles, the Contractor shall submit an installation
plan for review by the Engineer. This plan shall provide the
following information as a minimum:
Evidence of successful installation of auger-cast
piles under similar job and subsurface conditions, including
a job superintendent on site with a minimum of five years of
method specific experience.
List of proposed equipment to be used.
Details of mortar pumping and reinforcing steel
placement methods.

412 Brennan
Conclusions
The auger-cast piling industry has made several
advancements since this project was designed and constructed.
The use of an end-bearing component for bearing capacity
determination is now routine. Applied research is being
pursued to validate the use of auger-cast piles for uplift
resistance. Large diameter piles - 559 and 610 rom (22 and 24
inch) - have now become the largest standard auger-cast piles
available, and the development of a 762 rom (30 inch) diameter
pile is under way.
The experience of the Kansas Department of
Transportation with auger-cast piles has lead our agency to
accept these foundation elements for use in special
circumstances. Until a uniform method of analyzing
auger-cast piles is codified, the Kansas Department of
Transportation will continue to analyze these systems with
Nordlund's Method with mandatory load tests to confirm the
design.

Appendix I. References
Berry Jr., W., (1994). Personal Communication.
Bowles, J.E. (1982). Foundation Analysis and Design, 3rd
Edition. McGraw-Hill Book Company, New York, New
York.
Butler, H. D., and Hoy, H. E. (1976). The Texas Quick-Load
Method for Foundation Load Testing, FHWA-IP-77-B.
Federal Highway Administration, Washington, D. C.
Cheney, R. 5., and Chassie, R. G. (1982). Soils and
Foundations Workshop Manual. Federal Highway
Administration, Washington, D.C.
Drilled Shafts: Construction Procedures and Design Methods
(1988). Federal Highway Administration Publication
No. FHWA-HI-88-042. Federal Highway Administration,
Washington, D.C.
Hamilton, A. (1990). FHWA Region 7 Structural Engineer,
Personal Communication.

Lambe, W. T., and Whitman, R. V. (1976). Soil Mechanics.


John Wiley and Sons, New York, New York.
Poulos, H.G., and Davis, E.H. (1980). Pile Foundation
Analysis and Design. John Wi ley and Sons, New York,
New York.
Rollins, K. M., and Seed, H. B. (1990). "Influence of
Buildings on Potential Liquefaction Damage." J.
Geotech. Engrg., ASCE, 116(2), 165-187.

Appendix II. Conversion Factors


1 psi = 6.89476 kPa
1 ft.-lb. 1. 35582 Nm
1 ft. = 0.3048 m
1 inch 25.4 rom

413 Brennan
1 kip = 4.44822 kN
1 ton 8.89644 kN
10 F = (Degrees F + 459.67)/1.8

414 Brennan
Limit States Design for Deep Foundations
Bengt H. Fellenius, M.ASCE*

ABSTRACT
New foundation design codes have recently been proposed in Canada, USA and
Europe. The new codes are based on the Limit States Design, as opposed to the
conventional Working Stress Design. The 1983 Bridge Foundation Code published
by the Ministry of Transportation Ontario applied the Danish partial factor of safety
approach with fixed reduction of cohesion and friction. The new 1991 Code has
abandoned this approach in favor of applying a resistance factor to the ultimate
resistance of the foundation, differentiating the factor according to the method used
to determine the resistance. The paper summarizes the load and resistance factors
recommended by the new Code and indicates other new important aspects. An
example is presented on a pile group design according to the new Code.

INTRODUCTION
Several countries and regions are currently preparing for a forthcoming shift of the
foundation design approach from the proven "working stress design", WSD to a
Limit States Design, LSD. New limit states codes have recently been proposed in
Canada, USA and Europe. The Canadian efforts are contained in the 1991 Bridge
Design Code, which will be published in 1994 by the Ministry of Transportation and
Communication, Ontario, MTO. A further development is under way by the
Canadian Standards Association, CSA. The US development is led by the Federal
Highway Administration, FHWA, and a report has been published by Barker et al.
(1991).

The European Community, EC, has a committee' working on a limit states


foundation code to be applied to all countries of the European Community. A draft
has been published (Eurocode, 1990).

The working stress approach to geotechnical design, typically slope stability


problems, consists of establishing the soil strength and determining the allowable
shear by dividing the strength with a factor of safety-"global factor of safety
approach". In foundation design, the ultimate resistance (the capacity) of the
foundation unit is determined and the allowable load is obtained by dividing the

University of Ottawa and Anna Geodynamics Inc., 735 Ludgate Court, Ottawa, Canada, KlJ 8K8
I Chaired by Niels Krebs Ovesen, Danish Geotechnical Institute, Lyngby, Denmark.

415 B. H. Fellenius
capacity with a factor of safety. The particular value of the factor of safety to apply
depends on the type of foundation problem as guided by experience and ranges from
lows of about 1.3 applied to problems of slope stability of embankments to highs of
3 and 4 applied to bearing capacity equations, with values of about 2 applied to a
capacity determined in a loading test. Notice, the capacity expressed by the bearing
capacity equation does not just depend on the soil strength (cohesion and friction),
other aspects are also included in the equation. Moreover, soil strength is a function
of soil friction, tan ~', whereas the bearing capacity factors are far from linear
functions of tan ~'. Therefore, a factor of safety of 4 on the bearing capacity
calculated by means of the equation implies a factor of safety on soil shear strength
that is from about a third to half as large.

Often, the global factor is adjusted according to the type of load--dead or live,
common or exceptional-, but practice has developed toward letting those
distinctions be taken care of by applying coefficients to the load values. From this
basis, starting in Europe some years ago, a full "partial factor of safety approach" has
grown, in which each component, load as well as resistance, is assigned its own
uncertainty and importance. The design requirement is that the sum of factored loads
must not exceed the sum of factored resistances.

The partial factor of safety approach combines load factors, which increase the
values of the various loads on a structure and its components, with resistance factors,
which reduce the ultimate resistance or strength of the resisting components. This
design approach is called Ultimate Limit States, ULS.

Deformation of the structure and its components is determined in an unfactored


analysis (all factors are equal to unity) and the resulting values are compared to what
reasonably can be accepted without impairing the use of the structure, that is, its
serviceability. This design approach is called Serviceability Limit States, SLS.
Taken together, the ULS and the SLS constitute a Limit States Design, LSD, or, as it
is also termed, a Load and Resistance Factor Design, LFRD.

Initially, geotechnical engineers were rather unwilling to consider changing to a ULS


design approach as it applies to soils and foundations. Ten years ago, however, in
1983, a committee2 formed by the Ontario Ministry of Transportation, MTO,
produced a limit states design code for foundations of bridges and substructures. The
1983 Code very closely adopted the Danish system of partial factors of safety, where
all factors are larger than or equal to unity (loads and other 'undesirable' effects are
multiplied and resistances and other 'beneficial' effects, are divided by the respective
factors). In the 1983 Canadian version, all factors were multipliers and the resistance
factors were smaller than unity. Because the load factors were essentially already
determined (the same values as applied to the superstructure were used), the code
committee was left with determining what values to assign to the resistance factors.
Notice the importance distinction that these factors are applied to the soil strength,
only, the other features of the design condition were not to be factored.

Soil strength in classical soil mechanics is governed by cohesion, c, and friction,


tan~. After some comparison calculations between the final design according to the
WSD and ULS approaches, a process known as 'calibration', the committee adopted
the reductions used in the Danish Code of applying resistance factors to cohesion and
friction of 0.5 and 0.8, respectively. However, the calibration calculations showed

2 Chaired by Geoffrey G. Meyerhof, Technical University of Nova Scotia, Halifax.

416 Bengt H. Fellenius


considerable differences in the design end product between the 'old' and the 'new'.
A 'fudge' factor was therefore developed called "resistance modification factor" to
improve the calibration agreement. The idea was that once a calibration was
established, the presumed benefits of the ULS approach as opposed to the WSD
approach would let the profession advance the state-of-the-art. Such advancement
was apparently not considered to be possible within the 'old' system. Details of the
LSD approach used in the MTO 1983 Code are presented in the Canadian
Foundation Engineering Manual (1985).

Very soon after implementation of the 1983 Code, the industry voiced considerable
criticism against the new approach, claiming that designs according to the WSD and
the ULS agreed poorly in many projects, in particular for more complicated design
situations, such as certain high retaining walls and large pile groups. It is the
author's impression that many in the industry, to overcome the difficulties, continued
to design the most common and simple cases according to the WSD method and,
then, resorting to a one-to-one calibration detennined what the ULS values should be
in the individual cases! Hardly a situation inspiring confidence in the new code.

DESIGN OF A STRIP FOOTING USING A FACTOR ON FRICTION


The root of the difficulty in establishing a transition from the WSD to the ULS lies in
the strict application of fixed values of the strength factors to fit all foundation cases,
ignoring the existing practice of adjusting the factor-of-safety to the specific type of
foundation problem and method of analysis. Consider, for example, the very simple
case of a strip footing placed at the ground surface on a cohesionless deposit. The
capacity of such a footing is usually analyzed by means of the bearing capacity
equation and, thus, in this case, governed by the Ny -bearing capacity coefficient.
The allowable stress on the footing, qa' is simply the calculated bearing capacity, ru ,
divided by the factor-of-safety, Fs '

In the parallel ULS approach (partial factor of safety approach), a factored resistance,
rf' is calculated using factored strength values (in this case, the factor, flp' is applied
to the frictional strength), and the stresses on the footing are increased by load
factors. If we assume that the stress can be factored by applying a single load factor,
fq , ~hen, the design condition is that the factored stress must not exceed the factored
reSIstance: fq q ~ rf.

In a simple case, the mentioned single (combined; average) load factor, fq, will be
about equal to 1.3. A strict calibration of the working stress and the ULS approaches
then results in that Fs = 1.3 r u Irf' Thus, resistance factor, flp becomes a function of
two variables: the friction angle, <p, and the global factor-of-safety, Fs ' Fig. 1
presents a diagram over the relation between the resistance factor on friction, flp , and
the factor-of-safety, F s ' for friction angles ranging from 20 0 through 45 0 The
diagram shows that for the case of a friction angle of 35 0, the 0.8 value of f required
in the 1983 Code results in a 'calibrated' factor of safety equal to about 2.6. Current
practice, however, is to require a factor of safety closer to 4.0. A repeated calculation
(not shown here) for a footing placed at some depth below the ground surface and/or
in a soil with also cohesive strength indicates an even larger spread of the curves.
This is because that latter calculations of the bearing capacity include also other
features than the soil shear strength, such as the overburden stress and unit weight of
the soil below the footing.

417 Bengt H. Fellenius


It is obvious that the to-all-cases-applicable-one-value-resistance-modification-factor
approach is not workable. For the same reason, neither is the partial-factor-of-safety
approach favored in the draft European code.

5 1.00 , - - - - - - - - - - - - - - - - - - - - ,
i=
u
fE 0.90
Z
o
0::: 0.80 r-~~--~""
o
I-
~
u.
0.70
w
U
~ 0.60
I-
en
U5
w 0.50 '------'----"-----'-----'----~-----'

0::: 1.00 2.00 3.00 4.00 5.00 6.00


FACTOR OF SAFETY

Fig. 1 Resistance Factor on Friction versus


Factor-of-Safety as a Function of Friction Angle.

FURTHER DEVELOPMENT OF THE MTO CODE


In 1988, the Ministry decided to revise the Code. A Foundation Code Committee 3
reviewed the experience thus far and came to the conclusion that the partial-factor-
of-safety approach with fixed values on cohesion and friction should be abandoned.
Of course, the Code could not be returned to the WSD approach, nor would this be
desirable. Instead, it was decided to apply resistance factors to the ultimate
resistance of a foundation rather than to the soil strength and to differentiate between
types of foundations and methods of determining the capacity of the foundation.
Equally important, the Code continued the significant development of advising on
the overall approach to the design of foundations. In 1992, it was decided that the
MTO Ontario code should be further developed into a national code on foundations
under the auspices of the Canadian Standards Association, CSA, which work is now
underway by a newly formed CSA committee4 .

The new bridge code puts forward many new aspects of importance going much
beyond the simple changing of the approach to the resistance factors. It is presented
in two separate documents: A Code with mandatory requirements and a
Commentary with explanations to the Code and with recommendations for design.
The following presents some of the details which pertain to the design of deep
foundations.

3 Chaired by Roger Green, University of Waterloo, Waterloo, Ontario.


4 Chaired by Francois A. Tavenas, McGill University, Montreal, Quebec.

418 Bengt H. Fellenius


LOAD AND RESISTANCE FACTORS IN THE MTO CODE
The Code specifies numerous loads and load factors, such as permanent (dead),
transient (live), and exceptional loads; making differences between loads due to
weight of building materials, earth pressure, earth fill, wind, earthquake, collision,
stream flow, etc., with consideration given to the effect of various load combinations,
providing minimum and maximum ranges for the load factors. For example, the load
factor applied to dead load from concrete is 1.20, from earth fill it is 1.25, and a
factor of 1.25 is applied to a dragload due to negative skin friction. Earth pressures
are assigned load factors of 1.25. The factors combine and it is not easy to come up
with an estimated average factor; the average value for a typical design appears to
hover around 1.25 on dead load and 1.40 on live load. The unit weights of the soil
backfill material, such as sandy soil, rock fill, and glacial till, are all given with
values that assume that they are fully saturated. Because most backfills are drained,
this is an assumption of the safe side.

The resistance factors imposed by the Code pertain to the type of foundation. Table 1
lists the factors that apply to axial resistance of deep foundations.

TABLE 1
Resistance factors in the MTO Bridge Design
Code as Applied to Deep Foundations
Compression Tension
Static Analysis 0.4 0.3
Static Test 0.6 0.4
Dynamic Analysis 0.4
Dynamic Test 0.5

The lower resistance factors for tension loading as opposed to compression loading is
due to a consideration of the more severe consequence of a failure in tension as
opposed to in compression. Dynamic Analysis refers to wave equation analysis.
Dynamic formulae are stated to be "fundamentally incorrect and their use is
discouraged". Dynamic Test refers to field measurements using the Pile Driving
Analyzer combined with wave equation analysis of the data.

< PARALLEL DEVELOPMENT IN THE USA


Independently of the MTO, a US Committee 5 working on a contract from the US
Federal Highway Administration, FHWA, developed a limit states design manual for
bridge foundations (Barker et aI., 1991) employing the same approach as that used by
the MTO second committee. (Because the Eurocode has stayed with the partial
factor-of-safety approach, there exists now a fundamental difference between the
Eurocode and the Canadian and US approaches). For piles, the recommended
resistance factors (called "performance factors") to apply to a static analysis are
separated on shaft and toe resistances, and on the method used for the analysis:
whether a-method, p-method, or, even, A-method is used for the analysis of the shaft
resistance, as well as what particular field investigation method used for the analysis.
The factors are summarized in Table 2.

5 Headed by J. Michael Duncan, Virginia Polytechnique and State University, Blacksburg, Virginia.

Bengt H. Fellenius
419
TABLE 2
FHWA Resistance factors as Applied to deep Foundations
Shaft Toe Combined
Compression
Static Analysis 0.50 - 0.70 0.35 - 0.50 0.45 - 0.55
Static Test 0.80
Dynamic Analysis
Dynamic Test 0.70
Tension
Static Analysis 0.35 - 0.60
Static Test 0.80

The FHWA document is somewhat imbalanced between the furnishing of the many
details on static analysis and indicating no factor on resistance determined by wave
equation analysis. The numerical values of the FHWA factors given in Table 2 are
larger than those in the MTO Code (Table 1). However, they are to be combined
with load factors that are about 10 % to 20 % larger than the MTO load factors and
the combined effects are about the same in both documents.

OTHER NOVEL ASPECTS OF THE MTO CODE


The MTO Code emphasizes serviceability limit states, SLS. In the SL.S design
calculations, all loads are unfactored, that is, the load factors are equal to unity. At
first glance, this appears to be the same approach as used in the working stress
design, WSD. However, several important novel requirements are imposed, as
follows: First, the structural design engineer must provide not only the values of the
loads at ULS and SLS to the geotechnical engineer, information on what settlement
(and/or horizontal deformation) that the structure can accept must also be provided.
In tum, the geotechnical engineer must supply the structural engineer with not only
the value of the calculated settlement for the loads, but also the load values that
would bring about the particular limit settlement or deformation. Obviously, this can
not be achieved by one simple cycle of two-way communication. Therefore, the
code mandates that "consultation between the structural engineer and the
geotechnical engineer shall take place during planning and design, including
construction stages" and "final values of deformations at SLS shall be developed by
the structural and geotechnical engineers". In other words, the Code desires and
directs a collaborative effort between the design professionals. The Code also
requires that they both be involved also in the construction phase of a project.

The Code spells out that the following aspects shall be considered at LSD, singularly
and in combination: overall stability of foundation and adjacent slopes; bearing,
sliding and structural resistance; groundwater table and seepage; frost penetration,
erosion, and scour; as well as effect on existing adjacent structures of the
construction of the new foundation.

The requirements on geotechnical investigation report are detailed and must include
procedures, details on geology, subsurface conditions, and groundwater table, that is,
the usual information. However, the Code also emphasizes that potential fluctuation
of the groundwater elevations must be established and it is clear that it is not

420 Bengt H. Fellenius


sufficient to just identify a water table, also the vertical and horizontal gradients must
be determined. Furthermore, the geotechnical report shall provide values of
geotechnical parameters and not just recommend the values, but also include a
discussion of the values; and it shall include discussion and recommendations
concerning the foundation design and construction with special consideration for
inspection during construction and it shall identify aspects influencing maintenance
relating to the performance of the structure. Indeed, a tall order, but one that is
intended to improve the quality of the geotechnical report.

CODE REQUIREMENTS FOR DEEP FOUNDATION DESIGN


The MTO Code requires that the design shall consider several aspects that are
common for both the WSD and the LSD, such as the "ease of installation"
(drivability for driven piles), potential bending during driving, driving stresses,
slippage on sloping bedrock, minimum spacing of piles, etc. The Code does not
allow any contribution to the capacity of a pile foundation from the contact stress
below the pile cap. As to static analysis, it recommends the effective stress analysis (
~-method), suggesting appropriate ranges of ~ and Nt coefficients, but leaves it to the
geotechnical engineer to choose the specific parameters and, for that matter, the static
analysis method to use. The Code recognizes that all piles are subjected to negative
skin friction and applies the unified method of pile group design, as follows.

The ultimate soil resistance is considered by applying a resistance factor to


the full capacity of the pile and relating the factored resistance to the sum of
the factored dead and live loads. Dragload is not to be included.

The structural strength of the pile is considered by determining the location


of the neutral plane, which is where the maximum permanent load occurs in
the pile. The location of the neutral plane it is to be determined using
unfactored loads and resistances. The factored maximum load in the pile
consists of the sum of the factored dead load and the factored dragload.
Live load is not included. It is stated that the load at the neutral plane is
rarely of any consequence for the design of piles having an aspect ratio
(diameter to embedment depth) smaller than 1:80.

The settlement of the pile group is determined considering that the pile and
the soil settle equally (no relative movement) at the neutral plane. The
Code recommends that the settlement be determined as the settlement for an
equivalent footing equal in area to the pile cap and placed at the location of
the neutral plane.

PILE GROUP EXAMPLE


The design approach is illustrated in the following example. A group of twenty-five
355 mm pipe piles are to be driven at equal spacing and in a square configuration at a
site where the soil profile consists of, in sequence, an upper 4 m thick layer of
normally consolidated sandy silt, 17 m of soft, compressible, slightly
overconsolidated clay, 6 m of medium dense silty sand, and a thick deposit of
medium dense to very dense sandy ablation glacial till. The groundwater table lies at
a depth of 1.0 m and the pore pressure in the sand and till is hydrostatically
distributed but artesian with a phreatic height above ground of 5 m. An earth fill will
be placed symmetrically around the pile group (not inside the group) to a height of
1.5 m over a square area with a 36 m side. The pile cap is 9.0 m square and placed

421 Bengt H. Fellenius


level with the ground surface. Each pile will be subjected to unfactored dead and
live loads of 800 KN and 200 KN, respectively. The soils investigation has
established a range of values of the soil parameters necessary for the calculations,
such as density, compressibility, consolidation coefficient, as well as a range of
parameters for the effective stress calculations of load transfer (P and NV. The
structural engineer states that the limit deformation at SLS is 40 rnm and that the
load factors on the dead and live loads are 1.25 and 1.50, respectively.

Although the example is only for purpose of illustration, it is quite realistic. For
example, the indication of artesian pressure is in recognition of that it is the rare site
that has hydrostatically distributed pore pressure all through the profile. Actually,
more often than not, real cases involve considerably more variations than given in
the example. Yet, even for a straight forward case such as this, to be able to perform
the design calculations (usually by hand), the geotechnical engineer often further
simplifies the case to only relate to the service conditions. However, the new Code
requires that also the installation conditions be addressed.

The design must include several steps in approximately the following order.

Determine the range of installation length using the given range of effective stress
parameters as based on the at-least capacity: The calculated factored resistance must
be at least equal to the sum of the factored loads determined from an analysis of the
structure, that is, 1.25-800 + 1.5-200 = 1,300 KN. Therefore, with a resistance
factor of 0.4 on static analysis, the ultimate resistance (the capacity) of the pile
must be at least 1,30070.4 = 3,250 KN. Let us assume that the calculation results
show that, to obtain this capacity, the piles have to be installed to a penetration into
the sand layer of 2 m (upper boundary of P and Nt) through 6 m (lower boundary of
p and NV, i. e., to depths ranging from 29 m through 33 m, as shown in Fig. 2.
Fig. 2 shows a load-transfer diagram over the calculated resistance distributions
along the pile for the installation depths for achieving a capacity of 3,250 KN, as
determined for the boundaries of p-ratios and Nt-coefficients. The diagram also
shows the load distribution during service conditions and the location of the neutral
plane. Notice, the boundaries of p and Nt resulting in the range of installation depth
are not the extreme boundaries but the reasonably expected values.

Often overlooked in similar design calculations is that the uncertainty in the design
can work both ways. The designer may choose to specify the safe, deepest
embedment, for the purpose of ensuring the minimum capacity. However, if the
upper boundaries of p and Nt are those that prevail, the actual capacity at this depth
can be much, much higher, in the example case no less than 4,400 KN rather than
3,250 KN, which can cause, and has caused in actual cases, considerable problems
for the pile driving contractor. Normally, nature takes a forgiving approach, though,
because equally often overlooked is that the surcharge fills may not have been placed
at the time of the driving, which reduces the soil resistance, and, more importantly,
the driving induces large pore pressures in the soil. Therefore, the pile is driven
against a soil resistance that is much, much smaller than the resistance during the
service condition.

To analyze the conditions during the construction phase requires a drivability


analysis. The static analysis performed according to the preceding step is repeated,
but with large pore pressures assigned to the clay layer and also some in the sand and
till. The influence of the fill and embankment is excluded (assuming that neither
influence is present when the piles are driven). This analysis establishes the range of

422 Bengt H. Fellenius


ultimate static resistance to overcome during the driving. In the example case, the
static resistance at End-of-Initial-Driving, EOID, ranges from about 1,800 KN
through 2,100 KN. Restoring the pore pressures and repeating the calculation,
establishes the resistance that acts during a restrike after full set-up. These
calculation results are then used as input to a wave equation analysis (Goble et al.
1980), usually GRLWEAP (GRL 1993), to determine the basis for hammer selection,
the driving stresses (not too important in the example case, but would have been very
important had the pile been a concrete pile instead of a steel pile), and the bearing
graph. The GRLWEAP analysis (drivability option) also provides important
information on the expected number of blows and time required for the driving.

RESISTANCE AND LOAD (KN)


o 1000 2000 3000 4000

10

-----
E

:r::
~
CL
w 20
Cl
" "

. -:-
". "-
" I.
30 RESISTANCE CURVES a
Q, - Q. - fA.. pa'.dz I.
= Q 11.

Fig. 2 Load-transfer Curves and Bearing Capacity


for Embedment Depths of29 m and 33 m

Fig. 3 shows the bearing graph determined by the wave equation analysis. The graph
is in the form of a band consisting of envelops to the bearing curves that result when
considering the boundaries of the input parameters, such as hammer efficiency,
quake values, and damping factors. The illustrated range may seem narrow, but it is
not.

423 Bengt H. Fellenius


4000 .........- - - - - - - - - - - - - - ,

3000

Z
~
2000
~
....J
:J
I
c:
1000

o 10 20 30 40
PRES (Blows/25 mm)

Fig. 3 Bearing Graph (GRLWEAP Analysis)

One purpose of the bearing graph can be to indicate at what range of penetration
resistance (blow count) the at-least capacity can be expected. That the MTO Code
assigns the same resistance factor to both static and dynamic analysis is a good help.
Notice, it is not the intention of the Code that the designer should use either static or
dynamic analysis, the two must be combined. Furthermore, during the inspection of
the pile installation; the actually observed penetration resistance values are compared
to the calculations and assist in determining when to terminate the pile driving.

As evidenced by the bearing graph, the wave equation analysis on the example case
indicates that the hammer can not drive the pile against the at-least capacity of.
3,250 KN. However, this does not mean that the hammer can not drive the pile so
that it has a service condition capacity of 3,250 KN. The soil profile and the
'presumption' of excess pore pressures developing during driving suggest that there
will be a set-up of capacity making up for the difference between the EOID and the
resistance at a restrike after dissipation of the pore pressures. (Frequently, though,
soil set-up does not occur and its presence must always be proven before it is relied
on).

As mentioned, the soil resistance at End-of-Initial-Driving, EOID, is only about


1,800 KN through 2,100 KN, which suggests that the EOID capacity can be reached
at a mere 4 blows!25 mm. On the other hand, it may not occur until the penetration
resistance is 16 blows! 25 mm. Most of the time, the designer plays it safe and
specifies the termination criterion to be the higher value. Provided that this value is
not reached gradually during a prolonged driving, which can cost much in terms of
time, as well as wear and tear on equipment, this is acceptable.

Bengt H. Fellenius
424
In this context, the designer engineer should consider that the piling contractors size
their equipment according to the capacity that may be established by a field test,
whether or not the design includes such a test. That is, a piling contractor approaches
the project with a factor of safety in mind that may be very different to the factor
considered by the design engineer. In this difference in approach lies a seed of
conflict.

Notice, the Code could be interpreted to mean that the full array of resistance factors
shown in Table 1 should be applied to a project depending on the methods and
techniques applied in the design. This interpretation would be incorrect, however,
and, if implemented, it could be the cause of a serious dispute with the contractor. It
is, therefore, very important that the same factor is applied to all types of analysis
and tests of a project. That is, the smallest resistance factor governs the design. The
individual uncertainties associated with the methods used are then handled by
determining boundaries of applicable ranges.

The size of the example pile group (length, size, and number of piles, including the
pile cap represents an investment well over $100,000.00. It would be foolish and
wasteful to anchor the design to an at-least capacity of 3,250 KN. A project of this
size (and most smaller ones, also, for that matter) should include field testing for
capacity and proper functioning of the installation equipment. If the analyses are
supplemented with dynamic measurements during the outset of the pile driving by
means of the Pile Driving Analyzer, the Code requires that a factored resistance of
1,300 -:- 0.5 = 2,600 KN, a value that the PDA testing probably could establish.
Therefore, in the example case, if during the design phase it is already known that a
field test will be carried out at the site, the factor to apply on the results of that field
test will govern. Thus, if the Pile Driving Analyzer will be used, the static and the
wave equation analyses, as well as the range of embedment depth should be geared
to that the at-least capacity is 2,600 KN, not 3,250 KN. Moreover, if a static loading
test, a proof test, is carried out toward the conclusion of the pile driving contract, the
at-least capacity reduces to 2,170 KN, which is 1,300 divided by the resistance
factor of 0.6).

The final design step is the SLS analysis--the settlement analysis. The static
analyses show that the neutral plane will be located in the ablation till. Calculations
of settlement can be by means of a sophisticated finite element method or simplified
methods, provided that the fact is recognized that the pile loads do not enter the soil
above the neutral plane. In this case, calculations for an equivalent footing placed at
the neutral plane, as suggested by the Code, show that the pile group will settle less
than the limit value of 40 mrn, about 20 mm. The permanent pile load at SLS, that
is, where the full 40 mm occurs, is more than about twice the load indicated
(800 KN). As the neutral plane is located in the competent till layer, it makes little
different whether the equivalent footing is located at the neutral plane or at the toe of
the piles.

This brief expose of the new code is not exhaustive, but has aimed to present a few
of the interesting aspects of the code, giving some facts as well as illustrating the
principles which have guided the code development. The new code is a modem code
inasmuch that full compliance will not be possible using conventional hand
calculations and judgment calls. In particular because of the need to analyze several
"what-if' situations, the new code will hasten the transfer to computer aided designs.
Therein lies a challenge and a danger, but a discussion on this topic lies outside the
scope of this article.

425 Bengt H. Fellenius


References

Barker, R. M., Duncan, 1. M., Rojiani, K. B., Ooi, P. S. K, Tan, C. K., and
Kim, S. G., 1991. Manual for the design of bridge foundations. National
Cooperative Highway Research Programme, Report 343, Transportation Research
Board, Washington, D. C., 308 p.

Canadian Geotechnical Society, 1985. Canadian Foundation Engineering Manual,


Edition 2, BiTech Publishers, Vancouver, British Columbia, 460 p.

Eurocode, 1990. Eurocode, Chapter 7, Pile Foundations (preliminary draft), 25 p.

Goble, G. G., Rausche, F., and Likins, G., 1980. The analysis of pile driving-a
state-of-the-art. Proceedings of the 1st International Seminar of the Application of
Stress-wave Theory to Piles, Stockholm, Edited by H. Bredenberg, A. A. Balkema
Publishers Rotterdam, pp. 131 - 161.

GRL, 1993. Background and Manual on GRLWEAP Wave equation analysis of pile
driving. Goble, Rausche, Likins Associates, Cleveland.

Ministry of Transportation Ontario, MTO, 1991. Ontario highway bridge design


code and commentary (two volumes). MTO, Quality and Standards Division,
Toronto, Ontario, 713 p.

426 Bengt H. Fellenius


SOIL RESISTANCE FACTORS FOR LRFD OF DRIVEN PILES

Jay Berger' and George G. Goble 2

Introduction

The use of Load and Resistance Factor Design (LRFD) for highway bridges
has been specified by the Bridge Committee of the American Association of
State Highway and Transportation Officials (AASHTO) and this design
philosophy will be implemented in the very near future. Foundations are
included and, of course, this includes driven piles. In the LRFD design
approach, the "Safety Factor" is split between factors on the loads, the Load
Factors, and factors on the strength, the Resistance Factor, or the ql-Factor.
The load factors have been generated for the various loads in selected load
combinations by structural engineers using probabilistic concepts. They
have also developed the necessary ql-Factors for the various structural
elements and failure modes and the results of research studies have been
extensively published and discussed in the structural design community. The
ql-Factors for foundation design have also been selected (Barker et ai, 1991),
but a similar discussion has not taken place, nor has the data selected for the
evaluation, or the methods used in determining the selected values, been
reported.

In this paper, the general method used in driven pile design will be
reviewed and ql-Factors for the various failure modes and construction
control procedures will be developed from available statistical data. These
results will be discussed in the light of practical experience and realistic
values will be recommended for use in design. Also, factors of safety will
be generated for use in working load design.

Background

The concepts used in LRFD were first brought into practice in the United
States in the design of reinforced concrete buildings (ACI 1956) in 1956 as
an optional tool and in 1963 as the specified method (ACI 1963). At that

, Branch Manager, Goble Rausche Likins and Associates, Inc. 5398


Manhattan Circle, Boulder, CO 80303
2 Principal, Goble Rausche Likins and Associates, Inc., 5398 Manhattan
Circle, Boulder, CO 80303

Berger/Goble
427
time, the method was known as "Ultimate Strength Design" and the
motivation for the change was primarily associated with problems associated
with the use of linear elastic analysis in reinforced concrete. The method
adopted was, from the standpoint of the designer, identical with what is now
known as LRFD and can be stated as

where
Vii is the load factor for the ith load condition and the jth load combination
Q ij is the load effect for the ith load condition and the jth load combination
<P k is the resistance factor for the kth failure mode
Rk is the nominal strength of the element in the kth failure mode

The load and resistance factors contained in the ACI Code design method
were selected based on experience method. Since two quantities, load and
resistance factors, were used to replace a single safety factor the selected
values had to be based on judgement coupled with extensive "calibration"
studies to make sure that the new design approach did not produce
substantially different results than the previously used method. After the
ACI Building Code was changed, research was completed that placed a
theoretical basis under the method (e.g. Cornell 1969) and a great deal of
additional research expanded the concept and caused it to be more widely
adopted.

The expanded research on LRFD was based in probabilistic methods and


this orderly approach was the basis for implementation of the method into
additional codes and practice. In a study by the National Bureau of
Standards, Load Factors were set for buildings (Ellingwood et al 1980) based
on load inventory work and further probabilistic research. The design of
steel structures was studied and recommendations for a methodology for
determining Resistance Factors were made. This was followed by the
adoption of LRFD by the American Institute of Steel Construction (1986).

Problems were encountered in the design of building foundations because


the footing, usually a reinforced concrete element, and superstructure were
designed by LRFD while the foundation was designed based on working
loads. This implied that two sets of loads were necessary, working loads for
the foundation design and factored loads for the structural design of the
footing. The need for two sets of loads caused problems and confusion
between the structural and geotechnical engineer.

Load factors have been developed for bridges based on a knowledge of the
traffic and environmental effects. The basis for the structural element
resistance factors for bridge design were available from previous research.

An approach to LRFD for driven piles was presented by Goble et al (1980),

428
based on a research study sponsored by the Federal Highway Administration
(Goble 1979). The approach p{oposed recognized the practical aspects of
driven pile foundation design and installation. A methodology was
developed and Q:l-Factors proposed based on experience and calibration
studies. In a later study, the necessary Q:l-Factors for a number of
geotechnical applications including pile design were proposed by Barker et
al (1991). After this effort a LRFD-based bridge design code, including deep
foundations, was accepted by the AASHTO Bridge Committee.

Design of Driven Piles

The design of driven pile foundations is a very unusual process in that the
final design of the foundation is usually not complete until the driving criteria
is established after pile driving is underway. Furthermore, dynamic methods
have been the dominant approach used to determine capacity. Thus, every
driven pile is effectively subjected to a quality control test in the process of
its installation. The design and installation process is shown in the attached
Flow Chart of Figure 1. In this chart, it has been assumed that the decision
to use a driven pile foundation was already made and all considerations
concerned with the appropriateness of that decision are outside the scope
of this discussion. At the beginning of the design process (Block 1) the
designer has available a knowledge of the site geology and some subsurface
investigation results. For pile design, the subsurface investigation will
almost always be soil borings with Standard Penetration Test (SPT) data.
The designers should also have a knowledge of local pile driving practice and
the availability of the various pile types. They must also have some
knowledge of the design loads. This will usually be information on the
column or pier loads, not the load on an individual pile.

The design process (Block 2) involves the selection of the pile type, size,
length, number, and arrangement. This effectively implies that the load is
"selected" by the designer since the number of piles is designer determined
and this will control the magnitude of load applied to individual piles. Group
load considerations must be evaluated by some analytical technique and it
is assumed here that this analysis has been made and a required single pile
strength determined. This approach represents the typical practice. The
design also involves the determination of the driving criteria including the
effects of strength change with time after driving. Thus, the designer
selects the pile type, size, and length such that it will carry the required load
and be driveable. If it is determined that blow counts will be excessive, or
driving stresses will be outside the range that is considered tolerable, a
different design must be used. The new design may select a different
hammer or even a different pile type, size, or length with different design
loads. When this phase is complete the designer will state or specify a
required driving resistance at the end of driving and also at the beginning of
restrike (if restrike testing is performed), in addition to the other design
parameters mentioned above. The process of making a successful design
may require several iterations through selections for the various parameters.

429
DRIVEN PILE DESIGN

CD CrvEN
SITE GEOLOGY. SUBSURFACE INVESTIGATION,
LOADS, LOCAL PRACTICE

(3) SELECT
PILE TYPE, SIZE, LENGTH, NUMBER, LAYOUT
REQUIRED DRIVING CRITERIA. RESTRIKE,
BLOW COUNT

OFFlCE
------------ ----------
FIELD

9 REDUCE
DRIVING CRITERIA >-_N_O ..., MODIF'f
DESIGN

NO

YES

NO
YES

L..- ----j
DRrvE PRODUCTION
PILES WITH QUALITY
CONTROL

Figure 1 Flow Chart of Pile Design Process

430
After the selection of a design and a construction contractor, the field
portion of the design process can begin. In some cases, a preliminary test
program may be undertaken prior to finalizing the design selection. Such an
approach can often produce substantial savings on a large job. A "Test Pile"
is driven (Block 3) and the capacity is evaluated (Block 4). The method used
to evaluate capacity can vary, ranging from the observation of blow count
to the use of static load tests and dynamic monitoring or some combination
of methods. The selection of the capacity evaluation procedures will depend
on the size of the job, the variability of the site conditions, and past
experience in the region. It is most unusual for no dynamic method to be
used in the practices of the United States and most of the rest of the world.
In some countries, dynamic formulas are still in use but this practice is
changing rapidly. If no field capacity determination procedure is used this
implies that the pile is driven to depth and the strength determination is
accomplished exclusively by a static capacity determination procedure based
only on the subsurface investigation information.

After completion of the capacity evaluation procedure, the question is


asked, "Is the capacity adequate (Block 5)7" If the answer is negative, some
aspect of the design must be changed (Block 6). The most common solution
is for the pile to be driven to a greater depth and the capacity evaluation
repeated .. If this approach is unsuccessful, more difficult changes must be
considered such as an increase in the number of piles, an increase in the pile
size, a change in the pile type, or even a complete redesign of the
foundation. One less obvious solution when difficulties of this type are
encountered is to reduce the factor of safety while at the same time
increasing the strictness of the quality control procedures. Often this is the
simplest and least expensive solution and the realities of the field situation
dictates that it is quite frequently used.

When the check for adequate capacity is satisfied it is then advisable to


ask the question: "Is the capacity excessive (Block7)?" The real question
here is: Can the contractor substantially reduce costs with a relaxed driving
criteria. If the answer is yes, a reduced driving criteria should be established
(Block 9) and verified. The production piles are then driven according to the
established criteria and quality control procedures (Block 10).

Pile Design Limit States

Before considering LRFD Resistance Factors the failure modes for driven
piles need to be discussed. Three failure modes must be evaluated. There
are two obvious conditions, structural failure of the pile and failure of the
pile-soil system (penetration of the pile into the soil). In the first case, the
pile structure fails. If the pile is fully embedded, then the failure will be
based on the ultimate strength of the material. For most cases, the loads
used in evaluating individual piles are assumed to be axially applied so the
pile load can be treated as uniform on the cross section. If bending loads are
applied, the evaluation of the strength is still well defined from traditional
structural design procedures. In some cases, piles will be driven with the

431
upper portion extended above the ground surface to a pile cap. A portion of
the pile will be unsupported and the available strength must be evaluated
based on column strength analysis procedures.

The other obvious failure mode is the pile-soil strength. As described


above there are several methods for evaluating soil failure and these methods
are linked to the construction methods employed. In addition, when a static
load test is performed, the failure load must be defined from the load test
curve since many such curves do not have a sharply defined failure point.
Therefore, the safety margin that is supplied by a pile foundation is partially
defined by the failure load definition. A failure load definition such as the
Davisson criteria (Davisson 1972) will give a much lower failure load than a
procedure such as the Chin method (Chin 1970). Therefore, if the strength
is defined by the Davisson criteria (as was the case in this study) a pile is
credited with a lower capacity and a larger effective margin of safety results,
unless the q>-Factor is dependent on the failure load definition.

The last Limit State is the driveability. If a pile is designed so that during
driving it has an excessive maximum blow count or suffers damage due to
the driving operation, then this condition is unsatisfactory and the design
cannot be accepted. This case is similar to serviceability limit states
commonly used in structural design such as deflection or vibration limits.
Driveability can be evaluated using wave equation analysis. When
driveability requirements are violated it may be possible to solve the problem
by adjusting the driving system. Sometimes it is necessary to change the
pile selection or the design loads.

The pile structural limit states have been studied by Berger and >-Factors
generated for most failure modes (Berger 1989). The >-Factors for the
structural limit states will not be discussed further here but are given in
Table 1.

Table 1: Resistance Factors for Pile Structural Limit States

I Pile Material I Resistance Factor (


I
Steel 0.82

Prestressed Concrete 0.72


Timber 0.65

Determination of Pile-Soil Limit State CP-Factor

Prior to determining resistance factors for the most common methods of


predicting the pile-soil failure load, a calibration process is undertaken. The
calibration process determines the intrinsic safety in current design methods.
Safety indices, P, are calculated for existing designs utilizing the variability
and accuracy in both load and resistance predictive methods. After
reviewing the calibration results, a target P is selected. In most cases, there

432
will be a range in the resultant magnitude of the safety indices that requires
an averaging and interpretation before arriving at the most rational target
index. The discovered range in variability of P in present design methods
provides comparative values of reliability in, and underscores the necessity
of, developing design codes that insure a relatively uniform level of safety.
Although a necessity in developing reliability-based design criteria, the
process of calibration is not the sole basis on which the target indices rest.
Required are tradeoffs between simplification and accuracy, utilization of
imperfect data bases and the costs of improving them, the fit of safety
indices within the overall design index, and others. As a result, the reliability
of a pile design should be consistent over the wide range of capacity
prediction methods, pile types and quality control methods.

The computer program A58LF (used in Ellingwood et al 1980) was utilized


to compute reliability indices for the four methods of capacity prediction
under investigation. These methods are static analysis using SPT data, the
ENR formula, CAPWAP analysis, and wave equation analysis using
GRLWEAP. An extensive data base (GRL-FHWA) was available that
contained subsurface and capacity information at multiple U.S. sites. In this
data base the capacity was determined by the methods listed above. All
variability in the accuracy of each method was assumed inherent in the
coefficient of variation determined in the statistical analysis performed on
the ratios of static load test capacity to predictive method capacity. Table
2 presents the statistical results for the four predictive methods. Program
A58 LF, utilized for the NBS 577 study (Ellingwood et al 1980), computes the
reliability index associated with a given limit state function that represents
the particular design method under investigation. It is a first-order, second-
moment procedure that gives values of the reliability index by transforming
non-normal variables into equivalent normal variables and finding the
shortest distance between the surface of the limit state function and the
origin in a reduced coordinate system with zero mean and unit variance. This
approximate technique can often provide good agreement with more exact
solution methods (Rackwitz and Fiessler 1976),

An estimate of resistance factors for each predictive method was made


utilizing a derivation of resistance factors for structural members proposed
by Galambos and Ravindra (1973), By applying a linear approximation that
permits the separation of load and resistance variables and selecting the
linearization factor, a, through a minimization process, the resistance factor
may be calculated:

where
<t> is the resistance factor
a is the linearization factor, 0.55
P is the target reliability index

433
VA is the coefficient of variation for the predictive method
Rm /R n is the average of the ratio of the measured to predicted resistance

The determination of the various resistance factors implied by the design


procedure described above requires a volume of data regarding the accuracy
and reliability of the strength determination procedure used. The available
data base contained the following information at each site:

1. Subsurface investigation information including a soil boring log and SPT


N-values at least as deep as the toe of the pile.

2. A complete description of the pile.

3. The pile driving hammer and driving system together with the driving
record including restrike blow counts.

4. A capacity from a static load test to failure as defined by the Davisson


criteria. This capacity was the baseline to which the other predictive
methods were compared.

5. Dynamic measurements made during restrike after a substantial wait


period and generally made near the time that the static load test was
performed.

The data base used here consisted of 100 sets of data containing the
above information. When the data was assembled it was examined by
experienced field engineers to assure that it was not only correct but
reasonable and sensible. From this information capacity predictions were
made according to:

1. A CAPWAP analysis of the restrike dynamic measurements.

2. A capacity determination based on the subsurface investigation


information (static analysis).

3. A capacity determination using the a wave equation analysis with the


GRLWEAP program.

4. A capacity prediction using a dynamic formula together with the the


restrike blow count.

For each data set, the capacities from the static load test were divided by
the capacities from the above methods. Then the means and standard
deviations were found for each prediction method. These values are
tabulated in Table 2. The q>-Factors were calculated using the procedures
described above. The variation of q> with the safety index, P, is given in
Figure 2. The structural design profession has come to accept safety
indicies in the range of 2.5 to 3.0 based on extensive calibration studies.
Calibration studies were also performed in this study. Some of the results

434
are shown in Figure 3. For foundations, relatively small values of live load-
dead load ratios are appropriate since these conditions are typical for
bridges. It should be noted that for the large values of coefficient of
variation typical of foundation design the safety index is not very dependent
on live load-dead load ratio.

Table 2: Statistical Results from Calibration Study

Predictive Method Mean COV % Sample Size

Static Analysis 0.98 52 93

ENR 1.16 62 100

CAPWAP 1.24 32 89
WAP 0.99 41 100

Results and Discussion

Based on the above analysis the CP-Factors and the p-Values,of 2.5 and
3.0 are tabulated in Table 3. The CP-Factors contained in the LRFD version
of the AASHTO Bridge Code are tabulated in Table 4. The two sets of values
are not completely comparable since the two tables do not give values for all

0.9
.............. -......
0.8 .-..-.., .., CAPWAP
0.7
........ - ... ..-.......
ENR ,,~ ~ .-..-.._.., ..__._.
0.6
~ 0.5
...... ~ ~ ~ ~
..........
.......
... .... - - -.
GRLWEAP
....
~~~
~~
0.4 -~
STATIC ANALYSIS ....
.....................
0.3
0.2
0.1
o+ - - - - r - - - , . . . - - - , . . . - - - , . . . - - - . , - - - - - 1
1.5 2 2.5 :5 3.5 4 4.5
BETA

Figure 2 Relationship between cP and Safety Index for Various Methods

435
of the same cases. For the case of "skin friction and end bearing in sand"
with the SPT method, the AASHTO Code specifies 0.45 compared with our
value of 0.42 for a safety index of 3.0. However, the general practice in
almost all of the world would use some information from the driving
operation to improve the safety of the foundation. The Code specifies a
Resistance Factor of 0.70 for the "Pile Driving Analyzer". Our study
obtained 0.73 for a safety index of 3.0. Today the general practice would
normally use a CAPWAP analysis with the Pile Driving Analyzer
measurements. A Resistance Factor of 0.8 is specified in the Code for the
use of a "Load Test". In general, these numbers agree quite well. The
primary problem with both sets of values is that they did not take into
consideration the number of tests to be used in construction control. This
factor deserves further study.

Conclusions

In this paper, available data has been used to generate lP-Factors for the
design of driven piles. The values obtained agree quite well with the new
AASHTO Bridge Code for those cases where comparable values are available.
In both sources, no consideration was given to the number of construction
control tests performed. Certainly, if every pile is tested a larger Resistance
Factor would be appropriate than if only one test is performed. The
Resistance Factors should be based on some consideration of the number of

LL/DI.. ::: 1.0


3..------------------------,
2.5 ....................CAPWAP
........
....
2
......
............. _ - -GRLWEAP
-<
t:i 1.5
-- STATIC o\NALYSIS
m
_ _ EI'fi
..., .. -
0.5

o +--......,....-__,r----r----,.--..---~-__,r-----l
1 1.5 2 2.5 3 3.5 4 4.5 5
FACTOR OF SAFETY

Figure 3 Safety Index Compared with Factor of Safety for a Live-Load Ratio
of 1.0
436
Table 3 Resistance Factors for p = 2.5 and 3.0

Predictive Method Safety Index Resistance Factor

Static Analysis 2.5 0.48


3.0 0.42

ENR 2.5 0.50


3.0 0.42

CAPWAP 2.5 0.80


3.0 0.73

WEAP 2.5 0.56


3.0 0.50

Table 4: AASHTO Code Resistance Factors for Skin Friction and End Bearing
in Sand

I Predictive Method I Resistance Factor I


SPT-method 0.45

Load Test 0.80

Pile Driving Analyzer 0.70

tests performed. This study has shown that available data bases can be
used effectively to arrive at appropriate values for Resistance Factors for
driven pile foundations.

References

ACI Committee 318 (1956). "Building Code Requirements for Reinforced


Concrete (ACI 318-56)." American Concrete Institute, Detroit, Michigan.

ACI Committee 318 (1963). "Building Code Requirements for Reinforced


Concrete (ACI 318-63)." American Concrete Institute, Detroit, Michigan.

American Institute of Steel Construction (1986). "Manual of Steel


Construction, Load and Resistance Factor Design." First Edition, American
Institute of Steel Construction, Inc., New York, NY.

Barker, R. M., Duncan, J. M., Rojiani, K. B., Ooi, P. S. K., Tan, C. K., and
Kim, S. G. (1991). "Manuals for the Design of Bridge Foundations." Report
343, National Cooperative Highway Research Program, Transportation
Research Board, National Research Council, Washington, D.C.

Berger, J. A. (1989). "Development of Probability-Based Resistance Factors


for Driven Piles." Thesis submitted in Partial Fulfillment of the Requirements
for the Masters Degree, Department of Civil, Environmental, and

437
Architectural EngineeringUniversity of Colorado, Boulder, CO.

Chin, F. K. (1970). "Estimation of the Ultimate Load of Piles not Carried to


Failure." Proceedings 2nd Southeast Asian Conference on Soil Engineering,
Singapore.

Cornell, C. A. (1969). "A Probability Based Structural Code." Journal of the


American Concrete Institute, Proceedings Vol. 66, No.1 2, December.

Davisson, M. T. (1970). "Pile Load Capacity." Proceedings - Design,


Construction and Performance of Deep Foundations, American Society of
Civil Engineers Seminar, University of California, Berkeley, CA.

Ellingwood, B., Galombos, T. V., MacGregor, J. G., and Cornell, C. A.


(1980). "Development of a Probability Based Load Criterion for American
Standard A58." Special Publication No. 577, National Bureau of Standards,
Washington, D. C.

Galambos, T. V. and Ravindra, M. K. (1973). "Tenetative Load and


Resistance Factor Design Criteria for Steel Buildings." Structural Division
Report No. 18, Washington University, St. Louis, MO.

Goble, G. G. (1979). "A Pile Design and Installation Specification Based on


the Load Factor Concept." Report Submitted to the Federal Highway
Administration, Department of Civil, Environmental, and Architectural
Engineering, University of Colorado, Boulder, CO.

Goble, G. G., Moses, F., and Snyder, R., (1980). Pile Design and Installation
Specification Based on the Load Factor Concept," Transportation Research
Record 749, National Research Board, National Academy of Sciences,
Washington, D. C.

Rackwitz, R. and Fiessler, B. (1976). "Notes on Discrete Safety Checking


When Using Non-Normal Stochastic Models for Basic Variables." Loads
Project Working Session, MIT, Cambridge, Mass.

438
Seismic Retrofit of Foundations for a Double-Deck Viaduct
Matthew E. Fowler, P.E.1, Robert E. Johnston, G.E.1,
1
and Galen S. Nagle, P.E.

ABSTRACT
The seismic retrofit of foundations of heavy, double-deck viaduct
structures of the Alemany Interchange in San Francisco, California is
described. The focus is on the design of drilled shaft foundations which were
either added to existing pile groups or used to replace existing pile groups.
Large lateral loads and moments resulting from the application of seismic
design forces necessitated the added foundations. The design was
complicated by the differing characteristics of the new shafts and the existing
piles. The design was further complicated by changing soil and rock
conditions beneath the lengthy project alignment, adjacent buried utility
mains, steep terrain, and the need to maintain traffic during construction. An
innovative base grouting treatment is being applied to the sandy soils at the
base of the drilled shafts to mobilize needed end-bearing capacity for the
heavily loaded 2.7 m (9 tt) diameter shafts which replace pile groups.

BACKGROUND
The October 1989 Loma Prieta earthquake severely damaged several
vital highway structures in the Oakland and San Francisco areas. In the case
of the collapsed Cypress Freeway viaduct in Oakland, it provided a tragic
example of what can happen when adverse soil conditions combine with a
heavy, non-ductile structure during a severe earthquake. Like the Cypress
viaduct, the Alemany Interchange that links 1-280 with U.S. Route 101 in San
Francisco (see Figure 1) is a double-deck viaduct structure built during the
early 1960s and designed for seismic accelerations of 0.06g,. considerably
less severe than the present design accelerations. Expected peak bedrock
accelerations of 0.5g are based on a maximum credible earthquake of
Richter magnitude of 8.3 on the San Andreas Fault located 12 km (7.5 mi)

Geotechnical and Underground Group,


Parsons Brinckerhoff Quade & Douglas, Inc., San Francisco, CA 94107

439 Fowler, Johnston, and Nagle


west of the site. The Alemany Interchange and the nearly 4.8 km (3 mi)
portion of Interstate 280 to the immediate north were damaged but left
standing by the quake. 1-280 was closed to traffic from the intersection with
Route 101 to its northern terminus near the San Francisco financial district.


To downtown San Francisco - - - - '

B-line (6 bents)

W-line (23 bents)

TO San Francisco
International Airport
\ and San Jose

Figure 1. Alemany Viaduct Site Location

Local engineering firms were called on by the California Department of


Transportation (Caltrans) to assist in preparing retrofit concepts for the
damaged freeway. Design teams worked closely with Caltrans and a Seismic
Peer Review Committee appointed by the Governor to develop retrofit
concepts suitable for the varied structural configurations and soil conditions
present along the route. Retrofit concepts for each segment evolved into
detailed designs and ultimately into contract documents.

By April 1993, the upper deck of 1-280 north of the Alemany Interchange
was reopened to limited traffic. Construction was completed in 1993 on six of
the double-deck B-Line bents of the Alemany Interchange and construction is
presently (July, 1994) underway on the remaining twenty-three bents of the
adjacent W-Line section that parallel and cross the Route 101,

Nine separate classes or types (described herein as Types I to IX) of


foundation retrofits were identified for the 68 separate foundations along the
approximately 1.2 km (4000 ft) of viaduct that comprise the retrofitted
sections of the Alemany Interchange. About 44 percent of the foundations

440 Fowler, Johnston, and Nagle


were modified using a Type VI retrofit, which consists of adding 0.9 m and
1.2 m (3 ft and 4 ft) diameter drilled shafts around the perimeter of the
existing groups of driven piles. Large-diameter (2.7 m, 9 ft) shafts were used
with the Type VIII and Type IX retrofits to replace 32 percent of the
foundations. Figure 2 shows a Type VI foundation retrofit and the finished
superstructure of a typical bent from the now-completed B-Line portion of the
viaduct.

This paper gives an overview of the general approach used to analyze


and design foundations for the Type VI retrofits and describes the rationale,
design assumptions, and preliminary load test results for the compaction
grouting techniques being used to improve end-bearing resistance for the 2.7
m-diameter shafts used in the Type VIII and Type IX retrofits.

Figure 2. Typical Retrofitted Structure and Foundations

EXISTING CONDITIONS

Su perstructure
The existing viaduct consists of single and double-deck reinforced
concrete box girders cast integrally with transverse pier cap beams. Figure 3
shows a typical double-deck bent. A variety of other bent configurations
including single-column cantilever bents are present in the viaduct. Built
during the early 1960s, the viaduct was designed primarily for gravity loads

441 Fowler, Johnston, and Nagle


with a minimal lateral load capacity of 0.06g applied as an equivalent static
load. Descriptions of the existing structure, its reinforcement details, and the
adopted Caltrans retrofit guidelines are reported by Jackson (Jackson,
1993). Structural inadequacies that led to the required extensive renovation
of the entire structure include:

Inadequate shear and torsion reinforcement in the bent cap beams,


Non-redundant load paths for lateral loads,
Inadequately reinforced column/foundation pinned connections,
Inadequate uplift capacity in foundations,
No shear reinforcement and top main reinforcement in pile caps.

CAP
RECTANGULAR BEAM
COLUMN

FOUNDATION PINNED, TYPICAL

Figure 3. Double-Deck Bent Before Retrofit (Jackson, 1993)

Foundations
As-built drawings were used to determine existing foundation types,
depths, and configurations for each of the 68 separate foundations contained
in the 29 bents and two abutments. Pile foundations were most common with
only two bents being supported on spread footings. Three types of piles were

442 Fowler, Johnston, and Nagle


placed in groups ranging from 5 to 47 piles: Raymond step-tapered concrete
(cast-in-driven-shell) piles, 0.4 m (16 in) diameter CIDH piles, and steel
H-piles (10BP42). With the exception of the abutments, all piles were driven
or drilled vertically. All piles were rated at 400 kN (45 ton) with lengths
ranging from 4.6 to 12.5 m (15 to 41 ft).

The compressive capacity of the existing piles was evaluated and found
sufficient to justify their continued use where feasible in the retrofitted
foundations. However, because vertical reinforcement in the CIDH and
tapered piles extends only 3.6 to 4.6 m (12 to 15 ft) below the pile head, no
useable uplift capacity was assigned to these piles in the design.

Subsurface Conditions
Subsurface conditions were determined from boring logs included in as-
built drawings and from five new supplementary borings (Parsons
Brinckerhoff Quade & Douglas, Inc., 1992). Dense to very dense, silty sands,
sandy silts, and stiff clays ranging from 2 m to 40 m deep overlie a weathered
bedrock of Franciscan Formation sandstone and shale. Compressive
strengths obtained on selected intact samples of bedrock range from 5.8
MPa (845 psi) in the shale to 78 MPa (11,300 psi) in the sandstone.
Groundwater depths range from 2 to 8 m below surface at each shaft
location. Liquefaction potential was judged to be low at all but the two
southernmost bents, where marginal liquefaction potential exists.

FOUNDATION RETROFIT OPTIONS

Methods of increasing foundation capacity include drilled shafts, driven


piles, spread footings, and soil improvement. Drilled shafts were selected for
the majority of the foundations based on constructibility in the low-overhead-
clearance settings, and the high lateral and axial capacity that shafts can
provide. Driven piles were rejected because of their relatively low lateral
capacity and the limited available vertical clearance. Spread footings were
only considered where they were already in use as foundations on rock and
where lateral loads were small. Soil improvement in the form of compaction
grouting was adopted for two conditions: (1) where existing foundations were
found to be structurally adequate but soils were judged to be susceptible to
liquefaction, and (2) as a method of increasing end bearing resistance on the
large-diameter drilled shafts constructed using slurry. Drilled shaft base
grouting is described later in this paper. Table 1 describes the nine retrofit
types that were ultimately utilized.

443 Fowler, Johnston, and Nagle


Table 1. Definition of Foundation Retrofit Types
Type Description Comments
I No modification of existing pile Pinned-end column connections deliver
group foundations minimal moment or lateral loads to
foundation. Capacity of existing pile groups
deemed adequate.
II Compaction grouting along Borings indicate potentially liquefiable
perimeter of existing pile group zones
III Enlargement of existing spread Existing column not being demolished and
footing added weight from new structure
insufficient to warrant new footings.
IV Multiple 0.9 m (3 ft) diameter New foundation required because of major
drilled shafts with abandonment of changes to superstructure
existing foundation
V Multiple 0.9 m (3 ft) diameter New abutment required because of major
drilled shafts with abandonment of changes to superstructure
existing abutment
VI Multiple 0.9 m (3 ft) diameter Shafts placed along perimeter of existing
drilled shafts to supplement pile group.
existing pile group
VII Double 1.2 m (4 ft) drilled shafts Superstructure changed to reduce torsion
with grade beam in beam/column joints on bents having
outrigger cap beams. New columns may
be constructed without prior demolition of
existing columns.
VIII Single 2.7m(9 ft) diameter drilled Requires demolition of pilecap and
shafts with base grouting extraction of portion of existing pile group.
IX Double 2.7m(9 ft) diameter drilled Used on tallest bents having outrigger cap
shafts with base grouting beams. Allows new columns to be
constructed without prior demolition of
existing columns.

FOUNDATION ANALYSIS SEQUENCE FOR TYPE VI RETROFITS

General Approach
A review of the literature yielded no practical methods for analyzing the
complex interactions between closely-spaced groups of piles and the
earthquake motions in the soil mass. Recognizing the simplifications
involved, a pseudo-static approach was adopted. Trial foundations were
developed by adding drilled shafts to the existing pile group in progressive
stages of increasing length, diameter, and number. Shafts were lengthened

444 Fowler, Johnston, and Nagle


until a condition of fixity or no lateral deflection at the pile tip was predicted
by the computer model.

Acceptance of a trial foundation was based on three considerations: (1)


not exceeding the bending capacity of the existing piles, (2) not exceeding
the uplift and compression axial capacities of the drilled shafts, and (3) not
exceeding the lateral and rotational limits established by the structural
engineers for the particular bent. A load-moment interaction diagram was
used to compare the induced bending moments in the existing piles with their
estimated ultimate moment capacity. Bending moments in the drilled shafts
were generally not a limiting factor because it was always possible to design
the drilled shafts to satisfy the predicted moment demands. Once a feasible
foundation scheme was determined for transverse loads, the same
foundation was then analyzed for the loads in a longitudinal direction parallel
to the viaduct.

Computer Tools
A combination of the computer programs listed in Table 2 and manual
calculation techniques were used to estimate the strength and deformation
characteristics of single piles and pile groups.

Calculation Steps
Figure 4 illustrates the iterative design process followed in performing a
typical lateral analysis for a Type VI foundation retrofit. Key steps are
described as follows:

Step 1: Describe Foundation Layout and Pile Connections

Four considerations governed the placement of new shafts: (1) the


presence of buried utilities, (2) overhead clearance for equipment, (3)
distance to existing piles, and (4) distance between drilled shafts. Shafts
were placed to:

Maintain three shaft diameters minimum between shafts.

Maintain a minimum of 1.5 shaft diameters between shafts and


existing piles.
Maintain at least 0.76 m (2.5 ft) of lateral clearance between new pile
caps and buried utilities.
Where possible, locate shafts outside demolition limits of decks
overhead.

445 Fowler, Johnston, and Nagle


Existing concrete piles were modeled with a pinned connection at the pile
cap and new drilled shafts were modeled with a fixed-end boundary
condition.

Step 2: Describe Soil Profile and Properties

Soil depths and properties at each bent were developed from boring logs
and laboratory tests from the five supplementary borings and from logs
contained in the as-built drawings. Some degradation in lateral stiffness and
strength is expected under seismic loading of the foundations. An allowance
for cyclical loading effects on lateral resistance of the pile group was made
by implementing the empirical cyclic loading parameters used by GROUP to
generate the P-y curves for each pile sub-group.

Table 2. Computer Programs


Program Use Features
LPILE Analysis of single pile Computes pile rotation, deflection,
Version 3.0 response to lateral loads and shear, bending moment, and soil
moments. LPILE was used for reaction with respect to pile depth. Non-
(Reese, et
Retrofit Types VII, VIII, and IX. linear soil response is based on
aI., 1990)
empirical P-y curves.
GROUP Two-dimensional lateral Computes: (1) pile cap settlement,
Version 2.0 analysis of pile groups with rotation, and deflection,(2) distribution of
mixed pile types, varied pile axial and lateral loads to piles in group,
(Reese, et
geometries, and non-uniform and (3) pile rotations, deflection, shear,
al.,1990)
soil profiles subjected to bending moment, and soil reaction. P-y
combined vertical and lateral curves can be reduced for pile-soil-pile
loads, and moments. GROUP interaction. Allows for fixed, pinned, or
was used for Retrofit Types IV, laterally restrained connections between
V, VI. piles and pile cap.
APILE2 Calculation of single pile load Combines elastic compression of a pile
Version 1.0 versus settlement curves with "t-z" curves for side load transfer
and "q-w" curves for tip load transfer to
(Reese and
generate load-settlement curves for
Wang, 1990)
driven piles and drilled shafts.
SHAFT1 Calculation of axial capacity Computes axial capacity and load
Version 1.1 and load versus settlement for versus settlement for a drilled shaft in
drilled shafts. clay, sand, rock, and mixed profiles
(Reese and
using empirical methods presented in
Wang, 1989)
FHWA Drilled Shaft Manual (Reese and
O'Neill, 1988).

446 Fowler, Johnston, and Nagle


Describe Foundation Layout and Pile Connections

2 Describe Soil Profile and Strength Properties

3 Determine Applied Foundation Forces


Change Shaft
Position Length 4 Define Geometry and Rigidity of Piles and Shafts
or Diameter

5 Estimate Axial Capacity of Piles and Trial Shafts

Develop Estimated Axial Load-Settlement (LS) Curves

Modify Single-Pile LS Curves for Axial Group Interaction

B Modify Single-Pile p-y Curves for Lateral Group Interaction

Not OK

Not OK
Check Axial Loads and Moments

Not OK

No

,-::::':"':'=CO--I10

r---,=,-,-=:o.:..:..-l11 Plot Moments, Shear, and Deflection; Piles & Shafts

Go to Next Foundation

Figure 4. Flow Diagram for Foundation Retrofit Analysis

447 Fowler, Johnston, and Nagle


Step 3: Determine Applied Foundation Forces

Foundation loads provided for design by the structural engineering team


conformed to the 'Weak column" concepts embodied in the Retrofit
Guidelines developed by Ray Zelinski of Caltrans. The cap beams, edge
girders, joints, and foundations were all designed to be stronger than the
columns, and thus, force the formation of plastic hinges at the columns where
repairs can be made with less difficulty (Jackson, 1993).

Load and moment combinations were given at the base of the column (top
of pile cap) as shown in Figure 5 for a typical transverse load case. Two
transverse load cases and two longitudinal load cases were analyzed for
each footing. The given vertical foundation forces were increased by the
estimated dead weight of the new pile cap. The applied lateral forces were
reduced by approximately 220 to 440 kN (50 to 100 kips), depending on the
width of the pile cap, to account for passive pressures mobilized by the soil
against the pile cap. Passive pressures assigned to the pile cap were scaled
down from a full theoretical passive resistance in proportion to the predicted
levels of lateral deflection.

Step 4: Define Geometry and Rigidity of Piles and Shafts

The existing tapered-shell and CIDH concrete piles lack vertical


reinforcement below the upper 4 m of the pile which greatly reduces the
usable lateral resistance of the existing piles. For modeling purposes, the
steel shell was ignored and a plastic hinge was assumed to form below the
reinforcement. Two adjustments were made to the pile rigidities to account
for the expected cracking under seismic load and the assumed formation of a
plastic hinge below the reinforcement in the existing piles: (1) Moment of
inertia values for the pile and drilled shaft cross-sections were reduced to
one half their intact values to account for cracking, and (2) the portion of
each pile below the longitudinal reinforcement was assigned a nominal
moment of inertia (I) equal to 10 percent of that of the gross intact section.

Step 5: Estimate Axial Capacity of Piles and Shafts

Axial capacities of the piles and shafts do not enter directly into the
GROUP input data but were used to verify the axial load-settlement curves
developed in Step 6 below. The allowed 800 kN (180 ton) ultimate axial
capacities of the existing tapered piles were verified using Nordlund's Method
(Nordlund, 1963). Axial capacities of drilled shafts were estimated using the
program SHAFT1.

448 Fowler, Johnston, and Nagle


P = Axial Force
M = Moment
I I =
V Shear Force

M
SEISMIC ACCELERATION
v
Passive resistance of pile cap
1 4 - - - Lateral resistance of piles and shalts

1 r
" ' " Side resistance of pile

~ : ~ Existing Piles

Axial reactions al pile cap

Figure 5. Foundation Response to Typical Seismic Loads and Moments

Step 6: Develop Estimated Axial Load-Settlement Curves

GROUP requires axial load-settlement curves for all classes of piles and
shafts in the group. The self-generated load settlement curves included with
GROUP are only valid for driven piles. Hence, two additional programs were
employed; APILE2 (Reese and Wang, 1989) combines elastic pile
compression with empirical tip and side load-transfer ("t-z" and "q-w") curves
for driven piles to produce load-settlement curves. SHAFT1 (Reese and
Wang, 1989) follows a similar approach for drilled shafts and also provides
estimates of axial capacity.

Uplift capacity of the drilled shafts was limited to 2/3 of the ultimate side
resistance estimated for axial compression. Under tension, the shafts were
assumed to follow a displacement path parallel to the initial portion of the
compressive axial load settlement curve up to the assumed load limit, after
which the shaft was assumed to move upwards with no increase or decrease
in tensile resistance. The dead weight of the shafts was neglected. Existing
concrete piles were modeled to provide no uplift resistance consistent with
the absence of longitudinal reinforcement below the upper 4 m.

449 Fowler, Johnston, and Nagle


TENSION COMPRESSION

I
L
TENSION - - _
...~
~
COMPRESSION

~
... ... ...
ill
SINGLE MODIFIED ,
r----{" \
MODIFIED
PILE

PILE IN A
FOR GROUP
INTERACTION
\
\
SINGLE
SHAFT IN A SHAFT
FOR GROUP GROUP GROUP
INTERACTION

EXISTING PILE NEW DRILLED SHAFT

Figure 6. Adjustment of Load Settlement Curves for Group Effects

Step 7: Modify Settlement Curves for Group Interaction

The proximity of the existing piles to one and other and to the added
drilled shafts leads to pile-soil-pile interactions that must be considered in the
analysis. A rigorous treatment of vertical pile interaction or group effects is
hindered by the different types and lengths of piles in a retrofitted foundation.
To allow for the expected interactions, single-pile settlements obtained from
APILE2 were multiplied by group settlement ratios ranging from 4 to 6 that
were estimated using elasticity-based methods (Poulos, 1979). A similar
allowance for axial group effects for the drilled shafts was made by using the
lower-bound (i.e., less stiff) load-settlement curves calculated by the SHAFT1
program. Figure 6 illustrates the effect of applying group settlement ratios to
the single pile load-settlement curves for the drilled shafts and piles.

Step 8: Modify P-y Curves for Lateral Group Interaction

Like the vertical pile interactions described above, lateral interactions


between different types of piles in a single group must also be considered.
The approach followed here amounts to reducing the ultimate lateral
resistance to account for shadow effects and superposition of elastic
deformations (Brown et al. 1988). GROUP conveniently allows the user to
modify the single pile P-y curves that it generates to account for group
effects. P-multipliers ranging from 0.3 to 0.9 were assigned to the piles
according to their position in the pile group. A Y-multiplier of 1.0 was used.
Figure 7 illustrates the concept of overlapping wedge shaped shear zones
from adjacent piles and the rationale for modifying the single-pile P-y curves
to account for lateral group effects.

BASE-GROUTING OF DRILLED SHAFTS

In general practice, end bearing resistance is often ignored or reduced


significantly when designing drilled shafts that will be constructed under

450 Fowler, Johnston, and Nagle


single-pile
- - - - - P-y curve

group pile
~-- p.y curve
P (Single pile in group)

Dellection (y)

Overlapping Passive Wedge-Type Failure Pmu~iplier concept

Figure 7_ Adjustment of P-y curves for Lateral Group Effects


(After Brown et ai, 1988)
slurry or water. The FHWA drilled shaft design manual (Reese and O'Neill,
1988) recommends reducing end bearing with increasing shaft diameter for
shafts larger than about 1.3 m (50 in) diameter. The practice is reasonable
given the difficulty of cleaning the hole and the disturbance that occurs in the
underlying soils. However, this necessarily leads to longer shafts for a given
axial load demand. The added material and drilling costs incurred by ignoring
end bearing on smaller-diameter shafts of 1 to 2 m diameter are generally
offset by lower drilling and installation unit costs. As shaft diameter is
increased and the end bearing capacity being ignored also increases, the
cost tradeoffs become less clear.

Cost considerations, limited availability of workspace, and the need to


minimize shaft construction time prompted the decision to specify compaction
grouting to improve the end-bearing resistance and load-settlement
characteristics of the 2.7 m (9 ft) diameter drilled shafts that end in soil. A
review of case histories (Bruce, 1986, Stocker, 1983, and Littlejohn et aI.,
1983) established the precedence for base grouting, confirmed the feasibility
of injecting grout below the base of drilled shafts, and provided a basis for
quantifying the degree of improvement. Base compaction grouting will be
done after the shaft has cured for at least one day but before significant
structural loads are delivered to the shaft. The low-mobility sand-cement
grout (Brown and Warner, 1974) will be delivered to a 1.5 m deep target
zone beneath the shaft base through steel pipes placed within five of the
PVC inspection pipes cast in the shaft. Specified grouting criteria call for up
to 0.75 m 3 to be injected at each of the five points at pressures of up to 5.5
MPa (800 psi) with grouting to cease if uplift of the shaft or hydrofracturing of
the soils are detected.

For design of the 2.7 m (9 ft) diameter shafts it was assumed that the
base grouting procedure would double the estimated ungrouted end
resistance of 1.7 MPa (18 tsf) at a settlement of 2.5 inches. Preliminary test

451 Fowler, Johnston, and Nagle


data shown in Figure 8 from a recently completed series of load tests using
an Osterberg cell (Loadtest Inc., 1994) indicate the specified base grouting
technique to be quite effective at increasing the available end resistance at
tolerable settlements of a pile constructed in sand using slurry methods.
Load tests run before and after base grouting a 1.4 m diameter x 17 m (4 ft x
55 ft) trial shaft showed the grouting to reduce the tip settlement to 15 to 20
percent of the settlements that were measured at equal plate pressures
before grouting.

10

o
~ - r-----.
-
::2 -10
I-
-------==== --
~
I BEFORE COMPACTION
GROUT
1-----
m-20
::2
I "'"
w
~ 30
AFTER COMPACTION
GROUT ~
~
::2
-40

-50
o 2 4 6
AVERAGE PRESSURE ON BEARING PLATE (MPa)

Figure 8. Osterberg Cell Load-Movement Curves Before


and After Base Grouting (after Loadtest Inc., 1994)

CONCLUSIONS

Analysis of foundations for the seismic retrofit of the Alemany Viaduct


required the development of a systematic methodology to handle a variety of
foundation types ranging from spread footings to 2.7 m (9 ft) diameter drilled
shafts. Commercially-available computer programs were found to be helpful
for analyzing the lateral response of single drilled shafts and pile groups
containing mixed pile types and pile geometries under combined moment,
axial, lateral load combinations. Published empirical and elasticity-based
methods were employed in conjunction with the computer programs to
account for lateral and axial group effects caused by pile-soil-pile interaction
between and among the existing piles and the new drilled shafts in the
retrofit. Load tests performed to verify the effectiveness of base grouting as a
method of improving end resistance and shortening the length of large-
diameter shafts have confirmed the design assumption that the available end
resistance of large-diameter piles constructed beneath water or slurry can be
increased significantly to allow designed shaft lengths to be shortened.
Experience thus far clearly suggests each seismic retrofit project is

452 Fowler, Johnston, and Nagle


accompanied by its own set of peculiar design constraints and challenges not
found with new construction. Design coordination and continuous dialogue
between the structural and geotechnical engineers are vital to the
development of retrofit methods and the success of a retrofit design project.

REFERENCES
Brown, D.A., C. Morrison, and L. Reese, 1988 "Lateral Load Behavior of Pile
Group in Sand," J Geotech. Engrg., ASCE, 114(11).
Brown, D.R. and J.Warner, "Compaction Grouting" J Soil Mech. and Found.
Oiv., ASCE, 99(8).
Bruce, D.A., 1986 "Enhancing the Performance of Large Diameter Piles by
Grouting", Parts 1 and 2, Ground Engineering May, 1986.
Goodwin, J.W., 1993, 'Bi-directional Load Testing of Shafts to 6000 Tons';
Design and Performance of Deep Foundations: Piles and Piers in Soil and
Soft Rock, Proc. ASCE Convention, October 1993.
Jackson, 1. B., 1993, 'Bridge Restoration': Proc. Fall 1993 Lecture Series on
Structural Rehabilitation/Restoration, Boston Society of Civil Engineers.
Littlejohn, G.S., J.L.lngle, and K.Dadasbilge, 1983,"lmprovement in Base
Resistance of Large Diameter Piles Founded in Silty Sand", Proc. 8th
European Cont. Soil Mech. and Found. Engrg., Helsinki.
Loadtest Inc., 1994, "CIDH Test Pile, US101 11-280 Retrofit, San Francisco,
California" Report No. LT-141, June 2,1994
Nordlund, R. L., 1963, "Bearing Capacity of Piles in Cohesionless Soils", J.
Soil Mech. and Found. Oiv., ASCE, 89 (SM-3).
Parsons Brinckerhoff Quade and Douglas, Inc., 1992, Geotechnical Data
Summary Report, Earthquake Retrofit, Alemany Interchange.
Reese, L. C. and M. W. O'Neill, 1988, "Drilled Shafts: Construction
Procedures and Design Methods," FHWA-HI-88-042, August, 1988.
Reese, L.C. and S-T Wang, 1990, APILE2 Version 1.0, " Axial Load
Settlement Analysis of Piles,"Ensoft Inc.,Austin, Texas 78718.
Reese, L.C. and S-T Wang, 1989, SHAFT1 Version 1.1, "Drilled Shafts
under Axial Loading," Ensoft, Inc.
Reese,L.C.,K.Awoshika,P.Lam, and S.1.Wang, 1990, "Documentation of
Computer Program GROUP, Analysis of a Group of Piles Subjected to
Axial and Lateral Loading," Version 2.0, Ensoft Inc.
Stocker, M. 1983 "The Influence of Post-Grouting on the' Load-Bearing
Capacity of Bored Piles", Proc. 8th European Conf. Soil Mech. and
Found. Engrg., Helsinki.

453 Fowler, Johnston, and Nagle


DIAGNOSTIC EXPERT SYSTEM
FOR DRILLED SHAFT FOUNDATION CONSTRUCTION

by Siddika Demir 1, Deborah J. Fisher2 and Michael W. O'Neill 3

AB.STRACT
This paper explores field diagnostics in drilled shaft foundation construction
and describes an expert system that contains representative heuristics that suggest
field changes in specific practices that will produce a safer, higher quality and more
productive construction program based on conditions actually encountered during
drilling. A knowledge base of problems and solutions is developed with regard to
field constructability and diagnostic analysis relative to a class of problems that arise
during excavation that were not anticipated during design, involving three slurry-
related problems: apparent fluid loss, sedimentation of slurry and collapse of the
borehole. The knowledge base is converted into a system of rules for programming
into a diagnostic expert system shell, which is built around IFrrHEN rules, to which
explanatory text and references are added. It also supports the use of uncertainty
associated with a rule or the user's response and presents the rules that are used to
reach a conclusion. In addition, the expert system gives the user the flexibility to
change the parameters and to get immediate recommendations and conclusions.

INTRODUCTION
According to the 1992 National Bridge Inventory more than 35% of the
575,410 bridges in the U.S. need replacement or rehabilitation. The FHWA also
estimates that $164.9 billion needs to be invested over the next 20 years to solve this
problem (Dumas 1994). Such an undertaking undoubtedly calls for automated
construction operations to improve and enhance p Jductivity, quality and safety.
Foundations are a major issue in both infrastructure rehabilitation and new
construction, and a cost-effective type of foundation is the drilled shaft, which is well
suited to rehabilitation and expansion of existing facilities (Demir 1993).

IRes. Asst., School of Civ. Engrg., Purdue Univ., West Lafayette, IN 47907-1284.
2Assoc. Prof. and AGC Chair, Dept ofCiv. Engrg., Univ. of New Mexico, Albuquerque,
NM 78713-1351.
3Prof.. Dept. of Civ. and Envir. Engrg., Uillv. of Houston, Houston, TX 77204-4791.

454 Demir, Fisher, and O'Neill


Drilled shaft construction, like that of other foundation systems, is often
impeded by unanticipated conditions, produced by a wide variety of soil, rock and
groundwater states, variations in construction details and similar factors.
Unanticipated conditions can produce uncertainties and disagreements between the
owner and contractor on how to proceed. It is nonnally the contractor's
responsibility to adjust hislher construction procedures to solve the problem created
by encountering the unanticipated conditions; however, lack of experience of a
specific field crew with a particular problem can result in decisions that may not
produce an optimum result. Furthennore, the owner needs to know whether the
solution chosen by the contractor is the most appropriate technically and least
expensive for purposes of authorizing changes in construction procedures and for
processing claims.
A method to reduce the chances that an inappropriate decision will be made
regarding the solution, while simultaneously serving as a means to facilitate
consensus between owner and contractor, is the use of a diagnostic expert system.
Diagnostic expert systems can analyze symptoms identified by a user in a non-
repetitive construction environment, such as drilled shaft construction, through an
easy-to-use computational environment to arrive rapidly and reliably at potentially
optimum methods to alleviate problems caused by the unanticipated conditions
(Maher 1987). They can also assist in the education of novice construction and
inspection personnel (Finn and Reinschmidt 1986) and can offset the effects of the
trend toward the use of less skilled entry level laborers and operators. It should be
emphasized that diagnostic expert systems are most appropriate for dealing with
relatively routine problems.. Some construction problems are unique and complex
and do not lend themselves to inclusion in expert systems. When such is the case,
the user must appeal to a human expert.
Diagnostic expert systems appear to have a ready application to drilled shaft
construction, since most excavating and concreting takes place out of direct sight of
the contractor and inspector (O'Neill 1991), with the result that only symptoms of
unanticipated conditions (hole collapse, reduction of drilling rate, racking of
reinforcing steel, etc.) are observed. This paper describes the concept of a
diagnostic expert system for drilled shaft construction, introduces the first
component of a diagnostic expert system for drilled shaft construction called DS"2-
DrAG and illustrates the concept for three of many unanticipated conditions that can
be encountered. This expert system is "operational" in the sense that it is partially
verified, is not completed and is not being used commercially (Maher 1987). The
purpose of the paper is to demonstrate the applicability of expert system technology
to drilled shaft construction diagnostics and not to provide a completed, ready-to-
use diagnostic expert system, which is still under development.

BACKGROUND
Although expert systems have been developed for foundation design, e.g.,
"CUFAD" for drilled shaft design developed by Kulha\V)' and his colleagues
(Trautman and Kulha\V)' 1987), only a few knowledge based systems exist for
foundation construction (Mohan 1990). One of the existing classes of expert

455 Demir, Fisher, and O'Neill


systems assists in the choice of the most appropriate type of driven pile for a given
environment and is based on fuzzy logic (luang and Ulshafe 1990; Elton and Brown
1991). Another diagnostic expert system (Yeh, Hsu and Kuo 1991) assesses
damage to driven prestressed concrete piles. Out of 37 expert systems for
construction surveyed by Mohan (1990), 11 were developed for project planning,
scheduling and control, 9 were developed for project management, 6 were for
construction methods, and the rest were for equipment management, legal issues.
human-resources management, concrete mixing and placement and temporary
facilities layout. None addressed drilled shaft construction. To overcome this void,
the authors have developed a family of expert systems for drilled shaft construction
called the Decision Support for Drilled Shafts or "DS"2."
The operational expert system prototype, DS"2-DIAG, is the initial effort for
a flith component of DS"2, which consists of related core modules for analyzing
drilled shaft construction, as depicted in Figure 1. DSI\2-GEO, DS"2-SPEC and
DS"2-DIAG are expert systems, whereas DSI\2-COST is an interactive database,
and DS"2-SIM is a simulation program (Fisher, O'Neill and Contreras 1994). DS"2-
GEO selects the most appropriate construction method, while DS"2-SPEC provides
specific recommendations on construction details for the method selected in the tirst
module. The third module, DS"2-COST, assists in estimating costs based on
conclusions developed in the first two modules (Abaya, O'Neill and Fisher 1994).
DSI\2-SIM simulates the drilled shaft technology that utilizes future automated
technologies to arrive at production rates for costing purposes (Rho, Fisher and
O'Neill 1993). Finally, DSI\2-DIAG uses a heuristic rule-based system to diagnose
problems and recommend solutions during construction.
DS"2-DIAG is presently limited to three construction problems associated
with the use of drilling slurry:
Apparent fluid loss,
Sedimentation of slurry, and
Collapse of the borehole.
DS"2-DIAG is envisioned as a real-time, field tool to guide construction
decisions as construction progresses. A logical decision methodology for acceptance
of drilled shafts after the completion of construction, based partially on design stress
conditions in the shaft, problems observed during construction and post-construction
structural integrity testing, has recently been developed by the FHWA (Baker et al,
1993). DS"2-DIAG could be used in the future, for example, to complement such
acceptance logic in addition to its uses in real time.

DEVELOPMENT OF DS"2-DIAG
DS"2-DIAG has been developed by using EXSYS Professional Expert
System Software, Version 3.0.9-W and Microsoft Windows, Version 3.1. The
entire expert system software occupies about 2.9 MB of hard disk space. The
knowledge base has been built around IFrrHEN rules, to which explanatory text and
references are added. It also supports the use of uncertainty associated with a rule
or user's response and can present the rules that are used to reach a conclusion
(EXSYS 1988).

456 Demir, Fisher, and O'Neill


INPUT OUTPUT
Step 1
Geologic and Site Type of
DSI\2-GEO
Conditions Construction Method
Q
Type of Construction Step 2 Construction Specs;
Method DSI\2-SPEC Taxonomy of Method
Q
Step 3 Cost Analysis
Resource Data DSI\2-COST Using Spreadsheets
Q
Bid and Costing Step 4 Expected Benefits for
Data DSI\2-SIM Various Opera tions

Construction Q
Step 5
Construction
Problem Scenario Problem Solution
DSI\2-DIAG

Figure 1. Decision Support Model for DSI\2

The knowledge acquisition for DSI\2-DIAG was accomplished by visiting


construction sites in Houston, Texas, by observing numerous tield operations in
various geological settings around the U.S. collected on video and by interviewing a
domain expert (Demir 1993). Through this research it was concluded that many
diagnostic problems can be implemented into an expert system. A few scenarios
associated with unanticipated construction problems and construction problems
involving slurry are displayed in Tables 1 and 2, respectively.
Foundation engineering often requires solutions for poorly or incompletely
defined situations for which the design procedures being used are usually governed
by "rules-of-thumb." Because of the heuristic nature of Foundation engineering,
formalization of rules has not been successfully implemented (Siller 1987).
Construction, especially deep foundation construction, is an experience-based
industry, and well-defmed problems are not often encountered in the field.
Variability and uncertainty of (and lack of) subsurface data often account for the
incompletely defmed problems. DSI\2-DIAG, in its present form, is designed to deal
with some routine problems encountered in the field while using slurry in the drilling
process, a topic not well understood by many inspectors and contractors.

OPERATl"JN OF DSI\2-DIAG
This section explores the procedures required by the user to execute DSI\2-
DIAG and demonstrates its application. DSI\2-DIAG operates on an 80386 or
80486 platform. When DSI\2-DIAG is executed, the name of the expert system and
the author appears, immediately followed by a starting text, that explains how to use
the system. The system then starts asking questions to the user, who is to give only
one answer to each question being asked (Figure 2a); however, it is possible to
change answers later for sensitivity analysis to observe how the solutions may

457 Demir, Fisher, and O'Neill


change accordingly (Figure 2b). The answers to the questions are chosen from the
listed multiple choices which also sometimes include an "I do not know" answer.

Table 1. Typical Unanticipated Construction Problems

Unexpected free ground water Loss of drilling fluid


Unexpected boulder fields Deviations from alignment
Hydration of water by formation Excessive raveling of excavation
Unretractible tools Buckled casing
Smearing of the borehole Insufticiently clean base
Unproductive rate of drilling Movement of adjacent structures
Unexpected gas Hazardous or toxic waste
Excessive vibration of adjacent Incomplete penetration of concrete
areas (sensitive to humans) through rebar
Budding of reinforcing steel Damage to adjacent shafts/piles
Unretractible casings Changed surface conditions
Sloughing of geomaterial during Undocumented utility encountered
concreting
Collapse of underreams Unremovable cores
Karstic bearing stratum

Table 2. Construction Problems Involving Slurry

Collapse of borehole Loss of fluid in the borehole


Improper rheological properties of Development of excessive mudcake
slurries thickness
Excessive soil content in drilling Excessive soil content in drilling
slurry slurry
Rise in elevation of base Improperly hydrated slurry
Improper soil conditions for use of Improper slurry handling equipment
slurry
Excessi ve open time for the
borehole

According to the first question answered, sequential questions are asked of


the user automatically. This is demonstrated by a decision tree particularly in
Figures 3a, 3b and 3c. Once all the questions are answered, an ending text appears
before the recommended solutions are displayed. The recommended solutions are
accompanied by confidence values that range in a scale from 1 to 10. The higher the
value, higher the certainty of success that is perceived by the domain expert.
Obviously, the more "do not know" answers given by the user, the lower the
confidence value. Low confidence values suggest consulting a qualified human
expert. In any expert system the domain expert has developed his or her own
definition of "success." In DS"2, success is understood to mean that the indicated

458 Demir. Fisher. and O'Neill


File ,Edit Bule .Qptions ~B Files Question Fl =Help
~: _".::: '.' ." .... :. EXSYS Pro '.; ~,:::;:.:". .. ~~..,.. -::-ii; ~" ,_:.....1'.:." ...,: .:'"' ":; .

Select ONLY ONE value:


The chloride content is (the chloride content less than
5,000 ppm means treat it as fresh; otherwise, treat it as
brackish)
1: less than 5,000 ppm.

=<,'~~~'~Y;'l"':":(<:-~
~B;\1;;Sf!!SRf.~l:
I2
L _

fife ,Edit Bule Options ~B Files Question Fl =Help


EXSYS Pro
Select ONLY ONE value:

The chloride content is (the chloride content less than


5,000 ppm means treat it as fresh; otherwise, treat it as
brackish)
1: less than 5,000 ppm.

The pH level of the slurry is (the range of the pH level is 7-11 for vinyl and 8-10 for
PHPA) in range.
OOiiWmWJidm.i,ijG3rtElll

Figure 2b. Sample Screen - Input Conditions Changed

459
ATTRIBUTES
Is pH Is chloride Is sand
Type of Type of Type of level in content content
RULE
problem soil slurry range? less than less than NUMBER
* 5,OOOppm? 8%?

1
2
3
4
5
6
7
8
9
NA
10
NA
11
NA
12
NA
Collap
13
of the NA
borehole
14
NA
15
NA
16
NA
17
NA
18
NA
19
NA
20
NA
21
NA
* The range for the pH level is 7-11 for vinyl and 8-1 a for PHPA
** Chloride content less than 5,000 ppm means treat as fresh, otherwise
treat it as brackish.

Figure 3a. Sample Decision Tree for the Collapse of a Borehole

460
Rule Recommended coded
Number solutions (confidence
values in parenthesis)

1 0(8), F(6), D(6), 1(3)

2 0(8), F(5), D(6), 1(3)

3 B(7), 1(5), D(5), 1(2), F(2)

4 D(7), B(6), 1(5), 1(2), F(I)

S A(7), K(5), D(S), 1(2), F(2)

6 D(7), A(6), K(S), 1(2), F(l)

7 DO), H(6), L(5), 1(2), F(2)

8 D(8), H(6), L(5), 1(2), F(I)

9 0(8), C(5), 1(1)

10 D(8), C(5), 1(1)

11 A(7), D(6), M(4), 1(1)

12 A(7), D(6), M(4), 1(1)

13 D(9), 1(5), F(2)

14 0(9), B(6), N(4), F( 1)

15 0(9), A(6), 0(4), F(l)

16 D(9), P(3), F(l)

17 D(8), C(S), 1(4)

18 D(9), N(5), C(5)

19 D(8), 0(5), C(2)

20 D(9), 0(5), C(3)


21 D(9), E(9)

Figure 3b..Possible Coded Solutions with Confidence Factors

461
Code name Solutions

A Add sodium carbonate or commercial pH increaser to


bring the fluid pH >8 but <10.
B Add environmentally safe deflocculant and remix.

C Exchange for mineral slurry.


D Control drilling with casing.
E Contact a qualified engineer immediately.
F Weight mud with barite.

G Increase the mineral content, viscosity (getting


adequate cake formation), and

contact a qualified engineer immediately.

H Add sodium carbonate or commercial pH increaser to

bring the fluid pH >8 but <10.


Add environmentally safe deflocculant and remix.

I Take a boring immediately before a solution can be reached,

and contact a qualified engineer immediately.

J Add environmentally safe deflocculant and remix.

Increase the mineral content, viscosity (getting


adequate cake formation), and
contact a qualified engineer immediately.
K Add sodium carbonate or commercial pH increaser to
bring the fluid pH >8 but <10.
Increase the mineral content, viscosity (getting

adequate cake formation), and

contact a qualified engineer immediately.

Figure 3c. Possible Coded Solutions for the Decision Tree

462
L Add sodium carbonate or commercial pH increaser to
bring the f1uid pH >8 but <10.
Add environmentally safe deDocculant and remix.
Increase the mineral content, viscosity (getting
adequate cake formation), and
contact a qualified engineer immediately.
M Add sodi urn carbonate or commercial pH increaser to
bring the Duid pH >8 but <10.
Contact a qualified engineer immediately.
N Add environmentally safe deDocculant and remix.
Take a boring immediately before a solution can be reached,
and contact a qualified engineer immediately.
0 Add sodium carbonate or commercial pH increaser to
bring the fluid pH >8 but <10.
Take a boring immediately before a solution can be reached,
and contact a qualified engineer immediately.
p Add sodium carbonate or commercial pH increaser to
bring the fluid pH >8 but <10.
Add environmentally safe deflocculant and remix.
Take a boring immediately before a solution can be reached,
and contact a qualified engineer immediately.

Figure 3c. (continued) Possible Coded Solutions for the Decision Tree

463
solution has the greatest chance to produce an acceptable drilled shaft with minimum
effort and cost. DS"2-DIAG, as now constituted, is designed to ask questions
related to the type of problem encountered, type of soil and slurry being used, and
properties of the slurry.
DS"2-DIAG also enables the user to ask why a specific question is being
asked. In this case, the system displays the rule it is examining to check the validity
of the current problem. Further, DS"2-DIAG makes use of notes and references
that can be referred to while answering the questions. Once the system offers a
recommended set of solutions, it not only shows how it arrived at them but also
allows the user to make changes to any previously asked question(s). In addition,
after the parameters are changed in the question(s), the system will display both the
old and new set of solutions so that a comparison can be made by the user (Figure
4). Therefore, these additional applications help fonnalize the effects of variability
and unavailability of data and help young engineers to learn about and judge the
effects of different unexpected conditions that may occur on site during drilled shaft
construction.
DS"2-DIAG is easy to use, even for those who know little about computers
and computer programming. Furthermore, it does not require reference to a
comprehensive manual. All the necessary information on the system works is readily
available on help screens.
In DS"2-DIAG the domain knowledge is represented as IFITHEN rules, and
there are now 63 rules available for testing against a few slurry related problems. A
sample rule for a common scenario is sho\VTl in Figure 5. In this scenario the
contractor is drilling under a lightweight bentonite slurry in a deposit of limestone of
unkno\VTl composition but of known poor quality, as illustrated in Figure 6. Upon
reentering the borehole after discharging a load of cuttings from an auger or bucket,
the base is encountered 0.6 m above the level of the previously cut base. This
triggers a suspicion that something is wrong and becomes the "symptom" of a
problem.
Two possible reasons for this problem are: (1) sedimentation of soil (either
sand from the overburden or debris from the joints in the limestone) from the column
of slurry and (2) partial collapse of the borehole. DS"2-DIAG explores two logic
paths: (1) solution for sedimentation and (2) solution for collapse of the borehole.
In the borehole collapse branch, Figure 5, several questions are asked of the user
regarding the type of geomaterial in which the contractor is drilling. These questions
are known as attributes and are assigned values by the user according to Table 3.
In this scenario the slurry parameters are within normally accepted limits,
except for the pH, which is slightly out of range, so that the slurry probably should
not be (but may be) flocculating. The slurry may not be dense enough to retain the
sand overburden and/or the highly fractured rock, especially because the slurry head
is at the same level as the groundwaterhead, or it may not be viscous enough to
suspend the sand from the overburden being mixed with the slurry during drilling.
Considering all of these factors, DS"2-DIAG recommends in Figure 4 that the
solution in which the domain expert would have the greatest confidence would be
simply to set casing into the limestone, although other solutions such as (1) adding a

464 Demir, Fisher, and O'Neill


Continue drilling with casing.

noneW

Figure 4. Sample Screen - Results and Confidence Values


::~::;~:;;;;F;:~)):;:E~~:::.i<:::::.::: ..:'" ::';":"'.. EXSYS Professional.Editor . ;..'.' -,.,-::: :':'.':.
file ~dit Bule KB Files Question

The encountered problem is collapse of the borehole.


and The type of soil is permeable rock
and The type of slurry Is bentonite.
and The pH level of the slurry Is (the range of the pH leve/ls 7-11 for vinyl and
B10 for PHPA) out of range.
and The chloride content Is (the chloride content less than 5,000 ppm means
W' treat it as fresh; otherwise, treat It as brackish) less than 5,000 ppm.
:x'%
' THEN:
'g
:*1 Continue drilling with casing. - Confidence=9/1 0
J21 and Add sodium carbonate or commercial pH increaser to bring the fluid pH>B
:1* but <10.. Confidence=6/1 0
.~ and Add sodium carbonate or commercial pH increaserto bring the fluid pH>8
.1~ but <10. Take. a boring. sam~le bef~re a solution can be reached, and
i11! contact a qualified engineer Immediately. Confidence=4/1 0
1~\ and Weight mud With barite. - Confidence=1I1 0
:~W$~;':;':~M:::Y/,::::::Y,;,:~::::. .?".4m"f~7%:: Vh '7';:' ~ ... ..

]iii;!ilii~~:m@i;;: <!;;~.Ij~t~t;

Figure 5. Sample Screen Showing Rule.

465
pH increaser to the slurry, (2) using a pH increaser and simultaneously taking a
boring next to the shaft to discover more about the quality of the rock (with analysis
by a qualified engineer), or (3) adding a weighting agent, such as bariteTIv1 , could
also be considered, but with lower levels of confidence.

Dense Clayey Sand


3.8 m
N = 35B/O.3 m

8.2 m Soft Voggy


Limestone
Marsh Viscosity = 31 sec/qt
Unit Weight = 64 pef qu = 2.8 MPa
Chlorides = 5500 ppm RQD = 0
pH = 7.5 REC = 10%
Sand content unknown

(a) (b)

Figure 6. A common slurry problem


(a) drilling scenario (b) nearby boring log

Table 3. Attributes and their Values

Attribute Value
Type of slurry Bentonite
pH of slurry 7.5
Chlorides content of slurry 500 ppm
Value of sand content Unknown
Position of slurry head 1.5m
Unit Weight of slurry Mpcf
Marsh viscosity of slurry 31 sec/qr.

The rationale for choosing the setting of casing over the other three alternate
solutions is that it will solve either the problem of collapse of the sidewall in the rock
or the overlying sand, and, even if no sidewall collapse is occurring, and sand is
sedimenting from the slurry, sedimentation will likely soon stop when the sand
overburden is cased off. Alternate solutions (l) and (2) are specific to sedimentation
and are not likely to work if the source of the problem is sidewall collapse. Alternate
(2) is attractive because it suggests a didactic approach, but if sidewall collapse is
occurring, time is of the essence, and a solution must be found quickly; hence, the
relatively low confidence value. Alternate (3) is likely to reduce the propensity for
sidewall collapse, but it is also likely to be time consuming.

466 Demir, Fisher, and O'Neill


CONCLUSIONS
Expert systems for decision support in foundation construction have not yet
been developed sufficiently to have had a signiticant impact on practice. A major
reason for this condition is that expert systems are most efficiently applied to routine
construction scenarios, which are less common in foundation construction than in
other fonns of construction. The occurrence of non-repetitive, anticipated or
unanticipated subsurface conditions and their potential effect on both the
construction process and the integrity of the completed drilled shaft has a
particularly strong impact on decisions made regarding drilled shaft construction,
which are traditionally based on the experiences of the field personnel of the
contractor in consultation with field representatives of the owner. Nonetheless,
expert systems for drilled shaft construction may have a place in the office and field
in the future in order to
assist in training new field personnel through simulation of solutions to problems,
assist in developing rapid consensus between owner and contractor by
developing recommendations quickly and in an impartial manner,
assist in making construction decisions for those cases that are routine but for
which a particular field crew has inadequate experience, and
point out to field personnel when a panicular situation is not routine and direct
them to expert human assistance.
A decision support system, DS"2, has been described briet1y in this paper. It
provides the user consultation in (1) determining the general construction method to
be applied to a given situation (DS"2-GEO), (2) specifications for construction
operations (DS"2-SPEC), (3) cost estimation (DS"2-COST), (4) assessment of
automation technologies (DS"2-SIM) and (5) tield diagnostics (DS"2-DIAG). All of
the above modules are still in a state of development, although DS"2-GEO has been
validated by several experts. The remaining modules have not been validated, and
the last module (DS"2-DIAG), which is the principal subject of this paper, is in the
very early stages of development. DS"2-DIAG addresses at present a few routine
problems that can be encountered during construction with drilling slurry and
presents solutions with associated confidence values. Low confidence values for the
best solutions are indicators that human assistance should be sought. An example of
a situation in which high confidence values are unlikely would be where both soil
and rock are found to cover the base of a borehole. This situation would involve
complex diagnostics requiring human assistance. The research team plans to
continue to expand and update all of DS"2 modules in the future and to obtain
veritication of the ones that are not yet verified. With .diligent updating, 05"2 in
general and DS"2-0IAG in particular should, in time, become very useful decision
aids in drilled shaft construction. The entire decision support system is portable and
is intended for use in the field.

REFERENCES
Abaya, E. L., O'Neill, M. W., and Fisher, D. J. (1993). "An overview of expert
systems for drilled shaft construction." Transportation Research Record 1406,

467 Demir, Fisher, and O'Neill


Transportation Research Board/Construction Robotics, Automations and
Foundations Engineering, Washington, D. c., 31-34.
Baker, C. N., Jr., Parikh, G., Briaud J-L, Drumright, E. E., and Mensah, F. (1993).
"Drilled Shafts for Bridge Foundations" Publication No. FHWA-RD-92-004,
Federal Highway Administration, August.
Demir, S. (1993). Diagnostic expert system for drilled shaft foundation
construction. Senior Honors Thesis, University of Houston, Houston, TX.
Dumas, C. (1994). "Lateral rotational stiffness of highway bridges." Transportation
Research Board Committee A2K03 Report, Washington, D.C., January.
Elton, D. 1., and Brown, D. A. (1991). "Expert system for driven pile selection."
Proc. Geotechnical Engrg. Congr. 1991, I, ASCE, McLean, F. G., et. al., eds,
253-263.
EXSYS PROFESSIONAL: Advanced Expert System Development Software User's
Manual (1988), EXSYS Inc., Albuquerque, New Mexico.
Finn, G. A., and Reinschmidt, K. F. (1986). "Expert systems in an engineering-
construction fInn." Proc. First Symp. on Expert Systems in Civil Engrg., ASCE,
Seattle, W A, 40-54.
Fisher, D. J., O'Neill, M. W., and Contreras, J. C. (1994). "DS"2: Drilled shaft
decision support system." J. Constr. Engrg. and Mgmt., ASCE, to appear.
luang, C. H., and Ulshafer, M. L. (1990). "Development of an expert system for
preliminary selection of pile foundation." Transportation Research Record 1277,
Transportation Research BoardlNational Research Council, Washington, D.C.,
153-160.
Maher, M. L. (1987). "Expert systems in structural engineering." Expert systems for
civil engineers: Technology and application, ASCE, 49-76.
Mohan, S. (1990). "Expert systems application in construction management and
engineering." 1. Constr. Engrg. and Mgmt., ASCE, 116(1),87-99.
O'Neill, M. W. (1991). "Construction practices and defects in drilled shafts."
Transportation Research Record 1331, Transportation Research Board/National
Research Council, Washington, D.C., 6-14.
Philip, G. c., and Schultz, H. K. (1990). "What's happening with expert systems?"
AI Expert, November, 57-59.
Rho, 1., Fisher, D. J., and O'Neill, M. W. (1993). "SIMBASE: A decision support
system for economic justifIcation of automated construction technology." Tenth
International Symp on Automation and Robotics in Constr., May 24-26,
Houston, TX, 141-148.
Siller, T. J. (1987). "Expert systems in geotechnical engineering." Expert systems for
civil engineers: Technology and application, ASCE, 77-84.
Trantman, C. H., and Kulhawy, F. H. (1987). "TLWorkstation Code: Version
1.1, Volume 1b: CUFAD." Electric Power Research Institute, Palo Alto, CA.
Yeh, Y. c., Hsu, D. S., and Kuo, Y. H. (1991). "Expert system for diagnosing
damage of prestressed concrete piles." 1. Constr. Engrg. and Mgmt., ASCE,
117(1), 13.

468 Demir, Fisher, and O'Neill


A PROTOTYPE EXPERT SYSTEM FOR FOUNDATION DESIGN
Paul N. Roschke 1, Jean-Louis Briaud z, and Erik G. FunegardJ

ABSTRACT
A knowledge-based expert system is being developed to aid in the design of
foundation structures. The prototype. version of this system is intended to
demonstrate important advantages of a design-oriented rule-network that
automatically considers numerous types of foundations. Graphical representation of
salient quantities is emphasized. The system runs on a microcomputer and is
compatible with Microsoft Windows. Several existing FORTRAN codes are called
upon during the reasoning process to calculate settlement of piles and shallow
foundations.

INTRODUCTION
The design of a foundation system is a relatively complex undertaking that
requires consideration of numerous alternatives. However, constraints of time and
resources often limit the search and initial design to a single type of foundation, such
as piles. The goal of this project is to develop a computer program, mnemonically
titled Foundation Expert System (FES), that automates the selection and design of
the most effecient foundation system. Although completion of the project will
require several years of additional work, the current prototype illustrates the complex
nature of such an undertaking and yet demonstrates the usefulness of the final
product.

Some of the requirements for FES are to ensure that spread footings are
always considered as a solution, to incorporate the knowledge, of .practicing
foundation engineers, to develop features allowing interaction with borehole data
bases including graphics, and to integrate the capabilities of existing analysis codes

I Associate Professor, Department of Civil Engineering, Texas A&M University; College station,
TX 77843-3136.
2professor, Department of Civil Engineering, Texas A&M University, College Station, TX 77843-
3136.
3Senior Research Engineer, Geotechnical Consultant, Amoco Corporation; P.O. Box 400,
Naperville, IL 60566.

P. N. Roschke, J. Briaud, E. G. Funegard


469
into the design process. Such a program will provide a more uniform way of
designing foundations and allow less experienced personnel to arrive at a reasonable
design with reliability.

The following describes the system and two examples demonstrate its
operation. Target hardware for the prototype version of FES is the Intel 80386-
80486 and Pentium family of processors.

SYSTEM DESCRIPTION

C CODE AND GENERAL CONCEPTS


FES consists of suite of executable codes and supporting data files that run
under Microsoft Windows (1992). A driver program (FES.EXE) written in the C++
programming language handles user interaction and "drives" the expert system by
invoking other processes (see Fig. 1). FES.EXE uses the Commonview application
framework that enhances portability to other graphical environments such as
PresC?ntation Manager. A series of pull-down menus and screens allows the user to
specify foundation types, location, soil data, and various constraints. This
information is saved in a data file that can be reloaded at any time.

Borehole Database

FES.EXE

FES data file

THINK.EXE

NEXPERT

RULE BASE

FIG. 1. Flow of Program Control

Soil data is specified by selection of a borehole with a pointing device or use


of the keyboard. A project map, including boreholes, is displayed on the screen. The
user may examine borehole data to determine which borehole is appropriate. The
borehole data is discussed further in the following section.
470
After basic data for the problem has been entered, the "Think" option is
selected from the main menu. This sets into motion the inference process of the
expert system. The first step in this process is execution of a separate program,
THINK.EXE, that has been written in the C programming language. This program
was written because of incompatibilities between Commonview and Nexpert Object.
It reads the FES data file and suggests data to the inference engine of Nexpert
Object. For purposes of modularity, the problem of foundation selection is broken
down into several steps of elimination. A different knowledge base is loaded for each
step. For each knowledge base, data is "volunteered," i.e. passed to the inference
engine from the executable C code. During these steps, borehole data is read from
the borehole database (see Fig. 1). Using this borehole data and data from the FES
data file, additional input data files are built in preparation for execution by auxilIary
analysis programs.

Two analysis programs are currently supported: SCHMERT (Tucker and


Briaud 1990b) and COYLE (Tucker and Briaud 1990a). These programs support
the analysis of spread footings and piles, respectively. SCHMERT and, sometimes,
COYLE are run by THINK.EXE and their output files are analyzed to determine the
feasibility of a given foundation.

After the "Think" step has determined the optimum foundation, the C code
generates a report and summary. These reports can be viewed by using the Report
menu that is accessible through the main menu in FES.

DATABASE OF EXTERNAL BORINGS


The borehole database currently exists as a set of ASCII files on a PC. Data
is downloaded from GTBASE (1988) which is a FOCUS database existing on an
IBM mainframe. There are two basic types of files: a borehole file and an elevation
file. The borehole file contains information such as location, type, and depth for each
borehole. The elevation file contains profile data for boreholes at the site. There are
seven types of elevation files that correspond to the seven types of boreholes.
Therefore, a search of the borehole file yields the type of borehole along with other
pertinent data. Using the borehole type, the code searches the appropriate elevation
file to determine soil data.

KNOWLEDGE REPRESENTATION

General
Development of FES centers around the inference engine ofNexpert Object.
In addition to various kinds of inference mechanisms (forward and backward
chaining, action-based propagation, strong and weak links, etc.), object oriented
programming is extensively implemented in FES. Nexpert Object also supports the
graphical environment of Microsoft Windows. Pop-up and pull-down menus
complement display of rule and object networks. Finally, FES executes codes that

471
are external to Nexpert Object. This is especially useful for integration of routines
from a number of existing FORTRAN and BASIC codes that analyze soil-structure
interaction. As described earlier, two external routines, SCITh1ERT and COYLE,
have been integrated into the prototype system.
Object Network
Nexpert Object uses objects and rules to represent knowledge of foundation
design. Objects describe the environment upon which the reasoning is perfonned. In
FES objects are given a name and divided into classes and subobjects. Fig. 2 shows a
portion of the taxonomy developed for foundation structures in FES. Classes and
subclasses of objects are marked with circular icons. For example, "Shallow" and
"Deep" are subclasses of the main class "Foundation_Types." The class "Piles" is
further subdivided into subclasses "Bored" and "Driven," and so on. Objects
themselves, such as an "H" or "Pipe" pile, are marked with a triangular icon.

Classes and objects are given properties such as cost, dimensions, etc. These
properties, or slots, are inheritable for all instances of a class. For example, the class
of steel piles might be assigned a property value of "poor" for corrosion resistance.
Each of the objects in the steel class, including H and pipe piles would also inherit
this property. Properties of classes and objects are indicated by a square icon in Fig.
2.

OFoundation_Types'---<:-

_Dimension

FIG. 2. Object Network of Foundation Types

472
Figs. 3 and 4 show taxonomies of superstructure loads and substructure
environment, respectively. Although these sets of objects are not currently integrated
into the rule network and logic used in the prototype, their classification is a useful
step toward development of a more complete expert system. A total list of all classes
and objects associated with the four knowledge bases used in FES can be obtained
from use of the Nexpert Object development version.
Rule Network
Reasoning about classes and objects takes place through rules. For the FES
prototype, four sets of networked rules are used. Fig. 5 shows an example of one of
the rule networks, SPREADEL, that determines whether or not spread footings
should receive additional consideration as a candidate for the foundation. This
network contains rules that disallow use of spread footings for a specific site,
structure, and subsurface conditions that are not consistent with their use. As an
example, if the soil type is sand and the blow count is less than 10, a spread footing
should not be considered. This step serves to quickly eliminate from consideration
foundation types that should not be pursued with the external design and analysis
codes. If the network tacitly approves consideration of a spread footing design, the
external analysis code for spread footings, SCHMERT, is called by the C code to
perfonn necessary calculations leading to a recommended design. A second rule
network, SPREADEV, evaluates the design of a spread footing after analysis by the
external routine SCHMERT. It checks dimensions of the recommended design to
see whether or not they exceed geometric limits due to column spacing of the
structure to be supported. In an analogous manner, PILEEL and PILEEV eliminate
and evaluate, respectively, pile types and pile designs suggested by the external
analysis code COYLE. All four rule networks operate on the same object network.

6 Combined
~-6Downward_Only
6 HorizontaJ
aJue

~
rectionof Load
OSuperstructure_ Load - -
Load_TIme_ History-=---~:------6 Static

aJue

FIG. 3. Object Network for Superstructure Loads

Timing for use of these networks is controlled by the main C code. In


general, rules for spread footing elimination are checked first. If a spread footing is
not eliminated as a candidate, the external design is carried out with SCHMERT.
The recommended design is checked using rules of thumb in the knowledge base. If
the spread footing design is successful, a report is generated and deep foundation

473
designs are not considered. If the shallow foundation design is not feasible, a similar
procedure is carried out for piles.

FIG. 4. Object Network for Substructure Environment

eck SF2,
- ,
eck SF3-_',
- -'l>-
Soll.type Is 'Sand~' Check SF4-~-:;:/ - ' ,
- / ' "
. . Soil.blow count < 1 .5------~---Check_SF5" /

= > Do FALSE Spreadfooting_Ok


Soil.type Is 'ClaY~'
IDirection _01_ Load I Is 'HOriZontal'~ . Soil.Su < 1. .1-
.4
= > Do FALSE Spreadfooting_Ok FALSE Spreadfooting_Ok

Soil.type Is 'Oay"

= >Do SQRT(Verllcal_Column_Loa ~.

Soil.type Is 'Sand

=>00 SQRT(Verlical_Column_Loa ~

FIG, 5. Rule Network for Elimination of Spread Footings

EXTERNAL ROUTINES
The purpose of SCHMERT and COYLE is to calculate the dimensions of the
foundation required to satisfy a given factor of safety against soil failure and to
ensure that the settlement will be within tolerable limits. Numerous other routines
exist to achieve this purpose depending on the soil type, the type of soil test
perfonned, the loading type, and the type of foundation element considered. The
final version of FES is likely to include a dozen or more of these routines.
474
SCHMERT is a program written in the BASIC language for analysis of
spread footings on sand. It uses the results of cone penetrometer tests (CPT) or
standard penetration tests (SPT) and analyzes bearing capacity and settlement.
Bearing capacity is handled by using Terzaghi's bearing capacity equation. The
bearing capacity factors are obtained by correlation with the cone penetration
resistance, qc, as recommended by Schmertmann. SPT data can also be used by
correlation with qc' For this prototype, a ratio of qcCtsf) / N(lb/ft) was assumed to be
equal to 3.5 in all cases. The settlement calculations are carried out according to
Schmertmann's method based on the cone penetrometer test or the standard
penetration test.

COYLE is a program written in BASIC that performs calculations for piles in


sands, clays, or layered soils. It is based on the method developed by Coyle. It uses
results of standard penetration tests in sands and the undrained shear strength of clays
typically obtained in unconfined compression tests. It considers both the ultimate
capacity of the pile and the settlement of the pile by generating a load settlement
curve.

SYSTEM OPERATION
FES runs under Microsoft Windows. A mouse pointing device and a hard
disk are assumed to be available. An icon with the name AMOCO FES appears in
the Windows Applications folder (see Fig. 6). Before starting the advisory program,
relevant data about the soil, structure, construction technique, and environment need
to be known or estimated. After running the advisory program results can be
displayed on the screen, saved to a file, or printed. Data for a given case can also be
saved for later retrieval. The following steps outline how to accomplish these tasks:

if 1.11 Ii.! ,/ 't.E i'N .i 4


Ell. QptlOR9 ~Rdow

Fole Manager Control Panel Pr.... Manage< Dl:board DOS Prompt Wmows Selup 'YS.. di1


hlSWOAD

~
e-Tech
~
SO KPainl >

~
PowerParoI
~:iii";:i'[:';:

!:!i:
Bf....

iii
eIMo',_
:J@ ~ ~
CiviE-, NEXPEAT E>oeeI

FIG. 6. Microsoft Windows with Icon of FES

475
1. If FES is correctly installed, it can be run from Microsoft Windows by
opening the Windows Application group. Then, Click twice on the AMOCO PES
icon or select the icon and choose Open from the File menu of the Program Manager
window (Fig. 6).
2. Select New from the File menu of FES (Fig. 7) After the input data has
been entered by means of the Structure, Soil, and Constraints menus this infonnation
can be saved to a disk using the Save or Save As options. As an alternative, existing
files are called into memory using the standard Windows Open option.
3. Select Plant from the Structure menu. A predefined input data fonn,
called a dialog box, appears on the screen (Fig. 8). Note that when any of the dialog
boxes are open, all other FES functions are disabled. Hence, only one dialog box can
be open at a time. When all questions have been answered, the dialog box is closed
by moving the mouse to the OK button and clicking once on the left mouse button.
This saves the entered data into memory. The Cancel button can also be used to
close a dialog box, but this selection results in losing any changes made to the data.
Answer the questions in the fonn by selecting and entering the relevant data for the
bridge site under consideration. When these selections are complete, choose OK to
save the data, or choose Cancel to abort the input process.

- .. :
SQiI !;onslralnls Ihlnk! Beport Qplions

Qpen ...
~ave
Save As ...
A!!oul
qi1

FIG. 7. FES Menu to Open a New File

nii' eW.pe 9.1 :


Ble ~lructure Soli Constraints Think! Report Qpl/ons

.,
Cedar Bayou
Cooper RIver

f--
I Add:1
Chol:. Bayou
Decatur
IDelete ~
Gee. Belgium
Texas City
Ulsan Korea
~
Fertrln Ammonia
loolandan 7" ICanCel I

FIG. 8. Plant Selection Menu

476
Rapid double-clicking on the name of a plant causes a map of the plant site to
appear (Fig. 9). Borehole locations are indicated with circles. Placing the mouse
pointer on one of the boreholes and double-clicking causes a table of data values for
this particular borehole to appear (Fig. 10). A close-up view of a rectangular region
of the plant site can be brought into view by selecting the Zoom option from the
Option menu and specifying two opposite comers of a rectangle with the left mouse
button (Fig. 11).
4. In any desired order select Columns, Walls, Mats, and Mise from the
Structure menu and enter case data in the same manner as for step 3. Selection of
Columns leads to a table of column coordinate locations being displayed. To enter or
modify column loads select the Edit option. This leads to the options listed in Fig.
12. Enter the desired values and select the Update button followed by the Cancel
button. Data concerning life expectancy and settlement is entered with the Mise
option.
5. Enter data for the Soil and Constraints menus (Figs. 13-15).
6. When finished entering or editing the case data select the Think! menu.
Several intermediate screens are displayed that serve to inform the user of progress
during the reasoning process (Fig. 16).
7. After execution, view the results (Fig. 17) by choosing the Report menu.
Use the scroll bar at the right hand side to proceed through the document.
8. To save the case data, choose Save Results from the File menu.
9. To print the results, choose Print Results from the Report menu.

-.- -'!f'_ --r.'r


_, '_L~ ~ I" s~

File Structure Soil Constrslnts Think! Report Qplions

ra .t'r""3~.! ".;;~" "" ,>,.. ... ~ ~. ~JJ..~>i: ..::.""~"'l::,,"~~~ pO- :-',~",,<;,~\",::.;.;,::.~. ~"":a

e_ ~
Q,
..

- Oi
e
f-

~ .. f-

. 'II
SJ@
e CJ
0

+ ......

FIG. 9. Plant Site Plan with Boreholes

477
3":".1=[
File Structure Soli Constraints Thlnld Beport .QpUons

Number: B7A Type: ST Loe: 1925.0 .1432.0


Depth USCS Dese PL LL WC SU TEST
1.00 FILL SA CL
2.00 FILL SA CL
5.00 FILL SA CL 40
7.50 SA CL
9.00 SA CL 22 .50 U 103

FIG. 10. Table of Borehole Properties versus Depth

File Structure Sgll Conslralnls Thlnld Report .Qpllons

a ',' " /. .. ".. '1.>';..;>::" ..' ':< ,.,,:<,':':.11]


~

=
- D =
0=:1
0=:1


I I .

I

FIG. 11. Zoom In on Map of Site Plan

478
Ric Slructure Soli Conslrainls Think! R.eport Qplions

, ,,
Iype-
-:-;
T
C @.!;.olumn
X I 600.00 I II. Y I BOO.OO I II. ....;

o Wall ><21 603.00 I fl.


o Mat Y21 803.00 I fl.
loads
~
Horizonlal I 0.00 IlbS' f-

Vertical 1900000.00 Ilbs. o Earthquake


Momenl 0.00 IlbS Ollibradon ,
I l-

- I Add, I Io.elele 1I !!pdale ,I Icaneell
~

FIG. 12.. Menu for Structural Components

""SEEtHE"
file l;ilructure SRII .!;.onslralnls Ihlnk! R.eport Qpllons

Number Type x y Depth


Bl AUGER -1025.0 0.0 45.0 +;
I-
1632. a
B2 AUGER
BCPTI CPT
-725.0
725.0 2632.0
45.0
45.0
t-' I Add I
BCPT2 OlliER 1725.0 3632.0 45.0
83
84
OlliER
PMT
225.0
2725.0
4632 .0
-632.0
45.0
45.0
I ,E:d,i,t I
I~eletel
'"
BMPTI PMT 0.0 0.0 45.0
BMPT2 SPT 3425.0 1632.0 45.0
85 SPT 4325.0 32.0 45.0
I I I
B7 ST -525.0 3251.0 6.0
B7A ST 1925.0 1432. a 13.0 ~ IYiew I
[CGl1cell

FIG. 13. Table of Boreholes for a Plant

479
_ 5
~I.:I
Rle Structure Soli Conslralnls Ihlnk! Report Qpllons

ConstrDinls

DAre slrUclures sensltlYt: to vlbrlldon nearby?

o Is the environment corroslYt:?

FIG. 14. Dialog Box for SoU Constraints

Elle Slructure Soli Constrslnls Ihlnk! Report Qpllons

Candidates

Choose foundation candidates:

I8l Wood Piles

I8l Or/lied Shalts I8l ~oncrele Piles

I8l S\eel Plies t8l Mst

FIG. 15. Foundation Candidates for Design

480
file Slructure SQiI ~onslraints Ihinkl Report Qplions

Spread looting analysis complete

FIG. 16. Message to User During Execution of Knowledge Base

File Structure Soil Conslrainls Think! Report Qptions


SHAlLOW FOUNDATIONS CAN BE USED
h
f-
The size 01 the spread footing under lhe column should
be 17.56 feel wide and 17.56 leellong. The base ollhe laDling
should be placed at a depth of 5.0 feel Under a column lod
of 900.0 kips the anticipated senlement a"er a period 01 20
years Is 0.9& Inches.

This design Is based on a bearing capacity analysis using a factor


of safety 01 3 and on a settlement analysis using an allowable
senlement of 1.0 Inches.

+
I:c)t( I
-. 7

FIG. 17. Report of Recommendations for Spread Footing

EXAMPLE 1: SHALLOW FOUNDATION RECOMMENDED


A concise list of input values for an example problem that leads to a shallow
foundation is given in the following subsections. Data is arranged according to menu
options.
File
EXAMPLEl.DAT
Structure
Plants
Capco, Taiwan

481
Columns
Column position: x = 600.0 y = 800.0
Vertical Load (lb): 900,000 (4.0MN)
No earthquake zone
No vibration loads
Walls
None
Mats
None
Mise
Life expectancy (yr): 20
Allowable settlement (in.): 1.00 (25.4 mm)
Soil
Borehole
BI AUGER -1025 0.0 45.0
Constraints
No underground structures are present
Structures are not sensitive to vibration nearby
Environment is not corrosive
Candidates: Spread footing, wood pile, drilled shaft, concrete
pile, steel pile, mat

After the input data has been entered and Think! has been selected from the
main menu, recommendations of the inference engine can be viewed by means of the
Report menu options. In this case, the vertical load on a column of 900,000 lb (4.0
MN) can be supported by a square footing that is 17.56 fF (1.63 m2) (Fig. 17). The
analysis program SCHMERT predicts a settlement of 0.96 in. (24.4 mm) after a
period of 20 years. Mats and deep foundations are not considered in this case since a
shallow foundation is determined to be adequate. It might be noted that a load
greater than 1,180,000 lb (5.25 MN) would result in a spread footing being rejected
as a foundation candidate due to a rule related to assumed spacing of multiple
columns.

EXAMPLE 2: DEEP FOUNDATION RECOMMENDED


This example problem uses the same input file data as the previous example
except that the load is increased from 900 kips (4.0 MN) to 1,300 kips (5.78 MN).
With this data the expert system first attempts a shallow foundation design. An
analysis is carried out with SCHMERT. Dimensions of the footing required from this
analysis are too large for the available column spacing. Instead, the advisory system
initiates a pile analysis with COYLE, recommends a group of four 2-ft (0.61-m)
diameter piles, and gives their dimensions and recommended materials (Fig. 18).
Details of the input data fields are identical to those of Example 1 except for the
magnitude of the vertical load that is to be supported. Also the input data file may be
saved using the name EXAMPLE2.DAT.

482
File
A
Structure Soli
'iMiBW'+'
Conslralnts Think! Report 2Ptlons

I .
DEEP FOUNDATIONS ARE NECESSARY

Shallow foundations were considered and could not be used


.'
r--
r--
without soil impr"""menL Therefore deep loundations are necessary.

Piles driven to a depth of 55.0 fL below ground level are necessary


to resist the column load of 1300.0 kips. Four plies should be used
under each column and should have a minimum spacing of 3 pile
diameters center to center. The plies should be 2.00 fL In dlame1er.

..
I

I :OK"I
FIG. 18. Report of Recommendations for Piles

CONCLUSION
The development of a foundation expert system called FES has been initiated.
The program resulting from this initial effort is considered to be a demonstration
version of the final product. The framework of this demonstration version of FES
includes a variety of options that appear on numerous screens. Only a few options
have been developed to the point where a foundation recommendation can be
obtained. As it stands, the current version of FES demonstrates the concepts
involved in solving a simple design problem for foundations on sand. It selects either
spread footings or piles. This demonstration version also displays the many options
that can be developed further. Indeed the framework of the final version is well
advanced. Phase 2 of the project will focus on completing the process for foundations
subjected to vertical static loading on flat ground.

REFERENCES
GTBASE User's Manual. (1988). Geotechnical Borehole Database, Amoco Research
Center, Naperville, IL.
Microsoft Windows User's Guide for the Windows Graphical Environment, Version
3.1. (1992). Microsoft Corporation, Redmond, WA.
Nexpert Object User Manual, Version 1.1. (1989). Neuron Data, Inc., Palo Alto,
CA.
Tucker, L. M.~ and Briaud, J.-L. (1990). User's Manual for Coyle, Department of
Civil Engineering, Texas A&M University, College Station, TX.
Tucker, L. M., and Briaud, J.-L. (1990). User's Manual/or Schmert, Department of
Civil Engineering, Texas A&M University, College Station, TX.

483
A Numerical Solution for the Dynamic
Pile Driving Problem

S. M. Mamoon 1

Abstract

A numerical solution is presented for the vertical vibration of a single pile


under impact loads. The formulation is presented in the time-domain, using the
Boundary Element Method (BEM) based solution scheme. The soil displacement
equations are obtained using an efficient step-by-step time integration scheme.
Treating piles as one-dimensional structures, solutions are obtained by discretizing
the differential equations of the motion using finite difference approximations. The
pile-soil systems are then coupled through the equilibrium and compatibility
relations at the pile-soil interface. The displacement and force time responses
along the pile during the driving process are presented. By comparing the results
with those obtained by other methods, the stability and accuracy of the presented
method are established.

Introduction

The procedures for the frequency domain dynamic analysis of pile


foundations are well developed (Mamoon, 1990, Mamoon et. al. 1990). The
process of pile driving however, is a transient phenomenon. Transient analysis
of pile foundations has received very little attention, although many dynamic or
seismic responses involve transient motions. Nogami and Konagai (1986, 1988)
were the first to present approximate procedures for axial dynamic analyses of
piles. Adopting a Winkler's hypothesis, the soil response to the pile motion was
formulated through a simplified mechanical model based on Winkler's assumption,
the parameters of which were determined from a consideration of plane strain
wave propagation.

I Senior Engineer, Bechtel Corporation, 50 Beale Street, San Francisco, CA 94119

484 S. Mamoon
In this paper, a very efficient but approximate hybrid boundary element
solution procedure is presented to analyze the transient response of a pile during
the driving process (Figure I). The piles are represented by compressible one-
dimensional elements, and the soil is represented as an elastic half-space. An
efficient step-by-step time integration scheme (Banerjee and Ahmad 1985) is
implemented using an approximate half-space integral formulation to get the soil
displacement equations. The governing differential equations for the piles are
discretized by finite difference expressions, taking into account the appropriate
pile-head and pile-base boundary conditions. These solutions are then coupled at
each time step with the system equations for the soil domain by satisfying
equilibrium and compatibility at the pile-soil interface.

This analysis procedure allows computations of the general response of


piles subjected to impact forces, at greatly reduced computational cost. The
analysis here is restricted to floating piles, as opposed to end-bearing piles, since
in many practical situations resistance to the impact load is derived mainly from
the adhesion developed at the pile-soil interface.

Analytical Fonnulation

The hybrid boundary element fonnulation used here involves the


construction of a simplified boundary integral representation for the soil domain,
which is coupled with the pile equations represented by linear structural
components.

Soil Equations

For dynamic loading of a homogeneous, isotropic, elastic solid, the


governing differential equation in tenns of the displacements can be expressed as

(1)

where A and 11 = the Lame's constants; p = the mass density of the solid; uq =
B2uJBt 2 = the accelerations; and the subscripts p and q range from 1 to 3.

Since the solid here is of semi-infmite extent, the equation is cast in tenns
of the Green function for the half-space, which reduces the .domain of integration
to the pile-soil interface only. The integral equation then can be written as

up(~. t) = Is LGpq(x, t; ~, ")cPq(x, ,.)d,. ds (2)

where up = the displacements of the soil; G pq = the Green function for the half
space; <1>q = the tractions at the pile-soil interface; and S and x = the spatial
positions of the receiver and source, respectively. This equation represents a
fonnulation involving integration over the surface S, as well as the time history

485
t. To obtain the transient response at time tj , the time axis is discretized into j
equal time intervals, i.e., tj = j .~t, where ~t = the time step. Therefore Equation
2 can be written as

up(E. tJ - ('i
JT"",,-I
1s
Gpq 4><{ds d-r = (/j-I ( Gpq 4>"ds d-r
JT=U Js (3)

where the integral on the right-hand side is the contribution due to the past
dynamic history. By dividing the pile-shaft-soil interface into n number of
cylindrical segments, considering the base as a circular disc (Figure 2), and taking
x successively at the centroidal point of each element on the boundary, the
interface displacements at time step j are obtained from

(4)

The actual distribution of surface tractions are replaced by piece-wise


constant distributions over each segment (Figure 3). The soil displacements at
time step j then becomes
j- I

Vi = GI~ + 2: G
;=)
j- i+ I<IY; = GI~ + Rj (5)

where G 1 = the coefficient calculated at the first time step; Rj S = the effects of
past time histories on current node. Superscript s denotes that the displacements
are obtained from a consideration of the soil domain. Details of the fonnulation
can be found in Mamoon (1990) and Banerjee and Ahmad (1985).

Fundamental Solution

The soil flexibility matrix G in previous equations represent the transient


equivalent of Mindlin's (1936) static half-space point force solution. In this work,
an approximate fundamental solution was obtained by superimposing two Stoke's
infinite space fundamental solutions (Bringen and Suhubi, 1975). Details of the
fonnulation can be found in Mamoon (1990) and Mamoon and Banerjee (1992).

Pile Equations: Driving Response

The impact loading time history is digitized at each time increment ~t


(Figure 4). The axial pile response at time tj is governed by the equation

(6)

where m = mass per unit length of the pile; uj Z = axial pile displacement at time
tj ; Ep and A p = Young's modulus and cross-sectional area of the pile; d = pile
diameter; and <pj Z = axial traction on the pile at tj (Figure 2). Employing finite
difference expressions for time and space derivatives, and imposing forced
boundary conditions at the top of the pile, the pile equation for the jth time step
becomes

486
Vj = O74c1f + Cj (7)

Further details of the formulation can be found in Mamoon (1990) and Mamoon
and Banerjee (1992).

Assembly and Solution

Equation (7) can be rewritten as

Vf = ~ + cr (8)

where superscript p denotes that the displacements are obtained from a


consideration of the pile domain. Satisfying equilibrium and compatibility
conditions at the pile soil interface, the tractions acting on the piles at the jth time
step are obtained from (5) and (8) as

(9)

The pile displacements at the jth time step are obtained from (8). The propagated
impact load at any depth Z in the pile at time step j are determined by integrating
the tractions along the pile and deducting the inertia components. Further details
can be found in Mamoon (1990).

Comparison and Examples

To demonstrate the accuracy and applicability of the developed formulation


(called method I), examples are presented. The accuracy of the developed
technique are compared with the solutions provided in Nogami and Konagai
(1986, 1988), and using a generalized axisymmetric boundary element formulation
presented in Wang and Banerjee (1990) (method II). Details of method II are
given in Mamoon and Banerjee (1992).

The axial pile head displacement responses obtained by method I and II


are compared with Nogami and Konagai's results (1986) in Figure 5. The
material properties chosen are: Young's modulus of piles Ep.= 2.488 X 106 ,
kN/m 2, Young's modulus of soil E s = 20,000 kN/m 2, (Ep/Es = 124.4), density of
pile material PP = 2.7013 kN.s 2/m 4 , density of soil material Ps = 1.0194 kN.s 2/m4,
(p~Ps = 2.65), and soil Poisson's ratio Us = 0.4. The length L of the pile is 75
m, and it's diameter d is taken as 2 m (LId = 37.5). The time history of impact
load P(t) applied at the pile head is shown in Figure 6. The time step (~t) is
chosen as 0.000975 secs. In general, in Figure 5 there is good agreement between
the three methods. At initial times, however, Nogami and Konagai's results differ
significantly from those of methods I and II. The oscillations at later times in
methods I and II may be attributed to the various wave interferences which are
apparently ignored in Nogami and Konagai's formulations.
487
The displacement and impact forces computed by method I along the pile
shaft at various times are shown in Figures 7 and 8, respectively. These figures
show the propagation of longitudinal waves along the pile.

Conclusions

A time-domain hybrid boundary element formulation is presented for the


analysis of piles under impact loads. By comparing the results with those
obtained by other methods, the stability and accuracy of the presented method are
established. This approach is the most accurate analysis currently available for
this problem, primarily because of its ability to take into account the three-
dimensional effects of soil continuity and boundaries at infinity.

References

Banerjee, P.K. and Ahmad, S. "Advanced Three-dimensional Dynamic


Analysis of Boundary Elements," Proc. ASME Conf. Advanced Boundary Element
Analysis, AMD-72. T.A Cruse, A Pifko, and H. Armen eds., ASME, New York,
NY, 65-81, 1985.

Eringen, AC. and Suhubi, B.S. "Elastodynamics," Academic Press, New


York, NY, 1975.

Mamoon, S.M. "Dynamic and Seismic Behavior of Deep Foundations," A


Dissertation Submitted to the Faculty of the Graduate School of State University
of New York at Buffalo, in Partial Fulfillment of the Requirements for the Degree
of Doctor of Philosophy, 1990.

Mamoon, S.M. and Banerjee, P.K. "Time-Domain Analysis of Dynamically


Loaded Single Piles," J. Engrg. Mech. Div., ASCE, Vol. 118, No.1, P. 140-160,
Jan. 1992.

Mamoon, S.M., Kaynia, AM. and Banerjee, P.K. "On Frequency Domain
Dynamic Analysis of Piles and Pile groups," J. Engrg. Mech., ASCE, Vol. 116,
No. 10, P. 2237-2257, 1990.

Mindlin, R.D. "Force at a Point in the Interior of a Semi-infinite Solid,"


Physics, Vol. 7, P. 195-202, 1936.

Nogami, T. and Konagai, K. "Time Domain Axial Response of


Dynamically Loaded Single Piles," 1. Engrg. Mech., ASCE, Vol. 112, No. 11, P.
1241-1252, 1986.

Wang, H.C. and Banerjee, P.K. "Axisymmetric Transient Elastodynamic


analysis by BEM," Int. J. of Solids and Structs., Vol. 26, P.401-415, Mar. 1990.

488
P(t)

~
SOIL PROPERTIES PILE PROPERTIES
Denaity "" p, DeDaity .. p
Young's modulll.ll .. E, YOUDg'S mbdulll.ll - E,.
PoiSllon'. ratio = II, PoiaaoD'. ratio .. II,.

PILE

BASE

Figure 1. Problem definition and material properties.

pet)

BASE

Figure 2. Discretization of pile-soil interface into boundary elements.

489
Pile
Axis

-+-..---r-.,..-----.x

Figure 3. Actual and discrete distribution of stresses at jth time step.

P(t)

tj = j.~t

Figure 4. Piecewise linear variation of impact load with time.

490
3 r--------------------,
P(.)

l
.2

/1
~c.i :' I
~::I!
I.

I
,IJ
~~
. "".~.~ __
\If
.1! I
: I,i - - METHOD I ~~----
_.- METHOD II
. ----- NOGAMI AND KONAGAI (1986)
.11 ~-----_=_=-----_7_:::__------:!
.11 2.11 4.11 S

t/tJ,:r

Figure 5. Comparison of time histories of axial pile head displacements CUd =.


37.5, EplEs = 124.4, pIps = 2.65, Us = 0.4).

pet)

1'..- ----:1 t
o ~:r

Figure 6. Time history of impact load considered for examples.

491
2 r----------------------,
... . t/t:1r =0.5
.. ,, t/ t:1r =1.0
, tft:1r = 1.5
, t/t1J.r = 2.0
.,..-',- =
"" "" ""- "'
' .. ~-
....... ~
.
.............
t/t1J.r 2.5

.'-.. . . - . '...
,""""""
", . . . - ~.~'---
~>,.----
~.-- .............. .--.

tI' _ "--.. _

.8 .2 .4 .& .8
....
-I ~---...L..:--- , ._.--.....L...,.__--.....L....,....--.....J
1.8

z/L

. Figure 7. Displacement time histories of axial pile shaft responses to impact load
=
at head CUd 40, EplEs 150, P~Ps 1.6, 1.>s = 0.4). = =

1 r---------------.. . . - ----.....,
=
t!t1J.r 0.5
=
t/t1J.r 1.0
t/ t:1r = 1.5
=
t/t:1r 2.0
.58
,--- ..... . =
t/ t:1r 2.5
. ... ..
/ . ~ "'-
.. I
1 ..--.....

,~',
--==
')
,

"'-'-
-" ...
',"-

/
"
-.__"..""",
.
~~.
~-._

...... - --..
---.;;
;:;.-.
---. _
-.

-.58 L.....---......."----'----...L..,,..----..........-=---~
.8 .2 .4 .& .8 I

z/L

Figure 8. Force time histories of axial pile shaft responses to impact load at head
(Lid = 40, Ep!Es = 150, PP/Ps = 1.6, 1.>s 0.4). =

492
PILE LOAD TEST IS A PROOF TEST, BUT?

1
Fathi Abdrabbo and Rouby EI-Hansi

ABSTRACf: Design of pile foundations always follows a certain procedure. Firstly,


the pile working load is calculated using sounding and/or laboratory test results,
employing a safety factor against failure, whilst the pile length is assessed from a
detailed borehole-logs and/or field measurements during pile driving. This design; the
pile length and the Pile working load, was regarded as being part of piling contractor's
mystique. Consequently, a tendency towards conservatism allied to a fear of failure
meant that, the number of piles tended to be over designed. Secondly, to prove this
design, vertical loading tests are conducted on either individual or group of piles. The
procedure of this test is well established and documented in many international codes.
Nevertheless, the interpretation of the test results is not well standardized yet. This
mis-standardization causes disputations between the engineer and the contractor. The
paper discusses through case histories, the interpretations employed in pile load test
results and gives a new interpretation criterion.

INTRODUCTION

Justification of the pile design load and the installation procedure which are a contractor
commitment, can be done by performing loading tests on representative single piles. Axial
maintained load tests(M.L.), are the most common field tests to be performed. These tests can
be conducted either static or dynamic [EL Naggar and Novak 1991]. Despite the numerous
static tests performed and reported in many papers and the analysis thereof, the understanding
of the test results in current engineering practice, leaves much to be desired. That is why the
engineers give their attention only to answer the question, if the test passes or fails with
respect to the design load, without any serious analysis of test results. In addition, the
engineers mix between conventional rupture load, failure load and limit load. Furthermore, the
numerous criteria for test interpretation in literature make it difficult to choose the best criterion
to use and give the contractor the right to adopt the criterion which gives the biggest failure
load, whilst the engineer uses the criterion which produces the smallest, without any
understanding from both sides of the suitability of the adopted criterion. This may cause a big

1 Professor, Faculty of Engineering, Alexandria University


2 Lecturer, Faculty of Engineering, Alexandria University

493 F.Abdrabbo and R.El Hansy


disputation between the engineer and the contractor and spoil valuable practical information
about pile-soil interaction response. Keeping in mind the fact that, the test is expensive and
consumes time, so, the maximum amount of information, should be obtained to justify the cost
involved. So, engineers are asked to give their attention for understanding the achieved results
and to interpret these results properly.
Furthermore, the available methods of interpretation of test results, rely on predicting the
failure load or limit load and applying a proper factor of safety to get the pile working load.
The value of the factor of safety is left to be chosen by engineers. Also, the interpretation
methods ignore completely the settlement associated with the pile working load.
This paper is devoted to interpret fifteen pile loading tests by different criteria to emphasize
the overall spread of results and gives a new method, for interpreting load tests.

DEFINITIONS OF TERMS

A conventional rupture load is the load at which the pile continues to penetrate into the
ground without any increase in the applied load. If the load-displacement of the pile is
represented, in such a way. the loads (p) are presented on a horizontal axis and the
settlement (s) on a vertical axis, the conventional rupture load is the load (P) corresponding
to (Llpl Lls) =0, except for loose and medium density sandy soil at the pile tip, which undergoes.
strengthening with increasing displacement of the pile and reflect a finite value of (Ll pi Lls) at
large displacement of the pile. Therefore, this definition of the rupture load is inadequate and
one is obliged to define the conventional rupture load by one or another characteristic. So.
the pile limit load and the pile failure load are introduced in literature.
The pile limit load is defined as the load corresponding to a point of the load-settlement
diagram separating the zone of small values of (Ll slLl p) from the load of rapidly increasing
values of(Lls/Llp), [Bengt 1980]. Also, the ultimate pile load which means pile failure load was
introduced by many researchers [Fellenius 1980 ].
The pile load test is mostly conducted on representative working contract piles. Therefore,
the test load is always less than the ultimate pile load. So, numerous criteria in literature are
developed for predicting the ultimate pile load using load-settlement relationship of the test pile.
It is clear that, the two definitions; the pile limit load and the pile ultimate load, are
introduced in literature to facilitate the determination of the pile axial load capacity. But. the
existence of these two definitions in literature, creates two major problems. The first is that,
most of the researchers and engineers do not distinguish between the ultimate load and the
limit load, and the second is the numerous criteria in literature defining the ultimate pile load,
are developed without any restriction for application and leave the chance to one 's past
experience to choose one of these criteria. So, different pile working loads may be predicted
for a pile, depending on the chosen criterion.
So, it is really needed to establish just a criterion for assessing directly the pile working
load, rather to define the ultimate pile load. Especially most of pile loading tests are carried
on piles which are not allowed to fail.

METHOD OF TESTING

The follOWing are the most popular testing methods implied in pile foundation:

Maintained Load Test


The testing procedure is established over the past decade and has become more

494 F.Abdrabbo and R.El Hansy


standardized. The method is given in the ICE specification[1978 and 1988] in the U.K. and the
loading procedure, given in ASTM [1974]. is generally adopted, internationally, through the
industry. These methods are linked by way or another to the maintained load tests (M.L)
which are currently regarded as the most suitable testing method.
In the UK [I.C.E. specification of piling 1978 and 1988]. loads are applied in stages up to
the maximum chosen test load, typically in increments of 25% for each of the design load.
The load is held constantly at each stage for a specified time or until settlement virtually
ceases. The limiting criterion is defined as less than 0.25 mm/hour and decreasing.
In ASTM designation 0-1143 standard method [1974]. the pile is loaded in eight equal
increments up to a maximum load equal twice the predetermined allowable load. Each
increment is maintained until the rate of settlement becomes equal to or less than 0.25
mm/hour, but not larger than 2 hours. The final load is maintained for 24 hours, if the rate of
settlement over 1 hour exceeds 0.25 mm/hour otherwise, the final load is maintained for 12
hours.
But, ASTM designation 0-1143, optional method, the constant time interval loading which is
proposed by Housel [Housel 1966]. each of the eight increments is maintained on the pile one
hour, regardless of the rate of settlement.
In Egyptian code of practice [Egyptian code of soil mechanics and foundations engineering
1991], the pile is loaded in six equal increments up to a maximum load of 1.5 times the pile
working load. Each increment is maintained until the rate of settlement measured at the pile
head becomes equal to or less than 0.15 mm/hour or maintained a specified time according
to Table(l) whichever occurs first. The pile is unloaded decrementally at a rate of 15 min each
then, the pile is left free from any loads for 4 hours.

TABLE(l): Load increment-Time


Load increment,% 25 50 75 100 125 150

Time.hrs. 1 1 1 3 3 12

Constant Rate Of Penetration Test


In the constant rate of penetration method (C.R.P.) which was proposed internationally by
Whitaker and Cook [Bengt 1980]. the pile head is forced to settle at a constant rate; of 30
mm/hour and the load is recorded. Readings of load are taken every two minutes and the
test is carried to a total penetration of 50-70 mm, regardless of the pile diameter. ASTM
designation 0-1143 stated that, the rate of penetration is 15 to 75mm/hour for piles in cohesive
soil or 45 to 150 mm/hour for piles in granular soil. The total pile penetration is 15% "f the
average pile diameter. In Canada [Canadian foundation engineering manual 1985], the rate
of penetration is 15 to 30 mm/hour. This type of test is recommended generally for testing
piles in clays and for all tests when only the ultimate capacity of the pile is to be measured.

Quick Maintained Load Test


The procedure of this test is published by the Federal High Way Administration (FHWA). The
pile is loaded in small increments. for instance 20 equal increments. Each load is maintained
for a period of 2.5. 5. 10, or 15 minutes. ASTM designatlun 0-1143 stated that, the load is
applied in equal increments of 10 to 15% of the proposed design load with a constant time
interval between increments of 2.5 minutes. This method also defines the ultimate load
regardless of the pile settlement.

495 F.Abdrabbo and R.EL Hansy


DISCUSSION OF THE TESTING PROCEDURE

It is clear that, in M.L tests, the overall test period differs from one specification standard
method to another. In addition, tt"\e number of load increments and the length of holding
period are independent of pile geometry, the length, the diameter and soil conditions.
Furthermore, there is no real concept or practical basis for the length of the holding period
also, there is no meaning for holding the test load for 12 hours or 24 hours according to the
rate of settlement. Therefore, different performance of the test pile is anticipated, depending
upon the adopted testing procedure. The duration time of M.L test is vastly inadequate to
obtain time-clependent or creep performance of piles also, it is too long to achieve a trUly
undrained condition of the pile-soil system. Consequently, the achieved results of this test is
in confusion. The quick maintained load test (a.M.L.) and (C.R.P) procedure, allow almost,
a unique analysis of settlement versus time which are closer to represent truly undrained
condition, of pile performance.
Presumably, engineers are familiar with the testing procedures and they can adopt one of
the above mentioned testing methods but, most of the international specifications give no
gUidance on allowable settlement of the testing piles and there is not any acceptance criteria
for test results. Consequently, the acceptance criteria are varied with no standard being
applied through the industry.

TEST PILES

All test piles are of varying diameter, length, design working load and method of
construction, are shown in Table(2). In addition. material of piles, conditions and method of
testing piles are given in Table(3). It is obvious from Table(3) that, five test piles are
constructed by a continuous flight auger (C.F.A) using concrete mix and the other piles by
driving, so bored piles are not considered in this stUdy. Five of these piles are constructed
at Alexandria city, C.F.A piles, whilst the results of ten piles are extracted from literature [Bengt
1980, Chin 1971, De Beer et at. 1979, Housel 1966 and Leach and Mallard 1979] and inevitably
constructed out side the home country. The load-settlement diagrams of these test piles are
illustrated in figures (1, 2 and 3). The generality of these tests are presented, such that, the
tests are conducted on piles installed in different soils, different construction manners, different
pile material and different experience.

PREDICTION OF ULTIMATE PILE LOAD

The ultimate load capacity (P u) for each pile test is predicted using the follOWing methods
implied in the evaluation of the pile load test. In fact, many different failure criteria are
proposed in the technical literature, a number of which are discussed by Fellenius [Fellenius
1975 and 1980].

The Method Of Terzaghi


This method [De Beer et al. 1979] which is the earliest one, was proposed by Terzaghi and
related to the conventional rupture load because of the large relative deformation to be
considered. The conventional rupture load in this method is the load corresponding to a pile
head settlement equal to 10% of the diameter of the pile base. This method is applicable for
10-pile load tests and the results are presented in Table(4). For other tests, the failure load
can not be defined, because of the shortcoming of the pile load-settlement relationship. Also,

496 F.Abdrabbo and R.El Hansy


TABLE(2): Characteristics Of The Test Piles

Base Diameter Length of Pile Method of


Pile No. Reference
Dmm Lms Construction
1 908.4 9.26
2 539.0 9.30
3 [De Beer et al. 615.0 9.50
Driven
4 1979] 815.4 9.46
5 406.0 9.33
6 558.8 11.39
7 500.0 24.00
8 500.0 25.75
Continuous over
9 The authors 500.0 24.30
flight auger
10 500.0 24.10
11 500.0 24.52
12 [Benat 19801 304.8 23.10
13 rLeach 1979 1 275.0 9.90
Driven
14 rChin 19711 400.0 unknown
15 [Housel 19661 300.0 13.20

TABLE(3): Material, Soil Profile And Method Of Testing

Pile
Material of pile Soil Profile Method of Testing
No.
1 In situ vibrated concrete The test piles were
subjected to stress
Prefabricated concrete controlled loading test,
2 shaft with eliminated shaft each loading step being
friction maintained as long as
Piles are driven through the settlement of the pile
3 In situ vibrated concrete very soft clay layers and head does not exceeded
penetrate at relative small 0.05 mm over a period of
Prefabricated concrete depths into very dense 20 minutes,
4 shaft with eliminated shaft but no
sand layer longer than 1.5 hour.
friction
The pile was further
5 unloaded in two steps.
Steel tube filled with
concrete after driving
6
7 Egyptian code of
8 Piles are drilled through practice
soft clay layers and
9 Concrete mix
penetrate five meters into
10 sand
11
Constant rate of
12 - - oenetration
Very weak, closely The pile is loaded in
Precast reinforced
13 fissured argillaceous increments up near to
concrete
siltstone failure
14 - -
Closed pipe filled with
15
concrete
-

497 F.Abdrabbo and R.EL Hansy


700
pile-1 ... pile-2 ~ pile-3 x pi/e-4 o pile-5 0 pile-6
600 ~ pile-15

500
Z
~ 400

i 300
~
... "
.9
200

100

0
0 20 40 60 80 100 120
Settlement, S-mm

Figure(1) Load-Settlement Diagrams

2000 ~----------------------,

1500

z
.lIl::
d..
1000
i pile-7
.9 ... pile-8

500 o pile-9
~ pile-1O
o pile-11
O..._....-I--L----'----'--'----'---'--"'--''---L--l---'--'----'---'--'---'----''--L---L--J----'--...J
o 2 468 10 12
Settlement, S-mm

Figure(2) Load-Settlement Diagrams

498
F.Abdrabbo and R.EL Hansy
the method is hardly to be applied on working contract piles.

3000

2500

2000
z
~
1500
~ 1000 o pile12
pile13
500
T pile14

0
0 10 20 30 40 50
Settlement, S-mm

Figure(3) Load-Settlement Diagrams

Brinch Hansen Method ( 90% criterion)


This method [Bengt 1980] is proposed by Brinch Hansen (1963) who defines the failure load
as the load that gives twice the movement of the pile head as obtained for 90% of that load.
This method is applied in the case, when the pile is loaded up near to the failure. This method
is applicable for only 9-pile load tests. The results are presented in Table (4) and the method
is illustrated in figure (4) for pile-6.

Brinch Hansen Method ( 80% criterion)


Brinch Hansen also, proposed 80% criterion defining the ultimate load as the load that gives
four times the movement of the pile head as obtained for 80% of that load [Bengt 1980].
Brinch Hansen 80% criterion postulates that, the load-movement curve is apPiOximately
parabolic which may not be. This method is applicable for only 8-pile load tests and the
results are presented in Table (4) and the method is illustrated in figure (5) for pile-6

Chin Method
This method is proposed by Chin (1971). The method assumes that, the load-movement
curve, when the load approaches failure of pile is of hyperbolic shape. In the Chin method,
each movement value is divided by its corresponding load value and the resulting value is
plotted against the movement. The plotted values fall on a straight line. The inverse of the
slope of this line is the Chin ultimate failure load. This method is applicable for all tests and
the results are presented in Table (4) and is illustrated in figure (6) for pile-1.

499 F.Abdrabbo and R.El Han~y


500
P
400

z
~ 300
r:i
-g
.9 200

100
28 /- pile-SI
o -'---'--...o-.-JL..l...-.............................&..-...L-......................................L-""""-----'-""""---.....L-................................-......-l

o 20 40 SO 80 100
Settlement, S-mm
Figure(4) Ultimate Failure Load According to 90%
Criterion by Brinch Hansen

500

P
400

z
~ 300
d..
i0 200
...J

100

45 I- pile-61
0
0 20 40 so 80 100
Settlement, S-mm

Figure(5) Ultimate Failure Load According to 80%


Criterion by Brinch Hansen

500 F.Abdrabbo and R.El Hansy


TABLE (4) : Results Of The Study

Terzaghi Brinch Hansen Brinch Hansen De Beer Chin


(90%) (80%)
No
P,kN SID, % P, kN SID, % P, kN SID, % P, kN SID, % P, kN SID,
%
1 608 10 - - - - 450 2.6 697 -
2 270 10 260 6.5 270 10 250 4.0 310 -
3 319 10 300 6.7 330 13 270 3.4 363 -
4 520 10 475 5.0 500 7.8 420 2.5 587 -
5 167 10 166 10 175 14.7 140 4.2 195 -
6 451 10 425 6.8 460 11.8 370 2.9 516 -
7 - - - - - - - - 2274 -
8 - - - - - - - - 3172 -
9 - - - - - - - - 2073 -
10 - - - - - - - - 2741 -
11 - - - - - - - - 2664 -
12 1950 10 2060 11.8 2070 12.1 1825 7.8 2513 -
13 2000 10 1975 8.7 - - 1700 3.8 2636 -
14 2300 10 2250 8.75 2450 11.25 1760 1.5 2528 -
15 475 10 490 12.6 500 14 370 2.0 571 -

TABLE (4) :Continued

French Proposed Proposed


Mazurkiewicz method Working Load
No method
--------
Load,kN Working
P,kN s/D,% P,kN s/D,% P,kN s/D,% Load
1 520 4.7 - - 168 0.5 175.33 0.96
2 275 12.6 - - 113 0.9 88.33 1.28
3 310 8.5 - - 104 0.7 101.93 1.02
4 490 7.0 - - 168 0.6 160.03 1.05
5 162 8.4 - - 62 0.8 54 1.15
6 420 6.1 - - 138 0.5 141.73 0.97
7 1950 - 1050 1.0 731 0.5 650 1.12
8 2300 - 1000 0.5 703 0.3 766.67 0.92
9 2060 - 1050 0.3 637 0.12 686.66 0.93
10 1720 - BOO 0.24 507 0.16 573.33 0.88
11 2400 - 1050 0.6 788 0.32 800 0.99
12 2040 11.5 - - 600 0.9 663 0.90
13 1970 8.7 - - 696 0.12 637.1 1.09
14 2250 8.75 - - 702 0.2 734.0 0.96
15 537 21 - - 144 0.3 158 0.91

F.Abdrabbo and R.EL Hansy


501
700 18
Failure Load =697 kN
16
600
z
14 .l!:
500 E
12 E
1/Pu
Z
~ 400 .. 10 -I

T-
o
i 8
e.
c..
300 -.
.9 en
6 0
200 ~
- 4 a:
100
pile-1 2
0 0
0 20 40 60 80 100
Settlement, S-mm

Figure(6) Ultimate Failure Load According to Chin


Mazurkiewicz Method
This method [Sengt 1980] based on the assumption that the load-movement curve is
approximately parabolic. In this method, a series of equal pile head movement lines are
arbitrarily chosen and the corresponding load lines are constructed from the intersectiOns of
the movement lines with the load-movement curve. From the intersection of each load line
with the load axis. 45 line is drawn to intersect with the next load line. These intersections
fall approximately, on a straight line, the intersection of which with the load axis defines the
ultimate failure load. This method is applicable for all tests and the results are presented in
Table (4) and is illustrated in figure(7) for pile-4.
Load, P-kN
sao r-----------------------~
Failure Load =490 kN

I- pile-4 1
20 40 60 80 100
Settlement, S-mm

Figure(7) Ultimate Failure Load According to Mazurkiewicz

502 F.Abdrabbo and R.El Hansy


De Beer Method
In this method [Bengt 1980). the load-movement values are ploned in a double logarithmic
plot. When the values fall on two approximately straight lines, the intersection of these defines
De Beer failure load which is considered a pile limit load. This method is applicable for only
10-pile load tests and the results are presented in Table (4) and the method is illustrated in
figure(8)tor pile-3.

1000 . - - - - - - - - - - - - - - - - - - - - - - - - ,

Failure Load=270 kN

100

10 L-..---'----'-..........
'-L.LJ../..l-_...L-~__L_L
I- pile-si
...........L.____L..___'_....l_L...I...I...J.........._ . . L . . _ . l _ _ L.............u...I

0.1 1 10 100 1000


Settlement, S-mm 009)
Figure(S) Construction of De Bee1r ~ Yield UmitLoad~

When the pile is not loaded up near to the failure, the ploned values of the load-movement
fall on one approximately straight line and the failure load is not defined, figure(9), this is really
a drawback of this method.

503 F.Abdrabbo and R.EL Hansy


10000 . - - - - - - - - - - - - - - - - - - - - - - - - - - ,

-a
g
z
d. 1000

100 L-_ _"""--_-'--...........-'-_'_.....................


1--- Pile-81
""'---_.L-~_'_~ ................,

Q1 1 10
Settlement, S-mm (log)

Figure(9) Construction of De Beer - Method.


The French Method
This method [Egyptian code of soil mechanics and foundation engineering 199 I] requires
that, each load increment must be maintained on the pile for one hour and the readings of the
pile movement should be recorded against the time during each load increment. Then, for
each increment, the movement-time values are plotted in a single logarithmic diagram wl,ich
fall on approximately straight line. figure(10). The slope of these lines (a) are plotted against
the corresponding loads, figure(11). When the values fall on two approximately straight lines,
the intersection of these, defines the limit load. This method is applicable for only 5-pile load
tests due to the shortage in outcome of other tests results and the results are presented in
Table (4) and are illustrated in figures (10 and 11) for pile-g.

DISCUSSION OF THE ULTIMATE LOAD PREDICTION METHODS

AI! the above methods are linked to failure load except De Beer and the French methods
which are linked to limit load which still has no clear definition. So, one should distinguish
between the failure load and the limit load to adopt the proper factor of safety. The pile failure
loads which are predicted from load-settlement relationships of piles loaded to pre-failure, are
based on assuming certain shapes of these relationships, independent of pile geometry, soil
conditions and rate of loading, which may not be verified. Furthermore, the ability and the
reliability of the method to predict the pile ultimate load, depend upon the magnitude of the
pile head movement and the applied load level with respect to pile failure load. Therefore,
some of these methods fail to predict this load. Davisson and Vanderveen methods are tested
with the results of the fifteen tested piles, but without any success. It is clear from the above
that, the only methods which are applicable to all pile loading test results are Chin and
Mazurkiewicz methods. The other methods require the pile to be loaded up near 10 failure
which may be not permitted. Also. some of these methods require long duration of time such
as, tile French method, beside that, the method requires accurate and precise measurements
of the applied loads and the corresponding pile head displacements.

504 F.Abdrabbo and R.El Hansy


1

0.9

E
0.8
~
(j)
~
cQ) 0.7
E
Q)

~ 0.6
(j)

0.5
I- Pile-9 1
0.4
1 10 100
Time, min. (log)

Figure(10) Creep Curve,P=1350kN..

0.35

0.3

0.25

0.2
a 0.15

0.1
I
0.05 J

OL-.......,:::.....-
Umit Load=1050 kN I- pile-9 1
-'--.....l---'--'----'----'---1-....L-<..----'----'_"-----.l-"'---'--................o.-.J
o 500 1000 1500 2000
load, P-kN
Figure(11)Determination of Umit Load According to French
Method

505 F.Abdrabbo and R.El Hansy


With respect to pile head movement, Brinch Hansen 90% method predicts pile failure loads
at a corresponding average pile head displacement (siD) of 8.5%, whilst the average pile head
displacement, (siD). corresponding to failure pile loads predicted by Brinch Hansen 80%
method is 11.8%. The average pile head displacements, (siD), corresponding to ultimate pile
loads predicted by De Beer, Mazurkiewicz and the French methods are 3.5%, 9.73% and
0.52% respectively. In spite of De Beer and the French methods which predict pile limit load.
the corresponding pile head displacement in the two methods are different by more than
600%.
The overall spread of ultimate pile load predicted by various methods for each pile are
illustrated in Table (5). The ratio of the biggest to the smallest load is also given. This ratio
varies between 1.24 for pile No.(2)and 3.42 for pile No.(10). By excluding Chin method which
predicts the biggest ultimate load. Table(6), this ratio varying from 1.1 for pile No.(2) to 2.50
for pile No.(S). So, one can predict different values of pile ultimate load. with big range. The
spread of (siD) ratio corresponding to the ultimate loads for each pile is drastically very wide.
Table (5).
So, as an example confirming the unavoidable disputation between engineer and contractor,
the engineer may consider 1000 kN as an ultimate load for pile No.(8), whilst the contractor
may insist on 2500 kN. The consequence of this is drastic.

PROPOSED METHOD

Undoubtedly, the available methods which are used for predicting ultimate pile load have
shortage in outcome and causing disputation and contradiction between different parties
involved in deep foundation industry. So, it is hoped that the proposed method may put an
end for this contest and solve the problem of the prediction of pile ultimate load. In this
method, each load value(p) is divided by its corresponding pile head settlement value (s) and
the resulting (pis) values are plotted against the pile head settlements (s). The plotted values
fall on two approximately straight lines, connecting together with a curve. The ordinate of the
intersection point of the extension of the two straight lines, point (A), figures (12 and 13).
represents the value of the pile working load Pw divided by the pile head settlement at that
load. Whilst the abscissa value of the intersection point, point (A), indicates the pile .head
settlement Swat the pile working load. So, the pile working load is obtained by multiplying the
ordinate of point (A); Pw/Sw by its abscissa value Sw' This procedure was used for the
determination of pile working loads and the pile head settlements for the fifteen pile load tests.
In these tests the initial and the final straight lines of each test are clear to define easily the
intersection point (A). But, if these two lines are not clear, point (A) can be defined by two
tangents to the beginning and to the end of the PIS -5 relationship. The method is applied,
easily, to all pile test results and the predicted pile working loads are presented ir. Table(4).
Because of the limited space, the method is illustrated In figures (12 and 13) for piles NO.(1
and 2). Figures (12 and 13) indicate that point (A) separating the stiffness movement diagram
into a zone of big values of PIS from zone of small values of PIS.
The authors feel that, the method is independent of jUdgement, scale of graph or change
of scale and depends merely on the achieved results without any prediction of the ultimate
load. The method also, has the advantages that, the pile is not necessary to be loaded up
near to the failure to determine this load also, the load increment can be maintained for any
time.
By comparing pile working loads obtained from the proposed method by the corresponding
average working loads which are obtained from the other methods considering a factor of

506 F.Abdrabbo and R.El Hansy


TABLE (5): Overall Spread Of Results For The Piles

Pile Overall spread of ultimate Ratio of biggest Range


No. aile load (kN) load Ismail est load of siD %
1 450-697 1.54 2.60-10
2 250-310 1.24 4-12.4
3 270-363 1.34 3.4-13
4 420-587 1.39 2.50-10
5 140-195 1.39 4.2-14.7
6 370-516 1.39 2.9-11.8
7 1050-2274 2.16 -
8 1000-3172 3.17 -
9 1050-2073 1.97 -
10 800-2741 3.42 -
11 1050-2664 2.53 -
12 1825-2513 1.37 7.8-12.1
13 1700-2636 1.55 3.8-10
14 1760-2528 1.43 1.5-11.25
15 370-571 1.54 2-14

TABLE(6): Overall Spread Of Results For The Piles Excluding Chin Method

Pile Overall spread of ultimate pile Range of siD, Ratio of biggest


No. load kN % load Ismallest load
1 450-608 2.6-10 1.35
2 250-275 4-12.4 1.1
3 270-330 3.4-13 1.22
4 420-520 2.5-10 1.23
5 140-175 4.2-14.7 1.25
6 370-460 2.9-11.8 1.24
7 1050-1950 - 1.85
8 1000-2500 - 2.50
9 1050-2060 - 1.96
10 800-1720 - 2.15
11 1050-2400 - 2.28
12 1825-2070 7.8-12.1 1.13
13 1700-2000 3.8-10 1.17
14 1760-2450 1.5-11.25 1.39
15 370-537 2-14 1.45

507 F.Abdrabbo and R.EL Hansy


5Or---------------------.....,
I- pile-11
40
E
E
Z 30
.::.:: \
\
ch \
a: \
i 20 \
\
a: \A
10 p; ~~T----- ----
ISW
--=-....------__-- . . . .
I
o ,--,--,,--,-I -'---'--'---'---'---'---'----'--'----'----'---'---'---'---'----'---..1.---'

o 20 40 60 80 100
Settlement, S-mm
Figure(12) Construction of the Proposed Method.

30.....----------------------.
I- Pile-21
25

E 20
~-
~ 15
.g
Cl1
a:

20 40 60 80 100 120
Settlement, S-mm
Figure(13) Construction of the Proposed Method.

508 F Abd rabbo and R. EL Han sy


safety equal to 3, and excluding Chin method, a good correlation is obtained, Table(4). Also,
the pile working loads, obtained by the proposed method are used for calculating the factor
of safety against pile ultimate loads obtained by Terzaghi, Brinch Hansen 90%, Brinch Hansen
80%, and Mazurkiewicz methods. The factor of safety was found to vary from 2.38 to 3.47 with
an average value of 3.07. So, the obtained pile working load Pw' by the proposed method is
associated with a factor of safety equal to 3 against ultimate pile load.
The factor of safety for pile capacity is based on many factors such as method of site
investigation, method of pile testing and analysis, method of pile load verification, functional
significance of the structure. level of confidence in foundation design parameters, adequacy
of the analysis tools and the level of construction controls. This factor of safety varies from
a maximum value of 3.0 to a minimum value of 1.15 (Design of pile foundations ASCE 1993).
If the engineer decided to use another factor of safety, F, rather than 3.0, the pile working
load obtained by the proposed method should be modified according to the adopted factor
of safety. This can be done by multiplying the obtained working load by the ratio (3jF).
The (SjD) ratio, for pile load tests under consideration, corresponding to Pw varies from a
maximum value of 0.9 to a minimum value of 0.12, Table (4) and the average (SjD) ratio is
equal to 0.46%. If the values of SjD less than or equal to 0.16 are omitted; three values out
of the fifteen values, the average SjD ratio becomes equal to 0.54. The values of SjD greater
than or equal to 0.9 are disregarded in addition to the values less than or equal to 0.16; total
five values out of the fifteen values, the average SjD ratio is 0.47. So, one can say that, the
pile working load, with a factor of safety equal to 3 against ultimate load. is the load
corresponding to pile head settlement (SjD) equal to 0.46% for piles having diameters varying
from 300 mm to 900mm.

EFFECT OF LENGTH OF LOAD INCREMENT PERIOD

Fellenius [1980], pointed out that a better loadjsettlement envelope can be defined for a
large number of load increments with corresponding shorter holding periods. In addition the
length of the holding period is not crucially important as long as they are of equal duration.
Keeping in mind, the settlement behavior of a single pile during test loading, may bear no
resemblance whatsoever to the overall behavior of pile group. So, the settlements of the pile
are not a crucial item, but, only the shape of (P-S) relationship. Therefore, the effect of the
length of the holding period on the working load predicted by the proposed msthod was
investigated. Unfortunately, The time-settlement readings, are only available for pile load tests
No. (7 to 11). The values of PjS versus S of these piles are drawn for different lengths of
holding period starting from 5 minutes up to 55 minutes with steps 5 minutes. As an example
for this presentation, figure (14-a) illustrates the achieved results for pile No.(ll). The
relationship PjS versus S is expressed as:

PjS = - 174.852 In (s) + 566.122 kN,mm (1 )

This figure also illustrates individual relationships for (pjs) versus (s) at holding length of 5
and 55 minutes, figures (14-b and 14-c) respectively. The relationships PjS versus S at
holding lengths of 5 and 55 minutes are also, expressed respectively as:

PjS = - 147.183 In (S) + 535.784 kN,mm (2) and


P/S = - 144.005 In (S) + 525.937 kN,mm (3)

F.Abdrabbo and R.El Hansy


509
Pile (11)
S 800
S 700
..........
~ 600
pis = -147.183 In (s) + 535.784 (8)

vi ClOO -
..........
A 400

300

settlement (8) ,mm.


1000
Pile (11) / 5 min.
voo -
8 800
S 700
'-... p/s=-174.852 In (8) + 560.122 (b)
Z
~ 600

iii GOO .

..........
Po 400

300

. 0

1000..-----------------,

900
Pile (11)/ 55 min.

S 800

S 700
'-... p/s=-144.005 In (e) +525.937 (c)
~ 600

{J] 500
..........
Po 400

300

... J. I , ... J. ' I I J. I '-'--J--.JL...L......I......L


z.oO B.do l.oO .
settlement (8) ,mm.

Fi~urc(14) (pjfJ) VA. (R) RclnU()n~hipA.

51 0 Fathi l\tx1ratm ard Ihby El-Hansy


It is clear from this figure and equations No.(1, 2 and 3) that, there is no effect of length of
holding time on the (pis) versus (s) relationships which agrees well with Fellenius conclusion
concerning the non importance of holding time period. Thus, equal incremental loads can be
applied on the pile, during testing. Each load increment can sustain on the pile for a period
of 5 minutes or shorter. This agrees also with the test procedure published by FHWA.
The (pis) versus (s) relationships of piles No. (7 to 11) may be expressed as:

PIS = -A In (5) + B (4)

The parameters A and B were obtained for each of the five piles and the average values are
calculated as: A = -150 and B = 585. 50, for these piles PIS versus 5 can be expressed as

PIS =-150 In (s) + 585 kN,mm (5)

This relationship may be used in the numerical analysis of pile groups.

CONCLUSIONS

The above study reveals the following conclusions:-


1- The available methods which are used for predicting ultimate pile load, based on load
versus pile head displacement, have shortage in a way or another and causing disputation
between different parties involved in deep foundation industry. These methods predict
a wide range of ultimate load for a pile.
2- The proposed method has the advantages that, the method predicts directly the pile
working load, and applicable to pile test results not necessary to be loaded up near to
failure.
3- The study recommends the testing procedure published by FHWA with length of holding
period equal to 5 minutes or even shorter, to be an international test procedure for
working contract piles. The study also agrees well with Fellenius conclusion concerning
the non-importance of holding time period.
4- The pile head settlement corresponding to pile working load predicted by the proposed
method is 0.46 % of the pile diameter for piles having diameters vary from 300 mm to 900
mm and installed either by driving or by a continuous flight auger through soft clay layers
to a dense sand.

ACKNOWLEDGEMENTS

The Writers wish to express their gratitude to the reviewers for their valuable comments.

APPENDIX. REFERENCES

ASTM, (1974). "Standard method of testing piles under axial compression load", Annular book
of ASTM Standard part 19, Designation 0-1143-71, pp.178-188.
Bengt, H.F. (1980). "The analysis of results from routine pile load tests ", Ground Engineering,
pp. 19-31.
Canadian foundation engineering manual, (1985).
Chin, F. K. (1971). Discussion, " Pile tests, Arkansas river project ", ASCE. J. SMFD. Vol. 97,

F.Abdrabbo and R.El Hansy


511
SM6, pp. 930-932.
De Beer.E, De Jonghe.A, Carpentier. Rand Wallays, M. (1979). Analysis of the results of
loading tests performed on displacement piles of different types and sizes penetrating at
a relatively small depth into a very dense sand layer ., Recent developments in the design
and construction of piles. ICE, London, pp. 199-211.
Design of pile foundations, ASCE. (1993). Technical engineering and design guides as adopted
from the U.S army corps of engineers, No. -1
Egyptian code of soil mechanics and foundation engineering, (1991). Part 4, pp. 163-165.
EL Naggar, M.H., and Novak, M. (1991). Analytical model for an innovative piletest",
Geotechnical Research Centre Report, GEOT-7-91.
Fellenius, B.H. (1975). ''Test loading of piles, methods, interpretation and proof testing, ASCE
Geotechnical Engineering Division, Vol. 101, GT9, pp855-869.
Fellenlus, B.H. (1980). "The analysis of results from routine pile load test", Ground Engng.
13(6).
Housel, W.S. (1966). Pile load capacity, estimates and test results, Proc. ASCE, Vol. 92, SM4,
pp.1-30.
I.C.E. (1978). Pillng model procedures and specification", Institute of civil engineers, London.
I.C.E, (1988). "Specification for piling", Institute of civil engineers, London.
Leach, B. and Mallard, D. J. (1979). " The design and installation of precast concrete piles in
the Keuoer Mar! of the Seven Estuary ", Resent developments in the design and
construction of piles. ICE, London, pp. 33-43.

512 F.Abdrabbo and R.El Hansy


OPTIMUM SPECIFICATION FOR STATIC LOAD TESTING

ABSTRACT.-

This paper reviews some of the foundation load testing methods and procedures. It considers what would
constitute an ideal test, and comments on recent developments in static load testing and result analysis
techniques which increase their technical value.

It is concluded that all foundation test results need some form of analysis at the interpretation stage, that the
standard Maintained Load Test still represents the best available test method and that its specification and
procedure would benefit from some minor amendment. This would specifically require defined constancy of
load at each load stage and would permit better interpretation of results.

After more than 5 years of using static load test analysis techniques. such as TIMESET & CEMSOLVE,
which model the characteristics of a pile in time and under load to determine the unique pile behaviour, a
reliable and proven test specification can be defined which allows the unique foundation behaviour to be
determined. In addition to the greater technical value obtained, it also allows the most rapid and therefore
cost effective testing procedure to be prescribed without detrimental effect on the quality of the test results.

These methods of modelling foundation behaviour have been proven by back analysis on many hundreds of
static load tests on a wide range of piles and foundation types in many different soils..

Different organisations around the world appear to favour different test schedules for their static load test
requirements. The merits of some of these are reviewed and compared with the proposed specification.

M. England & W.G.K.Fleming


CEMENTATION PILING AND FOUNDATIONS UMITED
APRIL 1994

513 England/Fleming
INTRODUCTION

As part of the foundation installation process subsequent testing plays an important role. This is mainly
because of uncertainties in soil parameter measurement and design models. In addition, the ground
conditions may vary across a site and the extent of site investigation may be limited. Testing can
therefore be a means of confirming the ground conditions and of proving that the design parameters,
installation method and technique are appropriate to the prevailing conditions. It serves also to check that
any subsequent events are not detrimental to required performance.

The introduction of sophisticated pile installation instrumentation has minimised the variability occasioned
by pile installation methods.

Although a considerable amount of test data has been recorded in the past, complete analysis has usually
been difficult because in general the conditions of recording impeded such analysis. Recent
developments in static load testing techniques allow better analysis and behavioural models are now
available to characterise foundation behaviour both in time (TIMESET) and under load (CEMSET) in a
manner independent of the testing programme. As a consequence it is possible to prescribe the most
rapid practical testing schedule without compromising the technical information that may be revealed.

These developments indicate shortcomings in all foundation testing techniques, and a need for a method
of analysis that provides consistency and accuracy, allowing interpretation of the results to be independent
of the method of testing. They also serve as a reminder that the limitations of each testing process should
be considered carefully to ensure that misleading interpretations are avoided.

THE IDEAL FOUNDATION TEST

When evaluation of foundation performance is required, a large variety of tests of differing types are
offered by the industry; and for each type of test, the conditions under which the test may be performed
can produce significantly differing results. It is therefore important to ensure that the selected test type
and specification are appropriate to the results required.

Foundations are generally called upon to carry axial static loads for a long period of time. In the civil
engineering context, during most construction the loads applied to the foundation system gradually
increase as work progresses and result in some final, near constant load being applied upon completion.
The best foundation test that could be employed would be one which replicated these conditions as
closely as possible. However, for practical reasons it is desirable to carry out tests expeditiously, ensuring
that any external influences are minimised, so that construction can progress with minimum interruption
and delay. A compromise has to be found, with the result that the duration of the load application is
necessarily curtailed.

AVAILABLE PILE TEST METHODS

A variety of test methods are to be found in the industry, ranging from full scale static tests, with
application of load and monitoring of pile deformation, to the measurement of associated properties of the
pile-soil system, as for example in low strain integrity tests. The list includes Static Maintained Load
Tests, Load/Settlement Equilibrium Tests (in which a pile is made to settle by stages and allowed to reach
an equilibrium load at each before moving on to another settlement value), Quick Maintained Load Tests,
Continuous Rate of Penetration Tests (C.R.P.), Statnamic and Pseudo-static Tests, Dynamic Tests (in
which a pile is struck by a falling hammer), and Integrity Tests (which basically use wave propagation and
acoustic impedance measurement techniques to look only at structural continuity and implied section
variation). Costs in general move downwards through the list.

514 England/Fleming
In view of the range of methods and the specific knowledge of different engineers, some may prefer one
method to another and, amid conflicting claims, there is often a genuine difficulty in making test
programme choices from combinations of the various techniques referred to above. Sound engineering
jUdgement is required to make a sensible and justifiable choice of the appropriate testing for a particular
site.

All testing is really aimed at finding out whether installed piles will perform according to expectation and an
engineer has not completed his task in regard to testing until he has decided how many tests of a
particular type are sufficient to satisfy him that all is well. He even has the option not to do any testing, but
if this is what he decides, he would be well advised to ensure that he has taken other steps by way of
complete pile installation records and enhanced Factors of Safety or other redundancy features, to ensure
that he can later justify his decisions if they are called into question.

Many engineers believe that, in general, pile load bearing behaviour will be absolute and not a function of
the test specification applied: this is not correct because the test results will in reality be determined by
the procedure selected. The true, long term, pile behavioural characteristic is unique at the time of the
test, and this is what pile tests generally should aim to reveal, though often they do not succeed.

CURRENT PRACTICES

The most visible and tangible proof of pile performance is provided by full scale load tests. In the United
Kingdom, the most common test is the Static 'Maintained Load' Test and indeed it has not only traditional
practice to support it, but is capable of wide application, save in a few specifically restricted cases like
off-shore work. It also has scope for extension of its usefulness.

In the quest for a rapid substitute for the static load test, a modification was developed as the Continuous
Rate of Penetration Test (CRP) (1). However, it has some disadvantages in practice, which at the very
least need consideration at the interpretation stage. It has been shown (Whitaker and Cooke(2), Burland
and Twine(3), Patel(4). and others) that in fact the effect of the rate of penetration (normally approximately
1mm per minute) is to enhance pile shaft capacities in clay soils, but the same is also probable with regard
to friction in a wider range of soils and also to base capacities. The enhancement effect can be reduced
by decreasing the rate of penetration to less than say 1 mm per hour thus allowing longer for pore water
pressure dissipation to take place.

Possible enhancement of base resistance has not been commented upon much in literature, partly
because there is some confusion over definitions of failure, but, particularly in clay soils where porewater
pressures are unable to dissipate rapidly, the results are undoubtedly affected if it is long term pile
performance which really is of interest. CRP tests have little to say about deformations at Specified
Service Loads.

All rapid pile testing methods and in an extreme case, dynamic methods, suffer from similar problems in
trying to relate the enhanced resistances usually measured to long term static capacities. Depending on
the use an engineer wishes to make of them, frequently if not invariably, there is a need for some static
correlation testing at the present time. Of course such tests may simply aim at approximate installation
control, in which case they are probably theoretically superior to traditional dynamic formulae in most
cases. However, it should always be remembered that if a pile cannot be moved far enough into the
ground under any type of loading, then its ultimate load cannot be determined: only the maximum
mobilised load can be found.

The traditional static load test still remains the most informative and reliable pile testing method. It has
much to commend it and with careful accurate measurement and load control, it can contribute greatly to

515
the understanding of pile behaviour, even to the extent of revealing the important soil properties governing
foundation settlement.

REFINEMENT OF SPECIFICATIONS FOR STATIC MAINTAINED LOAD TESTS

Although several engineering practices have their own individual specifications for Maintained Load tests,
the most commonly applied procedure in the UK is presently that recommended in the ICE 1988
Specification (5). This calls for the load to be applied in incremental steps of a quarter of the working load
and in general, the test is continued beyond the working load using the same load increments. The
authors have developed an improved system and consider this type of specification now to be enhanced
with minor modification as follows:-

1. Load Control.
Loads are required to be held constant. This is a problem if manual load control is exercised,
largely because the reason for this requirement is often not understood. Observation clearly
shows that accuracy is not regarded as too important and that operatives do not restore loads with
sufficient regUlarity. Loads can now be measured and controlled within very fine limits and
equipment is available which checks and restores load every few seconds automatically. This
means that displacement/time relationships can be defined with very high accuracy and the
behaviour can be modelled so that extrapolation of the measured displacement to infinite time is
readily possible. In doing this the results become consistent at all load stages, with the settlement
being independent of the test duration. These projected settlement results obviously represent
"fully drained" conditions.

2. Settlement recording.
Measurement of displacement of the pile head is conventionally by dial gauges which are read and
the results written down on site. Results are not easily checked for error and finally have to be
manually transferred to a report. The problem of potential errors can be solved by the use of
electronic displacement transducers, allowing all the data to be obtained simultaneously, to be
logged and stored digitally for computer processing. This obviates the need for double handling of
the data, ensures readings are taken when required, and minimises the chance of error. All
plotting of resulting displacement can be carried out on screen in a cabin on the site so that any
untoward events become evident at the time. The computer can also be instructed to reduce the
applied load. at any time for safety reasons if a very large deviation occurs and give immediate
warning of the anomalous condition. A diagram of such an arrangement of electronic logging and
load control is shown in Figure 1.

3. Fully drained test results.


Specifications for the execution of a Maintained Load test have traditionally involved a programme
of load application which employs differing periods of load holding as the test proceeds and often
include one or more cycles of unloading and reloading.

The problem with specified time periods is that, at lower loads, the pile movement rapidly
approaches a stable state, whereas at higher loads, when the shaft friction has been fully, or almost
fully mobilised and load is being transferred mainly at the pile base, the time required to reach the
settlement rate becomes more protracted. The period of observation is often curtailed at this stage
simply for practical and cost reasons. It is very difficult from short test durations to define what the
final settlement may be and therefore what the fully drained ultimate capacity of a pile is. The
nearer the ultimate load is approached, the more difficult becomes interpretation without accurate
computer modelling of the displacement/time characteristic and settlementlload behaviour.

516
Justification for the use of computerised systems is easy because errors can be minimised,
engineering attendance may be diminished and reports can readily be produced, thereby providing
a more cost effective and safer test. However, care must be taken to ensure the data storage
system is secure in the event of power supply failure and that transducers are suitably calibrated.

An accurate method, using a computer model called TIMESET(6). based on high grade
displacemenVtime results, has been developed which allows each relationship to be divided into
components due to shaft friction and end bearing on the basis that each component can be
represented by a hyperbolic function. The results from such a modelling technique in undisturbed
conditions are so close to observed behaviour in the vast majority of cases, in a wide range of soils,
that they may reasonably be described as identical. Temporary deviations sometimes occur due to
site traffic or other external environmental factors.

Using refined electronically controlled equipment and this interpretation method, a sufficient portion
of the time displacement relationship at a given load needs to be recorded so that the remainder of it
may be predicted with accuracy. This can satisfactorily be defined by the point in time when at least
90% of the shaft displacemenVtime relationship has been mobilised. For most concrete piles of
average dimensions, this is between 1 and 3 hours, so that if the load is held for say 6 hours,
providing additional redundancy of data, the unique solution for each component can be found.

The development of base settlement behaviour in time is generally so long that the duration of
previous lower loads applied has little influence on the result and it is found that a normalised time
characteristic for the base is practically constant from load to load, at loads above the ultimate shaft
capacity. Therefore, if the normalised time constant can be derived accurately at a higher load, for
example, lesser loads may not need to be held for such long periods. Consequently only the
highest load needs be maintained constant for long enough to allow the unique separation of the
two functions. This allows the duration of the test programme to be minimised without
compromising the accuracy of the results. The ideal time for maintaining the single long duration
load is overnight when the external influences are generally minimal.

4. Umiting Settlement rates.


A simple way of controlling the specification for the test
schedule has been and is to specify a minimum load duration of say 1 hour, and a settlement rate
of less than 0.25 mm per hour. Although the governing rate should obviously vary from load to
load, this specified rate is generally sufficient to ensure that once creep or consolidation becomes
significant, the duration of the load holding period is automatically extended. A more appropriate
procedure may be devised, basing the settlement rate on a proportion, say 5%, of the total
displacement recorded since the start of the test, but for practical reasons, with this form of
definition a limiting minimum value also needs to be specified for the condition when the total
settlement is small at low applied loads. If analysis of the accurately measured behaviour in time,
using the computer model TIMESET for example, is available, the duration of application of any
load need not normally exceed 6 hours.

If the applied loads get close to the ultimate pile capacity, it may then be advantageous to reduce
the load incremental steps to avoid premature rupture or slip of the skin friction and also to
determine limiting rates by practical observation of the displacemenV!ime curves as they develop.
However, at this stage, the total test may become unduly protracted.

5. Modelling results.
An example of the displacemenVtime results from a constant load
stage for a pile is shown in Figure 2, together with the mathematically derived separate
relationships for shaft and base. These results are of great interest, since they may be used for

517
example to identify the soil type on which the base of the pile sits. As soon as the defonnation
characteristics can be accurately identified to give a consistent final settlement projection, the test
may proceed to other loads.

Provided sufficient points on the load/settlement graph are produced, so that the relationship is
unambiguous, the behavioural characteristics can readily be defined. This might involve typically
eight or more load stages, but where possible and if good interpretation is required to define all the
parameters with reasonable accuracy, the pile should, in general, be made to settle by something of
the order of 10% of the diameter. Although this may not be possible nor necessary for piles on or in
hard soils or soft rocks, it would be applicable to a wide range of other soils.

Typical proof load tests to only 1.5 times the required working load would not generally cause
sufficient displacement for a detailed analysis, but it is often worth considering an increase of test
load to give a more comprehensive view of the pile and soil behaviour. A cost effective altemative
would be to test a smaller diameter pile, installed using the same technique and to the same depth,
thus establishing the governing soil characteristics so that they may be scaled to the appropriate pile
size.

The techniques of back analysis are very useful and have become of substantial value when
investigating pile failure mechanisms where for example, closure of small cracks can be identified.
The identification of such features is very dependent on the accurate maintenance of load at the
given stage, and this is an additional reason why refined load control is an absolute necessity.
Figure 3 illustrates the distortion of pile head behaviour that may result from just a 0.5% drop in load.

6. Loading and unloading.


It is universally observed, that, when a load is removed and
subsequently re-applied, the settlement at the re-Ioad stage is more than that at the initial stage. A
typical result of pile head displacement, under the load stages shown in Figure 4, is illustrated in the
load/displacement diagram of Figure 5; the corresponding displacemenVtime diagram is shown in
Figure 6. In this example, several loads have been reapplied after unloading and each have caused
effects in the pile/soil system which differ from each other in that different stress paths are being
followed. It is often the case that the initial load is held constant at such a point for only a short
period, whereas the re-Ioad stage may be maintained for a much longer period (e.g. ICE
Specification). Experience shows that for best quality interpretation according to modem methods, it
is the initial loading result which is the more valuable and representative. Indeed for most purposes,
the cycling of load at such a stage would seem to serve little purpose, but it may be useful in cases
where there is some specific reason, as for example in the case of silo foundations. Settlements
from initial and re-Ioad stages should not be mixed together when plotting results.

ELASTIC SHORTENING

The techniques outlined for the analysis of pile behaviour normally employ just the pile head
displacement, applied load and the time data; without the need for expensive internal pile instrumentation.
However, where distribution of skin friction along the pile length or detailed assessment of elastic
shortening is required, specific sensors need to be used to capture the required data. Most of the
methods for detennining the full elastic shortening require elements to be cast into the pile during
installation.

Elastic shortening of the material of a pile is an important element in load/displacement performance,


especially at lower applied loads and often up to the specified working load. It is also a significant
component in the behaviour of long slender piles.

518
A very useful purpose in analysis can be served by the insertion of a short extensometer into the head
region of a test pile at such a level that there is little difference in transmitted load between the top and
bottom of the extensometer tube. This can enable the modulus of the pile material to be derived with a
reasonable degree of accuracy, although for cast in place piles, it may be necessary to do a little
excavation after testing in order to check the exact pile dimensions. The elastic shortening information
may be used to further refine the accuracy of the mathematical separation of shaft and base ultimate
loads.

INTERPRETATION OF PILE TEST RESULTS

The TIMESET and CEMSET methods in conjunction may be used to determine the controlling parameters
for any particular pile or foundation test result with good accuracy, based on high grade testing and
subject to sufficient settlement data. Indeed the problem may be regarded as a three dimensional
representation of single pile behaviour, the dimensions being Time, Load and Settlement. Basically
TIMESET may be used to determine the final settlement under each load at infinite time, thus removing
entirely the time factor from the load/settlement solution and the "fully drained" load deformation model,
CEMSET, may then be used.

1. The TIMESET model has been derived on the basis that any pile displacement/time
relationship consists of three distinct components.

a) Elastic Shortening - this can be assumed to take place


immediately upon application of load -(in reality,
elastic compression will travel the length of the pile
as the final shaft friction develops)

b) Shaft Behaviour - Modelling of shaft behaviour


requires a hyperbolic function, and is found to
conform with many test results including those from
piles tested in tension.

c) Base behaviour - Modelling is, as above, based on the


use of hyperbolic functions.

2. The CEMSET model uses hyperbolic functions to describe the shaft and the base
load/settlement characteristics determined by the founding strata and these component5 are
added to the modelled elastic shortening to accurately represent the pile behaviour under
load. The characteristic of the base and of the shaft response can be linked directly to the
pertinent soil parameters, provided settlement has been such as to mobilise a significant part
of the pile base reaction.

519
DYNAMIC TESTING

The quest for rapid and low cost pile tests led to the development of dynamic load testing. In such tests the
pile response to a high energy impact blow is recorded. Back an.alysis of the data using stress wave theory,
allows a mathematical model to replicate the measured pile behaviour. The theory assumes that the
dynamic and static elements of the mathematical model can be identified individually, and their separation is
relied upon for assessing the static component. However, the true displacement of a pile under load, other
than elastic shortening, is governed by consolidation and creep which are significant and very much time
dependent. They cannot be measured using a dynamic test, although, in some controlled conditions the
approximate immediate values which it yields may be useful.

It is difficult to see how variants on the dynamic method, such as that using progressively heavier
successive blows, can provide any advance on the more basic dynamic test. This is because, if a pile has
been "rested" before testing, upon successive restriking, it will gradually return to the driving conditions
under which only a reduced part of the final static skin friction may be resisting the pile penetration.

Statnamic and Pseudo Static load tests are those in which a high energy blow of relatively long duration is
imparted to the pile. The induced compression is prolonged sufficiently so that the full pile is presumed to
be loaded at the same time. The duration is however, still so short (approximately 0.1 second), that the
maximum displacement does not always correspond to the maximum force applied, and interpretation of
the results becomes necessary. The method of assessment of the results, from which it is hoped that the
static behaviour may be deduced, is still under development.

In interpreting all dynamic load tests some specific difficulties arise. Elastic shortening is often a major
component of settlement, particularly for high strength precast piles. The elastic modulus is often derived
by wave velocity matching. so that acoustic reflections are made to coincide with recognised or expected
soil/pile features. Where these are not distinguishable, the pile length is simply assumed to shorten in
accordance with a presumed modulus of elasticity which has not been measured. In addition, with dynamic
tests it appear generally that rupture of the soil/pile friction interface takes place and that the dynamic base
stiffness, reacting to the impulse, is generally closer to that of water than that of the soil. The relation
between statically developed soil/pile forces and those due to sudden rupture is still poorly documented and
has to be empirically based.

One is therefore faced with many complex and often currently unanswerable questions in regard to the later
generation testing systems, and while it would be unfair not to recognise them as useful, correlation
information is definitely required unless, perhaps, there is extensive existing experience in specific cases
and applications. One should therefore be cautious about finding a low price more important than a correct
answer in respect of the adequacy of static pile capacity determined by a particular test.

DEFINITION OF PILEIFOUNDATION FAILURE

It is obvious from the large number of tests and test stages analysed to date, that arbitrary definitions of
foundation failure are but a source of confusion and that, in our opinion, the only sensible definition is to use
that which defines all ultimate conditions by an asymptote parallel to the settlement axis on the
load/settlement diagram. This is the definition advocated by Terzaghi (8) which has unfortunately been
forgotten by many writers in recent years. While it is true that at large strains pile capacity may diminish
because of soil particle reorientation along the pile shaft, this definition still stands and represents real pile
behaviour.

The derivation of ultimate loads by bearing capacity theory based on plasticity implies an asymptotic
definition of failure.

520
All other arbitrary definitions may be made to yield loads corresponding to specific settlements, but the idea
that they may have any reasonably universal application has to be discarded. Settlement dependent
assessments may of course be used as serviceability states and are a function of soil stiffness as well as of
ultimate load.

True ultimate loads can only be derived if any pile tested is made to settle beyond the stage at which the
shaft friction is essentially fully mobilised. It is also necessary to mobilise a reasonable proportion of end
bearing. This may mean that to conduct good analysis, settlements in excess of say 25mm will be
necessary for traditional pile sizes and would be considerably more for large or underreamed piles.

It may be observed that even if piles are well instrumented to show the separation of shaft and base load, it
is not possible to fully determine ultimate base load and the stiffness of soil under the base without an
accurate modelling system, otherwise the contribution from the base behaviour is often found to be
underestimated. It should also be noted, that piles in certain chalks and other jointed rocks require special
consideration and interpretation because of the particular mechanics of rock block displacement.

A related issue is that of testing piles to destruction. Unless the structure of a test pile is actually damaged,
the ability of the pile-soil system to perform adequately remains. Fears that overloading of the soil, by
applying more than typical proof load values, may damage the long term performance of a pile do not
appear to borne out in reality. Indeed a stiffer response will result on reloading. The unload and reload
characteristic behaviour can be predicted with reasonable accuracy using numerical methods such as the
Cemset model.

CONCLUSION

Research clearly indicates the need for a unique definition of ultimate pile capacity which is asymptotic and
emphasises the role of stiffness in controlling settlement.

All pile testing methods for determining bearing capacity, from Continuous Rate of Penetration tests to
Wave Analysis systems, appear to introduce complications related to the inability of soils to reach a stable
state, in terms of effective stress, during the load period. This is not to dismiss such methods as being
inapplicable, but the findings from current research emphasise a need for further understanding of basic
pile-soil interaction. Even static load test results need some form of interpretation to evaluate the influence
of time and soil stiffness on the long term behaviour.

Static Pile tests, to yield good quality and useful results, may be carried out according to any Maintained
Load specification which produces sufficient well defined points on a load/settlement graph to determine the
relationship unambiguously.

Cycling of load may serve a useful purpose in certain cases. However it is much to be preferred that pile
tests are carried out by increasing load consistently, from stage to stage until completion. If unload/re-Ioad
stages are a requirement, then only the first application of load at a given value should be used in analysis
of pile performance. The second application of a specific load may be of interest, but it is not to be
confused with the initial load/settlement relationship.

The final settlement at any given load stage, as analysed by the time based model described, is
independent of any previous loading history. During the application of load, the model can reveal any
anomalies in the development of shaft and base capacities.

521
Specifications for static load tests can be improved to reduce both test duration and cycling. Perhaps more
importantly, a standard method of interpretation of results is needed across the industry. The methods
described above can provide a basis for this.

It is useful to consider a practical specification ie

1. with at least 8 equal load steps, typically to twice the working load.
2. without interim load/unload stages.
3. with specific times for application of load to include a settlement rate say of O.25mm/hr, with
the possibility of also using variable maximum load holding periods determined according to
the observed behaviour.
4. with the maintained load test being used to cover all test stages and not just up to twice the
service load.
5. with concentration on holding the last load longer.
6. with recovery also specified by rate.
7. with insistence on high quality load control and data recording.

The combination of behavioural models now available can represent the pile load/settlement characteristics
with good accuracy. Interpretation of results is practically simplified and the fundamental goal of the pile
test can be accomplished.

These models can also be employed to study the recovery characteristic of a pile on removal of load.
thereby practically eliminating the requirement for confusing unloading and reloading schedules in the pile
test specification. They have been found to be applicable to all foundation types so far examined.

522
LIST OF ILLUSTRATIONS

FIGURE 1.- DIAGRAM OF ELECTRONIC SENSORS MONITORING PILE DISPLACEMENT AND


CONTROLLING THE LOAD APPLIED.

FIGURE 2.- PLOT OF DISPLACEMENTfTlME RECORDING AND MODELLED BEHAVIOUR DURING


CONSTANT LOAD APPLICATION.

Ds- Effective diameter of the shaft; Db- Effective diameter of the base; Ws- Asymptotic value for the
shaft component; Wb- Asymptotic value for the base component; Ts- Time for the shaft to reach 50% of
the asymptotic value; Tb- Time for the base to reach 50% of the asymptotic value.

FIGURE 3.- VARIATION IN PILE HEAD DISPLACEMENT WITH VARIATION OF NOMINALLY


CONSTANT APPLIED LOAD

FIGURE 4.- TYPICAL LOAD-TIME DIAGRAM.

FIGURE 5.- TYPICAL LOAD DISPLACEMENT DIAGRAM SHOWING UNLOAD-RELOAD RESULTING IN


GREATER SETTLEMENT.

FIGURE 6.- TYPICAL DISPLACEMENT-TIME DIAGRAM.

523
REFERENCES

(1) Whitaker T., (1963), "The constant Rate of Penetration Test for the determination of the ultimate bearing
capacity of a pile", Proc.lnstn. Civ. Engrs, 26 (Sept), 119-123.

(2) Whitaker T. and Cooke, R. W. (1966)," An investigation of the shaft and base resistances of large
bored piles in London Clay", Large Bored Piles, Inst Civ. Engrs, London.

(3) Burland, J.B. and Twine, D. (1988), "The shaft friction of Bored Piles in terms of Effective Strength",
Proceedings of the Seminar on Deep Foundations on Bored and Augered Piles, Ghent, 411 - 420,
Balkema.

(4) Patel, D.C. (1992), "Interpretation of results of piles tests in London Clay", Piling Europe, ICE, London.

(5) ICE Piling Specification, (1988), Section 10. ICE London.

(6) M. England (1993) "A method of analysis of stress induced displacement in soils with respect to time",
Deep foundations on Bored and Auger Piles, pp 241-246. Balkema, Rotterdam

(7) Fleming. W.G.K., (1992) "A new method for single pile settlement prediction and analysis",
Geotechnique 42, No3, 411 -425.

(8) Terzaghi,K, (1944), "Theoretical Soil Mechanics", Second Edition. New York, Wiley. p 265.

524
Effects of Plugging on Piles Installed in an Overconsolid-ated Clay

Gerald A. Miller I and Alan 1. Lutenegger 2

Abstract

A study of the behavior of displacement piles in overconsolidated clay is


being conducted at the National Geotechnical Experimentation Site (NGES) at the
University of Massachusetts Amherst. As a part of this study, 44 pipe piles with
diameters in the range of 60 to 168 mm and embedment lengths in the range of 1.5
to 10.7 m have been installed. The methods of installation included pile driving
and jacking. Some piles were installed with 60 apex cone tips while others were
installed open-ended. Results of the test program indicated that the method of
installation had a significant influence on the amount of soil plugging during
installation of open-ended piles. In addition, it was found that the pile installation
method and degree of plugging had a significant effect on the resulting axial
compressive skin friction capacity of the pile. It appears that this effect was mainly
the result of differences in lateral stresses that developed along the pile shafts. An
effective stress method of analysis was proposed for estimating ultimate pile skin
friction. The proposed method takes into consideration the influence of installation
factors on pile capacity.

Introduction

As part of a study into the behavior of displacement piles in


overconsolidated clay, a large number of prototype pipe piles have been installed
at the National Geotechnical Experimentation Site (NGES) at the University of
Massachusetts Amherst. The main purpose of the prototype pile study was to
investigate the effect of the method of installation and the mode of penetration (i.e.,

1Assistant Professor, University of Oklahoma, School of Civil Engineering and

Environmental Science, 202 West Boyd Street, Norman, OK 73019-0631.

2Associate Professor, University of Massachusetts Amherst, Department of Civil


and Environmental Engineering, Marston Hall, Amherst, MA 01002.

525 Miller and Lutenegger


open-ended or closed-ended) on the skin friction capacity for piles subjected to first
time axial compressive loading.

A total of 44 piles were installed by either driving or jacking. Driving and


jacking were accomplished with standard exploratory drilling rigs, a tripod and
cathead apparatus, and a mobile hydraulic pushing rig. The piles had diameters in
the range of 60 to 168 mm and lengths in the range of 1.5 to 10.7 m, however,
most piles were 60 mm in diameter and 1.5 m long. The range of embedment
lengths utilized placed most of the piles within the overconsolidated zone of the
clay deposit underlying the test site.

This paper briefly describes the geotechnical characteristics of the test site;
results of laboratory and in situ soil tests; results of pile installation and load
testing; and presents a proposed method for analyzing pile skin friction capacity
using an effective stress approach.

Background

During the past decade several researchers have conducted studies using
instrumented piles to measure the stress regime which develops around a pile
during installation, reconsolidation and loading in overconsolidated clay (e. g.,
Karlsrud and Haugen 1985, Coop and Wroth 1989, Bond and Jardine 1991). In
general these studies have focused on the behavior of closed-ended piles installed
by jacking. By comparison relatively little effort has been dedicated to studying the
effects of the method of installation or the mode of penetration on the resulting
behavior. Bogard et al. (1985) developed an instrumented pile segment tool to
study the differences between open-ended and closed-ended pile penetration.
Results from tests in normally consolidated clay indicated that closed-ended
penetration resulted in higher lateral effective stress on the pile shaft relative to
open-ended penetration with no plugging.

Nowacki et al. (1992) presented a summary of results from instrumented


pile tests conducted by various researchers in overconsolidated clay. The summary
revealed a large range in the reconsolidation lateral stress, cr'he' acting against the
pile shafts as reflected by Ke values (Ke=cr'hjcr'yo) from the different studies. In
addition, Ke appeared unrelated to stress history (overconsolidation ratio, OCR).
The summary included piles installed by both driving and jacking, and closed-
ended and open-ended piles. Nowacki et al. (1992) suggested that the large range
of Ke values and the apparent independence of Ke from OCR may have been in
part the result of the different installation methods utilized and/or the differences
in the mode of penetration, i.e., closed-ended versus open-ended and different
degrees of plugging.

The research presented in this paper was developed to address the


uncertainty regarding the effect of the installation method and mode of penetration
on pile skin friction capacity, especially for piles installed in overconsolidated clay.

526 Miller and Lutenegger


Test Site Characteristics

The NGES in Amherst is a well documented test site for which numerous
laboratory and in situ tests have been conducted. Fig. 1 presents typical site
characteristics. The site is overlain by approximately 1 m of fill below which lies
a varved clay deposit roughly 25 m in thickness and known as Connecticut Valley
Varved Clay (CVVC). The CVVC is a moderately plastic, illitic clay and the
thickness of individual varves are typically between 2 and 8 mm. As indicated by
the results of incremental oedometer tests on piston tube samples shown in Fig. 1,
the top of the CVVC deposit is highly overcons01idated with an average OCR of
about 9. The OCR. decreases sharply with depth to an average value of
approximately 3 at a depth of 5 m and below a depth of 5 m the OCR gradually
decreases with depth. Below a depth of 15 m the CVVC exhibits near nonnally
consolidated behavior. Accordingly, the undrained strength, su' values detennined
with the field vane are high in the crust and decrease with increasing depth through
the crust.

Most of the piles utilized in the current study were embedded between
depths of 1.2 and 4.3 m which encompasses the moderately to heavily
overcons01idated zone of the CVVC.

0
2 .-a
.aD.
4 _lD ,,
,

,.--..... 6 .-00
E 8
'--'"
--co
.-00
..c 10 .--00
-+-'
0.. 12 .-00.
W .-00
0 14
.-0 0
16 11-0 0
18
20
22
I-
PL
e
Wn L~] -Mean ----Remolded
-Peak
o Extrapolated
0 40 80 0 4 8 0 40 80
Water Content OCR (Su)FV
(%) (kPa)

Figure 1. Test Site Characteristics

527 Miller and Lutenegger


Laboratory and In Situ Testing

One of the primary goals of the present study was to evaluate the potential
of using laboratory and in situ test results in an effective stress framework for
predicting pile skin friction in clay. For this reason, in situ tests which provided an
estimate of the in situ coefficient of lateral stress, ~ =er' hofer' yO' were conducted
extensively at the test site and included the prebored pressuremeter (PMT), self-
boring pressuremeter (SBPMT) and dilatometer (DMT) test. In addition, to provide
a range of probable lateral stress conditions around displacement piles, full
displacement devices including the full displacement pressuremeter (FDPMT),
dilatometer and push-in spade cells, were used at the test site to determine ~.
Resulting values of Ko and Ke , as well as the interpreted profiles are shown in Figs.
2 and 3. The influence of the stress history of the site on the ~ and Ke profiles is
apparent. In addition, the values of interpreted Ko and Ke converge with increasing
depth as the OCR decreases and the soil becomes softer. Assuming that ~
represents an upper bound limiting stress condition, this observation implies that
for an overconsolidated clay the value of lateral stress acting on a displacement pile
will likely be in the range of stresses defined by ~ and Ke and will be a function
of OCR. Furthermore the results suggest that for piles installed in soft normally
consolidated soil, the lateral stress will likely approach the ~ condition as Ke
approaches Ko ' This is consistent with previous observations from pile load tests
in soft clay.

0 -:::: ~::::: ........... -~::... -


00 00
2 ( h,:'o 0 0

~
~ D'\lO)
4 CD
~&
I~
,--... 6 , 0

E 8
'--'"

...c 10
-4---J
0... 12 J
Q)
I 0 DMT
0 14 I
I
- - - DMT
'V FDPMT
16 I
0 Spade
\ PMT
18 \
Cell
I 0 SBPMT
I
20 I

22
0 2 4 6 8 0 4 8 12
K0 Kc

Figure 2. K o and K c Profiles from In Situ Tests

528 Miller and Lutenegger


0
2 ,;
:
4

;----..
6
E 8
"'----"
10 .-----.-. K
0
..c
4-'
12 Kc
Q.
Q)
14
0
16
18
20 (
22
0 4 8 12 0 2 3 4 5 6
K K C /K 0

Figure 3. Interpreted K o and Kc ProfIles

In order to provide an estimate of the coefficient of interface friction


between a steel pile and the CVVC, a direct shear box test (DST) program was
completed. The purpose of the testing was to provide a likely range of effective
stress-strength parameters for the undisturbed CVVC and for the soil-pile interface.
Tests were performed on samples oriented both parallel and perpendicular to the
direction of soil deposition to evaluate drained strength anisotropy and at the same
time provide some indication of effective stress-strength parameters applicable for
soil adjacent to a full-displacement pile. In addition to the standard DSTs, tests
were performed by completely remolding the CVVC, at its natural water content,
against a steel block and shearing at the soil-steel interface.

Drained DSTs were performed on samples from three different depths


representing initial soil conditions that ranged from heavily to lightly
overconsolidated, as indicated in Table 1. Some strength anisotropy is reflected by
the friction angles reported for the highly overconsolidated sample from a depth
of 2.4 m, however, for the sample from a depth of 3.4 m, there was no apparent
anisotropy.

The effective friction angles from undisturbed specimens from all test
depths fall in a fairly narrow range from 22.8 to 25.5. The measured cohesion
indicated in Table 1 is large at depths of 2.4 m and 3.4 m where it ranges from
15.6 to 26.6 kPa. At a depth of 5.5 m the measured cohesion is lower at 2.9 kPa.
The friction angles reported here agree well with values reported for CVVC at a

529 Miller and Lutenegger


Table 1. Direct Shear Box Test Results

Test q,/ or 0' c/


Depth (m) OCR Type (deg.) (kPa) ~

Parallel 22.8 20.3 0.9988


Across 25.5 26.6 0.9973
2.13-2.74 7
Interface 20.2 -1.0 0.9995
Parallel 24.9 19.4 0.9984
3.04-3.66 6
Across 24.8 15.6 1.0000
Parallel 24.0 2.9 0.9976
5.18-5.79 2.5
Interface 21.3 1.2 0.9987

nearby site as obtained by Lacasse et al. (1972) from direct shear box and drained
triaxial compression tests. They reported friction angles in the range of 21 to 24.

The effective friction angles reported for the steel interface tests are similar
for the two depths tested. The reported values of soil-steel interface friction angle
compare favorably with values reported by Clark and Meyerhof (1972) for
remolded clay having properties similar to the CVVC.

Borehole shear tests (BSTs) were conducted to provide additional data


relating to the soil-pile interface strength characteristics. Results for both standard
serrated and smooth steel plates are shown in Fig. 4. The BST results indicate that
the soil effective friction angle in the heavily overconsolidated clay is quite high.
The average friction angle, as determined from six standard profiles at a depth of
1.2 m is approximately 35 with a range of about 32 to 42. The friction angle
gradually decreases to an average value of about 2JO at a depth of approximately
2.9 m, with a range of about 22 to 29, as determined from three standard profiles.
Although the quantity of standard BST results decreases with depth, the friction
angle data generally shows less scatter with depth as expected given the crustal
variability observed in the results of other tests.

Values of the cohesion intercept shown in Fig. 4, determined from the


standard BST results are scattered throughout the depths tested, however, the
average profile shows a general decrease with increasing depth starting from the
top of the crust, at a depth of approximately 1.2 m. Most cohesion intercept values
were low which is characteristic of the BST, and varied between values near zero
to about 9 kPa.

Fig. 4 also shows values of the friction angle determined with BST smooth
steel plates. These values are fairly constant in the crust with an average value of

530 Miller and Lutenegger


o Std. BST
--Mean of Std. BST
-----Smooth Plate BST
o

~
0,
E
'---/
o
2
...c
--+-J
0.. 3 o
Q)
0
4 ~-
-
0 o
5
20 30 40 50 0 2 4 6 8 10
(degrees) c (kPo)
Figure 4. Borehole Shear Test Results

about 23 below a depth of 1.2 m which falls in the range of results from the
standard and interface laboratory DSTs.

As observed in Fig. 4, the friction angle profiles determined with smooth


and standard BST plates tend to converge as the depth increases which is again
consistent with the decreasing OCR with depth. The disturbance to the borehole
wall caused by the augering process is greater for softer clay. Thus, in the upper
crust the radial extent of disturbance is minimized and shear failure during a BST
probably occurs in relatively intact clay when using the standard serrated plates.
At greater depths the disturbance caused by augering is greater and so the friction
angles may be more representative of remolded soil. Observations of hand augered
cuttings seem to verify an increased amount of remolding with depth. Voight
(1973) presented a correlation of the residual friction angle and plasticity index
based on many different clays. For the PI values typical for the CVVC at the test
site, the correlation predicts a residual friction angle in the range of approximately
18 to 26 for the clay crust and generally decreasing with depth. This suggests
that the friction angles determined from standard BST results may be more
representative of remolded soil for tests performed at depths lower in the crust
which is consistent with the previous discussion. At the interface between the
smooth plates and the borehole wall there is likely a veneer of remolded soil, even
in the upper crust. It is likely that lower friction angles given by smooth plate BST
test results are more representative of the interface strength between steel and
remolded soil.

531 Miller and Lutenegger


Pile Testing Program

Pile Installation - The two different methods used to install piles at the test
site were driving and jacking. Piles were installed in holes bored through the fill
and at least 1.2 m deep, in order that the embedment length would be within the
CVVc. During installation of open-ended piles, driving or jacking was stopped at
0.3 m intervals of penetration in order to allow the depth to the soil plug to be
measured. Fig. 5 presents a plot of the incremental plugging versus depth for some
60 rom diameter piles. The incremental plugging is expressed as the Specific
Recovery Ratio (SRR) defined by Paikowsky et al. (1989) as the ratio of the
incremental soil plug length divided by the incremental pile penetration. The SRR
gives an indication of the mode of pile penetration, i.e., plugged or unplugged, for
each increment of penetration. For completely plugged penetration no soil would
enter the inside of the pile and the SRR would be equal to 0%, whereas for
completely unplugged penetration the SRR would be equal to 100%. The Plug
Length Ratio (PLR) is a measure of the total plugging determined at the end of
installation and is defined as the soil plug length divided by the pile embedment
depth.

The results shown in Fig. 5 indicate a significant difference in the plugging


behavior for driven and jacked piles. The driven piles never completely plugged
while the jacked piles plugged completely after only a few increments of
penetration. Paikowsky and Whitman (1990) presented a simple procedure for

o , - 1.52 m
"V .... - 3.05 m
o I - - 4.57 m
o , , I ,
4iE--Plugged
I- Unplugged~

J.,,_--v.. _~~-&J~IT
..... [)I _ _

I
~

2
\7 0....,....-0
...c 3
-+-'
0-
(]) 4
~u ~
o o "V
o
5 o 0 Driven -
oQ Jacked
6 ""l::-_...l:U=-----L1_----'I_---J''-----'

o 20 40 60 80 100
SRR (%)

Figure 5. Typical Plugging Data for 60 mm Diameter Piles

532 Miller and Lutenegger


predicting the depth at which a jacked pile will plug completely and they predicted
that for soil conditions representative of soft to stiff clay complete plugging would
occur at a depth between 10 and 20 pile diameters. As shown in Fig. 5, results
from the current study were generally consistent with these predictions.

The differences in plugging behavior observed for driven and jacked piles
were also apparent from computed values of the PLR. The PLR for jacked piles
was generally much lower than the PLR for driven piles of the same geometry and
installed in similar soil. As will be subsequently discussed, the differences in
plugging were found to have a very significant effect on the resulting pile capacity,
especially for the piles installed by jacking.

An important observation resulting from pile installation was that piles with
larger inside diameters tended to plug less than similarly installed piles with
smaller inside diameters. The area ratio, Ar=(OD 2-ID2)/ID 2, did not appear to have
a direct relationship to the degree of plugging, at least for the range of pile
diameter and wall thickness investigated.

Another important observation was that piles embedded between depths of


1.2 and 2.7 m plugged more than similarly installed piles embedded between 2.7
and 4.3 m. Apparently, this phenomena is related to the differences in soil
properties resulting from different stress histories. The piles installed in the soil
strata with the higher average OCR tended to plug more during installation.

In addition, it was observed that the higher the computed theoretical energy
per hammer blow, the less plugging that occurred. This was revealed by plugging
data from piles installed with hammers having different weights and using different
drop heights.

Pile Load Testing - During load testing, axial compressive loads were
applied to the pile butt in increments equal to approximately 5 to 10% of the pile
failure load. The load was monitored using an electronic load cell and strain
indicator readout capable of resolving approximately 70 N. A ball and socket was
placed between the loading jack and reaction beam in order to minimize eccentric
loading. Axial displacement of the pile butt was recorded using 3 dial gages
capable of resolving 0.025 mm. Displacements were recorded immediately
following application of a load increment and at 2, 5, 10, and 20 minutes following
application of a load increment.

Each pile was load tested to failure and the failure load was interpreted
from load-displacement cUrves using the Davisson Method of interpretation, as
described by Fellenius (1980). The average skin friction, f 5 , acting on the pile at
failure was backca1culated from the failure load by first subtracting the best
estimate of pile end bearing and then dividing by the pile shaft area. The end
bearing was estimated as nine times the field vane undrained strength, i.e., 9su '
Results from load tests on pile cone tips indicated a range of end bearing from 9su

533 Miller and Lutenegger


to 18su and so some error in the backcalculated value of fs may have been
introduced by assuming end bearing equal to 9s u ' The variation in observed end
bearing is likely due to the natural variability in the clay crust which was also
reflected in results of several different in situ tests performed within the clay crust.
Before and after load tests on open-ended piles the depth to the soil plug was
measured and in general there was no plug movement detected. Therefore the pile
tip capacity was computed on the basis of the area given by the pile outside
diameter (i.e., Area=n(OD)2/4), for both open-ended and closed-ended piles.

The age of piles on testing dates was variable because some of the piles
were installed for the purpose of studying the effect of age and preshearing (prior
load tests to failure) on the pile capacity. Although these results are not discussed
. in this paper it should be noted that the age at testing, for piles ranging in age
between 1 and 10,000 hours, appeared to have little influence on the skin friction
capacity.

Pile Load Test Results

A summary of the pile test program is presented in Fig. 6 which shows a


plot of PLR versus a, where a is equal to fs divided by the average initial Su acting
along the pile shaft. f s was normalized by (sJavg. so that a reasonable comparison
could be made between results of piles embedded within different soil strata. The
results in Fig. 6 clearly show the effect of pile installation method and mode of

o Driven 0 Jacked
--Linear Regression, Driven
- - - Linear Regression, Jacked
1 .4

1 .2
0 B
1 .0 0
...
0.8 ...
ex. 0 00 .........
0
0.6 0 0
0.4
0
0.2 00 0
0.0
0 20 40 60 80 100 120
PLR (%)
Figure 6. PLR Versus ~ Backcalculated from First Time Load Tests

534 Miller and Lutenegger


penetration on the resulting skin friction capacity. The trend of data from jacked
and driven piles indicate that the capacity is a function of the plugging and that as
the PLR increases the normalized skin friction capacity decreases. This observation
seems reasonable because it is expected that the lateral stress that develops next to
a penetrating pile would be related to the amount of soil displaced radially,
especially for piles installed in overconsolidated clay.

The trend of data in Fig. 6 also clearly indicates that the installation method
has a significant influence on the pile skin friction capacity. Apparently, there were
adverse effects on skin friction capacity that resulted from the driving process.
Tomlinson (1957) discussed the importance of transverse vibrations (pile whip) that
occur during pile driving in stiff clay and the effects of the resulting gap and/or
disturbed zone of soil which develops between the surrounding soil and upper
portion of the pile shaft. It would appear that this phenomena may have contributed
to the differences in capacity observed for the driven and jacked piles during the
current study.

Proposed Method of Analysis

A number of researchers (e.g., Chandler 1968, Burland 1973, Kirby and


Esrig 1979) have recognized the importance of developing rational effective stress
methods for pile design. An effective stress analysis appears especially applicable
for overconsolidated stiff clay where pore water pressures generated during pile
loading are expected to be low (e.g., Kirby and Esrig 1979, Coop and Wroth 1989,
Nowacki et al. 1992).

Assuming an effective stress analysis is reasonable, the question remains as


to the appropriate value of lateral stress and interface friction angle to be used for
design. As discussed previously, for the jacked piles installed at the test site in the
current study it is likely that the lateral effective stress acting on the pile shafts
following equilibrium of the installation stresses, lies between the Ko and K c
conditions defined using in situ test results. It was hypothesized that closed-ended
piles installed by jacking would be acted upon by reconsolidation lateral effective
stresses represented by Kc conditions and for open-ended piles installed by jacking
the lateral stress conditions would be a function of the incremental plugging as
given by the SRR. As a starting point for driven piles, this same hypothesis was
assumed to apply. Fig. 7 shows the same data plotted in Fig. 6 along with
predictions of a based on this hypothesis. The skin friction capacity was predicted
incrementally over the pile shaft length assuming a soil-pile interface friction angle,
b', of 21 0 (from interface DST results) and using values of Ko and K c from the
interpreted profiles shown in Fig. 3. The incremental unit skin friction, fsj ' was
computed using the well known effective stress equation:

[1]

535 Miller and Lutenegger


o Driven 0 Jacked
, Predicted
--Linear Regression, Driven
- - - Linear Regression, Jacked
..... -Linear Regression, Pred.
1 .4 ~==:::;:===::;:::=====::;:::===:::;::::===:::;::::=~
1 .2

1.0

0.8
0.6
0.4
0.2
0.0
o 20 40 60 80 100 120
PLR (%)
Figure 7. PLR Versus Backcalculated a and a Predicted Assuming No
Driving Disturbance

where: K = operative coefficient of lateral stress,


cr/ vo = initial soil vertical effective stress.

cr/ vo was computed using unit weight data from the test site and in situ pore water
pressures detennined with pneumatic piezometers on pile load testing dates. The
value of K was estimated assuming that the lateral stress conditions varied linearly
between Ko and Kc as a function of the SRR such that:

[2]

The trends of backca1culated and predicted a values in Fig. 7 for jacked piles are
similar and suggest the proposed method of analysis is promising, however, for the
driven piles there is considerable disagreement between the trend of predicted and
backca1culated values of a. As discussed previously, this discrepancy may be partly
attributed to the transverse vibrations and the resulting gap that occurs during pile
driving in stiff clay. Investigating this possibility further, the gap required to
achieve agreement between predicted and backcalculated a values shown in Fig.
7 was computed for all driven piles. The results of these computations indicated
a somewhat normal distribution of gap length and gave an average gap length
equal to 50% of the pile embedment length. The fact that the distribution of
computed gap lengths was approximately normal indicates that the factor causing
the observed lower capacities for driven piles is a random phenomena. This

536 Miller and Lutenegger


observation is consistent with the random effects of pile whip and suggest that gap
formation may in fact be the factor contributing to the reduction of capacity for
driven piles, relative to jacked piles. Assuming that a gap developed around the
upper half of all driven piles, the skin friction capacities were predicted using
Equations 1 and 2 and the results are shown in Fig. 8. As expected the agreement
between predicted and backcalculated a values are significantly better for the
driven piles.

The proposed method provides a rational effective stress approach to


evaluating pile skin friction in clay and incorporates the effects of installation
method and pile plugging on the skin friction capacity. One of the drawbacks
regarding the use of this method is that incremental plugging data must be utilized
and so for design purposes the SRR profile would have to be estimated. This can
be accomplished somewhat reliably for jacked piles as the plugging behavior is
somewhat predictable, however, for the more common practice of pile driving there
currently appears to be no reliable method for accomplishing this task.
Accordingly, more research is required regarding the plugging behavior of piles in
stiff clay. Nevertheless, the method can be used to analyze piles after they have
been installed, provided incremental plugging data is obtained during installation.

It appears that the proposed method is appropriate for estimating the


capacity of jacked piles in overconsolidated clay and furthermore, provides a

o Driven D Jacked
, Predicted
--Linear Regression, Driven
- - - Linear Regression, Jacked
..... -Linear Regression, Pred.
1 . 4 ~==:;:::::::==:::;::::===::::;::===::::;::===::::;===~

1 .2

1 .0

0.8
0'..
0.6
0.4 ......
0.2
0.0
o 20 40 60 80 100 120
PLR (%)
Figure 8. PLR Versus Backcalculated a and a Predicted Assuming Gap
Formation Around Driven Piles

537 Miller and Lutenegger


rational approach for estimating the skin friction capacity for driven piles by
accounting for the formation of a gap in stiff clay. Further study involving other
clay sites is required to validate the use of this method. In addition, more
investigation is required into the effects of pile driving on shaft capacity.

Conclusions

The results obtained from the pile study have shovm that for the test site
investigated and for the range of pile geometries used, the effect of the method of
pile installation and the mode of penetration were important to the resulting skin
friction capacity. From the study several conclusions were made which apply to the
pile types and soil profile investigated, as follows.

1.) The amount of plugging that occurs during pile installation depends
on the method of installation. Piles which are driven plug less than
piles which are jacked into the ground. For driven piles the amount
of plugging is a function of the energy delivered to the pile per
hammer blow. Generally, more energy per blow will lead to less
plugging.

2.) For the range of pile diameters and wall thicknesses investigated and
for similarly installed piles, the inside diameter is an important
factor governing the amount of plugging that will occur. In general
it was observed that the larger the inside diameter the less plugging
that occurred during installation.

3.) The skin friction capacity of open-ended piles is a function of the


amount of plugging which occurs during installation. More plugging
during installation, as reflected by lower PLRs, resulted in higher
skin friction capacity.

4.) The skin friction capacity of piles depends on the installation


method. For piles of similar geometry and similar plugging, driving
resulted in lower capacity than jacking.

5.) There are currently no existing effective stress methods for


analyzing pile skin friction which provide a rational means for
incorporating the effects of pile installation method and plugging.
Such a method of analysis was proposed in this paper. The method
utilizes values of Ko and Kc determined with in situ test results and
an estimate of the interface friction angle 8' determined from
interface tests. Based on the results of pile load tests conducted at
the test site, the proposed method of analysis appears to be a
promising approach for analyzing pile skin friction capacity,
especially for piles installed in overconsolidated clay.

538 Miller and Lutenegger


Acknowledgements

Funding for this project was provided in part by the Federal Highway
Administration (FHWA). The authors gratefully acknowledge the support of Al
DiMillio and the FHWA.

Appendix I. References

Bogard, J.D., Matlock, H., Audibert, lM.E. and Bamford, S.R (1985), "Three
Years' Experience with Model Pile Segment Tool Tests," OTC Paper 4848,
Proc. of 17th Annual Offshore Technology Conference, Houston Texas,
May 6-9.
Bond, A.l and Jardine, Rl (1991), "Effects of Installing Displacement Piles in a
High OCR Clay," Geotechnique, Vol. 41, No.3, pp. 341-363.
Burland, lB. (1973), "Shaft Friction of Piles in Clay - A Simple Fundamental
Approach," Ground Engineering, Vol. 6, No.3, pp. 30-42.
Chandler, RJ. (1968), "The Shaft Friction of Piles in Cohesive Soils in Tenns of
Effective Stress," Civil Engineering Public Works Review, Vol. 63,
January, pp. 48-51.
Clark, ll. and Meyerhof, G.G. (1972), "The Behavior of Piles Driven in Clay. I.
An Investigation of Soil Stress and Pore Water Pressure as Related to Soil
. Properties," Canadian Geotechnical Journal, Vol. 9, No.4, pp. 351-373.
Coop, M.R. and Wroth, C.P. (1989), " Field Studies of an Instrumented Model Pile
in Clay," Geotechnique, Vol. 39, No.4, 676-696.
Fellenius, RH. (1980), "The Analysis of Results from Routine Pile Load Tests,"
Ground Engineering, Vol. 13, No.6, pp. 19-31.
Karlsrud, K. and Haugen, T. (1985), "Axial Static Capacity of Steel Model Piles
in Overconsolidated Clay," Proc. of the 11 th International Conference on
Soil Mechanics and Foundation Engineering, Vol. 3, pp. 1401-1406.
Kirby, RC. and Esrig, M.1. (1979), "Further Development of a General Effective
Stress Method for Prediction of Axial Capacity for Driven Piles in Clay,"
Proc. of Conference on Recent Developments in the Design and
Construction of Piles, Institution of Civil Engineers, London, pp. 335-344.
Lacasse, S., Connell, D.H. and Ladd, C.C. (1972), "Shear Strength of Connecticut
Valley Varved Clays," Research Report R72-16, Dept. of Civil Eng.,
Massachusetts Institute of Technology.
Nowacki, F., Karlsrud, K. and Sparrevik, S. (1992), "Comparison of Recent Tests
on OC Clay and Implications for Design," Recent Large Scale Fully
Instrumented Pile Tests in Clay, Proc. of the Institution of Civil Engineers,
Westminster, London, Thomas Telford Services Ltd., Publ.
Paikowsky, S.G., and Whitman, RV. (1990), "The Effects of Plugging on Pile
Perfonnance and Design," Canadian Geotechnical Journal, Vol. 27, pp. 429-
440.
Tomlinson, M.l (1957), "The Adhesion of Piles Driven in Clay Soils," Proc. of 4th
International Conference on Soil Mechanics and Foundation Engineering,
Vol. 2, pp. 66-71.

539 Miller and Lutenegger


Voight, B. (1973), "Correlation Between Atterberg Plasticity Limits and Residual
Shear Strength of Natural Soils," Geotechnique, Vol. 23, No.2, pp. 265-
267.

Appendix II. Notation

The following symbols are used in this paper:

A, = pile area ratio;


c' = soil effective stress cohesion;
fs = average unit skin friction along a pile shaft;
fsi = incremental unit skin friction along a pile shaft;
ID = inside diameter of a pile;
K = operative coefficient of lateral stress along a pile shaft at failure;
K o = coefficient of earth pressure at rest;
K c = reconsolidation coefficient of lateral earth pressure;
LL = liquid limit;
OCR = overconsolidation ratio;
PL = plastic limit;
OD = outside diameter of a pile;
PLR = plug length ratio;
r = linear regression correlation coefficient;
SRR = specific recovery ratio;
W n = soil natural water content;
Su = undrained shear strength of soil;
a = normalized average unit skin friction along a pile shaft;
8' = interface friction angle;
~' = soil effective stress friction angle;
cr' ho = initial horizontal effective stress in soil;
cr' he = reconsolidation horizontal effective stress acting on a penetrometer or pile;
cr'vo = initial vertical effective stress in soil.

540 Miller and Lutenegger


Drilled Shaft Load Test Database and an Evaluation of the Program SHAFTUF

John L. Davidson l , Lawrence D. Spears2 and Peter W. Lai3

ABSTRACT

A drilled shaft database was developed at the University of Florida using the 3-D
version ofLOTUS 1-2-3. This paper describes the database set-up, the menu driven system
and how the database can be used to evaluate a drilled shaft prediction program. Macros
are used to generate FHWA, FDOT, Davisson and Fuller-Hoy failure capacities from the
drilled shaft field load test results. Macros are also used, with the prediction program, to
calculate the predicted capacities and to provide comparison plots and statistics.

The program SHAFTUF, written at the University of Florida and based on FHWA
design methods, is used to illustrate how the database has been structured to evaluate such
a code. Comparisons between SHAFTUF capacities and those of the four load test
methods can be made on different groupings of shafts from the database, e.g., based on
shaft diameter, length to diameter ratio and soil type.

INTRODUCTION

A nwnber of deep foundation databases have been developed at the University of


Florida as part of a continuing research program with the Florida Department of
Transportation. Such databases serve two purposes. First, they are an organized, detailed
and valuable record of particular foundations. They contain information on the pile or
shaft, e.g., diameter, length and method of installation; information on the site, usually an
SPT profile; and the results of the field load test. The databases can then be used, with
suitable caution, to evaluate new foundations which have characteristics similar to those
in the database. A second use of the database is in the evaluation and modification of pre-
diction methods. lfthe necessary foundation and site information is available in a record,
the predictive method, usually a computer code, can be used to predict behavior under

1 Professor, Civil Engineering Department, University of Florida, Gainesville, FL 32611


2 Project Engineer, Williams Earth Sciences, 12290 U.S. 19 North, Clearwater, FL 34624
3 Project Manager, Florida Dept. of Transportation, Sawannee St., Tallahassee, FL 32399

541
Davidson et al. (78)
load. Then, since the actual field load test results are available, a comparison can be made
and the particular method statistically evaluated.

This paper describes a Drilled Shaft database and how it has been used to evaluate
one particular prediction code, SHAFTUF.-TheSHAFTUF program was written by
Shanmugaraj Subramanian at the University of Florida in May 1991 and updated in May
1992. It follows guidelines set forth by the Federal Highway Administration (1988) for
the design of drilled shafts. The shaft database was used to evaluate the predicted results
from SHAFfUF and compare them with the FHWA (1988) limiting criteria and the FDOT
(1991), Davisson (1972) and Fuller - Hoy (1970) failure criteria, defined from the field
load test.

THE DRILLED SHAFT DATABASE

The database, Davidson and Townsend (1994), was created using LOTUS 1-2-3
Release 3.1 in order to take advantage of the software's macro and three-dimensional
capabilities. Individual drilled shaft records are stored on successive sheets in the
database. Menu driven macros are used to manipulate the data, making it a simple matter
to update the database as new data become available and to perform statistical analyses on
the records. Each LOTUS database file is limited to 120 records for diskette storage
purposes. The database currently contains over 200 data sheet records (in two files),
however, only the first 84 contain all the parameters necessary for running SHAFTUF.

A database file consists of four major parts; the Database Directory (Sheet A), the
120 Database Records (Sheets B - DQ), the Database Macros (Sheet DR) and the Database
Template (Sheet DS). Each drilled shaft record is listed in the Database Directory (Sheet
A), which serves as a Table of Contents as well as a Summary Table. The directory
consists of 21 spreadsheet screens - three groups of seven. The first group contains
Records 1 - 40, the second Records 41 - 80 and the third Records 81 - 120. The first
screen of each group contains the assigned database number, the sheet number on which
the record is stored, and the location and engineer or reference of the drilled shaft. The
second screen contains the test date, the diameter, embedded length, method of
construction and primary soil types. The third screen contains the predicted results from
the program SHAFTUF, while the fourth, fifth, sixth and seventh screens contain the
Davisson, FDOT, FHWA and Fuller - Hoy capacities.

When a drilled shaft record becomes available, it is added into the database with
the help of a template which is created using the menu system. The template is placed
after the last database record sheet and has yellow text where data are to be entered. Red
text, which is later removed, provides guidance information. A record contains detailed
information about the shaft, the site and the field load test. Once these data are added a
menu option is chosen to automatically create an input file for the computer program
SHAFfUF. Predicted values of skin, tip and total capacities from the SHAFTUF output
file are then copied back into the record sheet. The record is then complete with regard
to user data entry and a summary of the data is copied to the Database Directory.

542 Davidson et al. (78)


The Shaft Database Macro sheet contains all 173 macros used in the database. The
vast majority of these are used by the menu system and are not individually available to
the user. The last sheet in the database contains the drilled shaft template used in the
creation of new records and all the screens used in the menu system.

The Database Menu System

The database is menu-driven. Local menus operate on a single database record


while global menus operate on a selected range of database sheets. When the database is
retrieved into LOTUS 1-2-3, an introductory screen is presented for approximately six
seconds, after which the Main Menu appears. The Main Menu lists seven options (labeled
othrough 6), Figure 1. The first option (0) allows the user to exit the menu system to the

o EXIT MAIN MENU TO CURRENT DATA SHEET.


1 SELECT SPECIFIC DATA SHEET BY SCROLLING.
1 SELECT SPECIFIC DATA SHEET BY DATABASE NUMBER.
3 GO TO LOCAL DATA SHEET MENU.
4 GO TO GLOBAL MENU.
S CREATES TEMPLATE FOR NEW SHAFT DATA SHEET.
6 EXITS DATABASE.

CURRENT DATA SHEET NUMBER 1

SELECT MENU OPTION: ISm~~;'~lit~

Figure 1 Shaft Main Menu

current data sheet, the number of which is shown on the menu. Options 1 and 2 allow
the user to select a database sheet number to work with by either scrolling through the
data sheets or by entering a specific database sheet number. Option 3 accesses the
Local Menu operations and Option 4 the Global Menu operations, Option 5 places the
data sheet template after the last recorded data sheet and makes this template the
current data sheet. The last option, (6), provides an exit from the database.

543 Davidson et al. (78)


Local Menu Functions

The Drilled Shaft Local Menu is shown in Figure 2. The menu functions operate
only on the current data sheet, i.e., on a single shaft - in this case Shaft Number 1. Option
1 provides a checklist of available data on the current data sheet. Option 2 creates an input
file for SHAFTUF. Option 3 creates load-settlement failure capacities based on the insitu
load test data. Option 4 updates the database Shaft Directory Sheet with information from
the current data sheet. Options 5, 6, and 7 are used for accessing the Local Print, Load-
Settlement Plot and Comparison of Capacities menus, respectively.

o EXIT TO CUIlRENT DATA SHEET.


I VIEW CHECKLIST OP CURRENT DATA SHEET.
2 CREATE INPUT FIT.E POR SHAPTUP PROGRAM.
3 CREATE LOAD SETTLEMENT CAPACITIltS.
.. UPDATE SHAFT DATABASE DIRECTORY.
5 GO TO LOCAL PRINT MENU.
6 GO TO LOAD SETTLEMENT PLOT MENU.
7 GO TO COMPARISON SELECTION MENU.
S RETURN TO MAIN MENU.

SELECT MENU OPTioN: 5 I

Figure 2 Shaft Local Menu

The Local Print Menu is shown in Figure 3. Option 1 prints all five pages of data
sheet shaft information with report quality. Options 2, 3, 4, 5 and 6 print specific pages,
with the information as stated on the menu. Option 7 prints all data sheet information with
draft quality (3 pages).

The Load - Settlement Plot Menu, Figure 4, allows five different plots. Option 1
plots only the load-settlement data while Options 2 through 5 add the failure criteria of
choice. Option 6 prints a hard copy ofthe last viewed plot. Figure 5 is an example, a load-
settlement plot showing the Davisson capacity construction, i.e., Option 2.

The Comparison of Capacities Menu, Figure 6, provides options of plotting


SHAFTUF results versus the methods of failure criteria in different formats. Figure 7 is
an example bar diagram.

Davidson et al. (78)


544
LOCAL PRINT MENU

o EXIT TO CURRENT DATA SHEET.


1 PRINTS ALL DATA SHEET INFORMATION.
2 PRINTS ONLY SHAFT DATA INFORMATION..
3 PRINTS ONLY LOAD SETTLEMENT DATA.
PRINTS ONLY INSITU TEST RESULTS.
S PRINTS ONLY INPUT DATA FOR SHAFTUF PR.OGRAM.
6 PRINTS ONLY SHAFTUF PR.OGRAM RESULTS.
7 PRINTS ALL DATA INFORMATION (DRAFT QUALITY).
8 RETURN TO LOCAL MENU.

SELEC~ MENU OPTION: LI ~~-,,5'------J1

Figure 3 Shaft Local Print Menu

o EXIT TO CURRENT DATA SHEET.


I LOAD-SETTLEMENT PLOT FROM LOAD TEST.
2 LOAD-SETTLEMENT PLOT WITH DAVISSON CRITERION.
3 LOAD-SETTLEMENT PLOT WITH FDOT CRITERION.
LOAD-SETTLEMENT PLOT WITH FHWA CRITERION.
S LOAD-SETTLEMENT PLOT WITH FULLER-HOY CRITERION.
6 PRINTS CURRENT PLOT.
7 RETURN TO LOCAL MENU.

~: ": ' : ' , ~ , r, ~ ' : ': '.

, ~-;
SELECT MENU OPTION': 'L[,-,---"-',-',:,,;sL..Jl

Figure 4 Shaft Load-Settlement Plot Menu

545 Davidson et aL (78)


----III

lII~iT h,
nJO}T
.'!'i"'' JI'' '.'
ii'n'A 11t1IilW'l',
lu,,,it"i'!iil' """ '....;ili"O",
lIj,gT ii"'~""'i!j1 Yii, Fit.:l'l\
.i.QT .. I>o\,i','<I' ev,;, ItRWA-,
'''i)T PH....'.'Ii' . . Ii, .. If/,...... ~ Ii,,;v,
J'i,i)T ""i", v"n,IjlOiil (!"i'A.,jl~ii'"
, ..QT ...1',1' ':'lIlil.",,, I2A!PA~iiHl!Ii
.""Ii . . .
i,,~" ('AAPRj,
"vit;.
JiAiN'i' ",,,-"/tii;';'" J;4,.i'l',

II:I:1iili:!iiil;:ii;i!:i!iii~l:li'i!!II":i:,i:T;'1,/~~;\;M~li;Tll;A
/ 'l 'I"{j ~6 M" 1\ "/il6'" "',i, .j(~i"i~ I'i

fjnvid~(l11 g( ul. (18)


LOAD (TONS)
o 100 200 300 400

Figure 7 Bar Graph for Comparison of All Failure Capacities

Global Menu Functions

The main function of the Drilled Shaft Global Menu, Figure 8, is to perform
statistical analysis on all the shafts (or on a selected group) in the database. Option 1 is for
plotting and evaluating data without any of the restrictions that are imposed in Options 2,
3 and 4. When Option 1 is selected, a menu appears and the user enters information
concerning the characteristics of the data to be analyzed, Figure 9. The database sheet
number range, method of comparison, criterion (failure or design values) and selection of
either analyzing capacity or settlement are prompted for. The data that fit the parameters
are collected from the Database Directory sheet and a plot of SHAFTUF predicted capacity
versus the chosen criterion capacity shown. Pressing Enter exits the plot and provides a
table of characteristics and statistics. Finally a menu prompts the viewer to either continue
without a print or print the statistics and plot on a single page with or without data-labels.
Figure 10 is an example print out.

Option 2 on the Global Menu is for plotting and evaluating data in a particular
diameter range. A menu, Figure 11, prompts the user to enter the desired diameter range.
The minimum and'inaximum diameters available are shown on the menu. The data that
fit the diameter range are collected and prepared for viewing. Again a capacity
comparison plot and statistics table can be printed. Option 3 is similar to Option 2 except
that length-to-diameter ratios rather than diameters are chosen.

547 Davidson et al. (78)


o EXIT TO DATABASE DIRECTORY.
1 PLOT BY DATABASE SHEET NUMBER RANGE.
2 PLOT SHAFT DIAMETER RANGE.
3 PLOT BY LID RATIO RANGE.
~ PLOT BY SOIL TYPE.
5 GO TO GLOBAL PRINT MENU.
, RETURN TO MAIN MENU.

SELECT MENU OPTION: 5

Figure 8 Drilled Shaft Global Menu

',. ;DATA~~SE SHEET NUMBItR RANGE

AVAILABLE DB SHEET NUMBER RANGE: I-no


METHOD CRITERION CAP /SET OPTION
1 DAVISSON 1 FAILURE 1 CAPACITY
2 FDOT 2 DESIGN 2 SETTLEMENT
3 FHWA
4 FULLER-HOY

ENTER MIN DB. tI RANGE: Im~""'':2~~iiji 'ENTER METHOD, C''''''"';21,: I

ENTER MAX DB 1# RANGE: m;r@J',LU:j ENTER CRITERIONf!2"iij :",-,j:!'j

ENTER CAP/SET:

Figure 9 Shaft Database Sheet Number Range Menu

548 Davidson et al. (78)


COMPARISON OF CAPACITIES
SHAFTUF PREDICTED Vs. FDOT FAILURE
1200
Vi'
z
0
t:- 1000 /
;>-<
-
t-<
U
<C 800
/
/
0... 0
<C
U 0
Q 600 0~

/
~
E-<
U
........ 0
Q '" IX] 0 0
~ 400

~
~ 0 0
0... o 0:;J
o 0
~
""'~

yo
:::::l 200 L
E-< "
~ 00
<C
::r:
if)

200 400 600 800 1000 1200


FDOT FAILURE CAPACITY (TONS)

. . . .. .. ,.. .. . , .,...
, .,
. ..
... .. . . , . . . . . ..
.- .... . . . .
... , . ,.., - , ., ... , . ,,
.. -... . ., .
- .
.... "CHA RAcl"IoA:l
.. .. .
Tics .sTA:li5T:i:C:so~
. .. ,
pLoT:: ..

METHOD: FOOT DATABASE # RANGE: 1 - B4


CRITERION: FAILURE
CAP / SET: CAPACITY

MIN SHAFTUF: 9.7 TONS NUMBER OF EVENTS: 55


MAX SHAFTUF : 832.7 TONS
AVG SHAFTUF: 325.8 TONS
MIN " SHAFTUF / METHOD:
MAX " SHAFTUF / METHOD:
38.6
217.4""
MIN METHOD:
MAX METHOD:
5.1
964.8
TONS
TONS
AVG (LOG) " SHAFTUF/METHOD:
STANDARD DEVIATION (lOG):
ERROR OF ESTIMATES:
98.2
17.1 ""
150.8 TONS
AVG METHOD: 362.8 TONS

Figure 10 Typical Comparison Plot and Statistics Table

549 Davidson et al. (78)


A V AILABLE DIAMETERS IN RANGE OF DB ~ RANGE

MINIMUM DIAMETER MAXIMUM DIAMETER

5.00 INCHES 54.00 INCHES

ENTER. MIN DIAMETER. FOR. RANGE: tE~,~t"l"a1fH

ENTER MAX DIAMETER. FOR RANGE: ru"Slf6tij'j1

Figure 11 Shaft Diameter Range Menu

Option 4 is for evaluating data by soil type. The procedure is similar to that for
Options 2 and 3. Three soil types are considered -- sand, clay and rock -- for both primary
side soil and soil at the base of the shaft. Sand has been given the designation "1".
Therefore, if a shaft were embedded in sand only (side and tip) it would have the
designation "11 ", indicating that the primary side and base soils were sand. Clay has been
designated as "2" and rock as "3". Although there are nine available soil type designation
options (see Figure 12), a maximum of six can be analyzed at one time. The first prompt
on the menu is to enter the number of different soil types to be analyzed, The cursor then
moves to the Soil Type Choice position, where the soil type designations are entered. If
the user enters "2" different soil types to analyze, then the cursor will move to the Accept
Entries position after selecting the second soil type designation. The available soil type
designations are shown in red while the unavailable soil types are shown in blue. A count
of soil type designations is shown to offer assistance. After entering each soil type, the
color of the number designation on the menu changes from red to blue, indicating that the
soil type is no longer available for selection.

Option 5 accesses the Global Print Menu which regulates printing of the Database
Directory, As noted previously, a complete printout of this directory totals 21 pages. The
menu allows the user the option of printing all or only certain ranges of the directory.
Another Option prints the macros names accessible to the user, the key strokes required
to run each macro and a brief description. The final Option prints all database macro
names and the location in the database where each macro can be found.

550 Davidson et ai. (78)


PRIMARY SIDE SOIL / BASE SOIL
11 SAND/SAND 21 CLAY/SAND 31 ROCK/SAND
12 SAND/CLAY 22 CLAY/CLAY 3Z ROCK/CLAY
13 SAND/ROCK 23 CLAYIROCK 33 ROCK/ROCK
RED ,,: SOIL TYPE IS AVAILABLE IN DB" RANGE.
BLUE ,,: SOIL TYPE IS NOT AVAIT..ABLE IN DB" RANGE OR
HAS BEEN PREVIOUSLY SELECTED.

ENTER II O F . 11~i':"i:;;):j'3.;j ENTER SOIL Ii":'fi,i: :f:23'" J .


DIFFERENT SOIL . TYPE CHOICE.
TYPES TO VIEW: . NUMBER. . . . 1.
(6 MAX)

Figure 12 Soil Type Range Menu

EVALUATION OF SHAFTUF PREDICTIONS

It is not the intent ofthis paper to specifically evaluate the program SHAFTUF but
rather to illustrate, using SHAFTUF, how the database has been structured in order to
evaluate any such code. As already noted, database macros are employed to generate, from
the drilled shaft field load test results, the FHWA, FDOT, Davisson and Fuller-Hoy failure
capacities. Macros are also used with the SHAFTUF program to calculate the predicted
capacities. The Global Menu system then allows comparison plots and their statistics to
be generated for different groupings of shafts, e.g., by diameter, length to diameter ratio
and soil type.

Figure 10 illustrated a typical output. In this case the SHAFTUF predicted capacity
was being compared to the FDOT failure capacity (a variant of the Davisson criterion).
Data points located above the 45 degree line represent unconservative SHAFTUF
predictions versus the method of failure used in the comparison. Points falling below the
45 degree line represent conservative results and points falling on the line display perfect
agreement. The statistics table shows that in this case there were 55 shafts available for
comparison. The table also provides a number of minimum and maximum values,
averages, a standard deviation and an error of estimates. The program SHAFTUF on
average predicted a capacity equal to 98.2% of the capacity determined from the field load
tests and using the FDOT criterion. The standard deviation was 17.1%.

The ratio of SHAFTUF predicted capacity to the capacity from the method being

551 Davidson et al. (78)


analyzed, designated JR, is not normally distributed but is skewed. If underprediction
occurs, JR has a value between zero and one. If overprediction occurs, JR can have a value
between one and infinity. Ratio JR is more closely log normally distributed. Average and
Standard Deviation values were therefore calculated in the database using the equations:

Avg JR = ~
10 n

STD L (log R - log Avg R )2 ]1/2


[ n - 1

Table 1 is a summary of SHAFTUF predictions compared with the four failure criteria.

Table 1 Statistics for SHAFTUF Predicted vs. Measured Failure Criteria for Data-
base Range 1 to 84

Number Error of
Min IR Max JR Avg JR Stand.
Method of Estimates
Dev.
Events (tons)
% % % %

FHWA 42 36.5 197.9 86.2 18.0 170.4


FDOT 55 38.6 217.4 98.2 17.1 150.8
Davisson 64 31.1 306.6 109.4 18.5 177.4
Fuller-Hoy 84 28.5 319.1 87.7 20.8 294.6

Comparisons by Diameter Range

The smallest diameter of the 84 drilled shafts examined was 14.0 inches and the
largest was 48.0 inches. This 14- to 48-inch range was divided into six diameter ranges.
Statistics for each range were collected to see if any trends in the data were evident based
on drilled shaft diameter. Table 2 is a tabulation of the results for all six diameter ranges
for the FDOT comparison. Figure 13 is a plot of average JR values (from Table 2) versus
diameter of shafts for the FDOT criterion. Based on the records available in the database
it would appear that SHAFTUF unconservatively overpredicts capacities for small
diameter shafts and conservatively underpredicts for large diameter shafts. Similar
comparisons and plots can be made based on shaft length to diameter ratio.

552 Davidson et al. (78)


Table 2 Statistics for SHAFTUF Predicted vs. Measured FDOT Failure
Criterion by Diameter Range

Error
Diameter Number Min lR MaxlR Avg lR Stand.
of
Range of Dev
Estimates
Events
(tons)
(in) % % % %

14 - 18 4 118.6 217.4 155.1 13.5 24.9


18 - 22 2 118.6 119.7 1192 0.3 33.9
22 - 26 16 57.0 177.9 116.4 13.5 82.3
26 - 32 18 38.6 183.5 95.3 19.0 189.7
32 - 40 15 50.1 115.6 78.8 12.0 173.4
40 - 48 2 59.1 81.3 69.3 9.8 143.6

250

,-..

~ 200
~
::>
...:l .\
<
;> 150
..: \
~
~
"
;2
100
"-
-----------.
><l
;>
< 50

0
o 10 20 30 40 50

AVERAGE DIAMETER IN RANGE (INCHES)

Figure 13 Plot of Average lR Values vs. Diameter Range for FDOT Comparisons

553 Davidson et al. (78)


Comparisons by Soil Type Range

There are a total of nine possible soil combinations using the Soil Type option on
the Shaft Database Global Menu. Currently, there are no data in the database that fit the
Rock/Sand or Rock/Clay soil types, and likely never will be. Therefore, seven soil type
combinations can be evaluated. Table 3 is a tabulation of the results for the FDOT criterion
and Figure 14 a bar graph comparison. The Sand/Sand, Sand/Rock and Clay/Clay groups
have unconservative lR values but within 10% of perfect agreement. The other four soil
type groups all have conservative average lR values: Clay/Sand 82.2%, Sand/Clay 79.6%,
Clay/Rock 78.1% and RockIRock 65.7%.

Table 3 Statistics for SHAFTUF Predicted vs. Measured FDOT Failure Criterion
by Soil Type Range

Error
Soil Number MinJR MaxJR Avg JR Stand.
of
Type of Dev.
Estimates
Range Events
(tons)
% % % %

Sand/Sand 18 38.6 217.4 107.3 205 170,3

Sand/Clay 3 59.1 106.7 79.6 12.8 121.1

Sand/Rock 5 87.9 130.6 105.7 7.9 77.7

Clay/Sand 6 57.0 104.3 82.2 9.2 141.4

Clay/Clay 20 47.0 183.5 102.0 17.4 140,6

Clay/Rock 1 78.1 78.1 78.1 0.0 210.9

RockIRock 2 50.4 85.8 65.7 16.4 217.6

554 Davidson et al. (78)


107.3

100
~

-0~

( oj
;;l
...l
-<
:>
t:rr: 50
(oj

"-<
t:rr:
(oj
:>
-<

0
=
z >-
-<
::Ill
= >-
-<
:Ill ::Ill
-<
~
...l
~ '"
0
Z
-< ...l
'"
0 '"
0

=
Z =
z
E!!!
=
z
~
>-
-<
'-<"
;;; E!!!
>-
E!!!
::Ill
-<
til -<
""'
-< ...l ...l -<
...l '"
0
til
'" '" '"
t:rr:

Figure 14 Average lR Values vs. Soil Type Range for FDOT Comparisons

CONCLUSIONS

A drilled shaft database has been developed. It forms a record of foundation


members, their sites and field load tests. Menu-driven macros are used to manipulate the
data, making it a simple matter to update the database as new data become available and
to perform statistical analyses on the records.

Computer prediction codes, such as SHAFTUF, can be easily .evaluated by


comparing predicted capacities with capacities determined from the field load tests.
Comparisons can be made on selected groupings from the database, e.g. by diameter,
length to diameter ratio or soil type. This allows the developer of the code to identify
specific shortcomings and modify his program for particular situations.

REFERENCES

Davidson, 1. L. and Townsend, F. c., Maintenance of Load Test Data Bases, Final Report
to Florida Department of Transportation, Gainesville, Florida, 1994

Davisson, M. T., High Capacity Piles, Lecture Series, ASCE, Illinois Section, 1972.

FDOT Publication 455, Structures Foundations, 1991

FHWA Publication No. FHWA-HI-88-042, 1988.

Fuller, F. M. and Hoy, H. E., Pile Load Tests, HRB 333, 1970.

Davidson et al. (78)


555
A New Approach to the Prediction ofDrilled Pier Performance in Rock.

Julian P. Seidel and Chris M. Haberfield 1

Abstract

The load-deflection performance of drilled piers in weak rock is a function of


many variables. These include the intact shear strength parameters of the rock and the
shear strength of the concrete-rock interface, the rock mass modulus, Poisson's ratio and
pile diameter, the initial normal stress imposed by the wet concrete on the rock, and the
surface roughness of the drilled pier. Research at Monash University has been aimed for
many years at developing a fundamental understanding of the contribution of each of
these factors to the performance of drilled piers in rock. This paper describes the theor-
etical aspects of a new approach to the prediction of drilled pier performance in rock,
and introduces a new computer program, called Rocket, which combines all these
aspects.

Introduction

The design of piles socketed into rock is traditionally based on local knowledge
derived from observation of full scale static load tests, empirical factors related to the
unconfined compressive strength of intact rock, or conservative city or state ordinances.
The uncertainties inherent in such approaches must inevitably lead to either a reduced
level of confidence that the foundation will safely or satisfactorily perform its function of
supporting the superstructure, or to overdesign, and thus additional expense to ensure
that a suitable level of confidence is achieved.

The design philosophy for piles socketed into rock has a relatively short history.
Initially, the tendency was to design drilled foundations to rock, allowing for end-
bearing only. Presumably it was recognized that the ultimate pile resistance, at least for
short sockets, was dominated by the end bearing component.

Specific allowance for shaft resistance in rock sockets start to appear in the
literature of the mid 1960's. Thorburn (1966), Freeman et al. (1972), Whitaker (1976)
and Tomlinson (1977) quote allowable shaft friction values of between 100 kPa (14.5
Lecturer and Senior Lecturer, Department of Civil Engineering, Monash
University, Clayton, Victoria 3168 Australia.

556 Seidel and Haberfield


psi) and 1000 kPa (150 psi) for different rocks with a variety of reported weathering and
jointing conditions. Although the importance of shaft resistance (especially to working
load perfonnance) was realized, recommendations were site-specific and unrelated to
any design philosophy whereby these results could be reasonably extrapolated.

From the mid 1970's, the technical literature reveals an attempt by researchers
and others to relate shaft resistance to the unconfined compressive strength of the rock,
qu - the use of a factors was borrowed from pile design in clay, and values were
extrapolated from these recommendations. Pells et al. (1978) and Poulos and Davis
(1980) suggested allowable shaft resistance of 0.05 qu (i.e. a =0.05), although data from
Thorne (1977) suggested substantially higher and lower values were possible.

Rosenberg and Journeaux (1976) suggested ultimate adhesion values of 0.05 qu


for high strength rocks [q u = 70 MPa (10 ksi)] increasing to as much as 0.3 qu for very
soft rocks [qu = 0.7 MPa (0.1 ksi)] based on their own tests. Williams et al. (1980) also
provided a reasonably comprehensive summary of selected results of socket resistance
tests in mudstone, shale, and sandstone, from around the world as reported in the techni-
cal literature. On this basis, adhesion factors as high as 0.8 for unconfined compressive
strengths of 0.6 MPa (0.09 ksi) or less, reducing to less than 0.1 for unconfined
compressive strengths of 30 MFa (4.4 ksi) or greater were proposed (see Figure 1,
where similar correlations by Horvath (1978) are superimposed).

The correlations developed were not entirely satisfactory, and researchers


(Horvath et aI., 1983 and Pells et aI, 1980) realizing the importance of socket roughness
attempted to empirically incorporate roughness into the shaft adhesion a factors - Pells

1.0

.9 0.8
u Williams, Johnston
<fl
c: 0.6
and Donald (1980)
.9
<I:l ,
1l
~ 0.4
"/,,-
0.2 Horvath ........
(1978)
o L---.l..-..L..-L......L.._--l...-_..L....-L.......l.-..L....---L._-J.--'---l-J

0.2 0.4 1 2 4 10 20 40 100


Uniaxial compressive strength (MPa)
Figure 1 Shaft adhesion factors after Williams et al.(1980) and Horvath (1978)

557 Seidel and Haberfield


by means of crude roughness classes and Horvath by a so-called roughness factor which
was determined by field measurement. Rowe and Annitage (1984) published the most
comprehensive correlations between socket shear and unconfined compressive strength
taking roughness into account. Figure 2 shows the correlations obtained for regular
sockets (classification < Pells roughness R4), together with th~ recommendations of_
Horvath (1982), Horvath et al. (1983) and Williams et al. (1980). The spread of data
about the least squares line of fit correspond to almost to an order of magnitude(!) -
hardly conducive to confidence and economy in design. Presumably, this wide spread is
a direct result of the empirical models not accounting for the range of parameters which
are critical to the prediction of available socket resistance.

It is finally noted that the period around 1980 saw the development of elastic
solutions by Williams (1980), Pells and Turner (1979) and others for side-only, base-
only and complete piles using the results of finite element analysis. These solutions

-0.43
a= O4
. qu 6.

....
9()
~ 0.1
I:: LEGEND
.9
<:Il
o Mudstone,shaJe
<U 6. Sandstone
.c::
~
Indicates test not to failure
Shaded symbol denotes a tension (pull-out) socket

CORRELATION FROM:
- - - - Linear regression
._ . _ . _. Horvath et al.(1983)
- - - - Horvath (1982)
- - - - - - Williams et al. (1980)

1.0 10 40
Unconfined compressive strength, q u (MPa)

Figure 2 Socket resistance correlations for class RI-J sockets (after Rowe and Annitage, 1984)

depend heavily on the constitutive relationship used at the interface between the pile and
the surrounding rock. Without a proper understanding of the mechanisms of shear
development at this interface, these solutions cannot be reliable. Rowe and Armitage

558 Seidel and Haberfield


(1984) conceded that the use of empirical cOlTe!ations would be more appropri.ate given
th~t the model predictions depend heavily on input parameters which are dift1cult to
obtain with sufficient reliability.

The potential advMtages of a rational, scientifically-based method for design of


rock-socketed piles to the cost of construction are obvious. However, because of the
difficulties associated with understanding the fundamental mechanisms of behaviour of
drilled piers, !\uch an approach has not been possible to date.

J'he Mechani.~ms of Shear BehaviolJLat the Concrete-Rock Interface

The load-deflection perfonnance of drilled piers in weak rock is fundamentally a


function of both the intact shear strength parameters of the rock and the shear strength
of the concrete-rock interface, the rock mass modulus, Poisson's ratio and pile diameter,
which together define the constant normal stiffness confinement around the shaft, the
initial nonnal stress imposed by the wet concrete on the rock, and the surface roughness
of the drilled pier. Research at Monash University has been aimed for many years at
developing a .fundamental understanding of the contribution of each of these factors to
the performance of drilled piers in rock. This research has been based on extensive
constant nonnal stiffness direct shear testing of interfaces between concrete and various
weak rocks. This testing program is described in more detail in a companion paper at
this conference (Haberfield et al., 1994b).

The approach of the research group has been to determine the mechanisms of
shear behaviour at the concrete-rock interface by careful observation, and to translate
these mechanisms into theoretical models which simulate the observed behaviour.
Behaviour of the interfaces has been recorded using time-lapse video cameras through a
25 mm (1 in.) gap between the split shear box halves. The important parameters wbjch
have been modelled are described in the following sections.

Constant Nonnal Stiffness

As noted previous.ly, a critical aspect of the perfonnance of drilled piers in rock


can be attributed to the roughness of the interface between the shatl and the surrounding
rock mass, In the process of constructing a dri.lled pier in rock, once the level of rock
has been attained. and any casing necessary to stabilize overburden ha~ been sealed into
the rock, a void, or socket is progressively drilled into the founding rock. The surface of
the socket so form~d has a roughness that may vary from minimal to significant depend-
ing on the drilling method, the rock type and quality, and the rock jointinR Grooving of
the socket may even be a specific requirement of the construction procedure.

On completion of the socket (and after any cleaning operations ofthe socket wall
or base, and installation of reinforcement) concrete is cast into the socket, either by use
oftremie or by allowing the concrete to free fall into the void. The concret~ may be cast
only to the level of the rock socket, or may be extended in one pour to the pile cut-off

559 Seidel and Haberfield


level, where connection with the superstructure will be made. The objective of the
concrete pour is to produce an intimate contact with the base and walls of the rough
socket.
..I
1

1:1 -:.:' .

(a) Pile before displacement

(c) Equivalent 2-D model (d) Equivalent 2-D model


for before displacement for after displacement

Figure 3 Idealized displacement behaviour or a drilled pier in rock

An idealized section of the rock socket after casting of the pile is shown in
Figure 3(a). The nonnal stress imposed by the concrete on the rock is shown as O"na' On
application of an axial load to the pile from the superstructure, the pile and rock mass
will displace elastically until such time as the shear stress at the interface causes slip.
Figure 3(b) shows the same pile section after a slip displacement of the pile relative to
the rock. Geometrical constraints require this sliding displacement to generate a dilation
of the interface, and an increase, (MJ = 2.1r), in the socket diameter. This dilation
occurs against a surrounding rock mass that must defonn to compensate for the enlarge-
ment of the socket diameter, and an increased normal stress at the pile/rock interface
results (Johnston, 1977; Johnston and Lam, 1989)

The expansion of the rock socket can be approximated to the expansion of an


infinite cylindrical cavity in an infinite elastic space. Accordingly, the increase in nonnal
stress, .1O"n' for a socket can be related to the interface dilation, .1r, as:

A
UO"n = 1 E+Vm m .1r
r
(Eq. 1)

560 Seidel and Haberfield


where r is the original rock socket radius and Em and Vm are the rock mass modulus and
Poisson's ratio of the rock, respectively. This expression can be rearranged to compute
the normal stiffness, K, as follows:

K = ~crll = Em
~r r (l +V m ) (Eq. 2)

As the increases in socket radius, ~, are much smaller than the initial radius, the
normal stiffuess, K, can effectively be assumed to be constant. The behaviour of pile
sockets is thus modelled as being governed by a constant normal stiffness (CNS) condi-
tion. It should be noted that the more commonly used constant normal stress or load
(CNL) condition in direct shear tests is only a particular case of the CNS condition with
zero stiffuess.

The eNS condition is extremely important to the work strengthening behaviour


of rock socketed piles, as progressive slip displacements of the pile prior to peak resis-
tance cause increasing normal stresses, and therefore increased interface strength. This
condition has been incorporated as a fundamental aspect of both the experimental and
theoretical work of the Monash Group.

Initial Normal Stress

The initial normal stress imposed on the socket walls is primarily a function of
the depth of concrete cast continuously above the socket. The concrete is assumed to
act hydrostatically against the walls of the socket, with a pressure proportional to the
total height of concrete poured, and the density of the concrete. In reality, the normal
stress applied to the socket is a complex function which is dependent on the rate of
placement of the concrete, arching effects of the concrete aggregate, the rate of harden-
ing, the degree of compaction, and any setting shrinkage of the cement (Taylor, 1965;
Clayton and Milititsky, 1983). However, in the absence of reliable methods to account
for these effects, the simple hydrostatic assumption has been adopted.

Current empirical methods of predicting socket strength do not explicitly incor-


porate the effects of socket dilation, which is roughness-dependent, and the variation in
normal stress caused by the constant normal stiffness condition. Furthermore, socket
diameter is an important factor in determination of the constant normal stiffness, but is
not considered in empirically-based socket design recommendations.

It is noted that the performance of drilled piers in rock may be improved by the
use of expansive concretes, which can substantially increase the initial normal stress
(Haberfield et al., 1994a).

561 Seidel and Haberfield


Initial Slip Displacements of the Interface

As noted previously, on application of an axial load to the pile from the super-
structure, the pile and rock mass will displace elastically until such time as the shear
stress at the interface causes slip. In detennining this critical shear stress, many
researchers (e.g. Patton, 1966; Ladanyi and Archambault, 1970) have idealized rough
rock surfaces as a set of constant angle triangular asperities. Indeed, early models of the
Moansh'research group were based on the same assumptions (Johnston and Lam, 1989).

The shear stress at which slip is initiated on these triangular asperities, 't, is a
function of the shear strength parameters relevant to the planar concrete/rock interface
(cAl), the normal stress acting across the interface, a", and the inclination of the inter-
face to the socket axis, e. The eNS direct shear tests of Seidel (1993) on interfaces
comprising simple triangular asperities of inclination, e, confirmed the shear models of
Patton (1966) and Ladanyi and Archambault (1970) for unbonded, purely frictional
surfaces, i.e.

(Eq. 3)

For actual rock sockets, surfaces will comprise asperities at a range of inclina-
tions rather than a constant value. It has been shown further by Seidel (1993) that for a
socket of total area A in an elastic medium comprising a distribution of n asperities, with
individual contact areas a j and local normal stresses orz.j the slip shear stress can be
computed as follows:

1 j=n
1: =A L[aj On,j tan(<j>s + 8 i )] (Eq. 4)
i=l

The tests performed in the Monash research program have all been on unbonded
concrete/rock surfaces. This has been considered necessary because of the unreliability
of bonding between concrete and rock due to construction procedures - wall smear,
softening of the socket and the use of stabilization fluids will all act to prevent the
formation of any bond. Research is currently being performed to evaluate the factors
affecting bond, and the effect of bonding on the interface behaviour. Bonding at the
interface may not improve ultimate capacity, as dilation of the interface may be
prevented.

Empirical shaft adhesion predictions do not incorporate any interface shear


strength parameters.

Failure ofInterface Asperities

After the initiation of interface slip, the contact area between the concrete and
the surrounding rock gradually reduces from full contact area, to smaller contact areas
as shear displacement progresses. This is demonstrated in Figure 4 for a simple 2

562 Seidel and Haberfield


~

~ i ~
concrete ~ concrete
~ ~

rock rock

Displaced Position

Figure 4 Reduction of asperity contact area with progressive shear displacement

dimensional model of an interface comprising regular triangular asperities. Local normal


stresses increase both as a consequence of the reduced contact area, and as a result of
the interface dilation in combination with the constant normal stiffness condition. A
critical normal stress is reached at which the asperity can no longer sustain the loading,
and individual asperity failure results. Observation of the video records clearly showed
this failure to be rotational. It was evident that individual asperity failure was analogous
to failure of a slope loaded by an inclined surcharge. A closed-form solution for an
inclined load on a weightless c,~ slope (Sokolovsky, 1960) was successfully applied to
the prediction of asperity failure for tests on regular triangular asperities.

The prediction of the failure stress for individual asperities was based on triaxial
test measurements of the drained intact shear strength parameters for the rock, c' and ~'.
Drained shear strength parameters were considered, and shown to be appropriate
because of the relatively slow rates of applied loading (0.5mm/min or 0.02"/min).
Drained shear strength parameters would also be appropriate to normal structural load-
ing. It is evident that the interface asperities fail under considerable levels of local
confinement, and that empirical correlations with the unconfined shear strength of rocks
cannot take the true aspects of this behaviour into account.

Elasticity and Normal Stress Distribution

The previous section considered the increases of normal stress due to progress-
ive shear displacement of interfaces comprising regular triangular asperities. In such
interfaces, the distribution of load between successive asperities is uniform. For rock
sockets, where a more random distribution of asperities may be expected, the distribu-
tion of stresses is highly irregular.

The distribution of nonnal stresses for a profile comprising a range of asperity


angles will be uniform as long as no shear displacement is applied. On application of
shear displacement, the interface will be constrained to dilate at the angle of the steepest
asperity. Shallower asperities may indeed lose contact, and the result will be a distribu-
tion of stresses that is highest on the steepest asperities, and lower or even zero on the
shallower asperities.

563 Seidel and Haberfield


Of course, the elasticity of the rock will result in defonnation of the loaded
asperity and depress the surrounding surface. The shape and extent of the defonnation
pattern will depend on the elastic parameters and depth of the rock, and the type of
surface load. In the constant normal stiffness tests, the depth of rock is given by the
depth of the sample, and the asperity strips can be approximated as.applying plane-strain
loading conditions. For this case, the surface deflection profile can be determined simply
by the Steinbrenner (1934) method for surface footings. For a rock socket, the situation
is obviously different, and somewhat less well defined. The socket is not surrounded by
rock of a finite and convenient depth, but rather by an elastic medium of infinite depth.
What is more, asperities will be 3 dimensional, and apply patch loads to the socket
surface rather than strip loads as for the CNS test (unless a roughening tool has been
used).

The Steinbrenner Method has been applied to the socket case by means of an
analogy. It is necessary to first detennine an equivalent elastic depth, in order for the
Steinbrenner method to be implemented to this case. Such a depth can be evaluated by
equating the constant normal stiffness for a rock socketed pile, as given in Equation 2,
to the stiffness, K, of an elastic medium (modulus Em) of finite depth, h, i. e.:

h = (1 +v).r (Eq. 5)

This analogy can best be visualized by considering the socket to be surrounded


by a "skin" of thickness, h. The skin is cut paralIel to the axis of the socket, and is then
opened out and laid flat. The Steinbrenner method can then be implemented as the
application of load patches (rather than strips) to the planar surface of an elastic medium
of finite depth, D. This concept is shown schematicalIy in Figure 5.

The determination of the distribution of nonnal stresses within either CNS


samples or rock sockets requires the fonnulation of a compliance matrix which must be
inverted and solved to satisfy the deflection boundary conditions imposed by the joint
dilation and the interface geometry. For rock sockets, a compliance matrix of 65,536
elements must be solved - a task which can be solved in less than 15 seconds in the
current generation of personal computers.

It is evident from the foregoing that the elastic parameters of the rock, Em and v,"
are important not only to the global deflection behaviour of the drilled pier, as demon-
strated by Williams (1980) and Pells and Turner (1979), but they are also criticalIy
important to the distribution of stresses within the concrete/rock interface, and ultimate-
ly the available shear resistance. This is a factor which is not incorporated in the consti-
tutive laws used in these finite-element programs for the interface behaviour.

Post-peak Behaviour

The critical nonnal stress for an asperity is a function not only of the intact shear
strength, but also the geometry of the asperity. As noted previously, the Sokolovsky

564 Seidel and Haberfield


I
I
I
I
I II

,,-. .---
II l
pen: II
II
II open
, "'
,,
I" II "' typical
II
II "load patch"
,,
-~-----

cut

Figure 5 Application of Steinbrenner's Method to a rock socket

(1960) method was found to closely predict the critical normal stress in direct shear
tests. Furthermore, this method predicts the critical failure surface geometry by means
of closed form solutions. Observations of eNS direct shear tests on triangular asperities
demonstrated a 'door-stopper' effect, whereby relative movement occurred both between
the concrete and a wedge of failed rock, and between the wedge of failed rock and the
underlying unfailed rock (see Figure 6). The solution to the net friction angle for this
dual sliding mode has been solved using the geometry defined by the Sokolovsky

concrete
displaced wedge of
compressed rubble

--
chord
curved faIlure
. 7',-"-"-"
.. --. ..

unfailed material
surface

Figure 6 Schematic representation of post-peak shear displacement

565 Seidel and Haberfield


method together with energy methods of analysis. Agreement of this solution with
experimental results is excellent.

Roughness

The final implementation of the various aspects of the models described previ-
ously to rock sockets depends on an effective characterization of socket roughness.
Seidel and Haberfield (1994) describe the use of fractal geometry in the characterization
of rough profiles, and determine relationships between the so-called fractal dimension
and the more common statistics of standard deviation of asperity length and angle. It is
shown that these statistics are not independent and that the scale-dependence of these
parameters can be predicted using fractal geometry. It is beyond the scope of this paper
to describe this work in more detail. In essence, however, the roughness of asperity
sockets is modelled using a quasi-probabilistic approach; the standard deviation of asper-
ity angle defines a probability density function for asperity angles. In the model, this
probability density function is implemented as a deterministic approximation, with asper-
ity angles randomly assigned to a grid of 16 x 16 asperity patches. The net response of
the highly complex surface roughness is determined by combining the interacting
responses of the many simple triangular asperities which make up the whole surface.

Implementation in Rocket

The results of the extensive laboratory programme (Haberfield et al., 1994b), and
the theoretical models outlined in this paper have been implemented in a Windows
program called Rocket. This program has been based on the models developed from
tests on simple test profiles, and has been used to predict the results of extremely
complex profiles tested under eNS conditions, as well as some full-scale shaft-only
pile load tests carried out by Williams (1980) in Melbourne mudstone, both with good
results.

The program requires input of the following parameters:

intact drained shear strength parameters


rock mass modulus and Poisson's ratio
pile diameter
initial normal stress
mean socket asperity length
mean socket asperity angle

For Melbourne mudstone, sufficient experience has been obtained with triaxial
testing that the drained shear strength parameters can be determined sufficiently accu-
rately from correlations with moisture content. The intact rock modulus can similarly-
be correlated with moisture content, however, what is needed is rock mass modulus,
which is preferably determined by in-situ pressuremeter testing. Pile diameter and
initial normal stress are easily determined, and the socket roughness parameters must

566 Seidel and Haberfield


800

-,
...... -
--; 600
~

--
~

Vl
Vl
QJ
....
-:;;
.... 400 ------
Cl:I
QJ
...= ./
-
~
Vl

QJ
U
I
I l
.'
0
ifJ. 200 I Pile Load Test
r. - "Rocket" average roughness prediction
._. "Rocket" lower - bound roughness prediction
- - "Rocket" upper - bound roughness prediction

o 10 20 30
Pile head movement (mm)

Figure 7 Comparison of measured and predicted responses for Williams' Pile M2

--
1200
.",
/
/
..... - -.
I
I
--; 900 1---i'----~~t------..,;::OO".....- _ _ I - - - - - _ _ _ 1
Q..
~

~
QJ .",..
-_._.-.
b ~.

.... 600
Vl 1-f-J1IIJ-~.----1f---------t------i
Cl:I
QJ
.c
Vl

Pile Load Test


- "Rocket" average roughness prediction
._. "Rocket" lower - bound roughness prediction
- - "Rocket" upper - bound roughness prediction

o 10 20 30
Pile head movement (mm)
Figure 8 Comparison of measured and predicted responses for Williams' Pile WG303/2

567 Seidel and Haberfield


be established either by individual measurement, or by experience with particular dril-
ling equipment I rock type combinations.

At the time at which the drilled pier load tests were perfonned by Williams
(1980), the importance of socket roughness was not fully appreciated Only minimal
socket roughness profiling was therefore perfonned. On the basis of these limited
measurements, therefore, a possible range of asperity angles and lengths has been
detennined - an average of 12.5 asperities with 50 nun (2 11 ) lengths ranging from a
minimum of 10 and 40 nun (1.6") to a maximum of 15 and 60 rom (2.4").

The range of predictions for Piles designated M2 and WG303/2 using these
roughness statistics are compared with the measured pile-head response in Figures 7
and 8. These piles had respective diameters of 1.3m (51 11 ) and 1.58m (62"). The
comparisons indicated in these figures are quite good; in particular it is noted that the
predictions are for the complete load-deflection response rather than for the peak shear
strength only. It is also evident from the three respective predictions representing aver-
age, upper-, and lower-bound roughness, that the predicted pile response is quite sensi-
tive to socket roughness - a factor which is often neglected in conventional design of
drilled piers in rock. Furthennore, the importance of accurate measurements of socket
roughness is underlined.

Summary

This paper has briefly described the methods of determination of shaft resis-
tance of drilled piers in rock in a historical context. The limitations of these various
empirical methods has been discussed. A new approach to the prediction of the
performance of drilled piers in rock, based on a detailed analysis of the interface mech-
anisms has been outlined. The importance of each part of this analysis to the prediction
has been emphasized, and the inability of empirical methods to capture these important
effects has been noted. A Windows program called Rocket, which incorporates the
proposed theoretical models, has been introduced. The pile load-deflection predictions
of Rocket for two full-size shaft-only drilled pier load tests have shown good correlation
with actual test measurements.

Conclusions

Empirical methods for detennination of drilled pier perfonnance in rock are inac-
curate because they exclude the many variables which affect their load deflection behav-
iour. The limitations are best demonstrated by the order of magnitude variations shown
in Figure 2 after Rowe and Armitage (1984). Empirical methods are therefore limited to
site-specific applications. The use of a theoretical approach to predicting the perfonn-
ance of drilled piers in rock offers the opportunity to encompass all the important para-
meters, and to extend the method to general application. Such an approach is outlined
in this paper.

568 Seidel and Haberfield


Acknowledgments

The authors gratefully acknowledge funding of the project described in this paper
by the Australian Research Council, and the financial support for Mr. Seidel's PhD
candidature by the Sir James McNeill Foundation. The previous research work by many
PhD candidates under the direction of Assoc. Prof. Ian Johnston is also acknowledged.

Appendix 1 - References

Clayton, C.R.I. and Milititsky, 1. (1983). Installation effects and the performance of
bored piles in stiff clay. Ground Engineering, March 1983 : 17-22.
Freeman, C.F., Klajnennan, D. and Prasad, G.D. (1972). Design of deep socketed cais-
sons into shale bedrock. Can. Geot. Jnl. Vol. 9, No. 11, February 1972 : 105-114.
Haberfield, C.M. , Baycan, S. and Chamberlain, T. (1994a). Improving drilled pier
perfonnance in rock. Proc. FHWA Int. Conf. on Design and Construction of Deep
Foundations, Orlando, December.
Haberfield, C.M., Seidel, 1.P. and Johnston, I.W. (1994b). Laboratory modelling of
drilled piers in rock. Proc. FHWA Int. Conf. on Design and Construction of Deep
Foundations, Orlando, December.
Horvath, RG. (1978). Field load test data on concrete-to-rock bond strength for drilled
pier foundations. University of Toronto, Dept. of Civil Engg. Publication 78-07 (ISSN
0316-7968)
Horvath, R.G. (1982). Behaviour of rock-socketed drilled pier foundations. Ph.D.
Thesis, University of Toronto, Toronto, Ontario, Canada.
Horvath, RG., Kenney, T.C. and Kozicki, P. (1983). Methods for improving the
perfonnance of drilled piers in weak rock. Can. Geot. Jnl. Vol. 20, 1983 : 758-772.
Johnston, I.W. (1977). Rock-socketing down-under. Contract Jnl., 279: 50-53.
Johnston, I.W. and Lam, T.S.K. (1989). Shear behaviour of regular triangular concrete-
rock joints - analysis. Jnl. Geotech. Engg. ASCE Vol. 115 No.5: 711-727
Ladanyi, B. and Archambault, G. (1970). Simulation of shear behaviour of a jointed
rock mass. Proc. 11th Symp. on Rock mechanics. Rock Mechanics: Theory and Prac-
tice: 105-125.
Patton, F.D. (1966). Multiple modes of shear failure in rock. Proc. 1st Congo Int. Soc.
Rock Mech. Lisbon: 509-513.
Pells, PJ.N., Douglas, D.1., Rodway, B., Thome, C. and McMahon. B.K. (1978).
Design loadings for foundations on shale and sandstone in the Sydney region. Aust.
Geomech. Jnl. 98 : 31-39.
Pells, P.1.N. and Turner, RM. (1979). Elastic solutions for the design and analysis of
rock-socketed piles. Can. Geotech. Jnl. Vol. 16, 1979: 481-487.

569 Seidel and Haberfield


Poulos, HG. and Davis, E.H (1980). Pile foundation design and analysis. John Wiley
and Sons.
Rosenberg, P. and Journeaux, N.L. (1976). Friction and end-bearing tests on bedrock
for high capacity socket design. Can. Geotech. lnl. Vol. 13, No.3, August, 1976 :
324-333.
Rowe, R.K. and Armitage, HH. (1984). The design of piles socketed into weak rock.
Report GEOT-II-84, University of Western Ontario, London, Canada.
Seidel, J.P. (1993). The analysis and design of pile shafts in weak rock. PhD. disserta-
tion. Dept. Civil Engg. Monash University, Melbourne.
Seidel, J.P. and Haberfield, C.M. (1994). Towards an understanding of joint roughness.
Submitted for publication Int. J. Rock Mech. and Min. Sci.
Sokolovsky, V. V. (1960). Statics of Soil Media. 2nd ed. Translated by Jones, D.H. and
Schofield, A.N. Butterworth Scientific Publications, London.
Steinbrenner, W. (1934). Tafeln zur Setzungsberechnung. Die Strasse, Vol. 1.
Taylor, W.H (1965). Concrete technology and practice. 1st ed. Angus and Robertson.
Sydney.
Thorburn, S. (1966). Large diameter piles founded in bedrock. Proc. of the Symp. on
large bored piles. Institution of Civil Engineers and Reinforced Concrete Association,
London: 95-103.
Thorne, C.P. (1977). The allowable loadings of foundations on shale and sandstone in
the Sydney region. Part 3. Field Test Results. Paper presented to Sydney Group of
Aust. Geomechanics Soc., Inst. Engnrs. Aust.
Tomlinson M.J. (1977). Pile design and construction practice. Viewpoint Publications,
Cement and Concrete Association, London.
Whitaker T. (1976). The design of piled foundations. 2nd edition. Pergammon Press.
Williams, A.F. (1980). The design and performance of piles socketed into weak rock.
PhD Dissertation, Dept. Civil Engg., Monash University.
Williams, A.F., Johnston, I.W., and Donald, I.B. (1980). The design of socketed piles in
weak rock. IntI. Conf on Structural Foundations on Rock, Sydney. Balkema: 327-
347.

Seidel and Haberfield


570
Improved Methods for Evaluation of Bending Stiffness of Deep Foundations

William M. Isenhower! , Member ASCE

INTRODUCTION

Incorrect evaluation of bending stiffness is probably the most common


mistake made when designing and analyzing laterally loaded drilled shafts and
concrete piles. This mistake results from many designers not recognizing that a
drilled shaft is a composite structure made from nonlinear materials and is subjected
to both axial load and bending moment. Bending stiffness of a drilled shaft is a
function of both axial load and bending moment. One must accurately evaluate
bending stiffness if reasonably accurate analyses are required when high loads are
applied to the shaft.
This paper will discuss how bending stiffness is evaluated, how drilled shafts
are analyzed for lateral loading, and modifications required for unusual cases. In
addition, examples of comparisons of incorrect analyses due to common mistakes and
correct analyses are presented.

BACKGROUND AND SIGNIFICANCE

Techniques for structural analysis of piles and drilled shafts have been
developed over the last 45 years. In the past, a majority of effort was devoted to the
development of the analytical techniques used for the calclliation of pile deflection
and bending stresses because these quantities are most important when selecting sizes
of piles and drilled shafts. Several techniques have become popular. The most
popular technique used in the United States is the p-y method, originally proposed by
McClelland and Focht (1955), and more fully developed by Matlock (1962), Matlock
and Reese (1962), and Matlock and Halliburton (1964). Other methods developed
include Broms (1966) method, the elasticity-based methods of Poulos (Poulos, 1971),

1 Asst. Prof, Department of Civil Engineering & Engineering Mechanics, University of Arizona,
Tucson, AZ 85721.

571 Isenhower
and the finite element method. Each of these methods requires, to some degree,
knowledge of the bending stiffness of the pile or drilled shaft.
Use of a constant bending stiffness can lead to somewhat erroneous results if
the foundation is supported primarily by side shear because of variation of axial thrust
forces along the length of a reinforced concrete drilled shaft. Chen and Atsuta (1976)
reported the effects of varying axial thrust forces on the moment-curvature
relationships for reinforced concrete beam-columns. They found that the initial
values of bending stiffness could increase by a much as 30 percent when the axial
thrust force increased from 0 to 31 percent of the axial squash-load capacity of the
beam-column. When axial thrust force was increased to above 31 percent of squash
load capacity, the maximum. moment carrying capacity of the cross-section decreased
and bending stiffness also decreased. This can be of critical importance for drilled
shafts anchored in rock where axial loads can be high. In this high range of axial
loading, initial values of bending stiffness do not vary as much with axial thrust as for
the case with axial loads below 31 percent of squash load capacity. For all levels of
axial loading, the reduction of bending stiffness for levels of moment near collapse
load is in the range of 60 to 70 percent below the low-strain elastic values.

EVALUATION OF BENDING STIFFNESS

The procedures used to calculate bending stiffness are based on the


relationship between bending moment and curvature. In linear elasticity, the
relationship between stress, cr, and axial strain, E , is described by Hooke's Law as
cr = EE (1)
where E is the modulus of elasticity. When a beam is subjected to pure bending, the
center of curvature is defined as the point of intersection between two initially parallel
planes as shown in Fig. 1. If the beam is not subjected to axial loading, there is a line
with zero extension located on the centroid of the cross-section of the beam. This line
is called the neutral axis. The radius of curvature, p, is defined as the distance from
the centroid of the cross-section to
the center of curvature. The
reciprocal of the radius of curvature
is the curvature, K.
Mathematically, K is defined as
d 2y
\ / K = dx 2 .. " ,.. (2)
p \ /
\ / At any depth in the beam other
\ / than the neutral axis, either tensile
\ / or compression strains are de-
veloped due to bending. The
0'
longitudinal strain in a beam
FIG. 1. Geometry of Curvature

572 Isenhower
without axial thrust, Ex" can be calculated from curvature using
YNA
Ex = --p- = -KYNA (3)

where p is the radius of curvature and YNA is the offset from the neutral axis (YNA will
be used in place of the common notation Y to distinguish between offset from the
neutral axis, YNA, and beam deflection, y). The negative signs in Eq. 3 are due to
different sign conventions begin used for curvature and bending moment as shown in
Fig.!. Combining the above relationship with Hooke's Law (Eq. 1), one obtains
ax = EE x = -EKYNA (4)
The moment in the beam, M, is equal to the product of bending stress and offset from
the centroid, Y C' integrated over the cross-section of the beam.
M = -fax Yc dA (5)

Using Eq. 4 in Eq. 5, one obtains an expression of bending moment in terms of


constant E and K values

M = f E K Y~ dA = EK f Y~ ciA = E K 1 (6)

where 1 is defined as the moment of inertia (second moment of area) and is equal to

1 = JY~A ciA (7)

Equation 6 allows one to obtain the relationship between bending stiffness moment,
and curvature.
M
E1 = - = MP (8)
K

The preceding equations demonstrate how the conventional definition of the


moment of inertia (Eq. 7) follows from the theory of elasticity. The above discussion
was for the case where the axial load on the pile or drilled shaft was equal to zero,
curvature is small, and the shaft was subjected to pure bending. When a deep
foundation is subjected to an axial load, the position of the neutral axis moves away
from the centroid of the cross-section. For this case, one may use the parallel axis
theorem to calculate moments of inertia offset from the centroid or use an alternative
numerical technique for computation. One alternative technique is the following.
Equation 8 can be used to calculate bending stiffness for drilled shafts and
piles subjected to inelastic bending and for which the stress-strain relationship of the
drilled shaft or pile is nonlinear. In application, the integrals must be evaluated
numerically because the cross-sectional areas are arbitrarily shaped and the stress-
strain curves of the structural materials may be nonlinear. The integrals are calculated
by dividing the cross-section into a number of thin slices parallel to the neutral axis

573 Isenhower
and assigning material properties to each area. The computational procedure used to
compute EI is:
1. Assume a value of curvature, K.
2. Assume a position for the neutral axis.
3. Calculate longitudinal strain versus offset from the neutral axis using Eq. 4
and the assumptions from steps 1 and 2.
4. Calculate stress in the structural material using the value of strain and the
appropriate stress-strain relationship for structural material. If the concrete is
cracked in a tensile zone, use a stress of zero in the tensile zone.
5. Integrate stress in the structural material over the cross-section of the drilled
shaft to obtain the axial thrust force in the foundation.
6. If the axial thrust force calculated in step 5 is not sufficiently close to the axial
thrust force acting on the foundation, adjust the assumed position of the
neutral axis and repeat steps 3, 4, and 5 until convergence is achieved.
7. Calculate the bending moment in the foundation using Eq 5.
8. Calculate the bending stiffness of the shaft using EI = M/K (Eq. 8).
The analytical procedure outlined above is basically the same as that used in
computer program PMEIX, developed by Reese and Allen (1977) and suggested by
Chen and Atsuta (1976). This procedure is repeated for all desired values of
curvature until the complete relationship between bending moment and bending
stiffness is defined. If desired, one may also evaluate the magnitude _of bending
moments at which drilled shafts initiate yielding of reinforcement or cracking of
concrete in the tensile zone.

Stress-strain Relationships for Concrete


In Step 4 above, stress in the structural material is calculated as a function of
strain. For concrete, several options for stress-strain curves are available. Many
programs use Hognestad' s (1951) second degree parabolic stress-strain relationship

f, ~ f:[{:)-(:,J] . n : (9)

where f: = compressive strength of concrete in psi, (10a)

f: .
Eo = 2-, and (lOb)
Ec

E c = 57,500.J7f = modulus of concrete (lOc)


In practice, a value of 0.002 is often assumed for Eo.
Hognestad's relationship is based on data for unconfined concrete where both
concrete and reinforcing steel are subjected to uniaxial stress. For the case of
concrete-filled tubular piles, the concrete is laterally confined so the strength and
ductility of the concrete is greater than that for unconfined concrete. Stress-strain

574
Isenhower
curves for confined concrete
are shown in Fig. 2. Curve 1
in Fig. 2 is Hognestad's for
f'c - - - - -=--------....;;....., .3
unconfined concrete. Curves
0.85 f~ ~::__----.., 2 2 and 3 are based on data
0.72 f~ reported by Burdette and
Hilsdorf (1971). Curve 2
represents a case vvhere
complete interaction betvveen
the steel and concrete takes
place, but the triaxial state of
E; stress in the concrete is
0.0023 0.0035 0.006 0.016 assumed to increase its
ductility only. A uniaxial
FIG. 2. Stress-strain Curves for Concrete state of stress is assumed for
the steel in Curve 2. Curve 3
represents the case vvhere complete inter-action takes place betvveen the steel and
concrete and the triaxial state of stress in concrete increases both the ductility and
strength. Again uniaxial stress is assumed for the steel. The curves in Fig. 2 are pre-
sented to illustrate that bending stiffness calculations based on Curve 1 (the
unconfined case) may be lovv because the stress in the concrete is underestimated.

Stress-Strain Relationship for Reinforcing Steel


For ordinary steel reinforcement, an elastic-plastic stress-strain relationship
vvill be assumed. Hovvever, significant nonlinear stress-strain behavior is observed
for high-strength reinforcement used in prestressed concrete. In this cases, it is
desirable to have a user-defined stress-strain relationship. Many engineers are
familiar vvith the Ramberg-Osgood (1943) formulation and the Duncan-Chang (1970)
model for soil. The Ramberg-Osgood formation is difficult to use because it
calculates strain in terms of stress and is not invertable to calculate stress in terms of
strain. The Richard equation (Richard and Abbott, 1975) is vvell suited for this
purpose because stress is formulated in terms of strain. The Richard equation for
stress in terms of strain is
E,f:
= -(-l-+--~"":"o-f:-n-J""",,""""/n-
............................................................................ (11)
0' +E p f:

vvhere Ep is the plastic modulus, E, is the difference betvveen the initial tangent
modulus and plastic modulus and is equal to E - Ep , n is a parameter that describes
the sharpness of curvature, and 0'0 is a reference plastic stress. This relationship
requires evaluation of four model parameters compared to only one for Eq. 9.
Hovvever, if Ep and E are vvell defined, 0'0 is the intercept of the Ep-line vvith the stress

575 Isenhovver
axis, 0'1 is detennined graphically as shoml
in Fig. 3, and n can be evaluated using Eq.
12.
-ln2
n = ----:,.....-------.,- ........................ (12)

fIl
fIl
Some designers may choose to use

-e...n
Q)
the relationship for bending stiffness
proposed by the American Concrete Insti-
tute. ACI 318-89 recommends
(EJg /5) + EsIse
EI = (13)
l+P d
Strain, E where I g is the moment of inertia of gross
FIG 3. Parameters in Richard concrete cross-section about the
Equation centroidal axis (neglecting
reinforcement),
Es is modulus of elasticity of
reinforcing steel,
Ise is the moment of inertia of reinforcement about the centroidal axis of the
cross-section,
E c is the modulus of concrete as defined in Eq. 9, and
Pd is the ratio of maximum factored axial dead load to maximum total factored
axial load.

When compared to an explicit evaluation, bending stiffness may be significantly


overestimated by Eq. 13.

COMPUTATIONAL PROCEDURES

Computational procedures used for analyzing laterally loaded drilled shafts


and piles are based on the solution of the differential equation describing the behavior
of a beam-column. If the drilled shaft or pile has a constant bending stiffness, the
differential equation to be solved is
Ely"" + Qy"-EsY =0 (14)
where Q is the axial thrust force in the drilled shaft or pile and E s is the soil modulus
defined as the ratio of ply.
Most analyses of laterally loaded deep foundations utilize a single value for
bending stiffness of the foundation and do not adjust this value for effects of
curvature and axial thrust. Many experienced designers take advantage of this feature
to closely scrutinize the results of an analysis to examine the location where the

Isenhower
576
drilled shaft or pile is highly stressed, then adjust the bending stiffness accordingly
for a subsequent analysis. This process can be automated for static analyses by using
the techniques developed by Kramer and Heavey (1988). This technique uses a
secant bending stiffness, E1s' for shafts at large curvatures and a modified version of
the lateral force equilibrium equation. The modified lateral force equilibrium
equation is
2
d 4y deE!), d 3 y [" d (E1)s ld 2 y
(E1)-4+ 2
. dz dz
-3+l
dz dz
2 +QJ-2-P=0
dz
(15)

As an alternative to solving Eq. 15, one may evaluate the moment-curvature


relationship for each structural section along a pile then compute the values of
bending stiffness versus bending moment for each section. These relationships are
then used in a conventional laterally loaded pile analysis routine using linear
interpolation along the E1s-M curve.

MODIFICATIONS FOR SPECIAL CASES

Most computer programs available for computation of bending stiffness of


drilled shafts can handle only the ideal cases of design where the shaft geometry is
symmetric and loading is along a principal axis. When conditions vary from the ideal
case, special modifications must be made to evaluate bending stiffness correctly. The
most commonly encountered unusual cases are:
1. Out-of-position reinforcing steel.
2. Off-axis bending.
3. Irregular cross-sectional shapes.
4. Concrete-filled casing.
5. Non-linear stress-strain curves.
Evaluation of bending stiffness of undamaged steel piles for loading along a
non-principal axis is easy if the cross-sectional shape of the pile is contained in the
AISC Manual ofSteel Construction. For this case, bending stiffness is the product of
the modulus of elasticity, E, and the moment of inertia in the direction of bending, 1.
If bending is not in a principal direction, the moment of inertia can be calculated
using the transformation equations based on Mohr's circle (Gere and Timoshenko,
1990).
Ix + 1y Ix _ 1y .
I = 2 + 2 Cos 28 - 1xy Sm 28 (16)

where Ix and 1y , are the moments of inertia about the x and y axes and 1xy , is the
product of inertia. If the foundation is a common structural steel shape, one may find
values of the moments of inertia and product of inertia in the AISC Manual of Steel
Construction.
Use of the above transformation equation is limited to the cases where the pile
is linearly elastic. When the pile is not linearly elastic, due to either yielding of the

577 Isenhower
pile or a nonlinear stress-strain
y y
relationship for the pile material, the
above is expression is no longer
valid. Consider the cross-section of
the pre-cast concrete pile shovm in o CD 0
Fig. 4. For this pile, Ix is equal to Iy , o L..-e--cJ--- x L..--B---.?-_ x
and I xy , is equal to zero because the 000
cross-section is symmetric.
Consequently the moments of inertia
FIG. 4. Off-axis Bending
calculated using the above
transformation equation (Eq. 16) are equal for any value of8. However the moment
of inertia for the rotated section on the right is not equal to the moment of inertia of
the section on the left. The reasons for this are the nonlinearity of the stress-strain
curve for concrete, yielding of steel reinforcement, and the influence of axial loading.
Another common case for which no simple technique exists for evaluation of
bending stiffness is a composite drilled shaft with an H-pile core. An example of the
ideal cross-section drawn by the designing engineer is shown in Fig. 5. Examples of
common construction flaws are shown in Figs 6 and 7. A case for which the H-pile is
rotated and off-center is shown in Fig. 6. A foundation with a damaged H-pile is
shown in Fig. 7.
The above cases considered only variations in the cross-section of the
foundation due to rotation of the direction of loading and off-center or damaged
reinforcement. Out-of-round shafts, and variations of bending stiffness due to
changing cross-section along the length of a deep foundation are difficult to handle
using existing methods. Tapered piles, whether wood or metal shell, require so much
additional effort to evaluate the variation of bending stiffness that they are seldom, if
ever, analyzed as tapered foundations.
y y y

1'1-----.-...- x '----.+--<_ x I'I----+-- X

FIG. 5. As Designed FIG. 6. Off-center FIG. 7. Damaged

Another important factor not routinely considered is the effect of varying axial
load due to axial load transfer along the length of a deep foundation. Many existing
programs for analyzing laterally loaded deep foundations can handle variations in
axial load with depth, but cannot calculate the variation of bending stiffness due to
varying axial load and moment versus depth. Instead the usual practice is to evaluate
a single value of bending stiffness under the pile-head load and a moment just below

578 Isenhower
the initiation of non-linear bending behavior and use this value for all design
calculations.
An additional factor affecting bending stiffness is pre-stress in a pre-stressed
concrete pile. Here the effective strains in the reinforcing steel and concrete are
unequal and must be calculated separately.
In summary, existing methods for calculation of bending stiffness of deep
foundations are deficient in five areas. These deficiencies are calculation of EI in
reinforced concrete shafts and piles as a function of:
(l) loading in a non-principal direction,
(2) non-symmetrical cross-sectional geometry due to faulty construction,
(3) variation of cross-section versus depth,
(4) variation in bending stiffness due to variation in axial thrust force with depth,
and
(5) unequal effective strains in pre-stressed concrete piles.
To handle the above cases, a programs must allow the user to input variable
cross-section geometry and the direction of loading. This requires that the program
be able to calculate segmental areas of a pile cross-section rotated at arbitrary angles.
The approach to be employed for this calculation will first transform cross-sectional
coordinates to a coordinate system orthogonal to the direction of loading then
calculate the cross-sectional areas. The coordinate conversion will use a conventional
coordinate transformation in which a matrix composed of direction cosines is
multiplied by a vector containing the x and y coordinates. Thus, the transformed
coordinates x I and y 1 relative to the X 1- Y1 axes are calculated using

{~J = [ ~~:~~=:~;~ ~~:~~=:~:~] {~} (17)

where Cos(X-;X1) denotes the direction cosine measured from the X-axis to the X 1-
axis and so on. The area of an arbitrarily shaped area is easily calculated by placing
the x and y coordinates in a two-row array, calculating the determinant, and dividing
by 2. Thus,
1 I Xl x2 ... x l
A = -Detl (18)
2 y, Y2 ... y"
n
J
The above equation can be used to compute the size of any polygon defined using an
array of coordinate points. Thus, the same procedure can be used to compute the
areas of slices through out-of-round drilled shafts or pipe piles and rotated or
deformed H-piles.
When a slice crosses a round object like a pipe or reinforcing bar, the area of
the circular segment above a slice line is computed using

1 2( .)
Asegmelll = :2 R a. - SIno. .................................................................................... (19)

Isenhower
579
where R is the radius of the circular object and a is the central angle defined by the
location of the slice line. In practice, the areas of segments defined by slices through
a reinforcing bar are first computed from the top down then starting from the bottom
slice, the area of the segment for the slice above is subtracted from the current slice.
For pipe sections, the area of segments defined by the outside of the pipe is computed.
first, then the segment areas defined by the inside of the pipe are computed and
subtracted from the values computed for the outside of the pipe. This procedure is
repeated for each reinforcing bar in turn. Thus, accounting for position of out-of-
position reinforcement or rotation of loading direction can be handled.
As mentioned above, many references contain misleading information about
bending stiffness of deep foundations. Usually, a reference might report only the
moment of inertia about the centroidal axis for the gross section without commenting
on the modulus of concrete, or a reference might report that bending stiffness is the
product of the low-strain concrete modulus and the moment of inertia of the gross
section. Three references that demonstrate the correct way to evaluate bending
stiffness of deep foundations are Reese and Allen (1977), Reese (1984), and Reese
and O'Neill (1988).

EXAMPLES

The following examples use either a 0.76-m (30-inch) diameter, round drilled
shaft or a 0.36-m (14-inch) square prestressed concrete pile. The drilled shaft has a
concrete compressive strength of 24.1 MPa (3.5 ksi) with steel reinforcement area of
1% with a yield stress of 414 MPa (60-ksi). The prestressed pile has a concrete com-
pressive strength of 34.5
1,000 ,------,,------,-----,.-----,-----, 500
MPa (5.0 ksi) and eight 11
Uncracked Stiffness
N
mm (7/16 inch) reinforcing
E cables prestressed to 609 kN
E BOO 400 '
I
Z
Z ~
(l37kips).

-
~

l/)
c:: Bending Moment
600 300 ~ Variation of Bending
---
Q)
E c::
0 Stiffness with Bending
~ CJ) Moment and Axial Thrust
Start of Rebar Yield
Cl 400 200
c:: Cl The variation of
c::
"0 Bending Stiffness bending stiffness and
c:: "0
c::
Q)
a:l 200 100 Q) bending moment with
a:l
curvature for the example
drilled shaft under zero axial
o 0 thrust is shown in Fig. 8.
o 0.005 0.010 0.015 0.020 0.025
The initial bending stiffness
Curvature rad/m
is high until the concrete in
the tensile zone cracks at a
FIG. 8. Variation of Moment and Stiffness with bending moment of
Curvature for Example Drilled Shaft approximately 130 kN-m

Isenhower
580
1,000 ,------,------,----,..------,---------, 500 (1,150 in-kips). After
cracking, the value of
C\l
E bending stiffness drops by
E BOO 400 I
I Bending Moment Z about 80 percent from 437
Z ~ 2 2
.:JIt. kN-m to 91 kN-m. At
en
~ 600 300 :g values of curvature above

-- this point, bending stiffness


l::
E
o
~ ...... is approximately constant
en
Cl 400 200 until the reinforcing steel
l:: Cl
"0 Bending Stillness l:: begins to yield at a bending
l:: "0
Ql l:: moment of 410 kN-m. As
OJ 200 100 Ql
OJ reinforcing steel yields,
bending stiffness gradually
0'-------'-----'-------'-----'------'0 declines as values of
o 0.005 0.010 0.015 0.020 0.020 curvature increase.
Curvature rad/m The general effect of
increasing axial thrust is to
FIG. 9. Effect of Axial Thrust on Variation of increase both bending stiff-
Stiffness and Moment with Curvature ness and maximum bending
moment developed in the
250 ,-----,----,-------,,-----r---,-----,----,--,----,
shaft. The reason for this is
axial compression reduces
the size of the tensile zone
E I
200
Maximum Moment in the shaft, thereby increas-
Z
~ ing the contribution of con-
...... crete in developing bending
l:: 150
Ql
E
moment. Since bending
o stiffness is bending moment
~
Cl 100
divided by curvature, if
l:: bending moment is in-
"0 Moment at Cracking
l:: creased for a given level of
Q)
to 50 curvature, bending stiffness
will be correspondingly in-
creased as shown in Fig. 9.
o'-------L--'------'-----'----'--------I.---L--L.----J
o 10 20 30 40 50 60 70 80 90 Effects of Direction of
G--- [j"" cr
Loading ~
Rotation of Loading, deg. The effects of the
direction of loading on
FIG. 10 Effects of Rotation of Loading on circular drilled shafts is
Example Square Prestressed Pile minimal for most cases.
However, when the cross-section is square, reductions in strength and stiffness are
possible. An illustration of this effect is shown in Fig. 10. The orientation of the
cross-section to the direction ofloading is shown at the bottom of the figure. For this

581 Isenhower
250 50 example, the pile is
weakest when the
Bending Stiffness C\l
E direction of loading
E 200 40 I

Z
I
Z crosses the section
..x: ~ diagonally. At levels of
-c:: 150
Q)
30
rn
l/)
moment below the
E
0
~
C) 100
Bending Moment

20
---
Q)
c::

CfJ
point of cracking, the
bending stiffness of a
prestressed, square
c::
"0
C)
c:: concrete pile is
c:: "0 relatively constant for
Q)
c::
OJ 50 10 Q)
CO
any direction of
loading. However,
0 0
once cracking begins,
0 0.01 0.02 0.03 0.04 0.05 the moment-curvature
Curva tura, rad/m relationship as shown in
Fig. 11 is similar to that
FIG. 11. Moment-Curvature Behavior of Example for a drilled shaft as
Square Prestressed Pile shown in Fig. 9.

Comparisons of Analytical Results for Laterally-Loaded Drilled Shafts


A comparison of two analyses of drilled shafts using incorrect and correct
(nonlinear) values of bending stiffness was made to illustrate the effects of using the
wrong value of bending stiffness. The lateral load analyses were made using
COM624P from FHWA (Wang and Reese, 1990) modified to utilize nonlinear
bending stiffness as a function of computed moment along the shaft. These analyses
are for the example 0.76-m drilled shaft in a sand with an angle of internal friction of
35 degrees. The shaft used in this example had free-head conditions and was
subjected to a pile-head shear load combined with a 334 kN (75 kip) axial load. The
incorrect analyses were made using an elastic bending stiffness computed as the sum
of the product of the gross moment of inertia times the low-strain concrete modulus
plus the moment of inertia of the reinforcing steel pattern times the elastic modulus of
steel.
A comparison of maximum computed moment versus pile-head shear for the
elastic and nonlinear analyses is shown in Fig. 12. In this example, use of incorrect
bending stiffness results in an over-estimation of maximum moment in the drilled
shaft. A comparison of shaft-head deflection versus pile-head shear loading is shown
in Fig. 13. Similar results are obtained for shaft-head rotations.
This comparison shows that deflections computed using the nonlinear bending
stiffness can be significantly larger than those obtained when elastic bending stiffness
values are used. This is a potentially unconservative situation that might lead bridge
designers to under-design the widths of bridge seats and expansion joints.

582 Isenhower
500 r------.--------,----,--------.-----,

400
Z
~

....
CIS
Q) 300
.!::
CfJ
"C
CIS
Q) 200
..c
I
Q)

a..
100
Elastic EI
Nonlinear E I - -

150 300 450 600 750


Maximum Bending Moment, kN-m

FIG. 12. Maximum Moment and Shear in Example Drilled Shaft

500
/
/
/
/
400 I
Z
~ I
I
..... I
CIS I
Q) 300 I
..c I
CfJ I
I
"C I
CIS I
Q) 200 I
..cI I
Q) f
.- f
a.. f
100 f
l
Elastic EI
Nonlinear E I - -
0
0 0.01 0.02 0.03 0.04 0.05

Pile-head Deflection, m

FIG. 13. Ground-Line Deflection for Example Drilled Shaft

583 Isenhower
SUMMARY
The computation of bending stiffness of drilled shafts and piles was the main
focus of this paper. A review of the mechanics related to bending stiffness showed
that the common definition of bending stiffness follows from linear elasticity and how
bending stiffness is computed from moment and curvature.
Several procedures to handle cases involving complex geometry, rotation of
loading, and nonlinearity of materials were discussed. Use of the Richard equation to
model the nonlinear behavior of high-strength reinforcement is recommended for
prestressed concrete piles. The stress-strain relationship used for concrete depends on
the magnitude lateral confinement.
Most errors made in practice arise from not considering the effects of the
nonlinear stress-strain curves of concrete and of cracking of concrete in tension.
Several examples bending stiffness calculations were presented to demonstrate the
effects of axial thrust loading and bending curvature. Lastly, a comparison of
analyses of a laterally loaded drilled shaft using nonlinear moment-curvature behavior
and incorrect analyses using elastic bending stiffness were presented to illustrate that
over-estimation of bending stiffness may lead to unconservative evaluations of shaft-
head displacements and rotations.

ACKNOWLEDGEMENTS
Financial support for this research was provided by Information Technology
Laboratory (ITL), Waterways Experiment Station (WES) of the U.S. Army Corps of
Engineers. This work was performed under the supervision of Dr. Reed L. Mosher,
WES. The support of Dr. Mosher is greatfully acknowledged.

REFERENCES
American Concrete Institute (1989). Building Code Requirements for Reinforced
Concrete (ACI 318-89) and Commentary (ACI 319R-89), ACI, Box 19150,
Redford Station, Detroit, Michigan 48219.
American Institute of Steel Construction (1990). Manual of Steel Construction, 9th
Edition.
Broms, B. B. (1964a). "Lateral Resistance of Piles in Cohesive Soils," Journal ofthe
Soil Mechanics and Foundations Division, ASCE, Vol. 90, No. SM 2, pp. 27-63.
Broms, B. B. (1964b). "Lateral Resistance of Piles in Cohesionless Soils," Journal of
the Soil Mechanics and Foundations Division, ASCE, Vol. 90, No. SM 3, pp.
123-156.
Burdette, E. G., and Hilsdorf, H. K. (1971). "Behavior of Laterally Reinforced
Concrete Columns," Journal ofthe Structural Division, ASCE, Vol. 97, No. ST2,
pp.587-602.
Chen, W. F. and Atsuta, T. (1976). Theory of Beam-Columns, Vol. 1 In-plane
Behavior and Design, McGraw-Hill, New York, 513 p.

584 Isenhower
Focht, 1. A., Jr., and McClelland, B. (1955). "Analysis of Laterally Loaded Piles by
Difference Equation Solution," published in three parts in The Texas Engineer,
Texas Section, American Society of Civil Engineers.
Gere, 1. M. and Timoshenko, S. P. (1990). Mechanics of Materials, 3rd edition,
PWS-Kent, Boston, 807 p.
Hognestad, E. (1951). "A Study of Combined Bending and Axial Load in Reinforced
Concrete Members," Engineering Experiment Station Bulletin Series No. 399,
University of Illinois, Urbana, Illinois.
Kramer, S. L., and Heavey, E. 1. (1988). "Lateral Load Analysis of Nonlinear Piles,"
Journal ofGeotechnical Engineering, ASCE Vol. 114, No.9, pp. 1045-1049.
Matlock, H. (1962). "Correlations for Design of Laterally Loaded Piles in Soft
Clay," Report to Shell Development Company, 71 p.
Matlock, H., and Halliburton, T. A. (1964). "A Program for Finite-Element Solution
of Beam-Columns on Non-linear Supports," Report to the California Company,
Shell Development Company, 171 p.
Matlock, H., and Reese, L. C. (1962). "Foundation Analysis of Offshore Pile-
Supported Structure," Proceedings, Fifth International Conference, ISSMFE,
Paris, Vol. 2, pp. 1220-1251.
Poulos, H. G. (1971). "Behavior of Laterally Loaded Piles: I - Single Piles," Journal
of the Soil Mechanics and Foundations Division, ASCE, Vol. 97, No. SM5, pp.
711-731.
Ramberg, W., and Osgood, W. R. (1943). "Description of Stress-Strain Curves by
Three Parameters," Technical Note No. 902, National Advisory Committee for
Aeronautics, Washington, D.C., 22 p.
Reese, L. C. (1984). Handbook on Design ofPiles and Drilled Shafts Under Lateral
Load, U.S. Department of Transportation, Federal Highway Administration,
Report No. FHWA-IP-84-11.
Reese, L. C., and Allen, 1. D. (1977). Drilled Shaft Manual, Vol. II, Structural
Analysis and Design for Lateral Loading, U.S. Department of Transportation,
Implementation Package 77-21, Washington DC.
Reese, L. C. and O'Neill, M. W. (1988). "Drilled Shafts: Construction Procedures
and Design Methods," National Highway Institute, Course No. 13214.
Richard, R. M., and Abbott, B. 1. (1975). "Versatile Elastic-Plastic Stress-Strain
Formula," Journal of the Engineering Mechanics Division, ASCE, Vol. 101, No.
EM4, pp. 511-515.
Wang, S.-T., and Reese, L. C. (1990). COM624P Program User's Manual, Version
1.0, Report No. FHWA-IP-90-005, 247 p.

585 Isenhower
Statnamic Tests on Steel Pipe Piles Driven in a Soft Rock

Tatsunori Matsumoto! and Mokoto Tsuzuki2

ABSTRACT

A series of Statnamic test are employed on open-ended steel pipe piles in


saturated diatomaceous mudstone at Nanao, Japan in cooperation with the Study
Group for Rapid Load Tests. This paper first describes the comparative test results of
the Statnamic and static load tests and then discusses the reliability of an interpretation
technique of the Statnamic test. Application of the Statnamic test results for pile
design is discussed in comparison with the static load test result and representative pile
design codes. The high potential of the application of the Statnamic test result to the
serviceability limit state design will be discussed.

INTRODUCTION

The static load test (SLT) is used for pile design with the highest reliability in
Japan as well as in other countries. However, since the static load test is costly and
time consuming, it is common practice to perform one or two static load tests at a site,
or none when load test experiences are available near the site. In case the static load
test is not carried out, the bearing capacity is derived from empirical equations
prescribed in several codes in Japan in which the N-value from the Standard
Penetration Test (SPT) and the unconfined compression strength, qu, are usually used
for the soil parameters.
A method ofload testing the pile, called Statnamic, has been introduced in Japan
recently and applied to more than 10 piles this year. The load-displacement curve of
the pile is directly obtained from the Statnamic signals. The loading period of the
Statnamic test is usually lOOms and the penetration rate of the pile is higher than that
in the static load test. Hence, minor adjustment of load-settlement signal may be
required depending on the soil type to derive the static behavior of the pile.

1 TatSlIDori
Matsumoto, Associate Professor, Department of Civil Engineering,
Kanazawa University, 2-40-20 KodatSlIDo, Kanazawa, Japan.
2 Makoto Tsuzuki, Managing Director, Fugro McClleland Japan Corp.,
1-21-20 Jingu-mae, Shibuya-ku, Tokyo, JaDan

586 T. Matsumoto et al.


The Statnamic test has several potential advantages in comparison with
conventional techniques from the aspect of cost and mobility. This issue is one of the
topics in this paper. The Study Group for Rapid Load Tests was formed in Japan in
1992 led by Prof. Kusakabe, Hiroshima University. This group promotes the use of
load testing methods other than the conventional static load test for the design and
quality assurance of piles. The Study Group has carried out several comparative static
and Statnamic load tests for open-ended steel pipe piles, precast concrete piles and
ca st-in- situ concrete piles.
This paper describes the results of Statnamic and static load tests on steel pipe
piles driven in a diatomaceous mudstone classified as a soft rock or a stiff clay in Noto
Peninsula, Japan. Potential application of the Statnamic test to the design stage for
improving reliability of the pile design will be also discussed.

STATNAMIC LOADING DEVICE

Benningham Hammer Corp., Canada, and the TNO Building and Construction
Research, the Netherlands, developed the Statnamic loading method (Bermingham and
Janes; 1989, Middendorp; 1993). It may be appropriate to briefly review the
Statnamic loading method for the discussion which follows.
The principle of Statnamic is based on the launching of a reaction mass from the
pile head (Fig. 1). Launching takes place by generating high pressures in a cylinder,
caused by the burning of a special fuel. As a reaction on the launching the pile is
gently pushed into the soil.
TIle load exerted on the pile head is measured by means of a load cell. The
displacement ofthe pile head is registered by means of a special developed laser sensor.
Load cell and laser sensor are integrated components of the Statnamic loading device.
No instrumentation has to be installed on the pile shaft. The required reaction
mass for a Statnamic load test equals 5 to 10% of the design maximum load. As an
example: for a pile which to be loaded to 2 MN a reaction mass of 0.1 to 0.2 MN (10
to 20 tons) has to be launched. -.
The test set up of a Statnamic device has been presented in Fig. 1a. The characters
in the graph correspond with the following components:

A = pile to be tested G = reaction mass


B = load cell H = gravel container
C = cylinder & burning chamber I = gravel
D = piston J = laser
E = platform K = laser beam
F = silencer L = laser sensor

The set up consists of a load cell on the pile head to measure the force and a laser
sensor to measure the displacement of the pile head. On the load cell a cylinder is
placed. The piston forms an integral part with a platform on which the reaction masses
can be placed. Cylinder and piston form a burning chamber. The reaction masses may

587 Matsumoto and Tsuzuki


be shaped as blocks or disks and be fabricated from steel, concrete or lead. A silencer
is placed in the center of the platform and is connected with the burning chamber. A
gravel container is placed over this set up. The space between container and reaction
masses is filled with gravel. The gravel is used to catch the reaction mass when it falls
back after launching and to protect the cylinder and pile head.
Figs.la to ld represent successive stages of a Statnamic load test. Fig.la is the
situation just before launching. In Fig.lb the fuel has been ignited. The burning of the
fuel generates high pressures and the reaction mass is accelerated. At this stage the
actual loading of the pile takes place. The reaction force pushes the pile into the
ground. The upward movement ofthe reaction mass results in space, which is filled by
the gravel (Fig.l c). Gravity causes the gravel to flow over the pile head as a layer.
When the reaction mass falls back it is caught by the gravel and impact forces are
transferred to the subsoil (Fig.ld).
// /7777"7'
',,',".' - ,"

FIG.la FIG.lb

FIG.lc FIG. Id

FIG. 1. Statnamic Set-Up and Successive Stages of Loading (Middendorp, 1993)

The outstanding characteristics of the system are summarized as follows:


1) A couple of tests will be possible per day depending on the loading level,
2) No reaction piles and beams are required,
3) Any special treatments of the pile head are not necessary and the test is applicable
to any types ofpile.
588 Matsumoto and Tsuzuki
STATNAMIC TEST RESULTS

Site Arrangement

Three test piles, designated as Tl, T2 and T3, have been installed at the test site
at Noto Peninsula in 1991. The geometrical and mechanical properties of the piles are
listed in Table 1. The layout of the test piles, the reaction piles and the soil
investigations are shown in Fig.2. Open-ended steel pipe piles were driven with a
diesel hammer having a ram weight of 2.06kN (2.1tons). The Statnamic tests were
canied out on Piles Tl and T2 in 1992. The static load tests were perfonned in 1991
prior to the Statnamic tests.

TABLE 1. Geometrical and Mechanical Properties of Test Piles

Length L(m) 11.0


Outer radius fa (mm) 400.0
Inner radius n (mm) 387.9
Wall thickness tw(mm) 12.1
Cross-sectional area A (m2) 0.041
Young's modulus E(MN/m2 ) 2.06 x 10 5
Mass mp ton 3.47

o T Test pile

O RR eactlon
. .
pile

.8 Borehole

0 8 Borehole with SPT


oC CPT

o 1 23m

FIG. 2. Layout of Piles and Site Investigations

589 Matsumoto and Tsuzuki


Each test pile was instrumented with strain gages at 10 levels. Steel channels were
used for the protection of the strain gages, which increased the cross-sectional area of
the pile to O.04Im2 .
The pore pressures in the soil around and inside Pile Tl and the acceleration of
the top of the soil plug were measured during the Statnamic test of Pile Tl (Fig.3).
I Pile T 1

~ 11---,....----..,...- EI.+1.5m

t 4tm i r t
1

9.3 5m
I
.ct; i 1
7.10m I

9.10m

1 i--,-- 1
~o~~---=-
H

1m
o piezometer
X accelerometer
FIG. 3. Instrumentation of the Test Ground

Several investigations were conducted in the 8 boreholes. SPT's were


conducted at boreholes Bs and B6. Also a number of unconfined compression tests
were carried out using soil samples from the boreholes. Additional eleven CPT's were
performed on several location of the test site.
Pile T 1
SPT N- value qc (MPa) qu (MPa) E so (MPa)
o 5 10 15 20 25 0 1 2 3 4 S 0 0.4 08 1.2 0 20 40 60 80

'V 2 Clay

a::II:l> Din
0
o()~ ~oe
~

E -2 Vl orne aJg
N ::l
Ql
0 0
c:- - 4
0
Ql c: 85
.2 u 0
~,I'
0 0
-6 lllu;
iii E"tl
:>
Ql
o ::l 86 "'.
w -8 iii E
0
- 10 0 0

- 12

FIG. 4. Soil Prome and Results of Site Investigations and Soil Tests
(after Matsumoto, Kusakabe, Suzuki and Shogaki ; 1993)

590 Matsumoto and Tsuzuki


The soil profile and the results of SPT's, CPT at Point C2 and the unconfined
compression tests are show in Fig. 4. The test site was characterized as a tluck
deposit of fully saturated diatomaceous mudstone. The variations of unconfined
compression strength, qu, and the secant modulus, Eso, with depth are relatively
uniform to the depth of T.P.-12m (T.P. :Tokyo Peil). The variations of the N-value
and the cone resistance, qc, are also uniform to the same depth.

Test Result of Pile Tl

The Statnamic test of Pile Tl was carried out 14 months after the completion of
the static load test. The Statnamic signals are shown in Fig. 5. The Statnamic load on
the pile head, F\'fn, and the pile head displacement, UO, as function of time, t, are shown
in Figs. 5(a) and (b). The measured displacement uo was differentiated once and twice
with respect to time to obtain the velocity, v, and acceleration, a, of the pile shown in
Figs.5(c) and (d). The measured peak velocity v=O.6m/s and the peak accelration a=
30m/s2 were very small compared with those induced by the driving of the diesel
hammer. The peak values of v and a during driving were 2.0m/s and 2400m/s2 ,
respectively.
It can be seen that the pile is pushed gently into the ground during the Statnamic
test. The acceleration of the pile head is an order smaller than during driving while the
penetration velocity is one third compared to driving.
The displacement of the soil plug, Uplug, is also indicated in Fig. 5(b), which is
the second integration with respect to time of the acceleration of the top of the soil
plug. Although a time lag of Uplug behind uo can be seen at the early stage of the pile
penetration, the change of uplug with time was almost identical to uo. This implies that
the open-ended pile penetrated into the ground as plugged. It may be interesting to
note that Pile Tl reached ultimate capacity as a plugged mode also in the static load
test.
Fig.6 is a pile-soil model for the Statnamic test (Middendorp and Matsumoto,
1994). In the pile-soil model, the pile is modeled as an elastic spring with a
concentrated mass, because the elastic deformation of the pile is included in the
measured displacement, uo, of the pile head. Based on this pile-soil model, the
Statnamic force, Fstn, the inertia force of the pile, Fa, and the soil resistance, FSOil, act
on the pile mass. The equilibrium of these forces can be expressed as follows
(Middendorp et aI., 1992) :

Fstn = Fa + Fsoil = mp a + Fsoil .................................................................. (1)

where mp is the mass of the pile and Fsoil is assumed to be the sum of the static
resistance, F u , and the penetration rate dependent resistance, F v :

Fsoil = F u + F v = F u + C v (2)

in which C is the damping factor.

591 Matsumoto and Tsuzuki


Time, t(ms)
0 40 80 120 160 200 240
7
GJ Z
o~
-6
5
F stn
L.._

.E '=0 4 F soil=F stn - mp a


-c
ltllL'"
GJ-c
3
.L: c 2
GJ ltl
a: .s 1
lL'" 0

- E
-1
-5
-
.;
E
~
a 0
5
--------
c
GJ 10
E
GJ
0
15 Pile head
ltl
20 (b)
0.. ---- Soil plug
III
25
0
0.8
-
~
0.4
->.
E
>
...
"0
0 ---
0 - 0.4
Qj
>
--
Ol

c:l
0.8
6
4 (d)
c 2
a
".;::; 0
l'll
L..
GJ -2
Qj
0
0
-4
-6
0 40 80 120 160 200 240
Time, t(ms)

FIG.5 Signals of Statnamic Test on Pile Tl

592 Matsumoto and Tsuzuki


F stn = Statnamic force

----:>. ! Fa = In e rtia force

F soil = F u + F v
i i = S oil resistance
Fv Fu
FIG. 6. Pile-Soil Model for Statnamic Test (Middendorp and Matsumoto, 1994)

The F\'ln vs. liO relation and Fsoil vs. liO relation are shown in Fig.7. The peak
values of F sm and Fsoil were 5.80MN and 5.95MN, respectively, indicating that the
inertia force, Fa, was relatively small (Fa = 0.15MN). The point of maximum
displacement on Fmil vs. liO curve is called the unloading point (Middendorp et aI.,
1992). The velocity of the pile, v, becomes zero at the unloading point and this point
can be considered as a static point, and as a result Fsoil is equal to F u according to
Eq.(2). In the unloading method, Fsoil at the unloading point is thought to be the
maximum static resistance obtained during the Statnamic load test, P max . Thus
estimated P max is equal to 5.2MN.
The Fwil VS. liO curve is compared with the result of the static load test in Fig.8.
A cyclic loading method was used in the SLT. Each virgin load was maintained for 1
hour. The ultimate bearing capacity, Pult = 4.8MN, was attained in the 4th loading
cycle. The Statnamic test was earned out from the end of unloading stage of the 6th
loading cycle. Note again that there was a rest period of 14 months between the end
ofthe SLT and the Statnamic test.
For comparison the Fsoil vs. liO curve has been shifted to the load Po vs. liO
curve at the 4th loading cycle in the SLT (Fig.9). These curves are indicated so that
liO of each curve starts from zero for comparison. Fsoil vs. liO curve from the
Statnamic test and Po vs. liO curve from the SLT are identical until Fsoil and Po reach
a load of3MN. The Statnamic curve deviates from the SLT curve after 3MN.
The distribution of axial forces along the shaft was measured during the
Statnamic load test. The axial force distributions measured during the Statnainic test
and in the SLT are compared in Fig. 10. The axial force distributions from the
Statnamic test are comparable with those from the SLT until F sm reached the Pult =
4.7MN from the SLT. It can be 'seen from Fig. 10 that wave propagation phenomena
can be neglected in the Statnamic test and that a tension force, which appears often in
the dynamic load test, is not generated at any level of the pile. The latter feature is
major advantage for cast-in-situ concrete piles which are likely to set cracked due to
excessive tension stresses by a dynamic load test.
593 Matsumoto and Tsuzuki
Pile head force, F stn and F soil (MN)
00 1 2 3 4 5 6 7
- E
E 5
'-'"
0
:::J
- 10
a.
en
"'0 15
"'0
m 20 unloading
Q)
..c::: point
OJ
.- 25 F stn - Uo relation
a.. ------ F soil - Uo relation
30
FIG.7 Fstn vs. uo and Fsoil vs. uo Curves of Pile Tl

Pile head force, Po and F stn (MN)


00 2 3 4 5 6 7

- ~
4th loading
cycle

-
~
60
80 6th loading cycle

~
()
100 r-=- ~j~_~
'="':--=-=-=-::_-=--

co
a. 120
cJ) loading
"'0 140
I
unloading J
I

"'0
co 160 ,'STATNAMIC
Q)
~ (F stn VS. uo)
..c::: 180
Q)
-----
-__ _ s ,, I
c::: 200 -----------------~
Unloading point ~
220 _
"--_....I--_....L-_---l-_~ _. L _ _ _ . . . L . __ _J

FIG. 8. Load-Displacement Curves from Static Load Test


and Statnamic Test of Pile Tl

594 Matsumoto and Tsuzuki


Pile head force, Po and F soil (MN)
E 00 1 2 3 4 5 6 7
E
o STATNAMIC
::J 5 (F soil vs. uo)
.....
C
<Ll
10
E
<Ll
U
ctl 15
c-
III
"'C 20
"'C
ctl
<Ll
1
..c 25 Unloaing
<Ll point

a. 30 '------'----'----=j-'------'---'-------'--:-j-----'
Japan Road CPT method
Association
(SPT- N)
FIG. 9. Fsoil vs. uo and Po vs. uo Curves of Pile Tl, together with Ultimate
Capacities Derived from the Unloading Point Method and Codes

Axial force, px(stn) and P x (MN)


Pile T 1
0 1 2 3 4 S 6 7
0 0
I
I
E 1 I
---><- 2
\
it2'
2
l:l rf11l
ro 37;0.'''9~;'
Q)
3
..c
Q)
4 4

a. S 5
E 6 6
a
~
'+-
Q) 7 7
c..>
c 8 8
ro
+-'
en 9 9
0 STN (Px(stn)
10 ----- SLT (P x) 10

11 11

FIG. 10. Comparison of Axial Force Distributions Obtained from Statnamic


Test and Static Load Test on Pile Tl

595 Matsumoto and Tsuzuki


Test Result of Pile T2

The Statnamic test of Pile T2 was performed 13 months after the completion of
the SLT. The soil inside Pile T2 was excavated up to O.5m below the pile toe level to
ensure that a side friction acted on the outer shaft only. The signals from the
Statnamic test are shown in Fig. II. The Statnamic signals of Pile T2 were almost
similar to those of Pile TI except that the maximum Statnamic force Fstn of Pile T2
was 5MN.

Time, t(ms)
0 40 80 120 160 200 240
6
Z
GJ
e- 5 (a) - - F stn

-~
0
"C
cu
GJ
.s::.
'0
LL
"C
l/l
4
3
2
.----- Fsoil=Fstn- m p a

c
GJ cu 1
.- c:
0- 0;
LL
0
. 1

- 10
E

-E
c
GJ
E
GJ
-5
0
5
(b)

0
10
..!!! 15
c-
.!!! 20
Cl
25
~ 0.8
~
E (c)
0.4
">
~ 0
0
0
GJ
> - 0.4

~ 6
~ 4
t::l (d)
c' 2
0
:;:::; 0
cu
~

GJ
GJ
0
U
-6
0 40 80 120 160 200 240
Time, t(ms)

FIG. 11. Signals of Statnamic Test of Pile T2

596 Matsumoto and Tsuzuki


Po and F stn (M N)
0 1 2 3 4 5 6
0
........
E 10
E
.........
0
:::J 20 4th cycle of loadin
+-'
c
Q)
30 loading
E
Q)
() ----- unloading
C'Cl 40
0...
(/)

"0 50 I
... ...
"0 ... .... ......
C'Cl
Q)
.c
60 -.. .... ...... -- -..
Q)
.- 70
0.. Unloading point /
80
FIG. 12. Load-Displacement Curves from Static Load Test and
Statnamic Test on Pile T2

Po and F soil (MN)


2 3 4 5 6
E
E
'-' I
0
::J 5 STN II
-
c::
Q)
10
(Fspilvs. uo)
I
E
Q)
u
ctI 15
c..
III
"'C
"'C
20 I~
ctI
Q) Jnloading
J:: 25 I point
~
ll.
30

Japan Road CPT method


Association
(SPT-N)
FIG. 13. Fsoil vs. uo and Po vs. uo Curves of Pile T2, together with Ultimate
Capacities Derived from the Unloading Point Method and Codes
597 Matsumoto and Tsuzuki
The Fsoil vs. uo curve is compared with the results of the SLT in Fig. 12. The
ultimate bearing capacity Pult = 3.7MN was attained in the 4th loading cycle of the
SLT. The ultimate capacity estimated by the unloading point method was 4.5MN
which was 22% larger than Pult from the SLT. The Fsoil vs. uo curve is compared
with the Po vs. uo curve in the 4th loading cycle again in Fig. 13. The Fsoil vs. uo
curve is in good agreement with Po vs. uo curve until Fsoil reached 3MN.
The axial force distributions from the Statnamic and the SLT are compared in
Fig. 14. A good agreement can be noticed. It also validates the assumption of the pile
acts as one mass.

Axial force, PAMN)


1 2 3 4 6 Pile T 2
00r---r--..--rr---,:;:.....-......--...r----,.,,......,.r.-~_-.....; 0 (E I. 4.01 m)

..-..
1
E
........ 2 2
x
3 , .;~;:,/
- 1.18m)
1J 3
'I':.Y',
. (E I.
co
Q)
..c 4 4
Q)

c. 5
5
I
I
E 6 6

-
I
0 I
L.. I

Q) 7 7
u
c 8 8
co
.....
C/)

Cl
9
- - - STN (Px(STN))
9

10 ------ SL T (P x) 10

11 11

FIG. 14. Comparison of Axial Force Distributions Obtained from Statnamic


Test and Static Load Test on Pile T2

USE OF THE STATNAMIC RESULT IN PILE DESIGN

The pile capacities derived from an empirical equation proposed in the


Specifications for Highway Bridges (Japan Road Association, 1990 : called JHA code
in this paper) and from the CPT method (ISSMFE, 1977) have also been indicated in
Figs.9 and 13. The blow count from the Standard Penetration Test (SPT N-value) is
usually used for the soil parameter in the JHA code. The JHA code gives a lower
bearing capacity, while the CPT method gives a higher one among several
representative codes in Japan and gathered from US, Canada, Norway. This topic are
discussed in a paper by Y. Michi and the authors, 1994.

598 Matsumoto and Tsuzuki


In the case of Pile T 1, P ult calculated from the unloading method reasonably fits
with the measured Pult within an admissible deviation (Fig. 9). The JHA code
prescribes that the safety factor F s of 2.8 is applicable to calculate the allowable load
Pa when a SLT is carried out. When a safety factor Fs = 2.8 is applied to the
unloading point load in the Statnamic test result, P a becomes 1.86MN (5.212.8 MN).
On the other hand, P a from the static load test result gives 1.68MN(4.712.8MN).
As mentioned earlier, Fsoil vs. uo curve is remarkably identical to Po vs. uo curve
within the range of allowable loads. This may encourages the use of the Statnamic test
result directly for the serviceability limit design of foundation piles where an
appropriate estimation of the load-displacement behavior of the pile is required. A
good performance for predicting the load-displacement behavior of the pile within the
allowable load is also detected in the case of Pile T2 (see Fig.l3).
Estimation of the pile capacity in tension is also required for pile design in Japan,
because ealthquake loads may generate tension forces on the piles. Therefore,
measurements of axial forces are performed usually in the SLT's in order to estimate
the shaft resistance the toe resistance separately. The JHA code prescribes that the
shaft resistance measured in the compression test has to be divided by a factor of 1.2
to obtain the tension capacity of the pile. The test results in Figs. 10 and 14 clearly
shows that the allowable shaft resistance measured by the Statnamic test can be used
for the pile design without any correction.

CONCLUSIONS

Through the series of tests in Nanao, the results indicate some advantages of the
Statnamic load test which can be summarized as follows:
The unloading point method (Middendorp et at, 1992) seems to be reliable for
the interpretation of rate effect of the Statnamic test in comparison with the static
load test.
Within a range of 50 % to 60% of the ultimate capacity in this particular test, the
load-displacement behavior oftained from the Statnamic test perfectly agrees with
the static load test results.
Cost efficiency ofthe Statnamic test was remarkably good in comparison with the
conventional method. Multiple tests at one site will achieve further cost efficiency
and improve the quality assurance of foundation piles.

The authors suggest the following improvements for the Statnamic load test :
An integrated instrumentation to measure the load distributions along the pile.
Reduction of the present noise level (80 to 90db).

The Study Group for Rapid Load Tests continues to evaluate the interpretation
methods through a desk top study and site tests of the Statnamic test and also will
study wide application of the Statnamic tests on such as lateral loading of foundations
on pile groups, and the Statnamic load test to improve initial load-settlement behavior
ofbored piles.

599 Matsumoto and Tsuzuki


ACKNOWLEDGMENTS

The Statnamic tests were conducted by the Study Group for Rapid Load Tests
in Japan. The authors extend our appreciation to Ir. Peter Middendorp, TNO Building
and Construction Research, and Mr. Patrick Bermigha1ll, Berminghammer Corp. Ltd.,
for their supports in carrying out the Statnamic tests and discussions in summarizing
this paper.

REFERENCES

Benningham, P. and Janes, M. (1989). "An innovative approach to load testing of high
capacity piles." Proc. Int. Conf Piling and Deep Foundations, London, 409-413.
ISSMFE (1977). "Report of subcommittee on standardization ofpenetration testing in
Europe." Proc. 9th Conf, Tokyo, Vol.3, 95-152.
Japan Road Association (1990) : Specifications for higlnvay bridges, Part IV :
substructures. The Japan Road Association, Tokyo (in Japanese).
Matsumoto, T, Kusakabe, 0., Suzuki, M. and Shogaki, T (1993). "Soil parameter
selection for serviceability limit design of a pile foundation in a soft rock." Proc.
Int. Syrup. Limit State Design in Geotech. Eng., Vo.V3, Copenhagen, 141-151.
Michi, Y, Tsuzuki, M. and Matsumoto, T (1994). "Design parameters for steel pipe
piles driven in a soft rock." Proc. Int. Conf Design and Construction of Deep
Foundations, Orlando (to be appear).
Middendorp, P. (1993). "First experiences with statnamic load testing of foundation
piles in Europe." Proc. 2nd Int. Geotech. Seminar "Deep Foundations on Bored
and Auger Piles", Ghent Univ., BelgiUlll, 265-272.
Middendorp, P., Benningha1ll, P. and Kuiper, B. (1992). "Statnamic load testing of
foundation piles." Proc. 4th Int. Conf Appl. Stress-wave TheoI)' to Piles, The
Hague, 585-588.
Middendorp, P. and Matsumoto, T (1994) : Private communication.

600 Matsumoto and TSllzuki


ANALYTICAL STUDY OF STATNAMIC TEST OF A CAST-IN-
PLACE CONCRETE PILE

Kiyoshi Yamashita 1 , Yasunori Tsubakihara2 , Masaaki Kakurai 3 ,


Takuhei Fukuhara4

Abstract

Dynamic and kinetic pile testing methods have recently at-


tracted attention because they offer simple and cost-effective alter-
natives to static loading tests. Kinetic loading tests are conducted
over a longer time frame than dynamic loading tests and the be-
havior of pile-soil system is closer to static loading tests. A kinetic
pile testing method called the Statnamic test has been reported to
yield load-displacement behavior that agreed approximately with
those from static loading tests. However, the Statnamic test is es-
sentially a dynamic test and the dynamic phenomena have to be
taken into account.
The authors conducted a Statnamic test of a 18.2 m long east-
in-place concrete pile whose diameter varied from 1.0 to 1.2 m and
the maximum vertical load was 6.28 MN. After the completion of
the Statnamic test, a static loading test using a vertical load of 12
IvIN was carried out on the same pile. The load-displacement curves
from the Statnalnic and static loading tests were in close agreement
'\vithin the first-lim.it-load determined by JSSMFE Standards.
The authors then attempted to numerically simulate the load-
displacement behavior of the Statnamic test using a three dimen-
sional finite element model in which the pile was modelled as re-
maining elastic and the soil was assumed to be elastic and elasto-
plastic. The numerically synthesized load-displacement curves were
generally in good agreement with the measured curves for the

1 Chief Researcher, R&D Institute, Takenaka Corporation,


1-5 Ohtsuka, Inzai-rnachi, Inba-gun, Chiba 270-13, Japan
2 Researcher, ditto
3 Senior Chief Researcher, di t to
4 Engineer, Takenaka Civil Engineering & Construction Co.,
Ltd.

601 K. Yamashita et ale


elasto-plastic analyses.

Introduction

The excessive noise and vibration associated with driven piles


can in nlany instances hinder their use in urban settings and as a
result cast-in-place concrete piles have recently found more popu-
larity in Japan. However, various factors can lead to considerable
reductions in the toe resistance of cast-in-place concrete piles in
non-cohesive soils.
Static loading tests are most appropriate for investigating the
load-displacement behavior of piles but these tests are for the most
part cumbersome, time-consuming and not cost-effective. On the
other hand, dynamic pile testing methods, which include dynamic
and kinetic tests, are simpler and more cost-effective and conse-
quently have been attracting greater attention as viable alterna-
tives to static loading tests. In the kinetic testing method, inertial
forces within the pile are small compared to the applied force and
the interpretation of the testing does not make use of the wave
equation framework ( Holeyman, 1992 ).
A kinetic pile testing method called the Statnamic developed
in Canada and the Netherlands uses reaction of a mass launched by
means of fast, expanding high pressure gasses in a cylinder to gen-
erate vertical load on the pile head (Bermingham and Janes, 1989).
The duration of the Statnamic loading is several times longer than
the duration of dynamic loading and the behavior of pile-soil system
is closer to static loading. However, the Statnamic test is inher-
ently a dynamic test and the dynamics of the pile-soil system need
to be considered (Middendorp et aI, 1992). EI Naggar et al.(lg92)
have shown that the load-displacement behavior from Statnamic
pile tests could be successfully simulated by an one-dimensional
mathematical model which took into account slip at the pile-soil
interface as well as energy dissipation in the far field.
In this paper, the results of in-situ Statnamic and static load-
ing tests on a cast-in-place concrete pile are presented. Secondly,
we present some results of three dimensional finite element analy-
sis for investigating the dynamic phenomena under the Statnamic
loading. The authors chose the finite element method, even though
the one-dimensional model is more practical, because the finite el-
ement method is more powerful when one wants to take account of
local variation of soil parameters in the pile-soil system.

Sunlmary of In-Situ Testing

Soil and Pile Concli tion

The test site is located 20 km south of Osaka in Mihara-cho.


Fig. 1 shows a schematic of the test pile and results of a soil
investigation. To a depth of 8.1 m from the ground surface there is a
fill. Beneath the fill lie medium to dense sand layers with some thin
clay layers to a depth of 24 m. Under these layers a very hard sandy
clay layer and a dense sand layer appear alternately. Primary wave

602
and shear wave veloci ties of the ground were obtained by means of
a P-S logging method as shown in Fig. 1. Water level was deduced
to be lower than the pile toe.
A 18.2 m long pile with diameter varying from 1.0 m to 1.2
m was constructed by an earth drilling method using bentonite
suspension. Shaft friction resistance on the upper 14 m length
of the pile was reduced by means of a number of small-diameter
borings adjacent to the circumference of the test pile to ensure
that most of the pile-head load is transmitted to the pile toe.
Soi I Wave velocity
Test pi Ie SPT P wave, Vp(Km/s)
boring N-val ue S wave, Vs(Km/s)
log o 2.0
o 0i-r--;r--;,.......,--"iSO 0 0 .5

EI ~SIIa II-d i aleter


... 'f boring
10 ~

e .:Sectlen 8
(14.2.)
- '---

;: 15 Sect len C
Q.
06.21)

L LJ
ll)
C
\Sectlen 0
07.2.) Vp Vs
1.0111
20

.... r--
25

30

Figure 1. Test pile and summary of soil investigation

Table 1. Instrumentation layout'

~
Pile head Section A Section 0
Transducer
Load cell 1

Laser sensor 1

Strain gauge 2 2
Accelerometer 1 1
Settlement gauge 1

603
Statnamic and Static Loading Tests
A Statnamic loading test was conducted on the test pile in
May 1993 ( Tsubakihara et aI, 1994) . Table 1 shows instrumen-
tation layout in the S tatnamic test. A load cell and a laser sensor
were placed on the pile head to measure the load and vertical dis:,
placement. Fig. 2 shows load-time and vertical displacement-time
histories at the head and toe of the pile. The duration of the im-
8..----'--r---~-----,----......--------,
,,-,
Z 6
~
'-"
"'0 4
ro
32
,,-, 0 ~~~~~-~~:::::::=~:----1
E
E
"-"2
.......
C
(l)

S
(l)
4
u
~6"-----"""""'----'------'-----'------'
~ 0
.-o 40 80 120 160 200
Time (ms)
Figure 2. Load-time and displacement-time histories

Pile-head load (MN)


,,-,
0 2 4 6 8 10 12
E 0
E
"-"
.......
r::: 5
(l,)

8
(l,)

-:.a
u 10
ro
~
IJl
15
Static test

"'0
ro
(l,)
...c 20
-.....
I
Q.)

0... 25
Figure 3. Load-displacement curves

604
pulsive loading was about 120 msec and the maximum load and
displaceluent achieved were 6.28 MN and 4.9 mm respectively. The
residual settlernent was 0.1 mm; therefore the settlement induced
by the Statnamic loading was almost elastic.
After the completion of the Statnamic test, a static loading
test using a load of 12 MN was conducted on the same pile. Fig. 3
shows the load-displacement curves from the Statnamic and static
loading tests. In JSSMFE Standards (1993), the first-limit-load in
the static loading test is defined as the load at a clear deflection

0, 6m
II
om
Om
I CD Pi I e ........ CD
Pi Ie"-
(J)

5. 0m
~

I-f-

@ I-f- ~:
....
c:

...... 9, 5m
...
@ ~

I
c:
C>
17. 5m = ,-- @.

@ 13. 5m

'- - @.
<ID
, 7, 5m
\
30m !
Non-reflecting
boundary

(a) Whole of the model (b) Details around pile elements

Figure 4. Element discretization

605
point appearing in a 10gP-logS curve, and synthetically judged by
several methods such as the S - logt method and the S /logt-P
method, where P:load, S:settlement at the pile head, and t:elapsed
time at a new load step. The first limit load was determined to
be 7 NIN in this test. The ultimate pile load was also determined
to be 10.9 NIN using Davisson's method. The two curves from the
Statnam.ic and static loading tests agree closely with each other
within the first limit load determined by JSSMFE Standards.

Analytical Study

Finite Element Method

An attempt was made to simulate the load-displacement be-


havior of the S tatnamic test by a three dimensional explici t fini te
element code DYN A3D ( Hallquist, 1981 ).
Fig. 4 shows the element discretization of the pile-soil sys-
tem, which consists of a concrete pile and surrounding soil mass,
represented by eight-node solid hexahedron elements. Considering
axisynl.Inetric case, only one-twelfth of the entire pile-soil system
was modelled. Deeks ( 1992 ) has pointed out that fineness of
meshing should be carefully considered by taking into account the
velocity of wave propagation in the soil and the loading rate on the
pile head in order to obtain accurate solutions. The mesh used in
this analysis was determined on the basis of Deeks's observation
and the total number of nodes and solid elements are 2440 and
1500, respectively.
Non-reflecting boundaries were introduced at radial and lower
boundaries of the soil mass to prevent the erroneous reflections at

Yi e I d --Et
stress

strain

Figure 5. Elasto-plastic behavior of soil


606
the boundaries. Deeks ( 1992 ) has, also, indicated that boundaries
form.ulated on the principle of zero energy reflection give inaccurate
results when subjected to stress waves with non-zero time averages
and therefore their use is hindered in pile driving problems. The
inability of the boundary to support static load and the resulting
rigid body motion leads to inaccuracies in the calculation of resid-
ual stresses and absolute displacement. In our analysis, relative
displacement and acceleration of the nodes on the lower boundary
were calculated to ehnl.inate the effect of rigid body motion.

Analytical Nlodels

In our analysis the concrete pile and soil were modelled as an


elastic material and in some analytical cases the soil adjacent to the
upper part of the pile was asslll11ed to be a simple elasto-plastic ma-
terial, in order to take account of the reduction of the shaft friction
along the upper part of the pile due to the small-diameter borings.
Fig. 5 shows the elasto-plastic behavior of the soil with isotropic
hardening. The plastic modulus E t which represents hardening on
the yield concli tion is asslUned to be very close to zero in the anal-
ysis. The yield condition of the von NIises is given in Eq. 1.

where a y is the von Mises yield stress.


Young's moduli and Poisson's ratio of the soil at very small
strains were determined from the meas1.1red prim.ary and shear wave

Table 2. Material parameters of pile and soil

~
Den sit r Young's PoJsson's I Yleld PIas tic
p modulus E ratio 1.1 stress oy modulus Et
Mat er IaI (g/cm 3 ) (MP a) I (k P a) . (MP a)
PII e concrete 2. 4 2. 45 O. 1 6 7 - -
X 10 4
Layer CD 1. 8 235 O. 25 - -
Cas e 1
(2) 1. 8 305 O. 45 - -
1. 9 782 O. 45 - -
I
@ 2. 0 136 2 O. 30 - -
2. 0 1 3 2 3 O. 30 - -
So i I
@ 1. 9 890 O. 45 - -
I
Layer ~ [nput data are same as cas e 1
Case2 10 6
Zone crJ 1. 8 3 9 O. 25 32. '3 l.X\O-4

2. 0 136 2 O. 30 951 l.X\O-4

2. 0 136 O. 30 - -
607
velocities in the soil by means of P-S logging as shown in Fig. 1.
Table 2 shows a list of analytical models and material parameters.
The attached numbers to the soil layeI's and zones in Table 2 cor-
respond to the numbers in Fig. 4. .. .. .
Case 1 represents an elastic analysis in which Young's moduli
of soil corresponding to the very small strain levels was used. Case
2 represents an elasto-plastic analysis in which Young's moduli of
soil adjacent to the upper portion of the pile were reduced to about
one tenth of those for the case 1 and the maximum shear strength
of soil in the same zone was capped at 19 kPa to account the ef-
fect of the small diameter borings adjacent to the circumference
of the pile. The maximum shear strength corresponds to a half
of the measured frictional resistance in the static loading test at a
load of 6MN. Yamashita et al.(1991) has indicated that shear mod-
uli of soil just beneath the pile toe of cast-in-place concrete piles
were reduced to less than 10% of those at very small strain lev-
els by the decrease in mean effective stress and the resulting shear
strain through elasto-plastic finite element analyses. In the case 2,
Young's moduli of the soil beneath the pile toe were reduced to
10% of the moduli at very small strain levels to account for the
effects of the soil loosening during the construction of the pile.
An equally distributed vertical pressure was applied to the
seven solid elements representing the pile head. The load-time his-
tory in the analysis is shown in Fig. 6. The loading consists of three
parts, a linearly increasing loading followed by a constant loading
of 0.46 NIN in the second part and finally an impulsive loading iden-
tical to the load data measured by the load cell at the pile head. In
DYNA3D, the maximum time increment which depends on element
size, Young's modulus and density of the material is automatically
set to ensure accuracy of the solution. Time increments used in the
analyses was 1.46 x10- 2 msec.

8.------,------r----..,...---.,.---......----.

oL....:::::::::::::::=.t._ _...L-_-----l!-_---L_ _...L:::==:::::::j


a 50 100 150 200 250 300
Time (ms)
Figure 6. Input load-time history
608
Resul ts of Analysis

Fig. 7 shows the typical deformation of the pile-soil systems


of the elastic and elasto-plastic analyses at the time when the max-
imun1. vertical displacen1.ent occurred at the pile head. It is seen
in the elasto-plastic analysis that the shear deformation is concen-
trated to the soil adjacent to the upper part of the pile.

Pi I e

J
(a) Elastic soil model, easel

(b) Elasto-plastic soil model, case2


Figure 7. Deformation mode of pile-soil system
609
Fig. 8 shows the shear stress in the soil elements adjacent
to the pile elements at depths of 1.25 m and 13.25 m from the
ground surface. In case 2, the soil elements attached to the pile
have reached the yield condition.

120 r I I I I

.........
~ 80 f- '. -
~
'-'
CIj 40 f- .. -
CIj
Q) .. __ .-._._.-._._.-'~
/-----, I .

..... 0 \,.. ....


""'"
CIj \ __ .r. ..".-
l-;
~ -40 - Casel -
Q)
...c -80 !- Case2 -
C/J
I I I I
-120
a 50 100 150 200 250 300
Time (ms)
(a) At 1.25 m from ground surface

120 I I I

.........
~
~ 80 - -

~
'-'
CIj 40 - -
CIj .,0 .. _ _ ---.
Q)
0
................................ :..:.:./ ........ ".
.....
l-;
CIj '--_ .. _-
..
l-;
~ -40 f-
Casel -
Q)
...c
C/J -80 - Case2 -
1 I I I I
-120
a 50 100 150 200 250 300
Time (ms)
(b) At 13.25 m from ground surface

Figure 8. Shear stress in the soil


610
Fig. 9 shows the calculated and measured displacement-time
histories at the head and toe of the pile. There are some residual
displacernents after the impulsive loading in case 2 of elasto-plastic
soil model.

a --. --. ..,


-..
s 1
_o-
S
....
"-'

~
Q) 2 Measured
a
Q)
Casel
0 3
~
....... Case2
0..
.-
0
C/)
4

5
a 50 100 150 200 250 300
Time (ms)

(a) At pile head

a
-..
E
E 0.5
....l:::
'-"

(],) 1.0 Measured


a
(],)
Casel
(j
~
1.5
....... Case2
0..
.-Q
tr.l 2.0

2.5
a 50 100 150 200 - 250 300
Time (ms)

(b) At pile toe

Figure 9. Pile displacement-time histories

611
Fig. 10 shows the calculated acceleration-time histories at
depths of 1.5 In and 16.5 m from the ground surface. The
acceleration-time histories at the pile toe represent similar mode
to those at the pile head. The n"la..."'(imum values of the acceleration
at the pile toe are reduced to 20% of the values at the pile head.

3 I I I I I

,.-..., 2 I- /. -
Ol) , \
---~
1
I'
, \ -
.-......ce
~
0 I
1\ .
/. .... :~ . ,
1\
'\
I0-oI
aJ
a . J .-..../ ......

\ .. /." ....j I, ~

....... './
aJ
u -1 I-
Casel \ j \i -
u 'I
\.
<C -2 I- Case2 -
I I r I I
-3
0 50 100 150 200 250 300
Time (ms)
(a) At 1. 5 m from ground surface

1.0 I I I

,.-...,
Ol)

--- 0.5 I- -

-
~
0
.......
ce
I0-oI a ~
r.
... \ .. :,,
y ../., ........ \.
I'
;
'\
\ .\.~ .. ~..

-
v
aJ I I \"" I
aJ
\ I \.
u Casel "(
u -0.5 I- -
<C Case2
I I I I I
-1.0
0 50 100 150 200 250 300
Time (ms)
(b) At 16.5 m from ground surface

Figure 10. Pile acceleration-time histories

612
Fig. 11 shows the measured and calculated load-displacement
curves at the pile head. In case 1, the calculated displacements are
smaller than the nleasured values because Young's moduli used in
the analysis were those at very small strain levels. The calculated
load-displacem.ent curve for case 2 is generally in good agreement
wi th the measured data.

Pile-head load (MN)


0 2 4 6 8
---E,..
t:
0
'-"
.......
~ 1
Il)

E
Il)
u 2
-.- co
0..
r.rJ
3
.... ....... .:

'"0
'"0 Measured " " ......
~
Il) ......
..c:: 4 Casel .....
I

.--
~
Il)

5
Case2

Figure II. Load-displacement curves at pile head

Pile-toe load (MN)


0 0.4 0.8 1.2 1.6
0
---E,..
t:
'-"
.......
ca.> 0.5 - - --
S
a.> 1.0
--__-
......... -- .. -.... .....
. - ..
uco ~

-.-
0 ..
r.rJ
""0
1.5
a.> Measured
0
....... 2.0 Casel
I

.--
~
a.>

2.5
Case2

Figure 12. Load-displacement curves at pile toe

613
Fig. 12 shows the measured and calculated load-displacement
curves at the pile toe. Although the calculated displacements are
somewhat smaller, case 2 shows relatively closer curve to the mea-
sured one. The measured data show that there is a lag between the
time of maximum pile-head load and maximum displacement. This
phenomenon is, also, observed in the results of both the elastic and
the elasto-plastic analyses.
Fig. 13 shows the measured and calculated load-transfer along
the length of the pile at the maximum pile-head load of 6.28 NIN.
The load transfer obtained from the static loading test is also shown
in Fig. 13. In case 2 the calculated load at around the pile toe is
generally in good agreement with the Statnamic and static test
results.

Conclusions

In this paper an attempt was made to simulate the kinetic re-


sponse of pile-soil system under Statnamic loading by simple elastic
and elasto-plastic soil models.
The load-displacement curve obtained from the elastic model
was looped in shape as well as the measlli'ed curve, which indicated
that the effects of inertial force and radiation damping were pre-
cisely represented in the analysis. However, the elastic model lead
to smaller displacements at the head and toe of the pile.
The calculated load-displacement behavior using an elasto-
plastic model for representing the soil adjacent to the upper portion
of the pile and incorporating a reduction in the soil moduli during
construction of the pile was generally in good agreement wi th the
measured data.

Load (MN)
0 1 2 3 4 5 6 7
0
.
0 Statnamic test
. f
4
Case 1
. I

-..
. I
Case 2
. I
E
'-'
..s:::
......
0..
8
- Static test
.
.
f
f

Q
Q)
12 .
f
_ . oJ
.--

Figure 13. Load trnasfer along the pile shaft

614
Acknowledglnents

The authors would like to acknowledge the contributions of


Iviessrs. NI. Chosokabe and T. Yamada of Takenaka Corporation in
conducting the Statnamic and static loading tests and also would
like to acknowledge the contribution of Ivir. IvI. Sugimoto of Talce-
nalca Corporation to the nmnerical simulation.

References

Berminghcun, P. and Janes, M., An innovative approach to load


testing of high capacity piles, Proc. of the Int. Conf. on Piling
and Deep Foundations, 1989, pp. 409 - 413.
Deeks, A.J., NUlnerical analysis of pile driving, Thesis, University
of vVestern Australia, 1992.
El Naggar, IvI. H. and Novak, M., Analytical model for an innova-
tive pile test, Can. Geotech. J. 29, 1992, pp. 569 - 579.
Hallquist, ,J. 0., User's manuals for DYNA3D and DYNAP (Non-
linear Dynalnic Analysis of Solids in Three Dimensions), U ni-
versity of California, Lawrence Livermore National Laboratory,
Rep. UeID - 19156, 1981.
Holeyman, A. E., Technology of pile dynamic testing, Proc. of 4th
Int. Conf. on the Application of Stress- Wave Theory to Piles,
1992, pp. 195 - 215.
Japanese Society of Soil Iviechamcs and Foundation Engineering,
JSSfvIFE Standards for Vertical Load Tests of Piles, 1993.
Middenclorp, P., Bermingham, P. and Kuiper, B., Statnamic load
testing of foundation piles, Proc. of 4th Int. Conf. on the
Application of Stress- vVave Theory to Piles, 1992, pp. 581-
588.
Tsubakihara, Y., Y arnashi ta, K. and Kakurai, IvI., A S tatnamic load-
ing test of a cast-in-place concrete pile in Japan, 3rcl Int. Conf.
on Deep Foundation Practice, 1994, pp. 299 - 305.
Yamashita, K., Fukuhara, T., Analysis of soil loosening around pile
bases during bored-pile installation, Piletalk International '91,
1991, pp. 161 - 164.

615
A Comparison of STATNAMIC and Static Field Tests
at Seven FHWA Sites
I P. Bermingham, 2 C. D. Ealy, and 3 J. K. White

The results of STATNAMIC Uoad tests conducted between


December 1990 and April 1993 at seven FHWA sites are
presented. Using the Unloading-Point model, ultimate capacity is
measured and rate effects are quantified. The resultant load-
displacement behaviour is compared to corresponding static load
tests.

Introduction
In over one hundred tests conducted on driven and cast in-situ piles, the measured
STATNAMIC load-displacement behaviour compares directly with static tests up to
the limit of elastic behaviour. As the pile begins to fail, however, velocity increases in
the elasto-plastic region. Although results from STATNAMIC load tests have shown
that rate effects are negligible for piles in very stiff soils and piles end-bearing in rock,
rate effects for piles in soft soils have been relatively large and have significantly
influenced load-displacement behaviour. The Unloading-Point model, as reported by
Hovarth, Bermingham, and Middendorp (1993), is a simple method of analysis for
determining the static resistance from a STATNAMIC test. As well, rate effects
present during a STATNAMIC test can be quantified with the Unloading-Point
model. Here, twelve test cases are presented in various soil types, ranging from gravelly
soil to very soft bay mud. Static resistance and the influence of rate effects are
measured for each test.
Berminghammer Corporation Limited, Hamilton, Ontario. Canada
2
Federal Highway Administration, McLean, Virginia
3
Berminghammer Corporation Limited, Hamilton, Ontario, Canada

61 6 P. Bermingham, C. D. Ealy, and J K White


Background
OV"V"'-=:::---------------:---------,
z 51 I maximum;
~ j I -1
load I
~ 0+---"""""'~-------;'----="--------1 [2 t Fmax I
E I c I
.s I ~-3+
c I g j .
~ -51i maximum
.
aXlmum
.
%-4 tmaxlmu
~ I load displacement 0 -5 jdiSPlacement ~
B-10 J at 90 m~ I ~t 98 ms I -6 .~F~u!!m~a~~L,-=::==;:~===:;:==::!;:=-+--J
o 20 40 60 BO 1 DO 120 140 160 0 1 2 3 4 5
Time (msec) Load (MN)

Figure 1. Measured STATNAMIC Signals. Load-Displacement Curve.

~:; t Unloading Point: velocity =0 I :: ~


1- j
~ 02 -t
.s 0.1
j

I
'I i
C
10 t
.~ 0 2 0 -!=o---""""-------'r--------j'---=-------'
u ~
~~1 ID

> -0.2 ~ -10


:i.
-0.3 t -20 Unloading Point
-0.4 t1
-0.5 +-1--+-~+-----__+~___+_~+--~--+-________1 -30 +-1 ----+----~_t___<--+-----+--+---+---f----+--+-_---ii
o W ~ ~ M 100 1W 1~ 1~ o W ~ ~ W 100 1W 1~ 1~
Time (ms) Time (ms)

Figure 2. Calculated Velocity and Acceleration


The STATNAMIC load test was developed to meet the construction industry's
demand for an accurate and cost-effective method of determining the load bearing
capacity of caissons and high capacity piles. In STATNAMIC testing, solid fuel is
burned within a pressure chamber. As the gas pressure increases, an upward force is
exerted on a reaction mass while an equal and opposite force pushes downward on the
pile. Loading increases linearly to a maximum and then unloads by a controlled
venting of the pressure. During the test (approximately 120 milliseconds), 2,000 load
and displacement readings are measured directly by a built-in load cell and laser
sensor,

The measured values of load and displacement versus time and load versus
displacement for a 5 MN test (760 mm diameter, 13.8 m long drilled shaft in stiff clay)
are shown in Figure 1. Characteristic of STATNAMIC load testing for any
foundation, the maximum displacement lags the maximum applied load-here,
maximum displacement of 5.7 mm at 98 ms follows maximum load of 4.48 MN at 90
ms. Net settlement is 1.77 mm. The "unloading point" (point of inflection between
loading and unloading) occurs at maximum displacement and zero velocity. Pile
accelerations range from a maximum of5g during loading to a maximum of -1.5g at
the unloading point. Velocity and acceleration versus time, calculated from the
measured displacement, are shown in Figure 2.

617 P. Bermingham, C. D. Ealy, and 1. K. White


The Unloading-Point Model

Due to the relatively long duration ofSTATNAMIC loading, pile/soil behaviour is not
controlled by a stress wave travelling down the pile as in a dynamic load test Instead,
the pile is under constant compression during STATNAMIC loading, comparable to a
static load test, and can be modelled as a mass on which the following forces act:

Fstnl 1. STATNAMIC Force (F stn)


2. Inertial Force (F~

3. Soil Resistance (Fsoil = Fu + Fv + F p)

Soil resistance is comprised of


static soil resistance Fu ,
damping forces from soil Fv ,
iii
Fu Fv F p
and pore water pressure resistance F p .

Figure 3. Pile Forces


For equilibrium, the following equation is valid:

Fstn(t) = Fsoil(t) + Fa(t) [1]


where: Fsoil(t) = Fu(t) + Fv(t) + Fp(t) [2]

and Fstn(t) = STATNAMIC load (measured)


Fsoil(t) = soil resistance
Fu(t) = ku(t) (static soil resistance) [3]
Fv(t) = C.v(t) (damping force from soil) [4]
Fp(t) = pv(t) (pore pressure resistance) [5]
Fa(t) = ma(t) (pile inertial force) [6]
k = spring stiffness (N/m)
C = damping factor (Ns/m)
P = pore water pressure damping (Ns/m)
m = pile mass (kg)
u(t) = measured displacement (m)
vet) = du/dt = velocity (m/s)
aCt) = d2u/dt 2 = acceleration (m/s2 )

(For the purposes of this study, it is assumed that pore water pressure resistance is
included as part of the damping force from soil. Limited field observations have
shown that pore water pressure resistance is less than 5%. Tests have also shown that
inertial forces are minimal since pile accelerations are typically on the order of 1g.)

618 P. Bermingham. C. D. Ea/y. and.J. K. While


Thus, from [1] and [2]:

Fstn(t) = Fu(t) + Fv(t) [7]


or Fu(t) = Fstn(t) - Cv(t) [8]

where the damping coefficient, C, is calculated from [8] at the unloading point (UP)
and at the maximum applied load (F max ):

at UP: velocity = 0, Fu(t) is a maximum (F umax ) .'. F umax = Fstn(UP) [9]


at F max : Fu(t) is a maximum (F umax ) .'. Fumax = F max - Cv(F max) [10]

Fmax - Fumax
Thus, from [9] and [10]: C = [11 ]
V(Fmax)

From Fu(t) and u(t), a load-displacement curve can be drawn representing the static
soil resistance as a function of displacement. The resultant curve passes through the
"ultimate capacity" point, Fumax . The portion of the original STATNAMIC load-
displacement curve to the right ofFu(t) depicts the rate effects due to velocity. (Note:
At the unloading point, velocity equals zero and, thus, the force due to damping is
zero )

o ...,...,......,---------------------, Fmax =4.48 MN


-1
E
F umax =4.04 MN
rate effects
5-2 Fv =Cv(t)
c
Q)
v(F max ) =0.14 m/s
E -3
Q)
u
C'O
C =3.14 MNs/m
~-4
(5

-5
F umax
-6 -t---t---t---t---f-------1---+---+---+.---+---j
o 2 3 4 5
Load (MN)
Figure 4. u(t) versus Fstn(t) and Fu(t)

A good measure of the rate effects present in a STATNAMIC test is the percentage of
the force due to damping included in Fmax (or the percentage difference between
F max and F umax ). In the above case, 9.8% of the maximum applied STATNAMIC
load is the result of rate effects due to velocity.

61 9 P. Bermingham, C. D. Ealy, and J. K. White


Although the Unloading-Point model provides a readily available measure of the force
due to damping at the maximum applied load, it assumes linear damping over the
entire loading region and negligible inertial effects. Since comparisons between static
and STATNAMIC tests have shown that damping is very low in the elastic region,
linear damping under-predicts static capacity in this region. Furthermore, results have
shown that the stress-strain relationship of the pile during unloading provides valuable
insight into the next static loading cycle and, thus, the true static equivalent for a
STATNAMIC test, but more work needs to be done to quantify the effects of
damping during unloading.
Q-.r-<;,......,-----------------------,
low damping
-1 in elastic region
E
.s -2 yield
-c:
(l)

E -3 static behaviour
(l)
'-'
C'C
~-4
rate effects
(5 due to damping
-5

-6 +---+---+---+---+---+---+----t------1~-+_____l

Q 123 4 5
Load (MN)
Figure 5. u(t) versus Fstn(t) and Fu(t)

Load Definitions

The following load definitions apply for a STATNAMIC load test: The unloading
point (Fumax) can be considered the maximum static resistance measured by the test.
It is not necessarily the ultimate load as it does not include pore water pressure or pile
inertial forces. The ultimate load (Fu) is the maximum load which a foundation can
sustain under static loading conditions. The failure load is the load at which the load-
displacement curve begins to plunge (displacement begins to increase rapidly.) During
a static test, plunging failure occurs when pile velocity changes form zero to some
positive value. During STATNAMIC testing, we look to identify the transition from
positive velocity to zero velocity. The "design load" for the foundation is normally
taken as one-half of the failure load. The maximum applied load (Fmax) is the
maximum load applied to a foundation during a loading test. During a static test, the
maximum applied load normally corresponds to the maximum displacement. During a
STATNAMIC test, the maximum applied load may be substantially greater than the
maximum static resistance. The percentage difference between F max and Fumax
gives a measure of the rate effects present during a STATNAMIC test.

620 P Bermingham, C. D. Ea/y. and J. K. White


Results
STATNAMIC tests were conducted,
February 13 & 14, 1991, in Shreveport,
Louisiana on two drilled shafts which
0 j J r
had been previously tested to failure on
February 6 & 8, 1991. During each test,
g'j 7 '//~
1
I

load versus depth output from several


Mustran load cells, cast into the shafts at
tl~t. . . .' -e- Static ~I
'
different depths, was recorded (Figure
..... STATNAMIC
6). The correlation of measured load -15 I I
distribution along the shaft for static and 0 2 3 4 5
Applied Load (MN)
STATNAMIC was good while the
agreement at peak load was excellent. Figure 6. Measured Load Distribution

The Mustran cell data suggests that the STATNAMIC test method provides a
reasonable means of determining the shaft and end-bearing resistance of the
foundation. It is probable that residual stress due to the prior static loading effected
the behaviour of the shaft and load distribution to some degree, though it was not
measured.

The equivalence of the load distribution results highlights the fact that STATNAMIC
testing loads the foundation as a rigid body. Load at the toe of the pile may lag behind
the top by 2.8 milliseconds, however, the long duration ofloading (100 ms) mitigates
any stress wave phenomena. The duration of loading is ten times longer than the pile's
natural frequency as has subsequently been confirmed by similar test programs in
Japan by Matsumuto (1994) and Yamashita (1994)

The STATNAMIC test is shown as a reloading test in Figure 7. The load-


displacement results compare directly with the static test results. The test measured
very little damping (maximum velocity was .30 mls) as the 5 1v1N STATNAMIC
device did not load the pile beyond its elastic range.
r=e=<o=E;::::;;:::S:::;;;:::;;;:~------=~~=---
-e- Static 1-20/1-49 Interchange
- STATNAMIC Shreveport, Louisiana

900 mm diameter 11.3 m long


reinforced concrete
silty sand and clay
Maximum Load =5.93 MN
Unloading Point =5.65 MN
% Damping =4.7%
-50 +--+---1--+--+--+--+--+-+-+--1--+1-+--+--+--+----1 Velocity (F max) =0.22 mls
o 2 3 4 5 6 7 8
Load (MN)

Figure 7. STATNAMIC and Static Load-Displacement

621 P. Bermingham, C. D. Ealy, and 1. K. White


STATNAMIC tests were conducted at o ~:::----------------,
production shaft
Gallup, New Mexico on a 760 mm -1
.43 mm/MN
diameter test shaft and a 1,200 mm E
diameter production shaft (Figure 8). ~-2
The production shaft showed a much ~-3
greater stiffness than the test shaft (1.0 l.4 "'.,'\
mmJMN to .43 mm/MN). Since the 0 test shaft ~ )"
STATNAMIC test did not mobilize the -5
1.0 mm/MN --/
full capacity of the production shaft (there -6 +--_+--~--+--+___+_---+-~__+_-_+_----.-___4
was no net settlement), it was not 0 2 3 4 5 6 7
Load (MN)
possible to determine the ultimate .
capacity. A larger STATNAMIC device Figure 8. Test and Production Shafts
would be needed to mobilize the production shaft. The test shaft was mobilized,
resulting in a maximum static capacity of 4.04 MN and 1. 7 mm of net settlement.

The static and STATNAMIC results are shown together in Figure 9. Since the static
test was conducted prior to the STATNAMIC test (net settlement of 10.4 mm),
STATNAMIC loading is shown as a second loading cycle. The load-displacement
behaviour is virtually identical to that for the static load-displacement. At the start of
failure (3.70 MN), there is little perceptible change in the slope of the load-
displacement curve. However, shaft displacement continues to increase as the load
decreases At the unloading point (4,04 MN), the applied load to the pile head is
equal to the resistance of the shaft and the velocity of the shaft is temporarily zero. As
the load continues to decrease, the shaft rebounds as in the static test.

The measured static resistance at the unloading point is 4,04 MN, 10% higher than the
maximum resistance measured by the static test (3,60 MN), perhaps attributable to
some strain hardening due to reloading as described by Brown (1992). The
percentage force due to damping included in F max is 9.8%.

O~=------------------,
""fr Static 1-40 Interchange
- STATNAMIC Gallup, New Mexico
E -5 (test shaft)

--
E
C
QJ
760 mm diameter, 13.8 m long
drilled shaft
~ 0 F:::::::~-=--_
1
C,)
co
stiff clay
0.. Maximum Load 4.48 MN =
6-15 Unloading Point 4.04 MN =
% Damping 9.8% =
-20 +---+----+--+---+--t---+----+--I----+---I Ve locity (Fmax) O. 14 mls =
o 234 5
Load (MN)
Figure 9. STATNAMIC and Static Load-Displacement (test shaft)

622 P. Bermingham, C D. Ealy. and J. K. White


At Cupertino, California, two 900 mm
diameter shafts were tested with a 6 MN 1,.... 2
,....
3 4
-,....
5
STATNAMIC device. Shaft 2 and shaft
4 had previously been tested by reaction
beam and hydraulic jack. The test site is
shown in Figure 10. The results of the
STATNAMIC and static tests on shaft 4
are shown in Figure 11. The
STATNAMIC curve is offset by 15 mm L- '-

to reflect the initial settlement due to


static loading. As described above,
) , 9.1 m)
\. ) \.
STATNAMIC loading is shown as a 12.2m
second loading cycle when preceded by
a static test. Figure 10 Pile Layout

The STATNAMIC test results indicate a somewhat lower capacity than the static load
tests Due to the cramped test site, the reaction piles were installed at a distance of 3
pile diameters, measured centre to centre, and may have resulted in substantial load
transfer between the reaction piles and the test shafts. The results of the Cupertino
site indicate a possible over-prediction of the static test results by 15% and emphasize
the advantages of testing without the use of reaction piles.

Damping was relatively low in this dry, non-cohesive soil. The percentage force due
to damping included in F max is 15.3%. Due to very low velocities during unloading,
the unloading portion of the STATNAMIC test is virtually identical to the static test.

o '!k==------------------, Caltrans Yard


~ Static Cupertino, California
E -5 - STATNAMIC
(pile 4)
-S 900 mm diameter 9.1 m long
1:
11l
E-10 drilled shaft
11l
u gravelly soil
Maximum Load =5.81 MN
ctl
c..
<Il
0-15 Unloading Point =4.92 MN
i % Damping = 15.3%
-20 +--+--+-+--+--+---+-+--+-+--,+-I-+--+--+------1 Velocity (F max) =0.23 m/s
o 234 567
Load (MN)

Figure 11. STATNAMIC and Static Load-Displacement (pile 4)

623 P. Bermingham, C. D. Ealy, and J. K. White


At the Albuquerque, New Mexico test site, the STATNAMIC loading tests took
place approximately three weeks after the static loading tests (compression and
tension). Figure 12 shows very close agreement of the load-displacement behaviour
for STATNAMIC loading and static compression loading. The STATNAMIC test
resulted in a net settlement of 42 mm, returning the foundation back to the same depth
prior to the tension test. The results indicate a strain-hardening soil with little elastic
rebound as in the original static test and typical oftests in sand.
A second STATNAMIC test was conducted on the same test pile the next day to
measure the performance after the pile toe had been reseated (Figure 13). The second
test was stiffer than the first cycle as would be expected from a second load cycle.
The maximum measured static capacity of 7.82 MN compares well with the static
capacity of 8.12 MN (average of Brinch-Hansen and Chin failure criterion). The
percentage force due to damping included in Fmax was 18.6%.
These two tests clearly demonstrate the STATNAMIC test's ability to mobilize a large
diameter shaft, accurately measure the load-displacement behaviour, and determine
the ultimate bearing capacity of the foundation.
O~--.,.........=-------------, 1-40 Interchange
-e- Static Albuquerque, New Mexico
-10
- STATNAMIC test 1
~20
....... 760 mm diameter, 18 m long
t30 drilled shaft
E
~40 saturated silty sand
ctI
c.. Maximum Load = 6.60 MN
.!:Q-50 Unloading Point = 5.05 MN
o
-60 % Damping = 23.5%
Tension Compression Velocity (F max ) = 0.86 m/s
-70 +-t-t--+-+--t-t--+-+-t-t---+--I-+-fL-+-t---+-+-+-I-+---j
-2 -1 o 2 3 4 5 6 789
Load (MN)
Figure 12. STATNAMIC and Static Load-Displacement (test 1)
0.,----=--------------, 140 Interchange
-e- Static
-10 Albuquerque, New Mexico
- STATNAMfC
E test 2
E-20
....... 760 mm diameter, 18 m long
~30 drilled shaft
E
~40 saturated silty sand
ctI
c.. Maximum Load = 9.61 MN
.~50
o ----'- --'-.'--_ .. Unloading Point = 7.82 MN
-'-'-'
-60 % Damping = 18.6%
Tension Compression Velocity (Fmax) = 0.25 m/s
-70 +-+-+-I--+--+-+-+-+-+-+-I--+--+-+-+-+-+-+-f--Jl-+-+-+-t-+-I
-2 -1 0 2 3 4 5 6 7 8 9 10 11
Load (MN)
Figure 13. STATNAMIC and Static Load-Displacement (test 2)

624 P. Bermingham, CD. Ea/y, andJ K White


Two STATNAMIC tests were conducted on model pile groups at the Turner-
Fairbanks Highway Research Center (TFHRC) in McLean, Virginia. Both pile
groups were arranged in a 3 x 3 array. Pile group "A" consisted of nine 67 mm
diameter, 3.2 m long pipe piles. A smaller group, group "B", consisted of 42 mm
diameter, 2.0 m long pipe piles. Centre to centre spacing for both groups was 3
diameters and piles were filled with an epoxy concrete grout.
As in tests conducted on single piles, typical STATNAMIC characteristics were
observed: the maximum displacement lags the maximum applied load; at the
unloading point, velocity is zero; loading is linear in the elastic region; the pile group
does not plunge, as it would in a static test to failure, because the applied load
continues to increase rapidly after the ultimate load is reached.
Figure 14 shows three STATNAMIC tests and the corresponding static test for
group "A". The average maximum displacement of the 2nd and 3rd test yields a
maximum displacement of64 mm at a load of 0.35 MN, approximately 30% greater
than the static plunging load. For pile group "B", the maximum STATNAMIC
displacement of 25 mm at a load of .03 MN underestimates the static plunging load
of .086 MN, perhaps due to the high weight of the pile cap relative to pile capacity.
O~----------------,
-e- Static TFHRC Pit
- STATNAMIC McLean, Virginia
pile group "A"

1; -3 67 mm diameter, 3.2 m long


E clay
~-4
rn
a.
c5 -5
-6

-7 -f---+---t-----i---+----'I+---+--+---t---+------l
o 0.1 0.2 0.3 0.4 0.5
Load (MN)
Figure 14. STATNAMIC and Static Load-Displacement (group "A" 1, 2 & 3)
o j"i~~~~----=-e-=S~t~a~tic=---I TFHRC Pit
--5 - STATNAMIC McLean, Virginia
E pile group "A"
5
~10
Q)
67 mm diameter, 3.2 m long
E
Q)
clay
~15 Maximum Load = 0.44 MN
C.
(/) Unloading Point = 0.35 MN
_20 % Damping = 20.5%
Velocity (F max) =0.18 m/s
-25 -f---+--+----+--+-----if----+--+---+--+----l
o 0.1 0.2 0.3 0.4 0.5
Load (MN)
Figure 15. STATNAMIC and Static Load-Displacement (group "A" test 2)

625 P. Bermingham, C. D. Ealy, and J. K White


At the Caltrans 280 test site in San Francisco, California, STATNAMIC and static
tests were conducted on different 410 mm diameter 33 m long drilled shafts (Figure
16). Pile 49 was constructed in soft bay mud with the tip at an elevation of about 29
meters (2 meters above the foundation layer of dense sand). The large difference
between the maximum applied STATNAMIC load (3.28 MN) and the unloading point
(1.70 MN) reveals the magnitude of rate effects (1.58 MN) in the soft bay mud. The
initial portion of the STATNAMIC loading curve shows close agreement with the
static loading test up to the point of failure. The Caltrans failure criterion, designed to
limit pile cap rotation to a maximum of 12.5 mm during earthquake activity, was
applied to static and STATNAMIC. Both results indicated 1.3 MN at 12.5 mm.

The results of the static load test on pile 48 are included for comparison because pile
48 was first tested in compression (1.4 MN) whereas pile 49 was first tested in tension
(1.2 MN) prior to being tested in compression by a static and then STATNAMIC test
A static compression test was conducted on pile 49 immediately after the
STATNAMIC test, resulting in an axial compressive load of 1.2 MN or 8% lower
than the 1.3 MN measured from the STATNAMIC test conducted the previous day,
possibly due to the loss of skin friction.

Due to the relatively long piles (33 meters) and very weak plastic clay, STATNAMIC
results showed abnormally high pile velocities and accelerations. Maximum pile
velocity was measured as 1.4 m/sec and maximum acceleration during loading as 4g.
Had the piles been founded in the underlying sand layer, the expected velocities would
be in the normal range. Nonetheless, pile acceleration of 4g are two orders of
magnitude less than those experienced during driving of the piles (400g.)

0 ........=:--------------_
--e- Static pile 48 Caltrans 280 Test Site
- STATNAMIC San Francisco, California
-10
E Failure Criterion
E 410 mm diameter, 33 m long
~-20 12.5 mm
Q) .drilled shaft (pile 49)
E
Q)
Soft bay mud (weak clay over
aJ-30
i5..
UJ
sand)
i:S / Maximum Load =3.28 MN
-40
//.i Unloading Point =1.70 MN
% Damping = 45.7%
-50 +--_+--_+-_+--'----+-_-+---_--+-_-+-----4
o 2 3 4 Velocity (F max ) = 1.36 m/s
Load (MN)

Figure 16. STATNAMIC and Static Load-Displacement

626 P. Bermingham, C. D. Ealy, and.! K. White


Texas A&M presented the first
_1:t~Pile7
opportunity to test full-scale caissons in
silty sand and clay. The three shafts ......20
~
~
- , --- .... .......

showed very different responses to 1 l-~ ----_-----,L..-~


30
pile

STATNAMIC loading (Figure 17). The


test results were used to determine the 140 j'
/
//pile 2

inertial and damping effects of friction


150
g- I
piles in sand and clay. An initial 0.60 +
I
estimate of the failure load of the three -70 Ii
shafts was determined from the slope of -80 ~-+----+--+---+--+----+---+---+---+----.-,----II

the load-displacement curve, providing o 2 3 4 5 6


Load (MN)
accurate correlations for pile 4 (12%
Figure 17 Pile 2, 4 and 7
over static) and pile 7 (15% over static).
The failure point for pile 2, however,
was overlooked.

Subsequently, the Texas A&M pilot test


program has revealed that the onset of
failure is clearly visible if load and ~ 2
~Ht
1.3 M N : / \
I~\
velocity are plotted against time (Figure :~ t__ .. --\~'>
/==-/=7:- _
18). Using the Unloading-Point model, n t K.
developed after the Texas A&M testing, ~q
predictions of ultimate capacity and ;2+
derived static behaviour show better %3t
results, The unloading point is 4,05 MN >-_4 !--+--t--+--+--+--+--+--t--+--+--+--+--+--t--+----i
5 - I I II I I

for pile 4 and 3.0 MN for pile 7, within o 20 40 60 80 100 120 140 160
1% and 11 % predicted from static Time (msec)

testing (Figure 19 and Figure 20). The Figure 18 Load and Velocity
unloading point for pile 2 was 1.50 MN.
O,......--e=e="'l3='6>=e:::;;::-::------------,

-e- Static TexasA& M


E -50 STATNAMIC College Station. Texas
g
-
c
Q)
E-100
Q)
900 mm diameter 9.1 m long
(pile 4)

C,,) reinforced concrete


ell

Q.
5 r---~~~======::::==~-
150 I------_-_--_--_--_--_--_--d-.:~~_----
__
silty sand and clay
Maximum Load = 5.5 MN
Unloading Point 4.05 MN =
-200 +--+--+-+---+---+----+-+---+---+-f---+-----1 % 0 amping 26.4 % =
o 2 3 4 5 6 Velocity (F max ) = 1.09 mls
Load (MN)
Figure 19 Texas A&M pile 4

627 P Bermingham, CD. Ealy, and J K. White


-130 - r - - - - - - - - - , - - - - - - - - - - ,
-e- Static TexasA& M
,?135 - STATNAMIC College Station, Texas
E (pile 7)
'-'
'[;140
Q)
900 mm diameter 9.1 m long
E
Q)
reinforced concrete
~145 silty sand and clay
C.
rJl =
Maximum Load 4.50 MN
Q150 =
Unloading Point 3.00 MN
=
% Damping 33.3%
-155 +---+-+---+--1---+---1-----+---1----+--1 Velocity (F max ) = 0.48 mls
o 1 2 3 4 5
Load (MN)
Figure 20 Texas A&M pile 7

-100 . , - - - - , . . . . - - - - - - - - - - - - -__
Texas A & M
-150 -r----"------_ College Station, Texas
E (pile 2)
E 1.5 MN
E=200
Q)

E
~~'-

~250 p--==~
---~
900 mm diameter 9.1 m long
reinforced concrete
ctI
silty sand and clay
C.
CIl
=
Maximum Load 4.20 MN
0.300 -e- Static Unloading Point = 1.5 MN
- STATNAMIC % Damping = 64.3%
-350 +--+---t-+---+----+-+--+---t-+---+----+-----i Velocity (F max ) = 2.09 mls
o 234 5 6
Load (MN)
Figure 21 Texas A&M pile 2

Pile 4 compares very well with the static test results. Here, Davisson's criterion
defines failure at approximately 2.8 MN, whereas the standard method (DIl 0 +
PLIAE) defines failure at a much larger displacement and corresponding load; 4.0
MN. The 31 % difference between Davisson and the standard method is due primarily
to the fact that the pile capacity is increasing with depth of penetration If a second
load test was performed, the capacity of the pile using either method would increase.
The percentage force due to damping included in F max is 26.4%, Pile 7 was the
stiffest pile tested. The maximum applied load was 4.50 MN and the net settlement
was 6.1 mm. The percentage force due to damping included in F max is 33.3%.

The capacity of pile 2, as measured by the Unloading-Point model, falls right between
the first and second static load tests and accurately measures the increase in capacity
due to reloading.

628 p, Bermingham, C. D. Ea/y, and J K White


Conclusions

In total, 12 tests were conducted on cast in-situ foundations without damage to any of
the shafts. The STATNAMIC devices were able to mobilize capacities ranging up to
1a MN. The loading equipment was able to mobilize each foundation without
introducing large accelerations or tension forces in the pile-pile accelerations were
two orders of magnitude less than dynamic testing. As well, prior to failure, shaft
velocities were one order of magnitude less than those of dynamic testing-a.S mls
compared to 5 mls. The measured STATNAMIC load-displacement behaviour was in
all cases virtually identical to the static load-displacement behaviour up to the point of
yield or limit of elastic behaviour.
Damping resistance, which is dependent 3,--------------_ ,

/
~ Texas 2.71 m/s I
on the rate of loading, becomes 2.5

significant only as the pile begins to 2 Caltrans 1.38 m/s


plunge and displaces soil. The damping 1 .5 1 Shreve 0.30 m/s
Gallup 0.19 m/s I
forces, which vary according to soil type :?;o 1
and pile velocity were measured by 'g
STATNAMIC and typically ranged from ~o.: ~~~====:~L6----:::~~~-~
5% to 64% (Table 1). Where velocities
exceeded 1 mls (Caltrans and Texas -0.5
~

A&M pile #2 in Figure 22), model -1 +---+--+---+---+--+---+---+---+-~-+---+----1


assumptions of a rigid body pile, linear 40 60 80 Ti~OeO(ms)120 140 160

damping, and negligible inertial effects


over-predict static behaviour. More Figure 22 Velocities at 4 sites
research into the relation between
velocity and damping, inertial effects
during unloading, and pore water pressure are required.

lTest Fmax Fumax Fv %diff v(F max )


(MN) (MN) (MN) (%) (m/s)
Shreveport 5.93 5.65 0.28 4.7 0.22
Gallup 4.48 4.04 0.44 9.8 0.14
Cupertino 4 5.81 4.92 0.89 15.3 0.23
Cupertino 2 6.00 5.06 0.94 15.7 0.18
Albuquerque 1 6.60 5.05 1.55 23.5 0.86
Albuquerque 2 9.61 7.82 1.79 18.6 0.25
McLean (large) 0.44 0.35 0.09 20.5 0.18
McLean (small) 0.16 0.03 0.13 81.3 1.14
Caltrans 3.28 1.70 1.58 45.7 1.36
Texas A&M 4 5.50 4.05 1.45 26.4 0.48
Texas A&M 7 4.50 3.00 1.50 33.3 1.09
Texas A&M 2 4.20 1.50 2.70 64.3 2.09

Table 1. Damping Forces at Maximum Applied Load

629 P. Bermingham. C. D. Ealy, and 1. K. White


The Unloading-Point model provides an accurate and reliable method of estimating
the rate effects of the soil and determining the static load-displacement behaviour of
the pile. STATNAMIC accurately measures the displacement at design load and at
failure load even when the damping forces are high (see Caltrans). When
STATNAMIC testing follows static testing, the result must be viewed as a reloading
cycle and not as a prediction of initial settlement. It is evident, as well, that the
unloading portion of the test reveals much about the static loading behaviour. In 9
tests (excluding two not failed and one tested twice by STATNAMIC), the unloading
point yielded predictions of static capacity between 77% and 131 % of those by static
tests (Table 2).
Gallup
Cupertino
Albuquerque
McLean "A" B
McLean "B" ~
Caltrans r=
," """.".".""""",.,.,.".'. """"""",','1
Texas A&M 4 I. J
Texas A&M 2 r"""""""""'d l Illid Static
r"""""" ,.".,.. ,.,.,',.,.,."',,..1 D STATNAMIC
Texas A&M 7 1

o 2 3 4 5 6 7 8 9 10
Ultimate Capacity (MN)

Table 2. Static and STATNAMIC Capacities


Today, STATNAMIC load testing offers a very cost effective and accurate means of
measuring the load-deflection behaviour and ultimate capacity of pile foundations.
The fundamental advantages of direct measurement of load and displacement coupled
with a perfectly controlled means of applying a load to a foundation will in the future
be applied to testing spread footings and the lateral capacity of pile groups.

References
"The Equilibrium Point Method of Analysis for the STATNAMIC Loading Test with
Supporting Case Histories," Horvath, R.G., Bermingham, P., Middendorp, P., Proceedings of
the Deep Foundations Conference, Pittsburgh, October 18-20; 1993.
"Evaluation of Static Capacity of Deep Foundations from STATNAMIC Testing," Brown,
Dan A., South-East Geotechnical Conference, Natchez, Mississippi, October 4-8, 1993.
"Comparative Study of Static Loading Test and STATNAMIC on a Steel Pipe Pile Driven in
a Soft Rock," Tatsunori Matsumoto, Makoto Tsuzuki, Yuji Michi, Proceedings of the 5th
International Deep Foundations Institute Conference, Bruges, Belgium, June 13-15, 1994.
"Kinetic and Dynamic Loading Tests of a Cast-in-Place Concrete Pile," Kiyoshi Yamashita,
Takuhei Fukuhara, Proceedings of the 5th International Deep Foundations Institute
Conference, Bruges, Belgium, June 13-15, 1994.

630 P. Bermingham, C. D. Ea/y, and 1. K. White


A Promising Method for Improving Drilled Pier Performance in Rock

Chris M. Haberfield, Serhat Baycan and Timothy D. Chamberlain l

Abstract

The use of an expansive cement admixture to increase drilled pier perfor-


mance in rock is discussed. The admixture promotes internal sulphate attack of
the concrete fonning the pier, which, in turn, results in the concrete expanding
against the surrounding rock. If the rock is capable of withstanding this expansion,
then it is argued that substantial stresses nonnal to the concrete-rock interface can
be developed, leading to a corresponding increase in the frictional resistance of
the interface, and an increase in pier capacity. This paper describes the results of
laboratory studies into the influence of expansive additives on pier perfonnance.

Introduction

When large structural loads are to be supported on rock, a common form of


foundation is a drilled pier into rock. Generally between O.5m and 2.0m (1.4 and
6.6 ft.) in diameter, these piers are constructed by drilling a vertical shaft through
the weaker overburden soils into the stronger rock at depth. The drilling action
results in the shaft wall having a roughness which depends on the rock type and
strength and the drilling technique used. Once the shaft has been excavated,
reinforcement is placed, and the shaft is back-filled with concrete. The portion of
the pier that is located in rock is called the rock socket. Drilled piers in rock
derive their capacity from both side and base resistance, with side resistance
dominating at working loads.
The side resistance developed in a rock socket depends upon a number of
fact9rs including the rock properties, the roughness of the interface between the
concrete pier and the rock, the angle of friction of the interface, and the pressures
that develop across the interface. It is argued that an increase in pier capacity can
be achieved if the pressures, or normal stresses, developed across the
concrete/rock interface can be increased. One way of achieving this increase is by
inducing expansion of the concrete against the stiffness of the surrounding rock.

Senior Lecturer and research students respectively, Department of Civil


Engineering, Monash University, Clayton, Victoria, 3168, Australia.

631 Habelfield, Baycan and Chamberlain


In the current study, the expansion of the concrete is achieved by use of an
expansive cement additive called CSA. Many different expansive additives were
considered before CSA was adopted. This particular additive invokes expansion
via a chemical reaction known as sulphate attack. In unconfined situations,
sulphate attack is highly undesirable as it leads to the decay of the concrete and
loss of durability and strength. However, under reasonable levels of confinement
such as those associated with piers in rock, the addition of CSA does not appear
to have any detrimental effects on the strength and durability of the concrete.
This paper presents the results of an investigation to determine the feasibility
of using CSA to improve pier performance in rock. In particular, the paper
describes a series of laboratory tests that have been conducted to determine the
increase in normal stress, or prestress, that can be generated by CSA and the
effect of this prestress on the frictional resistance of the pier/rock interface. Tests
conducted to determine the influence of CSA on concrete strength are also briefly
described. Although the arguments and results presented herein are directed
towards piers socketed into rock, they are just as applicable to rock anchor
applications (Chamberlain and Haberfield, 1993).

Previous Research

It appears from the literature, that the use of expansive additives for this
purpose is essentially unexplored. In the past, some mining groups have investi-
gated using expansive grouts in rock anchors, but their efforts were abandoned
due to the unfavourable side-effects that the then available expansive agents had
on the steel reinforcement and the grout. In these studies the expansive agents
forced the grout to expand by generating hydrogen bubbles. These bubbles unfor-
tunately led to an increase in the porosity and hydraulic permeability, a decrease
in the compressive strength of the grout and may damage the steel through
hydrogen embrittlement. New forms of expansive additives, that do not have the
detrimental effects caused by gas bubble generation, have since been developed.
As a result, it is now relatively common practice for construction and mining
companies to use expansive additives to offset the detrimental effects of excessive
grout shrinkage, but none appear to have extended the process beyond this use.
One group of researchers (Sheikh, O'Neill and Mehrazarin, 1985; Sheikh and
O'Neill, 1986) have investigated using expansive concretes to increase the
capacity of large bored piers in clay. They reported moderate success with a 25%
- 50% increase in capacity by using expansive cement. Shiekh et al. (1985)
suggested that this increase was due to the small lateral expansion of the pier
which caused the clay around the pier to consolidate and strengthen. Sheikh et al.
also reported a 30% to 40% reduction in the strength of the concrete.
Due to the significantly stronger and stiffer properties of rock, the authors
expect a much greater relative increase in resistance for expansive concrete piers
in rock. The stiffer and stronger properties of rock will provide a greater level of
confinement. As a result, the pier will not be able to expand anywhere near the
extent that it could in clay, resulting in the generation of a greater prestress and a
correspondingly greater improvement in pier capacity. The reduced expansion of

632 Haberfield, Baycan and Chamberlain


the pier will also result in very little, if any, loss of concrete strength, and in fact,
as discussed below, may result in substantial increases in strength.

Expansive Cement Additives

After concrete sets and is allowed to dry, it undergoes a sometimes-destructive


phenomenon known as drying shrinkage. As concrete is relatively weak in
tension, especially at a young age, if it is restrained from shrinking (e.g. by
reinforcement, friction, connections or other boundary conditions) it will crack,
often causing severe structural damage. It is this problem that led engineers and
concrete technologists to develop expansive cements to minimise drying
shrinkage. In 1892, Michaelis termed the phrase "cement bacillus" to label the
disruptive expansive effect resulting from a reaction between Portland cement and
sulphates; now known as sulphate attack (Klein et al., 1961). In an attempt to
manipulate this chemical process, the Lafarge Company experimented with a
mixture of Portland cement, aluminous cement and gypsum and found that the
expansive reaction was erratic and difficult to control. Subsequently, Lossier
added a stabiliser - blast-furnace slag - to slowly take up the excess gypsum so
that the expansive process supposedly. ended when the desired expansion was
attained (Neville, 1981). In its simplest form, the chemical reaction that describes
this process is given by :-

(1)

Although this is not the only chemical reaction that occurs with these
reactants, it is this reaction that produces the expansive forces. The product of the
reaction (right hand side of Equation 1) is known as ettringite and it occupies
227% more volume than the sum of its reactants.
The expansive agent used in this study is an additive called Denka CSA or
Calcium ~ulpho-Aluminate. It is a fine white powder, containing significantly
more aluminates and sulphates than ordinary cement. CSA is added to the cement
at the time of mixing and requires no alteration to normal batching procedures.
The main practical use of Denka CSA to date has been as a shrinkage
compensating additive and recommended dosages have generally been less than
30 kg of CSA per m3 of grout (Grace, 1986). The effectiveness of the additive
depends on the batching process, curing conditions and cement content. It is
expected that doses substantially greater than 30 kg/m3 will be required to
increase the capacity of piers in rock.

Expansive Concrete Properties

Given the importance of concrete strength and durability on the structural


integrity of piers, it was necessary to determine the effects that CSA has on the
properties of concrete. As a result, a range of laboratory tests incorporating
several different mix designs and different curing conditions was undertaken.

633 Haberfield, Baycan and Chamberlain


Tests were carried out to detennine properties under both unconfined and confined
conditions. A complete description of this study is beyond the scope of this paper,
and hence only a brief selective summary is included here.
Initially, standard laboratory tests were carried out on cylinders of unconfmed
concrete containing varying levels of CSA. The results of these tests showed that
the extent of expansion depended on the curing conditions and the CSA content.
The availability of free water was found to be an important factor. Water is vital
for cement hydration and the fonnation of ettringite. Expansion increased with
increasing CSA content; for concrete containing less than 50 kg CSA/m3 and
cured in a saturated environment, expansion is very rapid for the first 10 days
after pouring, at which time the available reactants appear to be exhausted and
expansion ceases. Maximum unrestrained expansion varied from effectively zero
in nonnal concrete (with no CSA) to almost 1% of length for concrete containing
50 kg CSNm3 For unconfined grouts containing more than 50 kg CSA/m3 ,
expansion appeared to continue indefinitely, and was so great that the concrete
eventually fell apart. The unconfined compressive strength of the concrete
decreases with the degree of expansion, which in turn is highly dependent on the
CSA content and the curing conditions.
However these (unconfined) results are not applicable to the field (confined)
situation where the surrounding rock may provide adequate support to the
expansive concrete. As a result, a series of laboratory tests were conducted to
detennine the influence of CSA on confined concrete properties. As for the un-
confined tests, these tests investigated the influence of CSA content and curing
conditions as well as the level of confinement. The importance of the level of
confinement on pier capacity is discussed later. In all tests, confinement was
provided by forming and testing the concrete cylinders in tubes of a range of
diameters and wall thicknesses and made from either steel, aluminium or PVc.
The range of tube cross-sectional dimensions and the different elastic modulus of
each material provided confining stiffnesses ranging from 1 to 1500 MPa/mrn. As
is explained below, these levels of confinement are commensurate with the level
of confinement experienced by the majority of anchors and smaller diameter piers
in rock. Further testing covering confining stiffnesses below 1 MPa/mm are still
required to model larger diameter piers in weak rock.
As with the unconfined tests, the results of the confined tests revealed that the
confined compressive strength of the concrete depends on the degree of expansion
which in tum depends on the level of confinement and the CSA content. For the
range of confining stiffnesses tested. the confined compressive strength of the
concrete containing CSA, in all cases (independent of CSA content), was greater
than the design characteristic strength of nonnal concrete (i.e. free of CSA)
(Chamberlain, 1993). For high confining stiffnesses. a substantial increase in the
confined compressive strength of the concrete was obtained (Urquhart, 1993).

Expansive Potential in Drilled Piers

As implied by the discussion above, the extent of expansion of an expansive


concrete pier depends upon the CSA content and the availability of free water.

634 Haberfield. Baycan and Chamberlain


The CSA content is controllable at the mixing stage, but the amount of free water
depends on the availability of groundwater. An expansive concrete pier in dry
rock above the water table will not achieve full expansion potential. As a result,
the increase in prestress will be reduced.
The properties of the rock surrounding the pier, and in particular the stiffness of
the rock, also have an important role to play. From elastic theory it can be shown
that the radial stiffness, K m , of rock surrounding a pier can be approximated by
Eq. 2, where Em and v m are the mass Young's modulus and Poisson's ratio of the
rock, r is the average pier radius, ~an is the prestress developed and & is the
radial displacement (or the expansion) of the socket due to the prestress.

(2)
K In =
r ~r

Eq. 2 depends on the rock remaining elastic and is therefore not valid for
high levels of prestress which may cause the rock around the pier to fail in
tension. However, although Eq. 2 may not be completely appropriate, it is useful
in illustrating the important role that confining stiffness has to play.
Eq. 2 predicts that an expansive concrete pier will expand more if the
confining stiffness is lower, given the same CSA content and curing conditions.
Soil may not be capable of supplying enough resistance to the expansion, thereby
endangering the strength and durability of the concrete. Weak rock, with a much
greater Young's modulus, may be able to provide sufficient confinement to
maintain the structural integrity of the concrete, simultaneously improving the
pier's frictional capacity. The pier diameter also plays an important role, whereby
a doubling of the pier diameter will halve the confining stiffness.
To illustrate the significance of confining stiffness, consider, for example, a
1.3 metre diameter drilled pier in a moderately weathered siltstone of uniaxial
compressive strength of 4 MPa. Assume that the siltstone has a mass Young's
modulus, Em= 500 MPa, and a Poisson ratio, v m = 0.3. These values are typical
for a soft rock, called Melbourne mudstone, on which most of the large structures
in Melbourne, Australia are founded. According to Eq. 2, K:=::600 kPa/mm. This
implies that a diametrical expansion of 2 mrn (approximately 0.2%) can induce a
600 kPa increase in the lateral stress on the pier-rock interface. Expansions of this
magnitude will not significantly affect the integrity of the concrete (Chamberlain,
1993). As will be shown (see Fig. 6), a prestress (or initial normal stress) of 600
kPa has the potential to produce significant improvements in pier performance.
It should be noted that a thin layer of softer material on the surface of the pier
socket could substantially reduce the level of prestress, and that in such cases a
higher CSA content may be required to obtain the same level of prestress.

Quantification of Prestress

A preliminary programme of laboratory tests was carried out to determine the


prestress generated for a range of CSA contents and confining stiffnesses. As

635 Haberfield, Baycan and Chamberlain


described above, cylindrical tubes of aluminium, steel and PVC, with a range of
different diameters and wall thicknesses, were used to provide the range of
confining stiffnesses required. These tubes were filled with expansive concrete
and then placed in a 100 % humidity environment for 28 days to approximate the
curing conditions that were likely to exist below the water table in a rock mass.
After curing, each tube was placed in a compression testing rig and the force
required to push the grout plug out of the tube was determined. All of these
"push-out" tests were carried out at a displacement rate of O.3mm per minute.
Typical load - displacement curves obtained from the push-out tests involving
the steel tubes are reproduced in Fig. 1. The results for this series of tests indicate
a linear increase in resistance until a peak load was reached at approximately 1
mm displacement. For this application, the residual resistance, not the peak
resistance, is of interest. The peak resistance is dependent on the initial
cementitious bond and the surface roughness of the concrete tube interface, whilst
the residual value is independent of these two parameters and is purely a function
of the residual friction angle between the steel and the concrete and the normal
stress developed over the concrete/tube interface.

6
/"'""0.

c5: 5
~
'--"
~
u 4
c 150 k CSNm 3
~

.
en
~
3 100 k CSNm3
1-0

.....
::l
0 2 50kgCSNm3
.J::
en
::l
p.,
1 NoCSA

o 2 4 6 8 10 12 14 16
Shear displacement (mm)
Figure 1. Shear stress - displacement curves for steel pUSh-out tests

The residual friction angle of the concrete/tu be interface was determined by


conducting a series of direct shear tests using conventional direct shear equipment.
By using the residual friction angle derived from these tests, in conjunction with
the residual push out resistances described above, an estimate of the prestress
developed was obtained. These estimates are included in Fig. 2 which shows on a
linear - log scale the variation in prestress developed with confining stiffness and
CSA content. Although there is a need for further testing, Fig. 2 shows that for a
constant confining stiffness, an increase in the CSA content from 0 to 150 kg/m 3
results in an order of magnitude increase in the prestress developed. Tests

636 Haberfield, Baycan and Chamberlain


covering other values of confining stiffness and greater CSA contents are
currently being conducted.
As a check on these test results, a number of tubes were instrumented with
strain gauges. The gauges were aligned to measure the circumferential strain on
the outside of the tube. The circumferential strains obtained were analysed using
thick walled pressure vessel theory to obtain estimates of the prestress. Good
agreement was achieved between the two sets of results.

Increase in Pier Capacity

From the discussion above, it has been established that the addition of CSA to
a concrete pier can result in a substantial increase in the normal stress acting on
the rock socket portion of the pier shaft. However, the extent that this prestress
transfers to an increase in pier capacity still needs to be established. For a
perfectly smooth socket, the fundamental model of friction predicts that the
increase in pier capacity will be directly proportional to the increase in nonnal
stress and the friction angle between the concrete and the rock. However, rock
sockets are never perfectly smooth, but contain a significant roughness that can
influence the capacity of the pier. As a result this simple relationship between
prestress and pier capacity may no longer hold. A large number of Constant
Normal Stiffness (or eNS) direct shear tests on rough concrete/rock joints were
therefore carried out to detennine the influence of prestress on pier capacity.

25

/
J'
c
o
.~ 15 ~
C':l 150 CSA /
/ ;' 1/. .
0.. , /
o>< ,
/ /'
....o
o
='
10 - 100CSA-J,./)/
1
"0
CIl
CIl
o / 1.' /'
/ ' ,
==
CIl
5 ..... / ' / 0 ' / 50CSA
o
'-
0....

o
, /,
,/
'''//
., / ._:.::......__ .-.
JJ:.""-::: .-......-:;..-.-
. /,/,-t!
. .......
!' -
.- ' .
No CSA
-=.-=-....:-.z~;..-. I I
0.1 1 10 100 1000
Confining stiffness (MPa/mm)

Figure 2. Prestress determined from push-out tests

637 Haberfield, Baycan and Chamberlain


Constant nonnal stiffness testing

The CNS direct shear test has been used extensively in the past by other
researchers (e.g. Johnston and Lam, 1989; Carter and Doi, 1988) to determine the
perfonnance of piers socketed into rock. It is generally accepted that the CNS test
is the most appropriate laboratory test for this application. This is because the
interface between a concrete pier and it's rock socket is a discontinuity similar in
most respects to a rock joint This interface has a roughness, and when shear
displacements are imposed by structural loading, the roughness causes an interface
dilation, just as for rough natural rock joints. For piers in rock, this dilation causes
the rock socket to expand radially (assuming that the concrete is stronger and
stiffer than the rock). This behaviour was shown in idealised form in Johnston and
Lam (1989) and is illustrated here in Fig. 3.

Pile and I
Vertica I
I socket dla. D .. displacement 1_'_Pi_IC_d_ia_._D_....j
,(:,,?fPi~,
I ':(::'/' ,', ':}""" i?f8j':
i". ~~~:':J!!illm!Pk?~b
Shaft section 'toc,
:'=,'
I~ki!.er~lj!i !(
soc

of co~crete
pIle
,>,Noimal,
" , force,
"j' ,,
[.. ~~b:-'!:\.' ~:L
, : : ,:: force
- ; ;..:.:.;..;;.:.:..: :\,"::::::=/>:::::;';
~---''----l :or)"
. . ,....-..'.. ::{{ ,: " " ":" " , :, v-~--;---7:><//i:
trr(j\. : : : ;. :.\ I

I
(a) Pile before displacement (b) Pile after displacement

Figure 3. Idealised displacement behaviour of pier socketed in rock

The macro behaviour of the pier/socket system can be analytically modelled


by Eq. 2. Since r is much larger than ~r, and since both Ern and urn can be
considered constant for the stress range considered, the normal stiffness ~ is
effectively constant. The development of shear resistance in a rough rock socket
is, therefore, governed by a condition of constant normal stiffness (CNS) rather
than the more common constant nannal load (CNL) condition. The CNS condition
profoundly effects the performance of the joint by imposing additional
confinement, thus suppressing dilation and increasing the normal stresses acting
on the joint.
This process is very similar to that caused by the addition of CSA to the
concrete. However, there is one important difference. The prestress developed
from the addition of CSA occurs before any shear displacement of the pier; i.e. it
can be considered to be an initial normal stress. As the pier displaces, the normal

638 Haberfield, Baycan and Chamberlain


stress will increase further as a result of socket dilation and the CNS condition.
The CNS shear apparatus used to carry out these tests is described in a companion
paper included in these proceedings, by Haberfield, Seidel and Johnston (1994).

CNS test results

A number of CNS direct shear tests on rough concrete/rock interfaces were


carried out to assess the influence that initial normal stress and nonnal stiffness
had on the shear strength of the joint. The rock used in the tests was a synthetic
soft rock called Johnstone (Johnston and Choi, 1986) with a uniaxial compressive
strength of approximately 4 MPa. The engineering behaviour of the Johnstone
closely models the behaviour of the natural Melbourne mudstone, and as such has
been used extensively as a modelling material for Melbourne mudstone. The
advantages of using Johnstone lie in its reproducibility and homogeneity, both of
which make it an ideal modelling material for laboratory testing.
The CNS tests were earned out for a range of stiffnesses and initial normal
stresses that are considered appropriate for expansive piers of between 0.5 and 2.0
m diameter, socketed into weathered Melbourne mudstone. The values of normal
stiffness adopted ranged from 150 to 2000 kPa/mm while initial normal stresses
ranged from 30 to 2000 kPa. The range of nOlmal stresses correspond to those
values of prestress likely to be obtained by varying the CSA content between a
and 300 kg CSNmJ of concrete, for the range of stiffnesses considered.
As discussed earlier, the roughness of the joint also has a marked influence on
the shear strength of the concrete/rock interface. The CNS tests were therefore
can'ied out for two different roughness profiles designated Class A and Class C.
Both profiles are illustrated in Fig. 4. The Class A profile corresponds to a Joint
Roughness Coefficient or JRC of 2-4 (ISRM, 1978), while Class C corresponds to
a JRC of 8-10. A summary of the tests camed out is given in Table 1.

Class A

~,- ......,.".-._--_ _ --~---.-..... .............-...-r_"",, Class


-- --- C-

J I I I I I I I I I

a 50 100 150 200 250 300 350 400 450 500


Horizontal Dimension (mm)
Figure 4. Roughness profiles used in eNS testing (Horizontal:
Vertical Scale 1:4) =
639 Haberfield, Baycan and Chamberlain
Table 1. Constant normal stiffness tests

Roughness Stiffness Initial Nonnal


Class (kPa/mm) Stress (kPa)

A 300 300, 900


900 30,300,900

2000 300, 900, 2000

C 150 150, 300


300 300, 600, 900
600 30, 150, 600, 1500

900 300

The CNS interface samples were prepared as described by Seidel (1993). This
procedure basically involves using a bandsaw to cut the specified roughness into
the rock half of the joint. A layer of plastic food-wrap is then placed over the
surface of the joint, and concrete poured onto this surface to create the matching
half of the joint. The concrete is allowed to cure for at least 7 days before
testing. Prior to testing the food-wrap is removed. The food-wrap is used to
ensure that there is no chemical bond between the concrete and rock halves of the
joint. This procedure was adopted for two main reasons. The first is that the
chemical bond which could be obtained in the laboratory is likely to be stronger
than that obtained in the field, especially in argillaceous rocks such as mudstone.
The surface of pier sockets formed in these rocks are often covered with a thin
smear layer which impedes the formation of this chemical bond. Secondly, the
main purpose of these tests was to determine the influence of the initial normal
stress on frictional resistance. To make analysis and comparison less complicated,
the effects of chemical bonding were therefore removed. The influence of
chemical bonding is currently being investigated.
The complete results for a typical test are reproduced in Fig. 5. The results
have been plotted for a test on a Class C profile, with an initial normal stress of
600 kPa and a normal stiffness of 600 kPa/mm. Each test result consists of five
graphs, these being: shear stress, normal stress and dilation versus shear
displacement, shear stress versus normal stress and normal stress versus dilation.
The graph of shear displacement versus nonnal stress shows that the normal stress
acting on the joint increases substantially above the initial value of 600 kPa. As
discussed earlier, the increase in nonnal stress is due to the socket roughness
causing dilation of the interface against the normal stiffness.
In Figs. 6 and 7, the shear stress versus shear displacement curves for several
tests are compared. In Fig. 6, the results of tests on Class C profiles with a
normal stiffness of 600 kPa/mm and involving four different values of initial

640 Haberfield, Baycan and Chamberlain


nonnal stress, i.e. 30, ISO, 600 and 1500 kPa, are compared. As discussed earlier,
the stiffness of 600 kPa/nun corresponds to a 1.2 m diameter pier socketed into a
moderately weathered mudstone. The test result for an initial normal stress of
30 kPa, corresponds to a pier constructed without CSA. The tests with higher
initial nonnal stresses (ISO, 600 and 1500 kPa) correspond to piers constructed
with increasingly greater CSA contents. Fig. 6 clearly indicates that increasing
the initial normal stress not only increases the capacity of the concrete/rock
interface, but increases the stiffness of the load-deflection response as welL The
relative improvement in performance is increased further if the shear resistance
corresponding to typical serviceability requirements (e.g. the shear resistance at
approximately 5 nun shear displacement) is considered. It is also interesting to
observe that for low nonnal stresses the response is very ductile. As the initial
nonnal stress is increased, the response becomes more brittle.
Fig. 7 shows a similar comparison between tests on Class A profiles, a stiff-
ness of 900 kPa/mm and initial nonnal stresses of 30, 300 and 900 kPa. In this
case, much more pronounced increases in perfonnance were obtained. The relative
improvement in perfonnance is thought to be largely a function of the interface
roughness. For Class C profiles, the increased roughness of the concrete/rock
interface leads to greater dilation of the interface during shear displacement, and
as a consequence, leads to significant increases in the normal stress above the
initial nonnal stress. On the other hand, for the relatively smooth Class A
profiles (Fig. 7), the dilation of the interface is reduced, and the nonnal stresses
generated during shear displacement are relatively much lower.
This dependence on roughness is also illustrated in Figs. 8 and 9, which show
the variation of peak shear stress with initial normal stress as determined from the
CNS tests. The results have been grouped in accordance with stiffness and profile
roughness. Results for Class C profiles are plotted in Fig. 8 and for Class A
profiles in Fig 9. Both plots indicate substantial increases in peak shear stress
with increases in normal stress. Also plotted in each figure is a line inclined at
24. This line represents the maximum peak shear stress that would be expected
for a smooth profile without any roughness; <I> = 24 is typical for the interface
friction angle of smooth concrete/Johnstone joints. The difference in shear
strength between the CNS test results and the 24 line is due to the roughness of
the interface. As would be expected, this difference is larger for the rougher
Class C profiles (Fig. 8) than the Class A profiles (Fig. 9). Figs. 8 and 9 also
indicate that peak shear resistance increases with increasing stiffness.

Conclusions

The proposal of using expansive additives in concrete to improve the capacity


of piers in rock has been discussed. From the laboratory tests conducted to date,
it is clear that the use of expansive cement additives, such as CSA, may result in
substantial improvements in performance. A field testing programme is currently
underway to determine whether the predicted increased capacity can be obtained
in the field.

641 Habelfield, Baycan and ChLlITlberlain


1200 1200

,..... ,.....
'"
p., '"
p.,
~ ,.Ioi
....... 800 ....... 800
~ ....,.. .'...-., - . .... , . .... .... ~

...'"'"en ...'"'"en
-
: ......- ..... ....__~~...iJ\-,.-.

-
~. ~

Q) J U it

... ...co
'"<lJ
.c
<lJ
..d
CI)
400 CI)
400

O -'-"'-o..J---. -l.. ........- 0L................................"J...........................-..l..............................._......J


o 10 20 30 40 a 400 800 1200
Shear displacement (mm) Normal stress (kPa)

.......
<a
p.,
1200

,. .-
J
~---- ----- .......
ell
1200

.
/
,,
~ J
/ ~
'-'
I
/ I
~ 800 ~{.
~
0
800
/

-E
ben
<a

0
400
..,-
,
-E
b
'"cd
0
400
,:

Z Z

OL.....................L.....................J...o................J..........................l...J...........~
10 20 30 40 1.0 ~o I~ 2~ J~ 4.0
Shear displacement (0101) Dilation (0101)

4.0
TEST NO.: MJF_CI C4 DATE: 20/10/92
MONOTONIC Shear rate .. 0.5 mmlmio
3.0
JOHNSTONB Sample P2
FRACTAL Profile CI coarse apprOL
....... 2.0
E Oladwl1Ip
E
'-'
c::
.g 1.0
.......1- Initial NorlJllll SUess (kPa) : 600
~ /J"
i:5 /'
0.0
". Normal Stiffness (kPalmm) : 600
1.0
TUI No.: MJF C1 C4
o 10 20 30 40
Shear displacement (0101)

Figure 5. CNS test results for Class C profile,


0'0 = 600 kPa, K = 600 kPalrnrn

642 Habelfield. Baycan and Chamberlain


1200

00 =1500 kPa
1000

00 =600 kPD.
....-.. 800
ctl
0..
...:.:
'-"
en
en
I1.l
b 600
CZl
~
ro
I1.l
...c
CZl
400 00 =30 kPa

200

o
o 10 20 30 40
Shear Displacement (mm)
Figure 6. Influence of initial normal stress on interface shear stress-shear
displacement response of Class C profiles, K = 600 kPalmm
800

600 00 = 900 kPa


....-..
ctl
0..
...:.:
'-"
en
en
I1.l
.:: 400
CZl
~
00 = 300 kPa
ctl
I1.l
...c
CZl

200
0=30kPa

0
0 5 10 15
Shear Displacement (mm)
Figure 7. Influence of initial normal stress on interface shear stress-shear
displacement response of Class A profiles, K = 900 kPa/mm

643 Habenield. Baycan and Chamberlain


1200

1000
..-...
co
~ 800
'-"
til
til
0 0
.... 600
~

CI)
~
co
0
..c
CI)
400
x
co
::E
A= 150
kPa/mm
200

a 500 1000 1500 2000


Initial Normal Stress (kPa)
Figure 8. Innuence of initial normal stress and stiffness on peak shear stress
Class C profiles

1200

1000 K =2000 kPa/mm


-.
co
~
'-" 800
til
til

....0
~

CI)
~
C':l
0
..c
CI)

X
C':l
400
::E
200

a 500 1000 1500 2000


Initial Normal Stress (kPa)
Figure 9. Innuence of initial normal stress and stiffness on peak shear stress
Class A profiles

644 Huberfield, Baycan and Chamberlain


Acknowledgments

The authors gratefully acknowledge funding of the project by the Australia


Research Council.

Appendix I. References

Carter, J.P and Goi, L.H. (1988). Application of a Jomt model to concrete -
sandstone interfaces. Proc. 6th Int. Con! on Num. Meths in Geom., Innsbruck,
pp. 889-893.
Chamberlain, TD. (1993). Investigation of Expansive Cements and their
Influence on the Capacity of Socketed Piles and Grouted Anchors in Rock,
MEngSc Dissertation, Dept. of Civil Engng, Monash University, Australia.
Chamberlain, TD. and Habemeld, C.M. (1993). The use of expansive grouts to
improve anchor performance in soft rock, Proc. Int. Conf. on Geot. Engng of
Hard Soils - Soft Rocks, Athens, Greece, September, pp. 1101-1106.
Grace, W.R. Australia Limited (1986). Denka CSA - Shrinkage Compensation
Additive for Grout. Company Leaflet, pp 0-28.
Haberfield, C.M., Seidel, J.P. and Johnston, I.W. (1994). Laboratory modelling of
drilled piers in rock. Proc. Int. Conf. on Design and Construction of Deep
Foundations, Orlando, December.
ISRM (1978). Suggested methods for the quantitative description of
discontinuities in rock masses. Int. J. Rock Mech. Min. Sci., Vol. 15, No.6,
pp. 319-368.
Johnston, I.W. and Choi, S.K (1986). A synthetic soft rock for laboratory model
studies, Geotechnique, Vol. 36, pp 251-263.
Johnston, I.W and Lam, TS.K (1989). Shear behaviour of regular triangular
concrete - rock joints - analysis, 1. of Geor. Engng, ASCEngineers, Vol. 115,
No.5, pp. 711 - 727.
Klein, A., Karby, T, Polivka, M. (1961). Properties of an Expansive Cement for
Chemical Prestressing. A.Cl. Journal, July, pp. 59-80.
Neville, A.M. (1981). Properties of Concrete. 3rd Edition, Pitman, London.
Seidel, J.P. (1993). The Analysis' and Design of Pile Shafts in Weak Rock. PhD
dissertation, Dept. of Civil Engng, Monash University, Australia.
Sheikh, S.A., O'Neill, M.W., Mehrazarin, M.A. (1985). Expansive Concrete
Drilled Shafts. Can. 1. Civ. Eng. 12, pp. 382-395.
Sheikh, S.A., O'Neill, M.W. (1986). Long-Term Behaviour of Expansive
Concrete Drilled Shafts. Can. 1. Civ. Eng. 13, pp. 213-217.
Urquhart, G. (1993). The Influence of CSA on Confined Strength of Grout.
Internal Report, Dept. of Civil Engng, Monash University, Australia.

645 Haberfield, Baycan and Chamberlain


DESIGN GUIDELINES FOR SCREW ANCHORS

CHESTER A. CARVILLE, P.E. 1


ROBERT W. WALTON 2

ABSTRACT

The screw anchor is a deep foundation member


consisting of a steel shaft with helical plates welded to
the shaft. Screw anchors are installed into the soil using
mechanical rotational force, by adding extensions as the
assembly advances. Once installed, the anchor has bearing
capaci ty in both tension and compression. For over 70
years, screw anchors have been used by the utility industry
for power pole guying, transmission tower foundation
underpinning and pipeline supports. Today, the general
construction industry is discovering and using screw
anchors for a much wider variety of applications. This
paper provides information about screw anchor installation
cri teria as they relate to screw anchor design. It is
intended that this discussion will provide the engineers
with important insights and case histories so that they can
work more effectively with contractors in the evaluation,
design and implementation of screw anchor projects.

INTRODUCTION

The screw anchor consists of a helical steel plate or


series of helical steel plates fixed to a steel shaft
(Figure 1). The shaft is directed toward the soil and
mechanically rotated, advancing the screw anchor into the
soil. The soil around the shaft remains relatively

lForensic Consultants, 40631 Deluz Murrieta Road,


Fallbrook, CA 92028
2Engineering Contractor, Walton Property Services, 4154
Datcho Drive, San Diego, CA 92117

Carville, Walton
646
undisturbed. Once
installed, the anchor
has bearing capacity
in both tension and
compression in the
r 2-3/4" ,

3/8" thick
subsurface by trans- steel plate
ferring the structure's
load to the bearing 11/2"
stratum. The anchor I I
installation angle can ~
range from horizontal 7-1/2
u
......
I

(0) to vertical (90).


An unique application
for screw anchors is
their use as tiebacks
(Figure 2). 4" Helix
Plan View Radius
When installing screw
anchors the lead
section anchor can be
driven to greater 1-1/2"
depths by adding shaft I I .J,
5/S"
extensions (Figure 3). ~"""""'r+ f
The bearing capacity of
the anchor increases
with increasing
magnitude of applied
Various lengths-
installation torque. Typical are
The theoretical 3'e", 5'0", & TO"
ultimate capacity which
can be achieved is
limited by the
torsional strength of Profile
c:::::~~
the anchor shaft and
the single helix load
limit. The final 3" r--.,.__
i
318" Thick Helix Plate is welded
to the steel shaft

extension shaft is
connected to the
supported structure by
an appropriate termi-
nation device; for
underpinning, we
recommend a concrete
pier cap reinforced 1 4 - - - - 7-1/2" - - - - ' I
with rebar and doweled
into the existing
foundation of the
structure (Figure 4).
Figure 1 - Screw Anchor Lead Section
1. 5" Square Shaft Anchor with Single
8" Diameter Helix (Scale 3" = 1')

647 Carville, Walton


Figure 2 - Screw Anchor Threadbar Tieback Assembly
Example (Scale 1" = 1')

Oywidag ," grade 150 thrNdbar


coal with corrosion inhPlor grease
1" threadMu' 11/2' square shaft
adapter for 1-112' square screw anchor
anchor aasembly lead section

5"x5' x 1-1/4" bearing plate


and 21/2" sched 4Ox1S" steel pipe 1-1/4' PVC sheathing class 200
welded 10 plate covers unbonded
pack pipe with conosion length of threadbar
Inhibitor grease and seal end
Imbed plate and pipe in conaete ,1/2' square shaft
8lTUCture to be restrained screw anchor
extension

Screw anchors are constructed of very high-strength


materials and when sufficient installation torque is
applied, particularly with a small helix or helices, they
will advance well into firm formational soils. The screw
Figure 3 - Screw Anchor Extension Shaft -- Typical Design
Example of 1.5 inch Square Shaft (Scale 3" = 1")

.
518'
,....+-----131...J11~6~-+- f

Pro tile
Various Lengtha .. Plan View
Typical are
3'6',5'0", TO'

-i 1112" t-

The Upset coupeling receives the top end


01 the anchor lead section shalt or the
top end of another extension.

Use a single 314' X ~ machine bolt and


Iow-prolile 318"thlck hex nut. Bolts are to be
SAE J429Grade 5 or approved equivalent.
Boltsshall be Installed to approximately 40
I I

I . 7/16' .
1-618'
I
ft.Iba. of torque.

anchor is designed to obtain its capacity in soil or soft


rock and not in solid rock. A knowledge of the depth and
characteristics of formational bedrock is an important

Carville, Walton
648
aspect of screw anchor design. Where subsurface conditions
include the prospect of an encounter with dense formational
soil or solid rock, screw anchor testing can help to avoid
soil or helix failure during production. For example, if
a test anchor penetrates formational soils and grinds on
solid rock, the helix surface area may be increased and/or
the minimum required
installation torque may be decreased. If the adjustment is
made properly, the screw anchor will reach its designed
installation torque and required ultimate capacity before
it begins grinding on solid rock.

Figure 4 - Concrete Pier Cap for Screw Anchor


Underpinning (Typical 12" Cap, Scale 1" = 1.0')
. . 3 . .
--------------- ---------_.
~--------r-------~
110

2-#4 bars dowel into


existing fooling

1
...
I
I
;

7'
I
I
I

I -.
I
I
I
I
I

"'" Co

~
Outside Edge of extg footing C'l
h \ Screw Anchor Shaft
~ J
Outline 0f Pier Cap Excavation--< 1./ Sleave Termination Plate
f-' Screw Anc hor Helix
Reinforc ing Grid l0- t
extends beneath M
existing footin g

1..o 1l ----+-- 2' S" ------,~~I Plan View

r
Existing G e

Existing Conerete Footing


~ S" I>
I

I
max
... -+ 3" 4- I
_. Existing Rebar
M
2-#4 Bars 1a" center to c enter f
dowel into Extisting
Footing . Use Hilti HIT e-1 00
Doweling System.
T 2000 PSI Concrete
;..
Rebar Grid- 3-#4 bars sach way I .
Sleave Termination Piate
M ~
~f
1/4" Thick Steel Plate, bolted I
to Screw Anchor Shaft

-M
I
-+ 3" l-
f
Screw Anchor Assembly Side View
per Plans and Specifications /~v
2' 6"
I
.-tfIfI!!!r
Carville, Walton
649
SCREW ANCHOR DESIGN

In screw anchor engineering, it is just as important


to avoid over-design as under-design. To obtain a good
installation it is necessary to penetrate the undesirable
soil and establish capacity in a suitable bearing strata.
Instead of seeking maximum anchor depth, one should seek
capacity in the suitable bearing strata, by using optimum
helix configuration at optimum depth, for maximum economic
benefit. The best solution is to conduct screw anchor
exploration and testing. After reviewing available
subsurface information, select a reasonable helix
configuration for the testing. Drive the anchor to the low
range of the anticipated installation torque. Perform a
pullout test to failure and record the capacity achieved.
The load at which the anchor begins to fail is the ultimate
capacity of the anchor at the given installation torque.
A compression test is seldom necessary since the
compression capacity is normally greater.

Using the test results, an empirical torque factor


(Kt) is found according to the following formula:

Kt = Qt/T
Where T = Average Installation Torque (ft-kips)
recorded from the installation machine, Qt = Ultimate
Anchor Capacity (kips) achieved during load testing and Kt
= Empirical Torque Factor (11ft).
Repeating the installation at the same as well as
slightly higher andlor lower installation torque values
will allow the engineer to select a value for Kt to be used
in the design of the production anchors. For a group of
tests conducted within the same depth range and soil type,
the lowest of the calculated values for Kt should be used.

It should be noted that testing of production anchors


should not exceed the mechanical helix load limit
recommended by the manufacturer. The single helix load
limit for most square shaft anchors is between 30 and 40
kip. Therefore at higher capacities, it is not advisable
to test the anchor to failure.

In the case where a minimum Qt has been predetermined,


testing of a proposed production anchor need only verify
that the minimum ultimate capacity has been achieved at a
gi ven installation torque. Testing of the anchor to
failure would not be necessary.

Carville, Walton
650
As an example of underpinning a building with screw
anchors, we know from a subsurface investigation that
suitable strata occurs at a depth of 12 feet and below. We
also know that each anchor must have a working capacity of
15 kip with a factor of safety of 2.0. We select a two
helix anchor assembly with one 8 inch helix and one 10 inch
helix, and drive it to a depth of 15 feet where an
installation torque of 2.0 ft-kip is achieved. We then
perform a load test and find that the anchor fails at 24
kip. We then drive the same anchor to a depth of 20 feet
where an installation torque of 4.0 is achieved. The
anchor does not fail under a load of 35 kip. The testing
is repeated at two other locations with similar results.
The ultimate capacity for production anchors is to be
30 kip (15 kip working capacity x 2.0 factor of safety).
Based upon the above formula, Kt = 24 kip/2. 0 ft-kip =
121ft. Therefore an anchor installed to 4.0 ft-kip has a
theoretical ultimate capacity of 12 x 4.0 = 48 kip> 30
kip. Also installing a production anchor to 4.0 ft-kip we
have verified a minimum ultimate capacity of 35 kip > 30
kip. Therefore, the proposed production anchor is
acceptable providing it is driven to a minimum depth of 15
feet and a minimum installation torque of 4.0 ft-kip.

Several studies have been documented to verify the use


of the empirical torque factor. The method of predicting
screw anchor capacity which yields the most consistent
result is the installation torque correlation method (Hoyt
and Clemence, 1989). Their referenced value for square
shaft anchors is Kt=10. One study (Mitsch and Clemence r
1985) obtained a range from Kt=12 to Kt=26. Dixie
Electrical Manufacturing (Alumaform, Inc., 1988) publishes
guidelines which suggest a range from Kt=12 to Kt=17 for
their multi-helix anchors. A. B. Chance Company (1989)
recommends a value of 10. A paper by Rupiper and Edwards,
1989 states that the common Kt used by practitioners today
is 10. These authors commonly commence a project with a Kt
of 10 and adjust based on test and production field
conditions.

It is important to understand that the value for Kt is


a combination of soil and anchor properties, primarily
relating to friction during installation. Therefore Kt for
a dense dry sand would normally be less than for a hard wet
clay. The factors become more complex when an anchor is
driven through a wet clay material into a dense sandy
material. The wet clay provides lubrication to the helix
surfaces, permitting them to advance further into the dense
sand. Also, the shape and size of the anchor shaft and the
method of coupling can be a significant factor. The factor

Carville, Walton
651
for pipe anchors, such as the A. B. Chance 3.5 inch HS is
recommended to be around 7 for most soils. This factor is
lower because the pipe anchors create significantly more
drag as they are installed due to their larger diameter and
three bolt connection. We have not found any clear
consensus as to the impact of variations in size and number
of helices on the value for Kt. In general, it appears
that the value remains the same since an increasing helical
surface area results in a corresponding increase in anchor
capacity. Crouch, Stephenson and Clemence (1993) provide
a detailed discussion of the factors which must be
evaluated in predicting screw anchor installation torque.

Once the design has been approved and the project is


underway, the engineer should maintain flexibility and be
prepared to respond to possible unexpected subsurface
condi tions encountered by the contractor. This would
include being able to change screw anchor locations, modify
load distribution plans, adjust estimates of soil design
parameters and modify requirements for screw anchor
configuration, depth and minimum installation torque.

INSTALLATION EQUIPMENT

The Tortuga is a high torque, tight access driver


which is used to install screw anchors for foundation
underpinning. The unit may be used for repair or new
construction whenever deep foundations or anchoring is
required. High installation torque is often required to
achieve the desired capacities. The smaller, tight access
rigs which have been used in the past are often unable to
achieve the required torque. The Tortuga has been designed
to overcome this limitation. The powerhead is an Eskridge
Series 72 BA with a rated torque of 12,000 ft. lb. The
mechanical and hydraulic system has been designed and
tested successfully on a number of repair jobs under a
variety of access, soil and loading conditions. The
Tortuga can set up and install hardware inside as well as
outside of existing buildings. It can be wheeled through
narrow doorways and operate with only
8' of overhead clearance. Alternately, the Tortuga can be
mounted on a lift or tractor for work on level surfaces or
slope faces or where access will permit automated
positioning of the machine for quicker set up time. The
Tortuga is operated by a 40 gpm hydraulic pump powered by
a Deutz diesel engine.

652 Carville, Walton


SCREW ANCHOR INSTALLATION CASE HISTORIES

Case History No.1

Fifty screw anchor piers were used to underpin a


failing 2 story dining hall at a religious retreat in the
Santa Monica Mountains near Malibu, California. Subsurface
borings indicated that building failure was due to a
combination of compressible fills and under design of the
existing load bearing footings. Soil borings determined
that weak clayey fills and slopewash extended to depths
between 10 to 15 feet. Underlying these soils was a
formational sandstone unit which would provide good support
for the building. Screw anchor underpinnings were
specified to be installed into the formational sandstone to
support the interior load bearing masonry walls.

Test anchors were installed at two locations and the


test results correlated with the installation torque
applied. The test anchor lead sections were 1.5 inch shaft
and included both 8"-10" and 10"-12" duel helix
combinations. Both pull and compression tests were
performed.

Results found ultimate capacities in excess of 60 to


70 kip with an installation torque of 5,000 to 6,000 foot-
pounds (5 to 6 ft-kip). An empirical torque factor (ETF)
of 12 was specified and all anchors were driven a minimum
final installation torque of 5.0 ft-kip giving minimum
ultimate capacity of 60 kip per anchor. Using a factor of
safely of 1.5, this gives a working capacity of 40 kips
which was well in excess of the load requirement per
anchor.

Case History No.2

Twenty-five screw anchor piers were used to underpin


and relevel a failing two story wood frame residence in
Huntington Beach, California. The floor level survey found
severe differential settlement with elevation differentials
in excess of six inches across the building. The
subsurface was known to consist of a thick
alluvial basin. The near surface soils were known to be
weak clays with occasional peat lenses. The soils
investigation consisted of hand pit excavations and the
installation of two screw anchor probes, one extending to
a depth of 65 feet.

Test anchors were installed at selected pier locations


to eventually become permanent building supports. A
continuous record of installation torque was maintained and

653 Carville, Walton


pull testing was performed as preselected installation
torque levels were achieved.

At a depth of about 20 feet, installation torque was


averaging about 1 ft-kip and produced a tested ultimate
capacity of 13 kip, providing an ETF of 13. However, at
the 25 to 30 foot zone, installation torque increased to
between 2 to 3 ft-kip. This increase produced only a
slight increase in anchor capacity. ETF's were calculated
to be a minimum of 5.33. At greater depth, higher ETF's of
10 were again achieved.

Based on this data, it was estimated that the helices


had encountered weak soil (possibly peat) lenses at depths
between 25 to 30 feet, and it was decided to specify four
helices per anchor and to use a conservative ETF of 5.33 in
designing the other 23 anchor piers. This would have the
effect of reducing the average total depth (about 40
foot/anchor), providing helix embedment in the stronger
soils and still accounting for the weaker conditions in the
peat zone. An installation torque of 4.5 ft-kip was
specified providing a minimum ultimate capacity of 24 kip.
Using a factor of safety of 1.5 this provided a working
capacity of 16 kip per anchor. Since the average anchor
spacing was 8.0 foot, the allowable load was 2 kip per foot
meeting the requirement based upon structural analysis of
the building.

Case History No.3

Forty-nine screw anchor hold downs were used to


support a temporary 100 foot high tower erected in downtown
San Diego, California during the Americas Cup competition.
The tower would be covered with a. canvas mural and the
primary concern was wind loading. Because the tower was
constructed of modular tube elements, each of the forty-
nine ground level base pads of the tower had to be
restrained.
Individual anchor loads were relatively low, ranging
between 0.5 kip and 6 kip. Production and testing was
conducted simultaneously with production anchor
installation. The near surface soils were a hydraulic fill
material well known to the contractor. Therefore an
average ETF of 10 was assumed and anchor installation was
conducted accordingly. A hollow core hydraulic ram and
calibrated gauge was used to pull test representative
installations. Anchors were tested to 1.25 times the
required working load without failure.

654 Carville, Walton


CONCLUSION

Screw anchors can provide a simple and effective way


to deal with a wide variety of foundation requirements.
What is most valuable to the engineer is the inherent
quality control system. Once the minimum required depth is
established, each anchor installation must advance until
the established minimum torque is achieved. In this way,
unforeseen and unsuitable subsurface conditions are
automatically detected and immediately apparent to the
engineer. In most cases, these unforeseen conditions are
self-correcting as the screw anchor assembly continues to
advance into the suitable strata.
REFERENCES
A. B. CHANCE Design Examples of Helical Anchors.
Centralia, MO, 1989.

ALUMAFORM, INC. Dixie Anchors. Birmingham, Alabama,


1988.
CROUCH L. K., STEPHENSON R. W. and CLEMENCE S. P.
Installation Torque and Uplift Capacity of Helical
Soil Anchors in Sand. ASCE, 1993.
HOYT R. and CLEMENCE S. Uplift Capacity of Helical
Anchors in Soil. International Conference on Soil
Mechanics and Foundation Engineering. Rio de
Janeiro, Brazil, 1989.
MITSCH M. and CLEMENCE S. The Uplift Capacity of Helix
Anchors in Sand. Uplift Behavior of Foundations in
Soil. ASCE Detroit, Michigan, 1985.
RUPIPER S. and EDWARDS W. Helical Bearing Plate
Foundations for Underpinning. Foundation
Engineering Proceedings Congress. SCE, Evanston,
Illinois, 1989.

APPENDIX A

CONVERSION TO SI UNITS

1 KIP = 4,448 NEWTONS

1 FOOT = 0.305 METER

1 POUND = 0.454 KILOGRAMS

655 Carville, Walton


Hardware Solutions for Quality Control of Deep Foundations - Overview

Garland Likins,' Frank Rausche,2 Members ASCE, David Peterman 3

Abstract

Deep foundation construction equipment and installation methods


have undergone dramatic improvement in recent years. Larger equipment
and higher design loads are often specified to reduce the number of piles
and project cost. Therefore, performance of each foundation element is
more critical, requiring additional quality assurance for every element of a
project. Fortunately, modern electronics allow routine implementation of
electronic tests by civil engineers to monitor the installation and/or assess
the quality of cast-in-situ foundation installations. This paper presents an
overview of currently available electronic monitoring techniques for quality
control of deep foundations. Benefits include reduced liability, better
accuracy, more information with less labor, and often a reduced cost.

Introduction

Obviously, quality (or lack thereof) is involved in the success (or


failure) of any project. Projects built on deep foundations require that this
support system be properly installed; failure of any component could result
in failure of the entire project regardless of how carefully the above ground
structure is built. Since visual inspection of driven or cast-in-situ piles is
practically impossible after installation, good quality control during
installation is of paramount importance. Most construction codes thus
specify proper recording of installation observations. Many companies
require Total Quality Management for risk management to reduce liability.

1President, Pile Dynamics, Inc., 4535 Emery Industrial Parkway, Cleveland,


OH 44128.
2President, Goble Rausche Likins and Associates, Inc., 4535 Emery Industrial
Parkway, Cleveland, OH 44128.
3Partner, Pile Dynamics, Inc., 4535 Emery Industrial Parkway, Cleveland, OH
44128.

656 Likins, Rausche, Peterman


In the past, manual visual observations of blow count or drilling
progress, followed by static testing of a small sample of piles, were often
the only available construction control. However, manually recorded
observations were only as reliable as the observer, Errors were common
For example, counting blows during pile driving is monotonous, and lack
of concentration or interference with the inspector can result in inadvertent
errors in counting (e.g . ... 77, 78, 79, 70, 71 .. ). The field records were
often transcribed for legibility, potentially compounding errors, particularly
when the original field records were difficult to read. Obviously, such a
process is labor intensive and therefore expensive. A manual recording
system is also more subject to abuse; records can be easily altered. This
is perhaps why both contractor and owner's representative each produced
their own manual records, adding further to the effort and cost of the entire
project.

The accuracy of both blow count and/or pile penetration frequently


was very poor when reference marks were inaccurately drawn on the pile.
The blow count for pile driving was often recorded for relatively large
increments (blows per 250 mm, or blows per foot), and the pile driven
farther than necessary to assure consistent blow count. If the equivalent
blow count over a smaller interval (or several successive smaller intervals
to assess consistency) could be reliably taken, then the accuracy and
economy of the project could both be significantly improved.

Static loading tests are performed on a small number of piles


(typically one percent or less) to twice the design load (proof load) to
prove the foundation design. Because of the high cost of failure, test piles
are often overdriven and proof tests usually pass easily, with actual safety
factors then being higher than necessary. Production piles then use the
same very conservative criteria, resulting in higher than necessary costs.
In numerous cases the static tests are avoided due to high cost, unwanted
construction delays, or because they are practically impossible for piles
in deep water. Extra care is generally given in driving a test pile.
Unfortunately, production piles are often installed with less care, and thus
may not achieve the same quality.

Electronic monitoring systems are now available and easily applied


to the installation of every pile. As a consequence, more care would
typically be applied during production. In addition, the automatic
installation documentation with electronics can faithfully record the entire
installation process, eliminating virtually all human errors. Finally dynamic
pile testing, already routinely applied worldwide, could be conducted on
a random sample or fixed percentage of all piles as quality inspection
perhaps causing more care to be exercised during installation.

657 Likins, Rausche, Peterman


In the design of any electronic monitoring or test system, the first
consideration must be to isolate the problem to be solved, or what test is
to be performed. Another question is what level of skill is required to
operate the device; is an engineer required, or a technician, or a member
of the pile crew? Is the device to be used for monitoring every pile, or
just for testing selected piles? Will the test require selection, or extra
preparation, of the test pile prior to installation? Should the installation of
all piles be monitored? Can the test method be applied to selected or
even every pile at any time during or after installation is completed?

A variety of sensors are available, each with a particular function,


range of application and accuracy. All equipment and sensors must
operate in extreme environmental conditions and give reliable information.
If the sensors fail to give correct information, the device should alert the
operator. Alternately, the operator must review the results after each test
(rather than once per day or once per week); most errors will be quickly
spotted and corrected before proceeding to the next pile, The
components must be field worthy and include a reliable power supply.
Use of line power is generally discouraged due to lack of availability. Thus,
the device should be equipped to use rechargeable batteries or the battery
of the crane or test vehicle where higher power is demanded, The basic
results would be stored in non-volatile memory, and then printed or output.

Monitoring Considerations

The inclination of a pile (often called "batter" or "rake") traditionally


has been measured with a carpenter's level. This requires a careful eye
and cooperation between the piling foreman and crane operator to
position the leads and pile to the proper inclination. Angle measurements
and adjustment are alternated in two perpendicular directions before the
hammer is started. After several blows, the hammer is stopped and
alignment verified, a process often repeated several times with a great
productivity loss. Actually, the inclination angle can be measured in the
two perpendicular directions using electronic tilt sensors attached to the
piling leads. With the readout device located in the crane cabin, the
operator can adjust the alignment even during installation thus greatly
improving productivity.

All driven or auger cast pile projects require recording of blow count
or grout take, respectively, as a function of depth. Such a system must be
continuously active during the entire length of every pile installed, and
would replace the manual observations taken on all piles on all projects.

658 Likins, Rausche, Peterman


The sound of impact pile driving has been used in the Saximeter
(Likins 1988) to detect and count hammer blows, relieving the observer
from the monotonous counting task. The time between blows can be
converted to an equivalent blows per minute to evaluate all hammers, or
compute the stroke of single acting diesel hammers. If an observer
presses a single key for each penetration increment, a blow count log can
be automatically generated and saved in memory (to reduce errors) with
user comments or observations, for transfer at the end of the day to a
computer. The blow count logs can then be professionally printed, without
copy errors.

If a position detection device is combined with the Saximeter, the


penetration increments can be automatically detected and the blow count
log generated without human observer. Encoding wheels, linear position
sensors, ultrasonic or laser technology, or proximity sensors can measure
displacements with good accuracy over a large range consistent with
typical pile lengths. This has been done in the Pile Installation Recorder
i.ElBl. The PIR also accommodates information from additional sensors on
a "network" system which identifies both the type and location of the
sensor and its reading. For example, the distance measuring technology
can also be used to record ram stroke. The angle of inclination data is
also easily measured. Adding an accelerometer, quickly attached to the
pile at the end of driving, can yield the final displacement and temporary
compression (TC) of the pile. A typical system is shown schematically in
Figure 1. Given the flexibility of the equipment, the PIR is easily adapted
to drilled or auger cast (CFA) piles by placing torque, pressure and
concrete/grout volume sensors on the network.

IN FIELD IN OFFICE

Figure 1. Pile Installation Recorder Typical Setup

659 Likins, Rausche, Peterman


Modern electronic devices have an internal clock which can record
a "time stamp" for the beginning and end of installation so that the
installation time and thus efficiency of installation can be determined for
each pile. This feature is particularly helpful for the Pile Installation
Recorder and Saximeter. The complete output should be clear, concise,
and complete. Ideally, it would be highly automated with minimal user
input (e.g. "what is the pile name") and a very user friendly interface. All
measurements (time, depth, count, pressure, torque, stroke, pile
inclination, ... ) would be automatically converted to digital form, processed,
and stored by a microprocessor. A responsible piling crew member can
easily learn to operate this system. Output would be either directly printed
or plotted at the end of each pile or saved in memory and downloaded to
a personal computer at the end of each day. A typical result is shown in
graphical form in Figure 2; of course, results could be printed in tabular
form.

Testing Considerations

In contrast to monitoring during installation, testing is here defined


as limited to the already installed deep foundation element. Usually
testing every pile is neither necessary, nor appropriate, nor economical.
Testing a small sample is often sufficient. For example, an existing
foundation is to be reused or the design load raised. If original records
are lost and soil information is available, it would be necessary to assess
the pile lengths to estimate pile bearing capacity. The integrity of the
existing piles might also be questioned. The pile length could be found
with a so-called Parallel Seismic Test (sensitive probe lowered into a bore
hole installed near the subject pile); however, additional tests require
costly additional bore holes. Other systems to determine pile length
include the Pile Integrity Tester (P.I.T.) (Rausche et.al. 1992) conceivably
used on every pile at reasonable cost, and the Pile Driving Analyzer@
(PDA) which is usually applied to selected piles only (Goble et. al. 1980).
The PDA allows for the assessment of the pile bearing capacity, while both
P.I.T. and PDA detect pile length as part of their pile integrity assessment.
Both these tests are sometimes limited by pile length, or by the
requirements that the pile be concrete for P.I.T., or be impacted by a large
weight (i.e. pile driving hammer) for the PDA. Thus, there is no single,
universally correct, solution as different field conditions dictate a particular
approach.

High Strain Testing Considerations

Static loading tests apply load to the test shaft while measuring the
displacement of the shaft. Static testing is generally performed according
to ASTM 01143; static loading specifications usually require that an

660 Likins, Rausche, Peterman


PILING REPORT AHD CHECKLIST WAGSTAff PILING PTY LTD
Project BUILDIHG SITE 1 Job Ito. ')13Q
Pile I TEST PILE Z Set 1~."""1'.'BL ~ lINIlT tiL 1
!!.e 215tt1 - PRECAST ~ 15.8M SIIoe PLATE
Itote 1 DYlWtIC JOINT Stroke Ii....... ~ J. P. WAGSTAFF
Itote Z H. CUSH - NJUftSTElIE Loa. 188.ITonnel
HII....e,. 8ANlT 6TOH Rake UERTICAl .(;Z1
Cqshion PLYWOOD - 3X11i"" t.O.EleY .3.5..
weather ClOU&Y ~ 21"C 5.6..
Dlte "/1Z/15 85:38 - 16:58 (1:21 Totll,
- - BIOws/598....
1:" Drive)
Stroke (....)
. . '~"
~

e 5e lee 1ee B88 .a~G)


r-~
& & ...I.G'I
ELEU

5 5
...,
N

1e 18
N Il>
'~
NLI'I'OI'
,
.. -. .. -. .. .. .. .. .. .. .. .. -, ., ......... , "LI'IG'I
0
..I.CJ

...I:
~
15 15
c
-...
0

e 28 Z&
.,
.
'01'

...I.
U .. ,
N Il>
C
........ ,- . NN'OI'
U -0
".all'
...I.Col
CI.
25 Z5
...0 HIH .
PEM.
..
J:.

Q
~
II 3a 39
.,
..
N

35 35
. , '~
NLI'I'OI'
Il>

.... -. " ....... _, 0 .. . ., ...- ,. "LI'IG'I


...I.Col
37.&..
1& -+- -..,.... -+ -..1...

8 B&8 1698
-,----- CqPllllat I ve Ito. of Blows
Total Blows 119B IH.C.R.I
H-25.8.... E-15.8....
lliIll CO!l1E!fiS .... :1" 8.3:18'
a.a.. PITCH 12.1H PilE 188 r - - - - - - - ,
9.a.. MOHITOR WITH PDA
12.&.. Stop 85:18 to 85:58
12.8.. PITCH HEXT 12.2H PilE
12.8.. RELOCATE PDA GAUGES
23.5.. Stop 86:&5 to 81i:18
23.5.. PITCH HEXT 12.2" PilE
Z3.5.. RELOCATE PDA GAUGES
36.&.. Stop 86:25 to ali:38 -188 e 1e8
36.a.. PITCH NEXT 6.1" PilE TOl1.9:189-
36.8.. RELOCATE PDA GAUGES RES 8.5:1&&

Figure 2. PIR Result

661 Likins, Rausche, Peterman


engineer be involved in the test and interpretation. While manual
recording of measurements has been normal, this too can be automated,
and with additional effort in the future the test could be preprogrammed
through a personal computer (or microprocessor) to incrementally load
and record the results. Well designed tests already require electronic load
cells; digital displacement sensors are now available. Some tests have
additional strain (using weldable gages or sister bars) or displacement
(usually using telltales) sensors at various locations along the shaft, which
in itself increases the need for automated data collection.

ZERD

EXPANDING OSTERBERG
CELL

Q Q

CONVENTIONAL OSTERBERG

Figure 3. Schematic Comparison Between Osterberg Cell and Conventional


Load Tests (after Schmertman 1993)

In some cases, the cost of a static test can be significantly reduced


by using an Osterberg loadcell (Osterberg 1989). The savings are
achieved primarily by using the soil system as the reaction load thus
eliminating the conventional static reaction system as in Figure 3. The
Osterberg loadcell is basically a hydraulic jack installed usually at the toe
of a drilled shaft. Applying pressure to the loadcell tests the base in
compression (as in a conventional static test). while simultaneously the top
of the cell moves upward, testing upward shaft resistance (also in
compression loading). The loadcell is pressurized and expanded until
either a proof loading is achieved, either the shaft friction or end bearing
failure load is reached, or until the maximum extension of the jack
(typically 150 mm) is reached. The Osterberg loading rate is slow enough
to be considered static.

An engineer must be involved in the planning, and the Osterberg


device must be installed concurrent with the shaft or pile. The loadcell
placement above or at the toe must consider the actual balance of soil

662 Likins, Rausche, Peterman


resistance above and below the cell to achieve optimal performance
Otherwise, the solution will be only a lower bound proof load (twice the
lesser of the friction or end bearing), since both friction and bearing
usually do not fail simultaneously. For low shaft resistance cases, some
additional loading (but less load and cost than a conventional static test)
at the top may increase the ultimate capacity tested. In low end bearing
situations, the loadcell may be placed above the shaft bottom, or the test
performed in stages. The displacements are monitored with telltales both
below and above the Osterberg loadcell. The load is determined by
measuring the jack pressure. Additional instrumentation may include
telltales at intermediate locations and/or sister bars for direct
determination of strain. Data collection by automated methods becomes
more important as the number of measurements increases.

Dynamic testing involves attaching accelerometers and strain


transducers to the pile shaft and testing under the impact of a large falling
weight (ASTM 0-4945, 1989) for high strain testing of piles as shown in
Figure 4. The dynamic load can be provided by a pile driving hammer, or
any large drop weight when testing drilled shafts. Stress wave
measurements require knowledge of both acceleration and strain as a
function of time during the blow; reusable transducers have been
developed and have seen decades of routine use worldwide. The strain
is generally converted to force from the modulus of elasticity and cross-
sectional area of the pile. The acceleration is integrated to velocity, and,
with good quality measurement, also to displacement. The energy
transferred into a pile is computed from the integral over time of the
product of force and velocity. Measurements are conditioned, processed
and evaluated by a Pile Driving Analyzer (PDA) according to wave
propagation theory using the Case Method (Goble 1980), or a special
numerical analysis program (CAPWAP@) when further evaluation or a
simulated static test is desired.

IN FIELD I IN OFFICE
I
I
I HARD COPY
I
IL _
PDA

PILE

Figure 4. Typical High Strain PDA Setup

663 Likins, Rausche, Peterman


Capacity is calculated at the time of the test; tests during driving
determine the remolded strength while restrike tests investigate the soil
strength changes with time due to set-up or relaxation. Generally capacity
results by this method are conservative often due to set-up, but also when
the penetration resistance (set per blow; blow count) is high (analogous
to a non-failing proof test with minimal net displacement). When repiacing
or supplementing static tests with dynamic tests, an experienced engineer
should operate the PDA. In that case a variety of parameters need be
reviewed such as soil strength time effects (set-up or relaxation) I

resistance distribution including negative friction and uplift, settlement


concerns, and site variability. Of course, the PDA also provides a wealth
of additional information on driving stresses, and hammer performance
(poor hammer performance may cause low bearing capacity) which are not
easily assessed in any other way. The potential of the PDA to test more
piles for considerably less cost is highly attractive.

The same technology can be applied to measuring energy


transferred into the drive rod of a Standard Penetration Test (SPT), and
potentially to evaluate dynamic soil behavior. As an alternate assessment
of energy efficiency, the Hammer Performance Analyzer (H PA; Likins 1988)
uses radar technology to determine the velocity (and hence kinetic energy)
of the SPT hammer or most pile driving hammers at impact. The main
advantage of the HPA is that no sensor is attached and it can be easily
included in the installation procedure of any or all piles/drill strings as
shown in Figure 5. It should be noted that many modern hammers,
primarily of the hydraulic type, have built in proximity switches for
detecting the impact velocity; most of these hammers have energy readout
and recording capability.
DPTIDNAI. RADAR
ANTENNA POSITION

MOVING RAM

RADAR ANTENNA
HEI.MET _ _ _ _

-'I---;:::==='~'\II \,
\\\ \
\
\
,
\ HPA STRIP CHART

o P'LE

Figure 5. HPA Setup

664 Likins, Rausche, Peterman


Another proposed solution to determine capacity is a slow dynamic
(some call it "quasi-static") test. The most common example is Statnamic
(Janes 1991), A mass of typically two to three times the mass required for
a dynamic test is placed on the pile top, A fuel charge (selected in
advance to the desired force input) is then rapidly burned, propelling the
mass upward and the shaft downward, The load is applied as a gradual
ramp, but with a time duration of only about one tenth second (similar to
a dynamic PDA test), If the force applied is low, the load will be a proof
load only, If the force applied is larger than the pile capacity the pile will
I

achieve significant velocity (and large displacement), the imposed force


must be balanced (Newton's second law) by both static and dynamic
resistances, and the measured force and displacement curves cannot be
directly interpreted, Unlike the true static loading of an Osterberg test,
and in contrast to the sharp impact of a conventional dynamic test, the
Statnamic gradual loading does not allow clear distinction between static
and dynamic resistances and large overpredictions have been reported
(Baker 1993, Janes 1994), The test uses a load cell to measure force and
a remote sensor to monitor displacement; both are automatically recorded
and results processed by a computer, A similar method has been devised
in Europe with a large drop weight, but inserting massive springs between
the weight and the pile top to extend the loading duration,
IN FIELD IN OFFICE

PJ,T, GRAPHIC
DATA LASERJET
COLLECTOR PRINTER

~""'"""" ~"'"
ACCELEROMETER

PILE

Figure 6, Low Strain P.I ,T, Setup

Low Strain Testing Considerations

The signals generated by a small hand held hammer in a typical


P,I,T. low strain integrity test are of a low intensity, requiring special
instrumentation and processing equipment. The pile top is instrumented
by attaching a sensitive accelerometer with a thin viscous material (often

665 Likins, Rausche, Peterman


a wax) and then striking the pile top with a hand held hammer as in Figure
6. In this test, again the acceleration is usually integrated to velocity. This
record contains both motion produced by the hammer input, and motion
caused by reflected stress waves which are of particular interest. Due to
both soil and pile damping effects on longer piles, reflections from cross
section variations or the pile toe may require special amplification to be
observable. The time from impact to observed reflection determines the
location depth, and the reflection magnitude is related to the size of the
cross section variation. An instrumented hammer may also measure input
force. A system including two accelerometers can detect downward and
upward travelling waves for piles embedded in the structure.

The low strain integrity test signals are captured by a small battery
powered data collection and processing device, P.I.T. Collector, which has
data enhancement features (filtering, averaging, time amplification) and
can output finished graphical results to plotters or laser printers. It can
rapidly be taken from pile to pile to acqUire integrity evaluation data (Likins
1993), making it possible to inspect every pile on site for major defects at
a reasonable cost. Additional analysis can estimate the pile shape.

Additional Developments

C>SC"'oscopoe

Structure

Parallel SeIsmIc Tes~


- --

P'..,zo ",'ec:1trlc
receiver - _-_- ~=LJ~ =- ". ...=: =--7
~ -

~
=-~ - ~~~
Tube ftlled _ _ "":.. _
vvlth """"ateor -..:...-- - ----
- --
Figure 7. Parallel Seismic Test (after Stain 1982)

Other sensing devices finding increased use in the deep foundation


industry include velocity measuring geophones and hydrophones for
vibration or blast monitoring, and applications involving Parallel Seismic
or Cross Hole sonic logging, with both methods relying on automated data
acquisition and processing. For Parallel Seismic testing (Hertlein 1992) I

a small casing is installed adjacent to the pile in question, but to a deeper

666 Likins, Rausche, Peterman


depth as shown in Figure 7. The pile is struck and the time from impact
to signal pickup by a hydrophone at different depths in the casing is
measured; since the rate of travel in the pile and the soil is different, the
pile length is readily apparent from a series of measurements, Because
of the needed borehole near the pile to be tested, this test is usually
restricted to a few test shafts only, However, if the pile cannot be tested
by P.I,T. or a PDA, then this test may be the best solution for length
determination of some existing piles (e.g. piles embedded in an existing
structure, steel sheet pile walls, H piles ... ).

Cross Hole Tests require at least two access tubes in a shaft (Levy
1970) into which a transmitter and a receiver are lowered as shown in
Figure 8. The arrival time and magnitude of the signal provides further
information on the integrity of the concrete between the two tubes; large
shafts require several access tubes to investigate the full perimeter, The
time required to perform the test is longer than for the low strain integrity
test.
OSCILLOSCOPE RECEIVER
WITH POLAROID CAMERA

IMPULSE GENERATOR

ELECTRICAL IMPULSE SYNCHRo-


NISER

RECEIVED SIGNAL

WINCH WITH
SENSOR
VOLTAGE PROPORTIONAL TO THE DEPTH OF THE TEST

:;.~..:... -i
~~~.':~.'!:~
C!:."o~;.
~'.:~:~
'''- . '.
..
II
..... "...
'oCl? .
TRANSMITTER ~~~~ RECEIVER

Figure 8. Cross Hole Tests (after Stain 1982)

Conclusions: Advantages and Disadvantages

Improved quality assurance is increasingly demanded for deep


foundation installations, primarily from a liability perspective. Using
electronic hardware solutions to automatically monitor the entire
installation process, or dynamically test selected piles, also has economic
advantages; the new monitoring methods being overall less expensive
than manually collecting data, primarily because they are less labor
intensive.

667 Likins, Rausche, Peterman


The automatic monitoring of pile installations can provide more
information and with improved accuracy in the recording. Electronic
devices are impartial and never bored by repetitive tasks. Well designed
equipment is easy to use, yet powerful and measures during installation.
Results are professionally presented in a standard format. The device frees
up time for the crew, or replaces manual observers, and because there is
no need to wait, the productivity rate of installation also improves.

Under favorable site conditions, Osterberg testing can be an


attractive alternate to static testing. Dynamic testing methods such as the
PDA are significantly faster and far less expensive than static loading
tests, and provide information on capacity, hammer performance, driving
stresses, and structural integrity. Piles selected for dynamic testing can
be chosen at any time, even after installation has been completed, with
restrike tests allowing for investigation of capacity changes as a function
of time.

Low strain integrity testing, parallel seismic tests, or cross hole


testing can only be used to evaluate integrity or length of suspect shafts.
Low strain integrity testing can be economically applied to every pile if
necessary and piles to be tested need not be designated in advance.
Cross hole testing requires advance planning and installation of access
tubes in the shaft, and requires more time to complete the test. The cost
of test per pile is therefore higher than low strain integrity testing. Parallel
seismic tests require an extra borehole adjacent to the pile tested; this
extra effort and cost generally limits this testing to a few selected shafts,

Appendix - References

American Society for Testing and Materials (ASTM). (1989). "Standard test
method for high-strain dynamic testing of piles." 04945-89.
Baker, C et al. (1993). "Drilled shafts for bridge foundations." Federal
Highway Administration, DTFH-61-88Z-00040.
Goble, G.G., Rausche, F. and Likins, G.E. (1980). "The analysis of pile
driving a state-of-the-art." Proc./nternational SeminarofApplication
of Stress Wave Theory on Piles, Stockholm, Sweden, 131-161,
Hertlein, B, (1992). "Selecting an effective low-strain foundation test."
ADSC Foundation Drilling.
Janes, M., Bermingham, P. and Horvath, B. (1991). "Pile load test results
using the statnamic method." 4th International OFI Conference,
Rotterdam.
Janes, M., Sy, A. and Campanella, R.G. (1994). "A comparison of
statnamic and static load tests on steel piles in the fraser delta"
Proc. Deep Foundations, 8th Annual Symposium, Vancouver
Geotechnical Society, Vancouver, B.C.

668 Likins, Rausche, Peterman


Levy, J, (1970), "Sonic pulse method of testing cast-in-situ concrete piles,"
Ground Engineering,
Likins, G. and Rauscher F. (1988). "Hammer inspection tools." 3rd Int'I
Application of Stress-Wave Theory to Piles, Ottawa, Canada.
Likins, G., Rausche, F., Miner, R. and Hussein, M. (1993). "Verification of
deep foundations by NOT methods." ASCE Conference, Dallas, TX.
Osterberg, J.O. (1989). "New load cell testing device." Proc. 14th Annual
OFI Conference, 17-28.
Rausche, F. r Likins, G.E, and Shen, R.K. (1992), "Pile integrity testing and
analysis." 4th Int'l Conference on the Application of Stress Wave
Theory to Piles, The Hague, Netherlands, 613-618.
Schmertman r J. (1993). liThe bottom-up, Osterberg cell method for static
testing of shafts and piles." 13th Central PA Geotechnical Seminar,
Hershey, PA.
Stain, R. (1982), "Integrity testing." Civil Engineering.

669 Likins, Rauscher Peterman


The Development of A New Pile Load Testing System

Toyokazu Fujioka! and Kiyoomi Yamada 2

ABSTRACT

The most reliable method to determine the bearing capacity of a pile is by


a static load test. For time and economic reasons, only limited piles are tested
at a site. A new pile load testing system developed is perfromed by utilizing
the pile shaft as reaction to push down the separable toe without reaction
beams. The results are transformed to the load.:.settlement relation of the pile
axially loaded at the head. This paper describes the principles, testing devices
and examples of test results for various pile types.

INTRODUCTION

In view of the many uncertainties involved in analysis of pile foundations,


the most reliable method to determine the bearing capacity of a pile is by a
static load test. However, a conventional pile load test is not easily performed
since this procedure requires considerable time and expense for the preparation
of anchor piles and loading frames. Thus, pile load tests are considered only
for pile types and soil conditions for which there is little previous local expe-
rience and to special piles installed by newly developed techniques.

Although bearing capacities of bored piles including bored precast piles


are considerably varied due to the influence of installation techniques, load

1 Manager, Civil Eng. Dept. 2, Chiyoda Corporation. 12-1, TSUfurnichuo, 2-chorne,


Tsururni-ku, Yokohama
2 Professor, Dept. of Civil Eng., Nihon University, 8-14, Kanda Surugadai l-chorne,
Chiyoda-ku, Tokyo

670 Fujioka
tests for bored piles are being conducted only in rare instances. Additionally,
it may be physically impractical to obtain a plunging failure or a settlement
greater than 10 % of the pile base diameter when the loads are very large.

Under these circumstances, we have developed a new pile load testing


system (the new system) to overcome the diffIculties and to gain detailed infor-
mation on load transfer in shaft and base resistances to allow for an improved
design. This system is a method to measure shaft and base resistances separate-
ly by pressurizing a hydraulic jack installed near a pile toe and creating an up-
ward force on the shaft and an equal downward force to the toe. Therefore, re-
action beams and anchor frames are eliminated.

PREVIOUS DEVELOPMENT

The concept of the new system utilizing pile shaft resistance as a reaction
to push down the base was invented in 1969 and patented for bored piles by
Nakayama and Fujiseki (1973) and for driven piles by Sumii (1978). The
conceptual schemes are shown in Fig. 1.

A similar testing method for determining concrete to rock bond stress in a


bore hole was reported by Gibson and Devenny (1973). Furthermore, similar
techniques based on the same Pressure Gauge Dial Gauges
concept have been developed Grout Pump
by Cemak et al. (1988) and
Osterberg (1989). I'
Pump
I. . . . I
A joint reseach team I' :.' I
gathered from Chiyoda Corpo- I". I
ration and Daido Concrete Co. 1:- :'-':".: Bored Pile I
Ltd. was established to devel- ....
"
.
r
I I
op the new system in 1985. I> . I I
As a result of the first stage of I :~.:-: ' I I Precast
I " .', :., " ' I Pi Ie
the research including some I
I" '<:':: I 1
comparative tests, it was con-
HydraUlic Jack I
cluded that the new system I ", I Hydraulic
I rnr!1 I IG Jack
provides a realistic load-car- . Lu'U.U
" .. . . .
' ' ~
.1
~;;

rying capacity and valueable


11111111
detailed information for de- (a) A Bored Pile (b) A Driven Pile
sign of a pile foundation (Nakayama & Fujiseki 1973) (Sumii 1978)
(Fujioka et al. 1989). In add- Fig. 1 Conceptual schemes

671 Fujioka
tion, we have developed some hydraulic jacks enabling to install at arbitrary
depths in a pile and a tesing method applying
load reciprocally by a pair of jacks installed at
different depths.

PRINCIPLES

The new system consists of a hydraulic


jack installed near a pile toe, electrical-resis-
tance strain gauges (rebar transducers) and
steel-rod telltales instrumented at different
depths as shown in Fig. 2. The jacking load,
strains and displacements at the instrumented
depths including the upward and downward Jacking
displacements of the jack are measured. With Load, Pi
a result, the displacement, load and unit shaft Point Resistance
resistance distribution curves in a pile are ob- Fig.2 Scheme
tained at each load as shown in Fig. 3.

Upward Downward Compression Downward Upward

~~n~~
J'io 'YolTop Disp I --r
Element i Q (~l
, YrnW
:..t
i+1
~+1)

Disp. at Jack Jacking Load

~ 1~=~7
Tip. Disp.,sp Point Resist,Pp
(a) (b) Displacement (c) Load (d) Unit Shaft Resist.
Distribution Distribution Distribution
Jacking Load, Pj

a:.
Ul
o
"E
<tJ
3:
c
I 3:
I YlnC') (5
Pile Displacement,Yrn
(e) r -Ym Curve (f) PJ-s i Curve
Fig. 3 Typical curves from the new system

672 Fujioka
The loads in a pile at different depths are computed by multiplying the
strains by the axial stiffness of the pile. The slope of the load distribution
curve at any depth, divided by the perimeter length, yields the unit shaft resis-
tance at the depth. A family of load transfer curves is obtained as shown in
Fig. 3e. A curve showing the jacking load versus the downward displacement
is also obtained (Fig. 31).

The base resistance-settlement relation and the unit shaft resistance-dis-


placement relations enable to transform to the equivalent load-settlement rela-
tion of the pile axially loaded at the head by load transfer analysis (Seed and
Reese 1955). For the transformation, following assumptions are made.

(l) A pile is elastic.

(2) Unit shaft resistance-displacement curves for the upward movement in a


new pile load test are the same as those for the downward movement in
a conventional pile load test.

It is believed that the second assumption is on the safe side. In order to


establish a more precise analysis, this point should be studied further on a basis
of many comparative test results for various pile types and soil conditions.

TESTING DEVICES

In view of load testing purposes, testing devices of the new system have
been designed so as to enable a test pile to be constructed in the same manner
as a working pile. Hydraulic jacks developed are mainly divided into retrieval
and non-retrieval types as shown in Fig. 4.

A pair of specially manufactured steel notches (Fig. 5a) is welded inside


the test pile slightly above the upper separable pile tip. Prior to testing, a re-
trieval jack is lowered to the depth of the notches through inner space of a pre-
cast pile and then locked by rotating 90' clockwise. After testing, the jack is
unlocked by rotating 90"
counter-clockwise and recov- Retrieval Notched Type
ered for repeatedly use. On Pot Type
[
the contrary, a non-retrieval Non-retrieval-f Sheath Type
jack is built in a pile and is MUlti-acted Type
left following a completion of
testing. Non-retrieval jacks Fig. 4. Type of hydraulic jacks

673 Fujioka
are classified in three types according to the shape and configuration and the
typical non-retrieval jacks are shown in Figs.Sb and Sc.

Shear or tension bolts and slide bolts are equipped with a jack to prevent
from dropping a toe portion of a rebar cage or a precast pile during the installa-
tion. A pair of rubber or steel plates is also equipped with a jack to reduce
cross-sectional area of concrete so that a pile is easily separated under low jack-
ing load prior to testing. The separation is usually well defined as the sudden
loss of jacking load. With regard to use of a tremie, small diameter jacks are

HydraUlic Lines

, - '(FJ'i!t------
'I
Specially
"...>J
:: -----T!
Manufactured
: - : :.' .
I I -"" ~ -
:
<;
-: r .-
I _
Precast
PIle'
(b) Non-retrieval Sheath Type Jack

Steel Notches: ' : .~, - , . '.

Jackl
=" Tension
Bolts

lJ-'---- i:~
'- __ .../t--- .__' a
a
9
~
g
a
a a 9
Installed State Testing State D 0 a
1'3 D !II "~

(a) Retrieval Notched Type Jack (C) Non-retrieval Multi-acted Type Jack

Fig. 5 Schemes of hydraulic jacks

674 Fujioka
arranged at an appropriate distance from the pile center so that there is no dan-
ger of a projection becoming hooked on the jacks as the tremie is removed dur-
ing concreting. A swivel-head mechanism is placed atop the ram to minimize
ram friction.

Hydraulic lines connecting between a pump unit and a hydraulic jack are
installed in the protective steel pipes built in a precast pile or along longitu-
dinal rebars of a rebar cage.

For a reliable test and a proof of the accuracy of the load-measuring


system, a calibration test is perfonned with a 3,000 tf (29.42 MN) universal
testing machine designed to allow rams to travel several centimeters. The
hydraulic lines actually built in a test pile and the pump unit are employed in
the calibration test.

Each hydraulic jack is a site specific manufacturing undertaking. The


normal load of up to 2,500 tf/m 2 (24.5 MPa) and 1,200 tf/m 2 (11.8 MPa) can be
applied for precast and bored piles respectively and the rams are specifically
designed with a sufficient stroke. A post test grouting of a hydraulic jack can
be easily performed to use a test pile as a working pile.

Settlements of a pile head are usually measured directly by sensitive dial


gauges supported by reference beams. Settlements at different depths in a pile
are measured directly by means of steel-rod telltales built in the test pile. The
telltales are unstrained steel-rods, that are inserted into steel tubes, and are in-
stalled in pairs (2 or 4 to each depth on opposite sides of the shaft).

Loads in a pile are obtained from the use of rebar transducers. The
intemalload is computed by multiplying the axial stiffness of the pile (cross-
sectional area times mudulus of elasticity) by the strain.

CASE HISTORIES

Bored Piles by Overall Casing Method (Benoto Method)

In this case, it was desired to determine how far to drill into sandy gravel
with boulders to obtain the required bearing capacity for design. For founda-
tion design of the Hokuriku Shinkansen piers at Takasaki, 3 test piles of 13.5
m long and 1,000 mm in diameter were drilled through silty sand 6.6 m, sandy
gravel with boulders 2.7 m and clayey fine sand 2.0 m to sandy gravel with

675 Fujioka
bouders. Since various sizes

-
~
of boulders were anticipated J:::.
.....
-
.- SPT Pi Ie Pi Ie Pi Ie
a. N-Value No. 0 No.1 No.2
in a sandy gravel layer as a <Ll
C>
...
0
c... 11 ).() ~ k-T-.z. ,....,..,.1.
~

result of several borings at the 10


'tcI S r-p,- I"-

(,D f.L
site, 3 piles were tested as I..D

shown in Fig. 6. It was actu-


ally found during excavation
X FILL

that boulders varied consider-


ably in size from Pile No.O to
No.2.

A non-retrieval multi-
acted type jack with 6 rams,
.>t-
750 tf (7.35 MN) jacking ca- ~
-,
pacity and 250 rom stroke _\
was manufactured and welded ~~ ~l~
to a rebar cage. The rebar cage e- ... t1 ... Et..
with a jack was lifted and low- Telltale Points
Strain Meas. Points
ered into a drilled hole in the
same manner as a working pile Fig. 6. Hokuriku Shinkansen Pier, Takasaki
and concrete was placed by use
of a tremie without any trouble. E? Jacking Load, Pj (tf)
-5 100 20n 300 4(jJ 51 )0 R 1n 7(
60
Fig. 7 presents the down- y---
--1,.-.-. ---- ._" - .... - -o--~

,. ,. -
ward and upward displace- 4C
ments during the load test of ----- !r'--:'::: -.= ::=
-~-::
~~ 5')
Pile No. I. It is recognized that 20 ~:::-.::.-
~ ... " -.;::::. --
-- -=.
"'F~
~.-=--~

failure in shaft resistance oc- ~-~~~.....==-== t:::" ,~


0 -~
curred at about 25 mm upward
r:.~~~ Pi Ie No. 1
displacement whereas 117 rom a> ""'=-< =-.-: ..... 1--->1
r-~:.;
downward displacement oc-
curred at the maximum applied 40
~--::~
-- -- po- ...... "--..=<e.::
'-L
load of 600 tf (5.88 MN).
60 ~- -'''; ,-:-,':_~":'_~....
"1
c. ~--..::;
Cf.l .:..~.
From Fig. 7, the ultimate Cl
-..,
00
base resistance defined as the
load corresponding to 10 % of
"'0
lo...
co
I'>'<-.:::-..~
~
~...... -
f;.:,"'-= :. ...'':':''''... -..... - .b(
5')
~1 , \IV
--a-,
the diameter of the pile base is ~
o
'"
estimated to be 580 tf (5,69 Cl
;>- - -
--- ... ---.., 1'--- -- ... -0---
120
MN). In this case, both the Fig. 7. Upward and downward load-dis-
ultimate shaft and base resis- placement curves, Takasaki, 1 tf=9.8 kN

676 Fujioka
tances were obatained dramatically. Also, a similar curves, not presented in
this paper, were obtained for Pile No.2 indicating that failure in shaft occurred
at 752 tf (7.37 MN) and
Load in Pi Ie. 0, (to
the downward displacement Pi I e No.1
.... 00 200 AOO 500
was 83 mm. ~ ~ t-=--=--=--=--=--=-i-..;,------=---- --i-::'::-;.._-:--------i~
"~,,",,2
The distribution of .... E
t-I
load with depth in Pile No. .... .....J
<.:>
<tI
4 ::
1 is shown in Fig. 8. It is of ~ '--' teIIII~--+----I-----i
c:: N 6
interest to note that axial
load is substantially carried
by the shaft resistance in the
upper gravel with boulders.

Fig. 9 presents the fam-


ily of unit shaft resis- 14~-------_ ......
tance-displacement curves Fig. 8 Load distribution for Pile No.1,
for Pile No.1. It is interest- Takasaki, ltf=9.8 kN
ing to note that an gradual
loss of the unit shaft resistnce in the lower gravel layer (element 6) is occurred
after the peak is reached whereas the residual unit shaft resistances in other lay-
ers are gradually increased.

,....... 50
N Pi Ie No.1
E o
-......... o
'+- c..o
+-'
'--"' 40
.....

a.>
u 30
c
ro
+-'
(/)

(/) 20
a.>
0:::
+-'
'+-
ro
..c
(/)

+-'

C Point Resistance
:::l 10 20 30 40 50 60
Pi Ie Displacement. Ym (mm)
Fig. 9 Unit shaft resistance-displacement curves, Takasaki, 1 tf/m 2=9.8 kPa

677 Fujioka
The transformed, equiva- Load at Head. Po (tf)
lent top load-settlement curves 400 800 1200 1600
of 2 new load testing piles
(No.1 and No.2)are compared
with the curve of a conven- '"
(J)40J----+--+--+--j.~~,.......::::l""-__I-_1
tional load testing pile (Pile -d
C1J
No.O) in Fig. 10. It appears Q)
:I:
that the differences in the top +-'
C1J 80
load-settlement curves are +-'
c
Q)
resulted from the variations of E
Q)

boulder size. ::::120


Q)
(/.)

Bored Precast Piles Fig. 10. Comparision of test results, Takasaki,


1 tf=9.8 kN
For foundation design of a 5 story building for dowelling at Urawa, 2 pre-
stressed high strength concrete (PHC) piles of 36.5 m long and 450 rnm in
diameter were tested. Non-re-
trieval sheath type jacks with SPT Pi Ie Pi Ie
350tf (3.43 MN) jacking ca- a.Q) ;:0Q) N-Value
.I::.

No. A No.8
Cl c: o 10 20 30 4D50
pacity, 200 rnm stroke and the ~-p-
same outside diameter as the
PHC pile were manufactured. 5' \
Ii
II',

pt
t
l
jl~
,', NH
Due to the scatter in .
10 ' :
shearing strengths of silts and : MH
clays, the jacks were welded - I' I
Ii i
at different heights. The 15 . , MH
o Ii 1
greater the shearing strengths "
-
SM V)
l~" ~
are evaluated, the higher the
-
-
Il-f-+++-i I ~
C"'l Ii ~
20 - CL I
I
toe height becomes. Finally, SM
I
I
the toe heights of 2.0 and -- I
I
12.0 m are selected for Pile 25 I
CL
I:
No. A and No. B respectively. I'
I
The test piles were con- ~ i~ ~F-+-+-il I
I
structed by a preboring I
~
!~
CSL
method and high early stength
L-_

portland cement (type ill) was .'" Points


used. Details of the soil pro- SIt-~~~~-P~~: Telltale
40 Strain Yeas. Points
file and test piles are shown in
Fig. 11. Fig. 11 Bored precast piles, Urawa.

678 Fujioka
The load tests were performed 15 days after installation of the test piles.
The upward and downward dis-
Jak i ng Load. Pj (tf)
placement curves for Pile 1C )0 200 300 400
1000
No.A are shown in Fig. 12.
Similar curves, not shown in 9lr
~
U8')
~

this paper, for Pile No.B are


~80
also obtained. Failure in shaft
! 70
resistance suddenly occurred
at about 10 mm upward dis- ;:: I
_ 60
placement and the jacking load ci.
<f.)

of 330 tf (3.26 MN) whereas o 50 NO.A

~w ~
the downward displacement 0 Ie
was only 15 mm. The figures Oll~
0
in parentheses indicate the
IR~ Yi Jack (

holding time at the maximum


applied load. The upward dis-
e2 0
.
l~.
C'I
? J
J
1/
placement rate at the failue 0 ......
~
was approximately 4.5 mm per 0
ci. "'-v-
minute. Since the maximum ~
<f.)

downward load is below the 1 0:-. ~

failure load, the base resis-


"C
I... ~ (18')
~ 2
tance-settlement curve is ex-
c
iI:
g3 0
\
'.
trapolated by a hyperbolic ap-
proximation (Kondner 1963). Fig. 12. Upward and downward displacement
curves,Urawa, 1 tf=9.8 kN
The dashed line in Fig. 12 indi-
cates the extrapolation.
Transformed Load at Head, Po (tf)

The results have been


transformed to the equivalent
top load-settlement relation
using load transfer analysis.
The transformed, equivalent
top load-settlement curves are
shown in Fig. 13. Again, the
dashed lines are computed by
use of the extraporated base I'
I'
resistance-settlement relations.
It is interesting to note that the
transformed load-displacement 1[4()L---------..J.....--..J
curves of the 2 piles with dif- Fig. 13. Comparision of test results,
ferent toe heights are identical. Urawa, ltf=9.8 kN

679 Fujioka
An Underreamed Bored Pile by Rotary Drilling

Due to anticipated downdrag, the engineers selected an underreamed


bored pile foundation socketed into a underlying mudstone for the subway
tenninal building at Shin-Yokohama. To obtain the required base resistance
for design, an underreamed test pile of 32.3 m long and 1,500 rom (shaft) to
2,200 rom (underream) in diameter was drilled through silt 26.9 m and fill
sand 2.8 m into the mudstone by use of a belling bucket.

As the test pile was estimated to have less shaft resistance than the base
resistance, a reaction beam was connected to 2 adjacent piles to compensate
for the insufficient shaft resistance. Details of the profile and arrangement are
presented in Fig. 14.

A non-retrieval multi-acted type jack with 6 rams, 1,800 tf (17.65 MN)


jacking capacity and 400 rom stroke was manufactured and welded to a rebar
cage in the same manner as in Hokuriku Shinkansen case.

Fig. 14 Profile of Shin-Yokohama subway tenninal building site

680 Fujioka
Upward and downward load-displacement curves for the test pile are
shown in Fig. 15. Failure in shaft resistance occurred at the load of 916 tf
(8.98 MN) and 15 mm upward displacement, indicating that it takes only small
displacement to mobilize the shaft resistance. By utilizing the shaft resistances
of 2 adjacent piles, the underream was loaded up to 1,626 tf (15.95 MN). The
dashed line in Fig. 15 indicates the net shaft resistance after subtracting the
load at the pile head from the jacking load. As seen from Fig. 15, the shaft
resistance decreases gradually every cycle of loading. This strain softening

150
-- --, \
. )
I
1
(~J-PO)-Yj
I
I
120 -----
I,,'
I
E I
E I
I
~
I
I PJ-YJ
>- I

90
ci.
en ,
0
"U
"-
.
"
i
ro I
;=
ro
60
0-
::J

30

,-,
(I
E 1 00 200
E

en

ci.
en
0
10
"U
"-
co
;=
C
;=
0
0

20
Fig. 15 Upward and downward load displacement curves,
Shin-Yokohama, 1 tf =9.8 kN

681 Fujioka
can be distinctly found in the unit shaft resistance-displacement curve for
the mudstone (element 8) in Fig. 16.
""'30---..-----....----~--~---..

~'--' 24'("\.

:118H11---+-+--+--f--+----+--+...,..-;-~_
--+--.....=......,-----/
_-2J
c
CIl
+-'
Ul
\

,
I
Jacking
----jl' ------ Load
I

30 60 90 120 150 Point Resistance


Pi Ie Displacement, Yml (mm)
Fig. 16 Unit shaft resistance-settlement curves, Shin-Yokohama,
1 tf/m 2=9.8 kPa

Fig. 17 presents the unit base resistance-settlement curves at the top and
bottom of the underream. It is of interest to note that the effectiveness of the
underream in carrying axial load is accompanied substantially by the side resis-
tance. As the settlement of the underream under the maximum load is quite
small, it is difficult to find the ultimate base resistance.
Unit Base Resistances, qd, qp (tf/m l )
o 100 200 300 400 500 600 700 800
0rw- 0 I I I I

o
o 0
0.
(/) o 0
10 - o
."
(/)

301.-_..J...1_---J1_ _J....-1_....I.1 11...--_..L.....1_-11_........I

Fig. 17. Unit base resistance-settlement curves, Shin-Yokohama,


1 tf/m 2=9.8 kPa

682 Fujioka
The slightly weathered mudstone was found to have an average uniaxial
compressive strength (qu) of 36.2 kgf/cm 2 (3.55 MPa). The unit base resistanc-
es at the top and bottom of the underream have been normalized by dividing by
quo The settlements at the top and bottom of the underream have also been
normalized by dividing by the respective underream diameters. The normal-
ized curves showing load transfer in the underream as a function of the settle-
ment ratio are presented in Fig. 18. For reference, the uniaxial compression
test results are normalized and shown in Fig. 18. As may be seen, the trend of
the normalized curves is consistent with the normalized uniaxial compression
test results.
Unit Base Resistance
Uniaxial Compressive Strength
o0~~""""''''''''''-er0~.~5~-;=;-_ _....;1-r.O:'' ' - -i1'....;5 2-i-.O..;....., -..:..;2. 5

,....., Ie d=Sd/Dd
~ IC p=Sp/D p
'--'
: : Ax i a I Strain
'" a : Ax i a I stress
c-
0.5
It:
,;
It:

+-'
c .....
ID ID
~ ~1.0
- E
+-' ro
+-' . -
IDQ
(/)

1. 5 -L.... --L. ....-JL...- ....l...- -J


~
Fig. 18 Nomalized curves showing load transfer in the underream,
Shin-Yokohama.

CONCLUSION

Up to February in 1994, total 27 piles, including 20 bored precast


(injection) piles, 2 driven PHC piles, 4 bored piles and a slurry wall have been
tested at 13 sites in Japan.

The testng devices of the new system have been simplified and modified
to apply for all types of piles and slurry walls. Because of the simplicity and
low cost, the new system has become popular. It tends to be used for piles in

683 Fujioka
rocks, underpinnings and large diameter bored piles recently since it is more
advantageous in some instances than a conventional pile load testing.

ACKNOWLEDGEMENT

The new system could not have been developed without the assistance
and dedication of many engineers. The authors wish to express their sincere
gratitude to Mr. Hifumi Aoki, Japan Railway Construction Public Corporation,
for his valuable advice and help for the development. Thanks are also due
Transportation Bureau of City of Yokohama for the adoption of the new
system for the underreamed bored pile described, Mr. Isamu Sandanbata,
Hazama Corporation, for his efficient collaboration and advice and Messrs.
Yoshitaka Ito, Shoji Tanaka, Hirofumi Kato, Atsuo Arai and Kunihiko Arai
for their never-failing support and decisive contribution to the development.

REFERENCES

Cernak, B., Hlavacek, J., and Klein, K., 1988, "A New Method of Static Pile Load Test
System VUIS-P", Proceedings, 1st International Geotechnical Seminar on Deep
Foudations on Bored and Auger Piles, A. A. Balkema, Ghent, Belgium, pp.29l-302.
Fujioka, T., Arai, K., Arai, A., and Yamada, K., 1989, "The Development of A New Pile
Load Test Method", Proceedings, 25th JSSMFE Annual Conference, Okayama,
pp.1297-1300 (in Japanese).
Gibson, G. L. and Devenny, D. W., 1973, "Concrete to Bedrock Testing by Jacking from
Bottom of a Bore Hole", Canadian Geotechnical Journal, Vo1.lO, No.2, pp.304-306.
Kondner, R. L., 1963, "Hyperbolic Stress-Strain Response; Cohesive Soils", Journal of
Soil Mechanics and Foudation Engineering, ASCE, Vo1.89, No.1, pp.1l5-l43.
Nakayama, J., and Fujiseki, Y., 1973, "A Pile Load Testing Method", Japanese Patent
No. 1973-27007 (in Japanese).
Seed, H. B., and Reese, L. C., 1955, "Action of Soft Clay along Friction Piles", Proceedings
of ASCE, Vo1.81, Paper No.842, pp.1-28.
Sumii, G., 1978, "A Load Testing Method for Precast Piles", Jananese Patent No.1978-
12723 (in Japanese).
Osterberg, J., 1989, "New Device for Load Testing Driven Piles and Drilled Shafts Separates
Friction and End Bearing", Proceedings, International Conference on Piling and Deep
Foundations, London, A. A. Balkema, pp.292-302.

684 Fujioka
Polyethylene Coating for Downdrag Mitigation
on Abutment Piles

K. S. Tawfiq'

Abstract

To overcome problems associated with bitumen coating


in reducing downdrag forces in piles, an al ternati ve
method is suggested herein to coat abutment piles with
polyethylene sheets. The effectiveness of this method was
investigated in the laboratory using direct shear and rod
shear apparatus. Concrete blocks were prepared and coated
with various arrangements of polyethylene layers and
tested under different temperature and loading conditions.
The obtained results showed that changes in temperature or
rate of loading did not influence the shearing
characteristics of the concrete-polyethylene-soil samples.
Varying the number of polyethylene layers, however, has a
great effect on the friction characteristics of the
polyethtylene coating. Furnishing the concrete blocks with
two 6 mil (0.15 rom) polyethylene layers reduced the shear
resistance by 78%. More reduction could be achieved from
the two oil-lubricated polyethylene layers. This layering
arrangement provided the most effective performance (98%
efficiency) for the polyethylene sheeting method.

1 Associate Professor, Department of civil Engineering,


FAUjFSU College of ENgineering, Tallahassee, Florida,
32316.

685 Kamal Tawfiq


Introduction

Proprietary Mechanically Stabilized Earth (MSE) walls


are typically constructed at bridge abutments on Florida
Department of Transportation (FOOT) projects. Prior to
wall construction, end-bent concrete piles are driven
behind the proposed wall. Subsequently, as the wall is
constructed, fill is placed and compacted adjacent to the
piles. This process induces a relative settlement between
the pile and the adjacent soil causing the friction
resistance along the pile shaft to decrease and,
eventually, to reverse its direction (Figure 1). The
amount of settlement that causes this reversal is about 1%
of the shaft diameter. This negative skin friction gives
rise to downdrag forces on the pile shaft and represents
one of the major causes of pile settlement or pile failure
(Baligh et aI, 1981). To mitigate this effect one of the
following alternatives can be used: (1) preloading the
site to reduce soil settlement that follows pile
installation; (2) use casing to prevent direct contact
between the pile and the soil; and (3) use bitumen coating
as a friction reducer. The last method represents the most
common method in practice for downdrag reduction.

PILE SHAFT PROPOSED ELEVATION

.- .-
(A) During Filling ' . "9 .
WALL UNDER
lo. ~
-- - -- - --- .
CONSTRUCTION ~'_

II'N,:.:;.E;::::GA~TI.:.;;VE~F.:;;RI:::,CT~IO;:,:.N4'
",,'-..,..,.,..,..,.':'- BACKFILL

PILE SHAFT
(8) After FIlling

...------1', H F1LL
-BACKFILL
- '' -,'

Figure 1 Concrete piles driven behind Proprietary


Mechanically stabilized Earth (MSE) wall.

686 Kamal Tawfiq


According to the Florida Department of Transportation
(FDOT) specifications, pile coating requires asphalt type
bitumen conforming to ASTM D946, with a minimum
penetration grade (pen. 2SoC) of SO at the time of pile
driving. The primer should conform to the requirements of
ASTM D41. Also, bitumen coating should be applied
uniformly at a temperature of not less than 149 C (300F),
and not more than 176C (3S00F). The bitumen coating should
be applied to a minimum dry thickness of 0.317 cm (1/8
in.) but in no case should the quantity of application be
less than 0.03 m3 /m 2 (8 gallons/100 ft 2 ).
It has been proven, however, that pen. 2So C (1/10
mm), required by the FDOT to identify different bitumen
coatings, is not sufficient for predicting shearing
behavior of the bitumen coating. This is because for the
same penetration level bitumen materials exhibit a wide
range of rheological properties (Corbett & Schweyer,
1972). Moreover, for the same asphalt cement, AC-10 for
instance, a change in the temperature from 2So C to 60 C
can vary the viscosity of the bitumen from 300,000 poise
to 834 poise (Busby & Rader, 1972).

The temperature susceptibility of the bitumen


material and the rate of the shear deformation are the
major source of uncertainty when it comes to downdrag
prediction in bitumen coated piles. A small increase in
temperature may distort the coating due to bitumen
dripping, and may leave the pile bare or cause the coating
to become too thin to be effective in reducing downdrag
forces. Bush et al. (1991) conducted several flow tests to
determine the tendency of certain bitumens to flow or run
off a pile during storage in warm or hot climates. Their
results show that the flow of bitumen increases by
increasing the temperature, the storage period, and the
coating thickness. Moreover, they found that upon
increasing the shearing strain rate, the viscosity of the
bitumen drastically decreases.

Polyethylene coating for Downdrag Reduction

considering the uncertainties surrounding downdrag in


bitumen coated piles driven behind MSE walls, an
alternative method is needed to reduce negative skin
friction on piles and to overcome problems associated with
bitumen coating. The requirements for this method include
a friction reducer performing as (1) a durable pile
coating to reduce downdrag, and (2) a pile casing to
prevent direct contact between the soil and the pile
surface. Such a method may be possible by coating the pile
shaft with a smooth layer of polyethylene (Tawfiq and Ping
1992). This kind of fabric could ease the effect of

687 Kamal Tawfiq


relative settlement of the adjacent fill. If more than one
layer of polyethylene is wrapped around the cantilever
pile, the anticipated friction could take place between
the layers.

Scope of the Study


The effectiveness of the polyethylene coating method
in reducing negative skin friction in cohesionless soil
has been investigated in the laboratory using sets of
concrete samples covered with various arrangements of
polyethylene sheets. The samples were tested under the
direct shear and rod shear test conditions.

To draw a meaningful comparison between the test


resul ts, the samples were tested at different rates of
deformation and temperatures. The boundary conditions of
the polyethylene sheets were also varied to obtain the
most efficient arrangement that could produce the least
skin friction.

Material properties
1. Soil properties
Soil samples used in this study were prepared from
two types of soils. The first soil was reddish brown sand
with some fines. The natural water content, we' of this
soil was 2%. The second soil was white to light tan
coarse-grained crushed limestone. The two soils were
selected to fit within the required range of the backfill
materials recommended by the many state highway agencies
and to suit the size of the direct shear box. According to
the Unified Soil Classification System (USCS) the first
soil was well-graded sand (SW) with 2% fines and, the
second one was well-graded gravel (GW) with less than 1.4%
fines.

2. Polyethylene Sheets
Two types of sheets were selected for this
investigation. The first type was TEXTRUO 0.15 mm clear
polyethylene sheets, and the second type was the 1 mm
GUNOLINE HO, which is a high density polyethylene
containing approximately 97.5% polymer and 2.5% of carbon,
anti-oxidants and heat stabilizers. The purpose for this
selection was to investigate the effect of layer thickness
on the friction characteristics of concrete-polyethylene-
soil samples.

688 Kamal Tawfiq


Sample Preparation and Testing Procedure

1. Direct Shear Test

To simulate the friction mechanism between the pile


shaft and the coating material during downward movement of
the surrounding soi I, concrete blocks were prepared to
imitate the surface of a concrete pile. These blocks
replaced the lower half of the shear box in the direct
shear apparatus. The blocks were the same size as the
lower half of the shear box (12.7 cm x 12.7 cm and 1.27 cm
in thickness). A large number of these blocks were
prepared and cured before furnishing them with different
arrangements of polyethylene sheets.

The required concrete-polyethylene samples for the


direct shear test were divided into three main sets. The
first set consisted of concrete blocks covered with one
layer of polyethylene. Some blocks in this set were
furnished with loose layers, others were provided with
fixed layers. In the loose-layer samples, the polyethylene
sheets were placed unfastened on the blocks free to move
with the direction of the applied shearing force. In the
fixed samples, the sheets were wrapped around the concrete
blocks and fastened tightly to the blocks.

The second set of concrete-polyethylene samples


contained concrete blocks furnished with two layers of
polyethylene sheets. As for the first set, the boundary
conditions of the sheets were also varied. In some samples
the two layers were placed loose on the concrete surface;
in others, only the lower layer was fastened to the blocks
while the top layer was set loose.

The third set was similar to the second one except


that the two polyethylene sheets were loose and were
lubricated in between the layers using mineral oil. For
the concrete-polyethylene samples provided with 1 rom
sheets, only one layer of polyethylene was used. The
boundary conditions of these samples were of the loose and
fixed layers.

The testing procedure for a typical concrete-


polyethylene sample using direct shear apparatus was
started by placing the sample in the direct shear
container, and the upper half of the shear box was
positioned on the top of the polyethylene sheet. A pre-
weight soil specimen was then carefully poured into the
shear box and densified before adding the loading cap
(Figure 2). Upon assembling the loading frame, the
vertical and the horizontal dial gauges were attached to
the frame, and the shear box was set, ready for testing.

Kamal Tawfiq
689
NORMAL LOAD

SOIL SAMPLE

POLYETHYLENE
SHEETS

concrete;r:
block "i?;

SOIL PARTICLES
SHEARIN
FORC
POLYETHYLENE
SHEETS

SECTION A-A

Fiqure 2 Schematic of direct shear test on concrete-


polyethylene-soil samples.

The direct shear testing was started by applying


normal stress on the loading frame and then shearing the
concrete-polyethylene-soil samples at a predetermined rate
of deformation. Three normal stress increments (18.85 kPa,
27.02 kPa, and 40.65 kPa) were applied on the polyethylene
coated samples. These stresses correspond with lateral
stresses on a pile shaft induced by 4.5 m to 10 m depth of
backfill. Shearing stresses on the samples were applied at
a constant rate of deformation. For the purpose of this
study, three rates of deformation were used. These rates
varied from 0.015 mrn/min to 1 mrn/min. Also the samples
were tested at temperatures equal to 5C, 25C, and 40C.
At high temperature (40C), the samples were warmed by
continuously blowing hot air directed toward the
polyethylene sheets. At 5C, a typical sample was cooled
first and then placed in the shear box surrounded with ice
bags. The same procedure was followed in all the sets used
in testing.

To compare the results from the concrete-


polyethylene-soil samples with the shear resistance of
concrete versus soil without any friction reducer, some
concrete-soil samples were prepared and tested using both
types of soils. Results from these samples were taken as

690 Kamal Tawfiq


control data to determine the effectiveness of
polyethylene as a friction reducer.

3. Rod Shear Test

The rod shear test was devised in the laboratory to


simulate the negative skin friction induced by the
relative movement between the pile and the soil at very
low rates of shear deformation (p < 0.015 mm/min). A
schematic of the test is shown in Figure 3.

LOADING
RING CONCRETE
ROD

VERTICAL
LOAD
STEEL
MOLD

POLYETHYLENE SOIL
SHEETS SAMPLE

DISPLACEMENT
GAUGE
LATERAL
STRESS

STEEL BASE
STEEL
MOLD

Figure 3 Schematic of Rod Shear Test


In this test, a concrete rod of 30 cm in length and
3 cm in diameter was first prepared and wrapped with
different arrangements of polyethylene sheets similar to
those of the direct shear test. The rod was placed
vertically in a 15.2 cm diameter and 17.8 cm high steel
mold. A hole of 3.7 cm was drilled in the base of the mold
so that the rod could advance through during testing.
After centering the rod in the mold, the soil sample was
placed surrounding the coated portion of the rod. The soil
samples used in the rod shear test were the same as for
the direct shear test. The setup was placed on a loading
frame, and a predetermined vertical load was then applied
to the soi 1 surface to produce lateral stresses which
ranged from 2 kPa to 5 kPa. Vertical loading was applied

691 Kamal Tawfiq


on the concrete rod under controlled rate of displacement.

Discussion of Results

Shearinq Behavior of Unlubricated Polyethylene Sheets

The obtained results showed that when polyethylene


sheeting method was employed, the effectiveness of a
single 1 rom polyethylene sheet in reducing friction
resistance was about 60% to 64% depending on the type of
soil. This ratio was related to the friction coefficient
of the concrete-soil samples without any friction reducer.
For a single 0.15 rom polyethylene sheet, the effectiveness
of the polyethylene sheet was about 67% for the sand
samples and 62% for the coarse-grained soil. When two free
layers of polyethylene were used, the effectiveness
increased to 77% in both soils (Figure 4).
100 - - - r r - - - - - - - - - - - - - - - - - - - - - - n
Polyethylene Sheets Arrangement
*
80

S
I Lay... e mil. ~ C

~ 1 Layer, 15 mil, !5 C
I Layer, 40 mil, ~C
+
2 lJty.... emil. 2S C

I LAyer. 40 mil, 40 C

+CDnctele.sand
.2
. . 2 Lay


Lay
emil. eC
emil. 40 C

Con creteUm eatone


.6. I LaY8f, 40 mil. S C
* I Layer. emil. 40 C

--.
ctl
a..
~
60
(J)
(J) Concrete-Sand

-Q)
'-
( J)
'-
ctl 40
Q)
..c
(J)

20

o 10 20 30 40 50 60 70 80
Normal Stress (KPa)

Figure 4 Friction Resistance of Different Polyethylene


coating Arrangements.

Changes in temperature and rate of displacement did


not appreciably affect the shearing behavior of the
concrete-polyethylene-soil samples. An increase on the
order of 12% of the shear resistance was noticed when the

692 Kamal Tawfiq


temperature increased from 5 C to 40 C on the single 0.15
mm polyethylene sheet samples. This increase was slightly
higher when the samples were sheared with coarse-grained
crushed limestone. In the two 0.15 rom sheet samples, the
variation in the shear resistance was about 10%.

The effect of shear rate was found to be


insignificant (Figures 5 to 7). Increasing the rate from
0.0025 mm/min to 1 rom/min caused about 12% increase in the
shear resistance. This observation was typical for all the
arrangements of the polyethylene sheets used in testing,
except for the two 0.15 rom sheets. In that arrangement,
the shear resistance increased by about 8% for the range
of shearing rate used in the direct shear test.

100 ----r--~-.- . - . - .-.- . - . - - - - .-.-.-.-.-.---~r_---------,


Norma! Stress II< Temp.
X 18.85 kPa, 25 C
X 27.4 kPa, 25 C

-
cu
c..
~ 10
Unlubricated Double LayerO 15mm ' , , ,
Polyethylene Free Sheels ,., ,., .
*40.1 kPa, 25 C
+ Lubrloatad Shaata, 40C
Lubrlcatad Shaata, 25C

en
en
w
a:
ti .' _ I. '_' J '.' '_ l I. '.' J I

a:
LiS - - - , I
- .
I
- - - -
I I I I

J: ~ ~=-!-' :-: ~.~, ,-- , , 1-' ,., - ,- i ,',;,

en
~- - _.
".- '~.'_'
- . _.
'-- - - -I
- I - I - , ." "- - - - - - .. -

- - . I
- ,
..I '"
- - It
- Lubricated Double Layer-O 15mm ..
Polyethylene Free Sheets

0.1
0.001 0.01 0.1 1 10

SHEAR RATE (mm/min)

Figure 5 Shear Stress vs. Shearing Rate of Concrete-


Polyethylene-soil Samples; 0.15 rom Two-Free Layers

693 Kamal Tawfiq


Nprm,! SIr, & Tlmp
X 18.85 kPII, 25 C
X 27.4 kPII, 25 C
Unlubricated Single LayerO, 15mm
*'+
-
Polyethylene Fixed Sheet 40.1 kPII, 25 C
Lubricated She.t., 40C
ca
a. Lubricated Sheete, 2:1C

-
~

en
en
10

. ,.
Lubricated Sheet ~ C

.- ,-"
w ,- i ,., i i i

a::
l-
en - - '. '_ I .' '.' '_ _ _ _. _'. I ." '.' ~ I _

a::
<'C - '.

.--
J .L '. I l' ~ '. '.'.' '.".
w
:I: - . ' . ' . ' '.' '-'-
en
., - 1- ," ,., ", ,., ,- - ,- ., -, - I ., .. , r, - r ,

'. Lubricated Double LayerO,15mm


Polyethylene Free Sheets

0.1
0.001 0.01 0.1 1 10
SHEAR RATE (mm/min)
Figure 6 Shear Stress vs. Shearing Rate of Concrete-
Polyethylene-Soil Samples; 0.15 rom One-Fixed Layer.
100 -r"":""'""':"--:-:-:-""':'"""::-"'7""':""":--:-:-:--:-:--:-"':"'""C-__- - - - - , . . . - - - - - - - - ,
,.". - I - : I : ': I : I ! ' ~ I :
Normal Stress & Temp.
- - - - - - . X 18,85 kPa, 25 C

UnlUbrlcated Single Layer-' mm I':' ,. ~ X 27,4 kPa, 25 C

I
-
cu
a.
Polyethylene Fixed Sheet ' ; ,~

~~
..

~
, .' -' .'
+
'*' 40.1 kPa, 25 C

Lubricated SheQts, 40C

-
: : , " _.;;, : : : Lubrlceted SheQts, 2SC

c ~",:,~,:-:-:"'7";':,:"'7,-:::,"'7,:'7":,7",:-~
!".",~::,...,:-:-:-:-::,-::-:,~:"","il'!.,...,.. ,.~. ,.,:-~'-::~:"\''''':~',~
Y"':!:,....,.-~
~
10 - -: "'7 Lubri cat"d Shu's. 5 C
,. - ,. _ - - , - - ., i,"'i.- '\:." .,.,",
en r"",. r ,_,.,,"" ..co..,. _,.,.,,:"_, ,,,,' . ...:., _,." ,_"','
en
w
a::
'- "'" ' . " '" -'"..... '. .' - ':"'.' ~ ,: ' . ' ..... ' -' -' : '."~

l-
en - - - I _ I _ I ~ ~ I

a::
<'C
W
:I:
en
1 - f-.

s,,:;-;~:.
,."" , '~' '"
-C-1:T!
:"L> ~
,-,', ,', . - . ,. -:: ::-( L~i~i~~~lin~o~~i~'~a~i~-t~ ~~ f: :
0.1 - t - - - - - - - + - - - - - - - + - - - - - - l - - - - - - - . . j
0.001 0.01 0.1 10
SHEAR RATE (mm/min)

Figure 7 Shear Stress vs. Shearing Rate of Concrete-


polyethylene-Soil Samples; 1 rom One-Fixed Layer.

694 Kamal Tawfiq


Shearing Behavior of Lubricated Polyethylene Sheets

To gain more from the reduction in the skin friction


of the two polyethylene layers, samples of this particular
setup were provided with two 0.15 mm oil-lubricated
layers. A small amount of mineral oil was sprayed on one
of the sheets to ease the friction between layers. The
samples were tested in the direct shear test apparatus at
5 C, 25 C, and 40 C and subjected to normal stresses of
18.85 kPa, 27.0 kPa and 40.60 kPa. The rates of
deformation, p, were varied from 0.015 rnm/min to 1. 0
mm/min. Similar setups were used in the rod shear test,
except that all the samples were tested at 25 C and
subjected to lower normal stresses. The rates of
deformation in the rod shear test were 0.015 rnm/min and
0.0025 mm/min. The shear resistance at higher normal
stresses was extrapolated from the obtained results and
was found to correspond to that obtained from the direct
shear test. Testing results of these samples indicated
that the presence of a lubricant substance between the two
layers facilitated the sliding of the top layer and
decreased the shearing resistance to the lowest value
possible in the two layer arrangement.

The effect of normal stress, on' on the shearing


behavior of the lubricated sheets was found to be
insignificant (Figure 8). When the results were plotted in
Mohr-Coulomb domain, the shearing behavior of the samples
was found to be characterize~ by straight lines with a
very low amount of friction.

3--,--------------------r--------, Tlmp & Sh,Br Rptl


Normal Stress
-021 C, 0.011 mmtmln
2.5 Two oil-lubricated' .. ~ 25 C, 0.05 mmlmln

-"' rl:::=::::h sheets + 25 C, 0.25 mmfmln


.25 C, 1 mmfmln

-
~
Ul
Ul
f1.5
2

' - - '..
in
... Lubricated Polyethylene Sheet
"'
lSI
.c
en
0.5

o 5 10 15 20 25 30 35 40 45 50 55 60 65 70
Normal Stress (KPa)

Figure 8 Shearing Behavior of Two Lubricated Polyethylene


Layers at Different Shearing Rates.

695 Kamal Tawfiq


Nonetheless, the shear resistance in the case of
polyethylene sheets lubricated with oil was considerably
less than the resistance of the unlubricated layers.

The effect of the shear rate, p, on the shearing


characteristics of these samples was more pronounced than
on those without any lubrication. Depending on the shear
rate and the ambient temperature, lUbricating the
polyethylene sheets with oil reduced the shearing
resistance of the concrete-polyethylene-soil samples by
80% to 98%. At low shear rates (p < 0.0025 mm/min) and
high temperatures (40C), the lubricated polyethylene
sheets exhibited the lowest shearing resistance. At these
limi ts, the effectiveness of polyethylene sheeting in
reducing the friction resistance reached 98%. As the
temperature decreased to 5C and the rate of
deformation, p, increased to 1 rom/min, the shearing
resistance of the lubricated polyethylene sheets increased
and the effectiveness decreased to 80%. The effectiveness
of the polyethylene sheets in reducing skin friction at
the room temperature (25C) and shear rate of 0.0025 mm/min
was about 94%. (Figures 5 to 7).

The shearing resistance in the lubricated


polyethylene sheets can be estimated from the shearing
rate of the sheets and the viscosity of the lubricant
material between the sheets and can be expressed as:

(1)

in which m = the shear resistance at the reference shear


rate; n = the slope of the log~ vs. logp relationship; Po
=the reference shear rate chosen arbitrarily to be 1 x 10- 3
mm/min. m and n parameters are determined as follows:
m = 0.7 10-0.0065 (Tl (2)

and
n = 0.09 (3)

SUbstituting Eqs. (2) and (3) into (1) gives


L = 0.7 10-0.0065(T) (~)0.09 (4)
S
Po

Equation (4) can be used to determine the shearing


resistance of the two lubricated polyethylene sheets and
hence the amount of downdrag force that may be induced on
piles. It is expected that other values of parameters m
and n may be obtained for different lubricants. However,

696 Kamal Tawfiq


the variation in these values might not be very
significant, especially for the n parameter which has only
a small effect on the shear resistance, 1: s ' compared to the
m parameter.

Summary and Conclusion

The aim of this study was to conduct a laboratory


investigation to evaluate the effectiveness of
polyethylene sheeting in reducing negative skin friction
on piles. It is evident from the test results that using
the polyethylene sheeting method significantly reduces the
shearing resistance between the soil and the concrete
surface. The two 0.15 rom polyethylene sheets reduced the
shear resistance, 1: s ' by about 77%. Further reduction was
achieved by lUbricating the two sheets with mineral oil.
Depending on the temperature and the shearing rate, this
arrangement reduced the shear resistance by about 98%. The
tested samples showed no signs of damage to the normal and
shear stresses used in the study. Using polyethylene
sheets as a friction reducer in piles eliminates some of
the problems associated with the conventional bitumen
coating method. Disturbance of the bitumen coating due to
temperature and soil particle penetration is unavoidable.
However, in polyethylene sheeting, these problems do not
exist providing the proper selection of the polyethylene
sheets and the method of construction. Moreover, the
installation process does not require extra precautions as
in bitumen coating method. For instance, cantilever piles
behind MSE walls can easily be wrapped with more than one
layer of polyethylene sheets without any consideration for
the changes in temperature that may distort other coating
materials.

ACKNOWLEDGMENT

Financial support was provided in part by the Florida


Department of Transportation, and the Department of Civil
Engineering at FAMU/FSU College of Engineering. This
support is gratefully acknowledged.

REFERENCES

Baligh, M. M., Figi, H., Vivatrat, V., and Azzouz, A. S.,


(1981) "Design of Bitumen Coating to Reduce Downdrag on
Piles", Research Report R80-42, Department of Civil
Engineering, Massachusetts Institute of Technology,
Cambridge, Mass.

Kamal Tawfiq
697
Busby E. o. and Rader L. F., (1972) "Problems in
Determining Viscosity of Asphalt at Low Temperatures with
Shell sliding Plate Microviscometer," Seventy Fifth Annual
Meeting, American Society For Testing and Materials,
Special Technical Publication 532" Los Angeles,
California.

Bush, R., K., viswanathan, R., Jeong, S., and Briaud, J.,
L., (1991) "Downdrag on Bitumen-Coated Piles," Preliminary
Draft Interim Report for the National Cooperation Highway
Research Program, Transportation Research Board.

Corbett L. W., and Schweyer, H. E., (1972), "Viscosity


Characteristics of Asphalt Cement," seventy Fifth Annual
Meeting, American Society For Testing and Materials,
special Technical Publication 532, Los Angeles,
California.

Tawfiq, K., and Ping, W., (1992), "Effectiveness of


Polyethylene vs. Bitumen Coating on Downdrag Forces,"
Research Report WPI-0510628, C-3976, Florida Department of
Transportation, Tallahassee, Florida. 1992.

698 Kamal Tawfiq


CORRELATIONS BETWEEN THE STANDARD PENETRATION TEST (SPT)
AND THE MEASURED SHEAR STRENGTH OF FLORIDA NATURAL ROCK

Hernando R. Ramos, 1 Member, ASCE, Juan A. Antorena, 2, and


G. Thomas McDaniel, 3 Member, ASCE

ABSTRACT: The Standard Penetration Test (SPT) is the most common


method for field testing of soil and natural rock in Florida. The most
appropriate field sampling for natural rock is rock coring since it allows the
retrieval of intact sections of rock for observation and laboratory strength
testing. However, the market has forced the relatively costly rock coring
to be very limited on small to medium geotechnical projects. Data obtained
from limited rock coring and strength testing have been supplemented by
SPT correlations. Even large projects rely on a relatively small amount of
rock coring data for the geotechnical evaluation of deep foundations.
Therefore, it is very important to attempt to establish meaningful
correlations between the measured side shear strength of the Florida rock,
in particular with the South Florida limestones and sandstones, with the SPT
values.

In order to prepare these correlations, data from over 30 grouted anchor


pullout tests and several static pile load tests performed in rock were
analyzed. The results show a defined trend which establishes an acceptable
correlation between the measured rock side shaft resistance and the SPT
values. These correlations serve as an effective geotechnical tool to predict
preliminary design compressive and tensile capacities of drilled piles.
Instrumented static load tests, loaded preferably to failure, combined with
anchor pullout tests, should be included in the project recommendations to
confirm the rock shear strength used in design.

I Principal Geotechnical Engineer, Law Engineering and Environmental Services, Miami,

Florida
2 Project Geotechnical Engineer, Law Engineering and Environmental Services, Miami,

Florida
3 Chief Engineer, Law Engineering and Environmental Services, Miami, Florida

H. Ramos et al.
699
INTRODUCTION

The use of drilled piles (drilled shafts, augercast piles) is very common worldwide,
specially in Florida. In Florida for instance, drilled shafts are mainly used by The Florida
Department of Transportation and in less degree by the private sector for high-rise projects.
Augercast piles are mostly used by the County Public Work Departments and for
commercial developments.

Typical drilled pile design relies substantially on the side shear strength of the
natural rock and to some extent on the end bearing. However, the end bearing component
of the design (if any) is generally considered a minimal portion of the overall carrying axial
capacity of the piles due to uncertainties associated with the questionable quality of the rock
below the design pile tip and to the difficulty in providing a clean pile bottom prior concrete
or grout placement.

Fully instrumented static pile load tests and anchor pullout tests are the most reliable
methods for design of drilled piles in Florida, however these tests are very expensive. Static
pile load tests are used to confirm the design loads that have already been estimated using
other methods.

This paper presents an acceptable correlation between the ultimate rock side shear
strength with SPT N-values in blows/.3 m (blows/ft.).

TYPICAL TESTING USED TO ESTIMATE TIIE ROCK SIDE SHEAR TRANSFER

The design of drilled piles is based on field and/or laboratory test data. Typical
correlations to estimate the rock side shear transfer include unconfined compressive and
splitting tensile strength test data, anchor pullout tests, and SPT data.

In Florida, anchor pullout tests and laboratory strength testing on rock are mainly
performed for state and federal projects. In the commercial arena, anchor pullout tests are
practically not used and laboratory strength tests on rock are generally very limited to tall
high-rise structures. The SPT data is by in large the most common method used in Florida
to estimate rock strength.

DISCUSSION OF TIIE STANDARD PENETRATION TEST MEmOD

The Standard method ASTM D 1586 describes the procedure for driving a slit-spoon
sampler into the strata to obtain representative soil samples and measure the soil resistance
during the penetration of the sampler.

As described in the standard, the method was developed for testing soil, however,
it has been adapted to test the weak rock in Florida. The test results may deviate
significantly due to faulty equipment and lack of operator skills. In addition, variation in
N-values can be produced by using different drill rigs and hammer types. Substantial
differences (up to 40 percent) of N-values have been observed at the same site when using
both automatic and safety hammers.

700
For N values less than 60 blows/.3m, which corresponds to the soft to moderately
hard South Florida limestones and sandstones, the hammer type will show a substantial
difference. However, for hard to very hard rocks (50 blows/ .15m or greater) the difference
is not as important since rock coring is a more appropriate method for retrieving rock
samples for laboratory testing.

The majority of the N values used in this paper for correlation were obtained using
a safety hammer. Our experience with hammers in the South Florida area indicates that the
N values obtained using an automatic hammer must be multiplied by a factor of 1.4 to
convert them to safety hammer values. The safety hammer typically delivers less driving
energy to the SPT drilling rods, than the automatic hammer, thus causing high N values.

BACKGROUND

The data of 36 anchor pullout tests in rock were analyzed to develop the presented
correlations. Thirty four tests were performed to failure (until an excessive movement of
the cement grout plugs was achieved). Thirty two of these tests were conducted in South
Florida.

The diameter of the cement grout plugs ranged from 115 to 300 mm (4.5 to 12
inches). Some of the cement grout plugs were installed as shallow as 4.5m (15 ft.) and
others were as deep as 17m (57 ft.), measured from the ground surface. The length of the
grout plugs ranged approximately from 0.6 to 3.6m (2 to 12 ft.).

Furthermore, a total of 16 static pile load test results were included in the database.
Most of these tests were performed out of South Florida, and as we understand, were not
loaded to failure.

DATABASE DISCUSSION

Table 1 contains the results of the anchor pullout and static pile load tests. In
addition, available laboratory strength data was also included. Table 1 includes the following
information.

SPT Average Values - For the static pile load tests, SPT values were collected as an
average over the length of the test pile embedded in rock. For the anchor pullout tests, the
N values for the rock layer at the grout plug location was used. The N values were
obtained in a boring near or in the borehole where the cement grout plug was installed.

Observed Rock Side Shear Strength - The observed rock side shear strength was obtained
during the application of the axial tensile load, which was generally applied until failure of
the bond between the grout plug and the rock was achieved. The ultimate rock side shear
strength is estimated by dividing the total applied load by the shear surface of the cement
grout plug. It is very important to be able to pull the plug out of the ground in order to
better estimate the shear surface area. Anchor pullouts performed in South Florida rock
were estimated separately from the ones performed in the rest of Florida.

701
TABLE 1- DATABASE

SP I O~SlKVW UBSlKVW ,PLlI IINL


PROJECT TYPE AVERAGE SfI[AR STRENGTH SHEAR STRENGTH SHEAR STRENGTH RESIDUAL SHEAR COMPRESSIVE TENSILE REFERENCE
NAME OF N-VALUES PULLOUT TESTS PULLOUT TESTS FROM LOAD TEST STRENGTH FROM STRENGTH STRENGTH
TEST SOUTH FLORIDA REST OF FLORIDA PULLOUT TEST
Blows/.3 m kP, 1<>1 kP, 1<>1 kP, luI kP, I<> kP, I<>r kP. I<>

METRORAIL PULLOUT 10 512 10-7 40 .1 22 I 46_6 43 I 9.0 IA


METRO RAIL PULLOUT II 675 14_1 546 I L4 2231 46.6 43 I 9_0 IB
METRORAIL PULLOUT 13 474 9_9 345 7.2 2231 46_6 43 I 9_0 IC
METRORAIL PULLOU 15 651 13_6 445 9. 22.51 46_6 431 9.0 ID
METRORAIL PULLOUT 24 402 B_4 321 6.7 2231 46.6 431 9_0 IE
METRORAIL PULLOU 27 694 14.5 -610 1J-7 1/24 36_U 575 12.0 IF
METRORAIL PULLOUT 2B 60B 12-7 1724 36.0 575 12_0 IG
METRORAIL PULLOUT 33 694 14.5 637 13.3 2231 46.6 431 9.0 IH
METRORAIL PULLOUT 28 632 13.2 1724 36.0 575 12_0 II
METRO MOVER PULLOUT 21 6B7 14.4 594 12_4 1120 23.4 2A
METROMOVER PULLOUT 25 436 9.1 421 8.B 1149 24-0 0 6.9 2B
METROMOVER PULLOUT 29 603 12.6 555 I L6 2C
METROMOVER PULLOUT 3I 455 9.5 244 5_1 795 16.6 2D
METROMOVER PULLOUT 3I 594 12.4 450 9.4 1044 2LB 2E
METROMOVER PULLOUT 36 583 12.2 517 10_8 1120 23.4 249 U 2F
DADELAND PULLOUT 18 958 20.0 718 IS_O 4453 93.0 3447 72.0 JA
DADELAND PULLOUT 70 2298 48_0 1915 40_0 4405 92.0 1149 24_0 JB
DADELAND PULLO T 100 1173 24.' 1044 2L8 14412 30LO 160 66.0 .lC
FLORIDA KEYS PULLOUT 42 1053 22_0 4A
-...J FLORIDA I(E S PULLOUT 42 1341 28.0 4~
o FLORIDA KEYS PULLOUT 55 1149 24.0 4C
IV
FLORIDA KEYS PULLOUT 55 1341 28.0 4D
FLORIDA KEYS PULLOU 55 2298 48.0 4E
FLORIDA KEYS PULLOUT 63 1772 37_0 4F
FLORIDA KEYS PULLOUT 6 2011 42.0 4G
FLORIDA KEYS PULLOUT 63 2586 54.0 4H
FLORIDA KEYS PULLOUT 100 2729 57.0 41
FLORIDA KEYS PULLOUT 100 4118 86.0 41
FLORIDA KEYS(LT) LOAD TEST IS 144 3.0 4K
tLORIDA KEYS(L T) LOAD TEST 18 192 4_0 4L
FLORIDA KEYS(L T) LOAD TEST 41 862 18_0 4M
tLOKIDA KEYS(LT) LOAD TEST 48 1197 25.0 4N
FLORIDA KEYS LT LOAD TEST 54 1628 34_0 40
MIAMI BEACH (L T) LOAD TEST 20 203 43 SA
MIAMI LT LOAD TEST 89 900 18.8 6A
GAINESVILLE PULLOUT 38 622 13.0 5267 110.0 565 11.8 7A
GAINESVILLl PULLOUT 46 752 15.7 752 15_ 2595 54.2 5:16 IL2 8A
1595 PULLOUl 28 852 17.8 503 10.S 1101 23.0 670 14.0 9A
1-595 PULLOUT 10 397 8.l 335 7.0 1101 23_0 670 14.0 98
1-595 PULLOUT 29 608 12-7 589 12.3 1101 23.0 670 14_0 9C
1-595 PULLOUT 36 670 140 273 5.7 110 I 23_0 670 14.0 9D
JACKSONVILLE LOAD TEST 30 479 10.0 1724 36.0 lOA
TAMPA LOAD TEST 60 393 8.2 2298 48.0 326 6.8 IIA
jACKSONVILLE(LT) LOAD TEST 55 119 25.0 6703 140.0 12A
CLEARWATER PULLOUT 100 1494 31.2 9768 204.0 1034 21.6 13A
CLEARWATER PULLOU r 100 1570 32B 8044 168.0 1187 24.8 14A
TAMPA LOAD TEST 100 1532 320 15A
JACKSONVILLE LOAD TEST 40 862 18.0 15B
AMPA LOAD TES SO 646 13.S 15C
TAMPA LOAD TEsT 50 431 9_0 150
ALLAHASSEE LOAD E, 100 802 18.0 1st
ACKSONVILLE LOAD TEST 100 1436 30.0 15F
Observed Rock Side Shear Strength from Static Pile Load Tests - This rock side shear
transfer was estimated as a mean value over the length of the pile in direct contact with the
rock. Properly instrumented piles (with strain gauges) provide a good estimate of the actual
rock side shear transfer. Unfortunately, piles are seldom loaded to failure, especially in
South Florida, therefore, the average rock side shear values obtained from load testing are
not a good indication of the ultimate side shear capacity of the natural rock.

Observed Residual Rock Side Shear Strength - Immediately after achieving the maximum
shear strength during anchor pullout testing, continued pullout causes the plug-rock system
to assume a residual shear resistance. Table 1 shows several observed residual rock side
shear strength values estimated from anchor pullout tests performed in South Florida. No
data was available from test in the rest of Florida.

Unconfined Compressive and Splitting Tensile Strength Tests - These tests were performed
on cored samples retrieved from or near the boreholes where the cement grout plugs were
installed. These laboratory strength tests are generally performed on the best rock samples
retrieved. The results do not represent the average in-place strength properties of the rock.
The test results must be correlated with the properties determined from the field such as
percent recovery and RQD.

DISCUSSION OF CORRELATIONS

RESIDUAL SHEAR STRENGTH VS. OBSERVED SHEAR STRENGTH

Figure 1 presents the correlation between the residual and the observed ultimate
shear strength obtained from anchor pullout tests. This database consists of 21 anchor
pullout tests with available residual shear strength information. The residual shear data
range from 244 to 1915 kPa (5.1 to 40 ksf). A best fit line was drawn showing the
following correlation:

f, = 0.8 fu (1)

where,
f, = residual shear strength, kPa (ksf)
fu = observed ultimate shear strength, kPa (ksf)

UNCONFINED COMPRESSIVE STRENGTH VS. OBSERVED ROCK SIDE SHEAR STRENGTII

Reynolds et. al (1980) obtained correlations of the South Florida Oolitic Limestone
in the order of fJqu = 0.3. The Oolitic Limestone is a weak rock mainly encountered in
Miami, extending to a depth ranging from 4.5 to 7.5 m ( 15 to 25 ft.). Gupton and Logan
(1984) recommend a fJqu ratio of 0.2 after evaluating several static load tests installed in
several rock formations (Key Largo, Anastasia, Fort Thompson and Miami Formations) in
South Florida.

Figure 2 presents the correlations between unconfined compressive strength and


observed shear strength. The values of the unconfined compressive strength range from 795
to 1724 kPa (16.6 to 36 ksf).

703
--------~~---_._--- -------- - ----- ----------..
1- Residual Shear Strength Vs. Observed Shear Strength
Figure 1

[ RESIDUAL SHEAR STRENCTH, fr IkPal I


~. 240 480 720 960 1200 1440 1680 1920 I
50 2400

2000
40

;;: Ii
III D..
~ 1600 ~
...
:J
...
:J

-..J ... 30 ..... , ... ,." ......


- ...
0
~ I 0
Z
w
-_. 0
zw
...""en --
1200 ...""
III
""c(w --- '"
:r ""w
c(
en :r
.. 11 .................. .... -.............
Q
w
20
, '"
Q
> - II ~
""w 800 ""w
'"'" <)
'"0m
m
0 III ... IIII1B1B111

10 . IilotIf
II I 400

o a
a 5 10 15 20 25 30 35 40
RESIDUAL SHEAR STRENCTH. fr IkSfl

III Pullout - South Florida o Pullout - Rest of Florida A Load Test - 0.8 fu
Unconfined Compressive Strength Vs. Observed Shear Strength
Figure 2

20 r----------------- -------575----------------_
C-_U-_ncon~~d_c_o_m_p_r_e~s_i_ve_-~ten!~_~~~,!p_a_l-_J
--,cc,ccll-=-g----- 17~60

..

.-
15 ..... ~ ............ -l 720

;; ii
VI Q.
:!! :!!
...::r III
I11III
II
-'
/
...::r
......
r: r:
tl. tl.
0
I .
s::
......
....
c
(J1

10 480

.... ....
III Iii III
.c III
.c
'a 'a

. ...
~
~
VI
-- >
VI
J:I , J:I
/
0 0
. / :

5 . ,,,;. . .. .~ -l 240

o o
o 12 24 36
Unconfined compressive strength, qu (ksfl

[ l1li pullout - south FlorIda ~ 0.35 QU _ 0.5::"qu _ J


For the softer rock (qu < 1800 kPa, 36 ksf) the fJqu ratio has been estimated as:

best fit line (2)

lower bound, (3)

where,
fu = ultimate shear strength, kPa or ksf
qu = unconfined compressive strength, kPa or ksf

Figure 3 shows a correlation between unconfined compressive strength and observed


ultimate shear strength for unconfined compressive values ranging from 1724 to 9768 kPa
(36 to 204 ksf).

For harder rock ( qu > 1800 kPa, 36 ksf) the fJqu ratio was estimated as 0.12. A
substantial amount of data is required to provide a better correlation for rocks at this
strength level.

SPT VS. OBSERVED ROCK SIDE SHEAR STRENGTH

Figure 4 depicts the observed side shear strength obtained from anchor pullout and
static pile load tests in the State of Florida versus SPT N Values. Most of the anchor
pullout tests were performed to failure ( with the exception of two tests), therefore, ultimate
side shear strength values were obtained. None of the static pile load tests were loaded to
failure.

As shown in Figure 4, some of the lower values correspond to the observed rock
side shear strength obtained from static pile load tests (10 tests), anchor pullout tests
performed in South Florida (6 pullouts), and anchor pullout tests performed in the rest of
Florida (3 pullouts).

The field experience obtained by the authors indicates that the low anchor pullout
test results correspond to tests performed in near surface limestones having a relatively high
density (low porosity), therefore, the side shear transfer can be referred more to a "skin
friction" mechanism.

On the other hand, the observed upper values of shear strength correspond to a more
porous rock (and probably with more cavities). When these rock conditions are present, the
concrete or cement grout intrude into the cavities and porous rock zones, creating an
interlocking bond. Subsequently, the side shear transfer is mainly provided by a
combination of skin friction and the interlocking mechanism. The interlocking mechanism
is probably the major contributor of the relatively higher tensile capacity observed on anchor
pullout tests performed in these rock conditions.

It is well understood the difficulty of installing short, deep cement grout plugs in the
ground. Basically, the grout is placed without a "grout head" and/or pressure, therefore,
it is difficult to obtain a proper grout-rock bond.

706
Unconfined Compressive Strength Vs. Observed Shear Strength
Figure 3

----------- ------------ --------~

Unconfined Compressive strength, qu (kPal

E4-- 3456 5170 6895 8620 ---"0340J


40 1920

30 I- ... ~~
.. .. ~ 1440

..
;
~ ~
ii
Q.

...r.-
:::I
...
..
:::I

-...I ..
~

I ........
DI
0 DI

..........
c
C
-...I

. .
co
20 ............ ..... 111: ... 960
...co
r. -
... ...
r.
...
'a
-- - 'a
~
..
...
Q
.......
>
'II

.a .a
0 <> 0

IIIIl
10 ! - II 480
l1li&

o o
36 72 108 144 180 216
Unconfined Compressive strength, qu (ksfl

III Pullout South Florida o Pullout - Rest of Florida AI. Load Test 0.12 qu
----_._----

Standard Penetration Resistance Vs. Observed Shear Strength

I
Figure LI
100 1_ 4800
1------------ -- ------ - - - - - - - -
II1II Pullout - South Florida <) Pullout - Rest of Florida A Load Test _ Formula
. 4400

4000
80 1= 13600

;:;: ~
3200 'ii
Q.
'"
~ ~
.c
40' 60 .c
CI - 2800 ~

g! ....
s:: s::
QI
40'
III -
II1II e
l':I
2400 :l':I
QI II1II III QI
.c .c
III -- III
'tI
QI II1II 2000 ~
~ 40 ~
QI QI
III
'"
g'"
g
0 A 1600

1200
20 E -----II1II ---------~
AA
"" T
-f- 800
<)
to.
A 4. --t 400
to. ...
A
o II I I I I I I I I I I I I I I I I ! I I I I I I I I I I ! I I I I I I I I I I I I I! I ! ! ! I I! ! ! ! I I ! I I I I I I I! I ! ! I I I I ! ! ! I ! ! I ! I I ! I I ! I ! I I ! I I I I I I ! I I I 0
0 20 40 60 80 100
standard Penetration Resistance N-Values (blowS/O.3 ml
Production augercast piles are on the other hand, installed by pumping a sand-
cement water mix with typical pressures in the order of 2070-2400 kPa (300-350 psi) and
grout heads averaging 1.5 to 3m (5 to 10 ft.) and greater.

Drilled shafts are constructed with concrete. Concrete heads in the order of 1.5 to
3m (5 to 10 f1.) or greater are also typical.

Therefore, augercast piles may develop greater rock side shear transfer due to the
contribution of pumping the grout at high pressure when compared to drilled shafts.

Cement grout plugs, on the other hand, may predict less rock side shear transfer
since the grout is placed without a head or pressure. However, this in-situ testing method
may overestimate the in-place shear strength of the rock due to the size of the cement grout
plug (diameter, length) when compared to large diameter drilled piles and the size of the
cavities and porous zones of the rock formation. Therefore, anchor pullout test data used
to aid in designing large diameter drilled piles should be carefully analyzed.

SPT VS. SIDE SHEAR STRENGTH CORRELATIONS

Based on the data base presented in Table 1, the following correlations were developed:

Soft to Moderately Hard Rock - For N values ranging from 5 to 60 blows/.3m (blows/ft.),

fu = 192 + 19.2 Nave, in kPa or, (4)

fu = 4 + 0.4 Nave, in ksf (5)

for N values less than 5 blows/.3m (blows/ft.), the material must be considered as soil.

Hard to Very Hard Rock - N values greater than 60 blows/.3m (blows/ft.),

fu = 768 + 9.6 Nave' in kPa or, (6)

fu = 16 + 0.2 Nave, in ksf (7)

where,

fu = ultimate side shear strength, kPa (kst)


Nave = Average SPT value, blows/.3m (blows/ft.)

For N values greater than 60 blows/.3m (blows/ft.), the database does not show a
substantial number of insitu testing, therefore, correlations 6 and 7 should be used with
caution. Furthermore, correlations 6 and 7 should be used for preliminary design.

We do not recommend the use of rock side shear strength values greater than 1440 kPa (30
kst) obtained from these correlations unless those values are confirmed by anchor pullout
and/or instrumented static pile load tests at the site.

709
CONCLUSIONS

The Standard Penetration Test (SPT) is the most common method used in Florida
to estimate rock side shear strength. Therefore, the authors consider that it is very
important to attempt to develop correlations between the side shear strength of the Florida
rock with SPT N-values.

Data from 36 anchor pullout and 16 static pile load tests were used to develop these
correlations.

The database collected provide good correlations between the observed residual
shear strength fr with the observed ultimate shear strength fu. A best fit line developed with
the data available shows a correlation in the order of f/fu = 0.8. This ratio appears to be
appropriate for the porous South Florida rock. Rock formations in the rest of Florida may
show lower f/fu ratios.

Correlations between the unconfined compressive strength qu versus the observed


rock side shear strength fu presented a fJqu ratio of 0.5 for the best fit line and a fJqu ratio
of 0.35 for the lower bound, for rocks having qu less than 1800 kPa (36 ksf). For qu greater
than 1800 kPa (36 ksf), the fJqu ratio was estimated as 0.12.

The data provided by the anchor pullout and static pile load tests show a defined
trend which establishes an acceptable correlation between the observed rock side shear
strength fu and SPT N-values. Two correlations are provided; one for soft to moderately
hard rocks (for N values ranging from 5 to 60 blows/.3m), and the other for hard to very
hard rocks (for N values greater than 60 blows/.3m). The second correlation is
recommended to be used with caution since the database does not show a substantial number
of insitu testing. A proper factor of safety (2.5 to 3.0) should be used when using these
correlations.

REFERENCES

Crapps, D.K. (1986). "Design, construction and inspection of drilled shafts in limerock and
limestone." Proc. Annual Meeting of Florida Section, ASCE. (Ref. 4A thru 40)
Gupton. c., and Logan, T. (1984). "Design guidelines for drilled shafts in weak rocks of
South Florida." Proc. South Florida Annual ASCE Meeting, ASCE.
Law Engineering and Environmental Services, (1980), "Report of a Subsurface
Investigation, Metropolitan Dade County, Transit Improvement Program, Metrorail, Line
Section 7". (Ref. lA, lB, lC, lD, IE, lH)
Law Engineering and Environmental Services, (1980). "Report of a Subsurface Investigation
Metropolitan Dade County, Transit Improvement Program, Metrorail, Line Section 4" .
(Ref. IF, lG, 11)
Law Engineering and Environmental Services (1980). "Report of Drilled Shaft Foundation
Installation, Cement Storage Silos, Pennsuco Cement Plant". (Ref. 6A)
Law Engineering and Environmental Services, (1984). "Report of Anchor Installation
Testing Program, Florida Department of Transportation, 1-595 Interchange at U. S.
441". (Ref. 9A thru 9D)

710
Law Engineering and Environmental Services, (1990). "Report of a Geotechnical
Exploration, Metropolitan Dade County, Metromover North Extension". (Ref. 2A thru
2F)
Law Engineering and Environmental Services, (1990). "Report of a Geotechnical
Exploration, Florida Department of Transportation, Dadeland North Parking Garage".
(Ref. 3A, 3B, 3C)
McMahan, B. (1988). "Drilled shaft design and construction in Florida". Dept. of Civ.
Engrg., Univ. of Florida, Gainesville, Fla.
McVay M.e., Townsend F.e., (1992). "Design of Socketed Drilled Shafts in Limestone",
Journal of Geotechnical Engineering, Vol. 118, No. 10. (Ref. 7A, lOA thru 14A).
Reynolds, R.T., and Kaderabek, T.J. (1980). "Miami limestone foundation design and
construction." ASCE, New York, N.Y.
Schmertmann, J.H. (1977). "Report on the development of a limerock tension shear test to
guide drilled shaft foundation design of the Florida DOT Keys bridges project." Final
Report to FDOT, Florida Department of Transportation, Tallahassee, Fla.

711
AN EVALUATION OF PREDICTED ULTIMATE CAPACITY OF SINGLE
Pll..ES FROM SPll..E AND UNIPll..E PROGRAMS

Swamy Kumar V. Avasarala l , M.ASCE, J. L. Davidson2 , M.ASCE, M. C. MCVay 2,


A.M.ASCE

ABSTRACT: This is a summary of results from the evaluation of two of the computer
design programs available (SPILE and UNIPILE), which predict ultimate capacity of
axially loaded prestressed concrete piles based on in-situ test data. Over the past few
years a number of deep foundation load test databases have been developed by the
professors and graduate students of the Geotechnical Engineering Group at the
University of Florida. The most recent database was designed using LOTUS 123
version 3.1. The database provides a record of pile load tests and allows a comparison
of predicted ultimate capacities from SPT91, SPILE and UNIPILE programs and
ultimate measured capacities. Currently there are 62 piles in the database of which
only 24 piles were tested to ultimate capacity and were used to compare with the
predicted capacities from SPILE and UNIPILE programs. The analysis of the results
leads to conclusions on the precision of predicted ultimate capacities from SPILE and
UNIPILE programs. The scope of the study was limited to the performance of
individual piles and no consideration was given to the group behavior of pile
foundations.

UNIVERSITY OF FLORIDA Pll..E DATABASE

The University of Florida pile database (UFPILE) was developed using LOTUS 123
Version 3.1 (3-dimensional). This pile database contains a total of 62 pile load tests.
Each spreadsheet in the database is devoted to one pile load test. The pile load test
contains information of SPT and/or CPT data, soil data, pile data, load-settlement data,
and pile driving data. Of the 62 piles in the database only 24 could be used in this
comparison because the rest of the piles were not loaded to ultimate capacity. Of the
62 piles in the database only 51 piles have failed according to the Davisson criteria.
These piles were used to compare the ultimate skin friction between different programs.

I _ Senior Staff Engineer, Alamo/Saxena Consultants, Inc., 5675 New Tampa Hwy.,
Suites 1-3, Lakeland, FL 33801
2 _ Profs., Dept. Civil Eng., 345 Weil Hall, Univ. of Florida, Gainesville, FL 32611

Avasarala
712
Eleven piles were not included in the analysis because they did not fail according to
Davisson criteria. Even though some of the groups of piles were too small for
meaningful statistical analysis, they are included and discussed. As more piles are
added to the database these analysis will become more significant.

SPILE PROGRAM

The SPILE program is a microcomputer program for determining the ultimate vertical
static pile capacity. The program is based on methods and equations presented by
Nordlund (1963,1979), Thurman (1964), Meyerhof (1976), Cheney and Chassie (1982),
and Tomlinson (1979,1985). The program is coded in Turbo Pascal 4.0 for the IBM-
pc through the use of friendly input menus and data checking routines. The code
implements portions of the microcomputer program SAF-I developed by PROTOTYPE
Engineers, Winchester, MA.

UNIPILE PROGRAM

UNIPILE is a program used for the design of individual piles and pile groups. It
considers bearing capacity, settlement, and negative skin friction based on the Fellenius
Unified Method. The program includes aspects of drivability and residual stresses.
Load Transfer can optionally be made using effective stress analysis and only friction
(beta-method) or friction and effective cohesion (c'). It can also be made using total
stress analysis and undrained shear strength (alpha-method). The program also does
settlement analysis based on Janbu's tangent modulus principle using modulus numbers
and stress exponents. The calculations can also be based on E-moduli for sands and
conventional coefficients of consolidation and void ratios. The program allows for
consideration of overconsolidated soils and its value can be expressed in terms of either
overconsolidation ratio (OCR), or in terms of a constant value of stress added to the
existing effective stress. The program allows soil profiles made up of upto 15 different
soil layers. Provision for fill or excavation is also included in the program. The
program accepts piles of round, octagonal, hexagonal, and square shape.

SPT91 PROGRAM

The SPT91 computer program is used to estimate the static axial capacity of driven
piles. The methodology is based on Research Bulletin 121 (RB-121), "Guidelines for
use in the Soils Investigation and Design of Foundations for Bridge Structures in the
State of Florida", prepared for the FDOT by Dr. John Schmertmann in 1967. The
program is used primarily for analyzing concrete piles although plugged steel pipe piles
may also be accommodated. The analysis has been modified in recent years based on
load test database results compiled for the FDOT by the University of Florida. The
program requires the Standard Penetration Test "N" value to calculate the static axial
capacity of piles.

713 Avasarala
EVALUATION OF COMPUTER PROGRAM PREDICTIONS

Comparisons between the SPILE and UNIPILE ultimate pile capacities and the SPT91
and load test capacities was performed by statistical analysis of the ratio of the .
predicted ultimate load over the measured ultimate load indicated as "p". A perfect
prediction would have a mean of one (p = 1, i.e points falling on 45 0 line with predicted
capacity on Y-axis and measured capacity on X-axis), a standard deviation of zero and
a error of estimate of zero.

Figures 1 and 2 show that the SPILE and UNIPILE programs overpredict the ultimate
capacity compared to measured ultimate capacity from load tests, for most piles in the
database. Table 1 presents a summary of predicted and measured capacities for the 24
piles used in the above comparison. This table shows the pre-bored depth, width,
length/width ratio, pile soil number, measured ultimate load test capacity and the p-
ratio, for the 24 piles for SPILE, UNIPILE and SPT91 programs. The data for SPT91
program was obtained from the pile database developed by Pedro Ruesta, Graduate
Student, University of Florida.

The twelve combinations of soils used for the comparison according to pile soil number
were:

Clay / Clay (11) Silt / Clay (21) Sand / Clay (31)


Clay / Silt (12) Silt / Silt (22) Sand / Silt (32)
Clay / Sand (13) Silt / Sand (23) Sand / Sand (33)
Clay / Rock (14) Silt / Rock (24) Sand / Rock (34)

where the first integer of the soil type number indicates the predominant shaft or
friction soil, while the second integer indicates the pile tip or bearing soil.

An p-ratio greater than 1 is considered as unconservative i.e the predicted capacity was
larger than the measured capacity. An p-ratio less than 1 is conservative i.e the
predicted capacity was less than the measured capacity. The maximum and minimum
values of the p-ratio for SPILE were 5.55 (unconservative) and 0.28 (conservative) for
Pile #s 28 and 27 respectively. The maximum and minimum values of the p-ratio for
UNIPILE were 6.66 (unconservative) and 1.11 (slightly unconservative) for the Pile
#s 28 and 5 respectively. The ratio of average SPILE and UNIPILE predicted
capacities to measured ultimate capacities were 182 % and 243 % respectively. The
standard deviation of the 24 ratios for the SPILE program was 1.07 and that for
UNIPILE program was 1.27. The error of estimate for the 24 piles for SPILE and
UNIPILE were 2529.3 kN (284.32 tons) and 5504.22 kN (618.73 tons) respectively.

A regression analysis is conducted between the SPILE, UNIPILE arid Load Test
results. In case of SPILE program regression analysis produced a R squared value of
0.81 and x-axis intercept of 1.8 (Pile database numbers 25, 27, 28, and 47 were

Avasarala
714
eliminated for the regression analysis so as to achieve some correlation). Figure 1
shows the best fit line through the points, a slope 1 to 1. 8. In case of UNIPILE
program regression analysis produced a R squared value of 0.81 and x-axis intercept
of 2.99. Figure 2 shows the best fit line through the points, a slope 1 to 3. Out of the
24 piles, the SPILE program predicted fairly accurately for piles 4, 8, and 26, over-
predicted for piles 1, 3, 6, 7, 9, 19,20, 28, 30, 31, 34, 46, 48, 50, 53, 60, 53, 60,
and 61 and under predicted for piles 5, 25, 27, and 47. The UNIPILE program
predicted fairly accurately for the piles 5, 6, 8, 19, 31, and 53 and over predicted for
piles 1,3,4,7,9,20,25,26,27,28,30,34,46,47,48,50,60, and 61.

Comparison according to Pile Width or Diameter

A summary of the statistics for predicted and measured ultimate capacities according
to pile width for the SPILE, UNIPILE and SPT9l programs is shown in Table 2. This
table do not show any trend for any particular pile width and hence it seems that the
width of the pile does not influence the relationship between predicted and measured
capacities.

Comparison According to LID Ratio

A summary of the statistics for predicted and measured ultimate capacities according
to pile length to width ratio for SPILE, UNIPILE and SPT91 programs is shown in
Table 2. This table shows that both the SPILE and UNIPILE programs predict fairly
well for lower LID ratios 22), with the exception of piles 4, 8, and 26 for SPILE
and 4, 27, and 61 for the UNIPILE program, for all types soils.

Comparison According to Soil Type

A summary of the statistics for predicted and measured ultimate capacities according
to soil type for the SPILE, UNIPILE and SPT91 programs is shown in Table 4. This
table shows that the soil type influences the relationship between predicted and
measured capacities. The predicted ultimate capacities from both SPILE and UNIPILE
are not in agreement with ultimate capacities from load tests for piles for all types of
soils, with few exceptions.

Comparison of SPILE, UNIPILE, and SPT91 Results

Figures 3 and 4 show comparison of ultimate skin friction and end bearing resistance
from the SPILE, UNIPILE and SPT9l programs. The total number of piles compared
were 23. All these piles were loaded to their ultimate capacity. Figure 5 shows
similar comparisons for the 51 piles which met the Davisson failure criteria. These
figures show that the ultimate skin friction and end bearing from SPILE, UNIPILE
programs differ from SPT9l program in all types of soils, with few exceptions.

715 Avasarala
12 I PREDICTED = 1.8 x MEASURED 1..../
z
c 28
lIE
+ .//
W
--l
0:: 9
en 50
lIE
~
aa: Iii"
'tl
...\~
LL. c:: 30 /
III lIE ...
~ 6
:...
VI
~ 60/
(3 0
< ..c:: /
c. t::.
<
() 34 .. 4~ 25
~~i{~12
w lIE
!;;: 3
~ 47
~
--l
lIE 27 LEGEND
:J *
50 Pile database number
O~----r------r--=======::;::::::=======-1
o 3 6 9 12
ULTIMATE CAPACITY FROM LOAD TEST (kN)
(Thousands)

Figure 1. Comparison of Predicted vs. Measured Ultimate Capacity for the SPILE
Program.

20 "T"'-------..,......--------------"
i$ !,-/ --,
/f PREDICTED = 3 X MEASURED

46
lIE

LEGEND
50 Pile database number

5 10 15 20
ULTIMATE CAPACITY FROM LOAD TEST (kN)
(Thousands)

Figure 2. Comparison of Predicted vs. Measured Ultimate Capacity for the


UNIPILE Program.
Avasarala
716
PRE-BORED SPILE UNIPILE LOAD TEST
PILE
DEPTH' WIDTH 1 1
DB# LID SOIL # Pspl PuP PsPT
ml em l ULTIMATE CAPACITY, kNt

1 0.00 30.48 29.00 32 1908.64 2206.21 809.60 2.15 2.48 1.73


3 0.00 30.48 8.50 33 340.09 338.05 195.71 1.72 1.71 3.25
4 U.UU J~.~b I.I,} 33 3 J) .19 .'lUI.bI 166.88 1.18 1.88 3.41
5 0.00 30.48 14.00 33 564.27 894.05 809.60 0.63 1.01 1.41
6 0.00 30.48 13.00 33 1249.35 916.29 800.64 1.56 1.14 2.02
7 9..'l0 30.48 35.00 31 2838.71 2250.69 1067.52 2.66 2.11 1.73
8 16.00 JU.411 Ill.UU jj 114U.41 7b~.OO 1l1.b8 Ull 1.08 2.16
9 0.00 30.48 16.00 33 1905.08 1240.99 711.68 2.68 1.74 2.58
19 15.00 30.48 16.33 33 1724.49 1027.49 809.60 3.47 3.61 1.69
20 0.00 30.48 49.10 34 3084.51 3207.01 809.60 3.47 3.61 1.69
I.'l U.W 60.96 24.60 21 3422.65 17850.0 5337.60 0.64 3.34 0.66
26 0.00 50.80 28.38 21 2635.44 12361.0 2668.8 0.99 4.63 1.00
27 0.00 60.96 13.95 11 1.'l16.59 18143.0 5337.60 0.28 3.40 0.65
28 32.5 .'l0.80 28.20 14 9871.80 11849.0 1779.2 5.55 6.66 3.99
...... 30 0.00 50.80 1.1.bU ZJ b1.'}J .n 7045.bJ l.b~I.UU 2.37 2.66 2.00
..... 31 0.00 35.56 51.43 23 2599.23 2268.48 1654.66 1.56 1.37 -'
...... 34 0.00 3.'l ..'l6 28.29 34 3505.20 3046.88 1636.86 2.14 1.86 1.26
46 0.00 60.96 46.60 33 7113.95 8726.98 4287.87 1.66 2.03 1.00
47 0.00 60.96 1.1.JU Jj 11l~1..77 11Zbl.U J7111 ..'lJ O.~U J.03 0.93
48 0.00 60.96 31.00 13 3623.96 6427.36 2437.50 1.49 2.64 1.49
50 0.00 60.96 51.80 12 8111.82 11911.7 3851.97 2.10 3.09 0.94
53 0.00 35.56 57.43 32 2382.3.'l 1378.88 1245.44 1.91 1.11 0.85
60 U.W .'l0.1l0 27.72 33 54,}I.03 6,}34.43 2668.80 2.06 2.60 2.00
61 0.00 50.80 21.60 33 2860.42 4430.21 2206.20 1.30 2.01 1.44
11m = 3.28 ft., 1 cm = 0.3937 in., 1 kN = 0.1125 tons (US).
PSP' Pup, PSPr are the ratio's of the SPILE, UNIPILE, and SPT91 ultimate capacity to ultimate capacity from load test respectively.
data not available.
All the three programs donottake into account the pre-bored depth.

Table 1. Summary of Pile, Soil, Predicted and Measured Ultimate Capacities.


I t;;~ PILE WIDTH (em
35.56 I 50.80
l
)

I 60.96
D
6.3 - 20.0
LID RANGE
I 20.0 - 40.0 I 40.0 - 58.0
D
Ii - 14
PILE SOIL
I 2i - 24
~
~
~PILE PROGRAM
. UI" PILt;:oi 'J 4 ) II II 11 ~ 4 4 Ib
I
MAX. PsI" 3.47 2.14 ~.)) 2.1 2.bll ~.)~ 3.47 ~.)) 2.37 3.47
MIN. PSI" 0.63 1.18 0.99 0.28 0.28 0.50 1.56 0.28 0.64 0.50
AVE. PSI" 2.00 1.70 2.45 1.11 1.40 1.99 2.14 2.36 1.39 1.80
STANDARD DEVIATION 0.81 0.36 1.63 0.67 0.70 1.32 0.69 1.96 0.66 0.74
J<;KI{UK UI" (liN" 1!:l1l.4 11 'J2.U 4172.2 :l1l1l).U 1411U.ll 300):1 2~72.2 4'J'JU.~ 211U.b 1~U2.0

AVE. ULTIMATE CAPACITY (kN') 783.1 1202.7 2394.8 4162.7 1225.7 2460.1 2388.4 3352.3 3079.8 1442.6
~NIPILE PROGRAM
.U~ PILES 9 4 ~ 6 8 11 5 4 4 16
I
MAX. PsI" 3.61 1.88 6.66 3.40 3.40 6.66 3.61 6.66 4.63 3.61
MIN. PSI" 1.01 1.11 2.01 2.03 1.01 1.86 1.11 2.64 1.47 1.01
AVE. PSI" 1.78 1.55 3.71 2.92 1.64 3.09 2.24 3.95 3.00 1.91
0.80 0.33 1.72 0.47 0.74 1.34 0.97 1.59 1.18 U.12
ERROR OF ESTIMATE (kN) 991.0 779.4 6896.4 8926.0 4533.0 6553.1 4271.0 9296.3 8218.5 2630.4
-...I AVE.ULTIMATE~APA~ITY(kN'] 783.1 1202.7 2394.8 4162.7 1225.7 2460.1 2388.4 33)2.3 3~~2.4 1442.6
.....
~PT91 PROGRAM
(X)

UF PILJ<;S 6 II 11 4 4
..1
9 4 ~ 5 16
MAX. PsI" 3.25 3.41 3.99 1.49 3.41 3.99 1.69 3.99 2.00 3.41
MIN. PSI" 1.41 0.85 1.00 0.65 0.65 0.66 0.85 0.65 0.66 0.85
AVE. PSI' 2.UII 1.114 2.1u U.9~ 2.2U l.1l1l Ll2 l.f) 1.22 I.II~

STANDARD DEVIATION 0.53 1.12 1.02 0.28 0.85 0.85 0.33 1.32 0.57 0.73
ERRuR OF ESTIMATE (kN') 768.6 458.9 2971.3 1183.2 990.6 2140.7 306.1 2889.0 1855.1 968.8
AVE. ULTIMATE ~APAUTY (kN 783.1 960.7 2394.8 4162.7 122~.7 2460.2 2570.0 3352.3 3)~2.4 1442.6
I cm - 0.3937 m., 1 kN - 0.1125 tons (US).
PSP, PUP, Psvr are the ratio's of the SPILE, UNIPILE, and SPT91 ultimate capacity to ultimate capacity from load test respectively.

Table 2. Summary of Statistics for Predicted and Measured Ultimate Capacity According to Pile Width,
LID Range, and Pile Soil # for SPILE, UNIPILE, and SPT91 Programs.
Total number 01 piles = 23
1 - Clay; 2 - Mix; 3 - Sand; 4 - Rock 112
Z 6000 2/3/5 - Soil type(s) along pile length
:::.
z
o 5
i= 3/1
u
a:
u.
4
3/1/4
Z
s;::: 3
en 2 3
w /1
'/3
t-
3/1
~
i=
.-J ,
::J

, 3 4 5 6 7 6 9 19 :20 25 211 'Z7 211 30 34 46 47 48 50 53 60 6'

PILE DATABASE NUMBER


I_ SPT91 0 SPILE a UNIPILE I
Figure 3. A Graph Showing the Comparison of Ultimate Skin Friction Resistance
from SPILE, UNlPILE and SPT91 Programs for 23 Piles which are
Loaded to Ultimate Capacity.

'6
'5 Total number of piles 23 =
1 - Clay; 2 - Mix; 3 Sand; 4 - Rock
Z 14
2 - Soil type at pile tip
:::. 13

"
z
a: 11

UJ
CD
."
c
~
10

6
Cl
Z "
0
.c 7
t:
UJ
UJ
t-

~
i=
.-J
::J

0
, 3 4 5 6 7 8 9 '9 :20 25 2ll 'Z7 2ll 30 34 48 47 48 50 53 60 61

PILE DATABASE NUMBER

Figure 4. A Graph Showing Comparison of Ultimate End Bearing Resistance from


SPILE, UNlPILE and SPT91 Programs for 23 Piles which are Loaded
to Ultimate Capacity.

719
Avasarala
Total number of piles = 51
1 - Clay; 2 - Mix: 3 - Sand; 4 - Rock
2/4/5/2 - Soil type(s) along pile length
000

Z 000-
.:.::
z
a 1

~
u 1000-
a:u..
z
i:
(/) lOCO-
-..Jw
",I- I'J,
o~ 1

5:l 1000-
/
Iz 3/

'000-

0
J

]~ 'W ~
I 2 J
,!-Il.
4 8
J

7 8 g
h In
J
3/2 4

'0 11 12 13 14 '5 Ig 20 21 22 2'J 25 2ll V


~
PILE DATABASE NUMBER
/2

~
1
1/2

tt3
~-
,
29 30 32 33 34 35 3e 37 3e 40 41 42 48 47 48 4g 50 51 52 53 54 55 58 57 58 5g BO 81 82

, _ SPTg, 0 SPItE UNIPILE I


Figure 5. A Graph Showing the Comparison of Ultimate Skin Friction from SPILE, UNIPILE, and SPT91 programs for 51 Piles
which met the Davisson criteria.
The results obtained from the SPILE and UNIPILE programs can be influenced by the
following factors:

1) Reproducibility of the pile data in the database. This depends on number of factors
like the insitu test procedure, the load test procedure, the variability of the soil, human
error, etc. The Standard Penetration Test is very difficult to reproduce due to the
many variables involved. The N-values reported are very dependent on equipment type
and operator. To minimize the effect of soil variability it is recommended that the soil
boring be performed as close as possible to the pile test location. In addition it is
recommended that the testing depth interval be as small as possible (0.76 m (2.5 ft
in order to obtain a representative N-value for each 'SPT layer'.

2) In both the SPILE and UNIPILE programs the undrained shear strength of clay soils
is required. Such information is not contained in the database. Because of this the
SPT N-values had to be converted to undrained shear strength using Teng's conversion
charts. This could affect the results from the SPILE and UNIPILE programs. It is
recommended that the properties of clayey soils be determined wherever possible and
should be used for the analysis in the SPILE and UNIPILE programs.

3) Pile capacities can be overpredicted as a result of installation methods such as pre-


boring and jetting. The SPILE and UNIPILE programs do not account for such
effects. These can be taken into account by neglecting the pre-bored depth of soil, or
by reducing the N-values within the affected depth to one half their test values. This
seems to be the case in piles 7, 19, and 28 with p values of 2.66, 1.94, and 5.55
respectively for the SPILE and 2.11, 1.16, and 6.66 respectively for the UNIPILE
program.

4) The SPILE program uses the Tomlinson's a-values to determine the skin friction in
clayey soils. From the results presented above it seems that the Tomlinson's procedure
over predicts. However to confirm this more elaborate analysis needs to be done.
This can be achieved by adding more data to the database.

5) For the UNIPILE program the properties of cohesionless soils have to be given in
terms of the Bjerrum-Burland beta-coefficient and the properties of soil at the pile toe
have to be given in terms of the Nt factor. SPT N-value have to be converted to angle
of internal friction in order to use the suggested charts in the manual. The ranges
shown in these charts are too broad and hence this can effect the results from the
UNIPILE program considerably for cohesionless soils.

Avasarala
721
CONCLUSIONS

As a result of this investigation, the following conclusions can be drawn:

1) The pile design methods used and implemented by SPILE and UNIPILE
programs give unconservative estimations of the ultimate capacity of
piles in sands. The errors of estimate were 1466.4 and 2630.4 leN
(168.84 and 295.68 tons) and the standard deviations of predicted to
measured capacities (PsP and PlJN) were 0.74 and 0.72, respectively. The
average ultimate capacity from the load tests for such piles was 1442.6
leN (162.16 tons). The averages from SPILE and UNIPILE were
2373.5 and 3070.8 kN (266.81 and 345.19 tons) and an average "p" of
1.80 and 1.91 (slight tendency to over predict the capacity) respectively.

2) Out of the 24 valid piles only 8 piles were in clay, silt and mixtures.
Both programs tend to give very unconservative results for piles in clay,
silt and rocks.

3) The Tomlinson's 'a' method used in the SPILE program for clayey soils
seems to be very unconservative. The Teng' s correlations used to
determine the undrained shear strength also could be the reason for
overprediction of capacity of piles by the SPILE and UNIPILE programs
in clayey soils. However to analyze this aspect more elaborate research
needs to be done by adding more data to the database.

4) The ultimate skin friction and end bearing results from SPILE,
UNIPILE, and SPT91 differ with each other, for piles in all types soil.

5) The SPILE, UNIPILE and SPT91 programs does not show any trend for
different pile widths.

6) The SPILE and UNIPILE programs do not show any particular trend for
different LID ratios. Both these programs predict much higher than
SPT91 for all ranges of LID ratios.

7) The charts providing conversions between angle of internal friction, (3


and Nt, given in the UNIPILE manual, are too broad.

8) For practical purposes a factor of safety of 1.8 (R squared value = 0.81


and x-intercept = 1.8) and 3 (R squared value = 0.81 and x-intercept =
2.99) be applied for predicted ultimate capacities from SPILE and
UNIPILE programs respectively, to match the measured load test
capacities.

722 Avasarala
REFERENCES

Meyerhof, G. G., 1976. Bearing capacity and settlement of pile foundations, The
Eleventh Terzaghi Lecture, November 5, 1975, American Society of Civil Engineers,
ASCE, Journal of Geotechnical Engineering, Vol. 102, GT3, pp. 195 - 228.

Prakash, Shamsher., and Sharma, Hari D., Pile Foundations in Engineering Practice,
John Wiley and Sons, New York, 1990.

Ruesta. Pedro F .. Pile Load Test Database and An Evaluation of the SPT91 Program,
Department of Civil Engineering, University of Florida, Gainesville, 1992.

SPILE-Ultimate Static Capacity For Piles in Cohesionless Soils-User's Manual, Federal


High Way Administration (FHWA).

SPT91-Static Pile Bearing Analysis Program-User's Manual, Version 1.0, Structures


Design Office, Florida Department of Transportation (FDOT), 1991.

Swamy Kumar V. Avasarala.. Pile Load Test Database and An Evaluation of the
SPILE and UNIPILE Programs, Department of Civil Engineering, University of
Florida, Gainesville, 1993.

UNIPILE - Unified Design of Piles and Pile Groups Considering Capacity, Settlement
and Negative Skin Friction-User's Manual, Version 1.0, Bengt Fellinius Consultants
Inc., Ottawa, Canada, 1990.

ACKNOWLEDGEMENTS

The authors would like to thank the following for their help throughout the study: Dr.
F. C. Townsend, Prof.; Dr. D. Bloomquist, Asst. Prof.; and Pedro Ruesta, Graduate
Student, Dept. of Civil Eng., University of Florida, Gainesville, Florida.

723 Avasarala
Static Pile Capacity Predictions with SPT91

l.A Caliendo!, M. Bartholomew2, P.W. Lai3


F.e. Townsend4, M.e. McVay4

ABSTRACT

The computer program SPT91 is used extensively throughout the state of


Florida for estimating the static pile capacity of prestressed concrete piles. The
Florida Department of Transportation (FDOT) requires that SPT91 analyses be
performed and included in all Geotechnical reports submitted to the Department
that address the issues of pile lengths and/or axial pile capacities.
The methodology of the program is based on empirical correlations between
cone penetrometer tests and standard penetration tests (SPT) for typical Florida
soils developed by Schmertmann (1967) in Research Bulletin 121-A (RB-121-A)
developed for the FDOT. The relationship between unit skin friction resistance,
unit end bearing resistance, and the SPT blow count, N is provided for each of
the four common Florida soil types recognized by SPT91.
Correlations between predicted SPT91 results and the measured capacities
from full scale static load tests have been well documented. The University of
Florida is presently maintaining a large load test database on behalf of the
FDOT. As the database grows, a number of revisions have been made to the

1 Assoc. Prof. Civil & Environmental Engineering, Utah State University,


Logan Ut. 83422, prevo FDOT, State Geotechnical Engr., Tallahassee, Florida
2 Technical Manager, Bridge Division, Steinman, Boynton, Gronquist, and
Birdsall, Tallahassee Florida 32301
3 FDOT, Assistant State Geotechnical Engineer, Tallahassee, Florida 32301
4 Professors of Civil Engineering, University of Florida, Gainesville 32316

724 Caliendo
software code in order to improve the agreement between predicted and
measured values. These revisions as well as several pertinent algorithms in the
SPT91 computer code are discussed.

INTRODUCTION

The SPT91 computer program is the principal design and analysis tool of the
Florida Department of Transportation (FDOT) for estimating static axial pile
capacity. It is used by several other State Highway Departments as well. The
objective of this paper is to document the theory and computer algorithms on
which SPT91 is based, with the intent of providing the user with some insight into
the workings of the program.
The basis of the SPT91 program is work done by Schmertmann (1967) for the
FDOT and reported in Research Bulletin 121-A (RB-121-A). There are
definitive relationships between unit side friction and unit end bearing for
different soil types. Unit friction is an approximate constant percentage of bearing
capacity for given soil types. Since bearing capacity and standard penetration test
(SPT) blowcount values (N) are related it follows that a correlation exists
between unit side friction, unit end bearing, soil type and N values. These
correlations were developed by Schmertmann (1967) for typical Florida soil types
and are shown on Table 1. Clay soils exhibit the highest friction for a given
blowcount while cohesionless soil have the highest end bearing for a given N
value.
Table 1. RB-121-A, suggested values

Soil Type F r (%) Side Friction End Bearing


ult. (tst) ult. (tst)

Plastic clay 5.0 0.05N 0.7N


Clay silt sand
mixtures, silts 2.0 0.04N 1.6N
and marl
Clean sand 0.6 0.019N 3.2N
Soft limestone 0.25 O.01N 3.6N

Some reduction suggested for stiff clays and sand-clays

F r = ratio of unit side friction to unit end bearing

N = Standard Penetration Test blowcount (SPT) N > 60 use 60; N < 5 use zero

725 Caliendo
RB-121-A Methodology
Since the RB-121-A method has been familiar to many practioners for many
years, a discussion of several key features of the original methodology may be
useful and provide an introduction to some of the changes that have been
implemented and from which SPT91 has subsequently evolved.

Averaging of N values
The RB-121-A methodology requires that the average of all N values within
a distance of 3.5 pile diameters (3.5B) below the pile tip be averaged with the
average of all the N values within a distance of 8.0 pile diameters (8.0B) above
the pile tip but stopping at a distinct weaker overlying layer if this is encountered
first. This is an averaging of two averages and gives equal weight to N values
above and below the pile tip. The unit end bearing is then determined from
Table 1 for the soil type corresponding to the soil at the tip elevation. This is
irrespective of the soil ~ ROB above and 3.5B below, since only the N values
are averaged. An important change in the SPT91 program involves an improved
algorithm that was written for establishing weighted average resistance values.
Also, the actual unit bearing values themselves are averaged as opposed to the
N values. This is further explained below.

Critical Depth
According to Schmertmann, the ultimate bearing capacity of a foundation
increases as the foundation is buried deeper with respect to its least width. The
pile capacity increases with increasing penetration of the bearing layer until a
depth of embedment corresponding to a critical depth/width ratio (DIB) is
reached. The pile capacity is not fully mobilized until this critical depth is
reached. Originally, Schmertmann suggested that this critical depth be taken to
be as shown in Table 2. The DIB effect is assumed to be linear between a
surface footing and a footing founded at an elevation corresponding to the critical
ratio. The bearing capacity of a pile tipped at the top of the bearing layer is set
equal to that of the layer above. The side friction in the bearing layer is also
reduced by this depth of embedment effect and the unit value reduced
accordingly.

SPT91--AN OVERVIEW

Computer solutions for the RB-121-A methodology have evolved from the
original computer program, RB-121-B, developed by Nottingham and Renfro
(1972) and later modified (RB-121-C) by Ho and Webb (1987). This was
accessed through the FDOT's mainframe computer. A PC version (SPT89),
developed and distributed by the University of Florida, followed. The University
of Florida has an ongoing contract with the FDOT to maintain a large database

726 Caliendo
of static pile load tests: (Davidson, Ruesta, and Townsend 1994; Davidson and
Townsend 1993; Sharp, McVay and Townsend 1987). Comparisons between
measured and computed axial pile capacities showed that the computer model
could be improved. As a result, SPT91 was developed by FDOT personnel with
assistance from the University of Florida: (Bartholomew, Caliendo, Lai, and
Graham 1991)
The previous software code was re-written and a number of revisions made
to many of the computational algorithms. A pre-processor and post-processor
were also added to make the program more interactive and to provide both
tabular as well as graphical output. Several of the features and significant
revisions are discussed in detail below.

Soil Models
The original soil types described in RB-121-A are maintained in the program.
The side friction relationship between the soil type and SPT N value remain.
However, as is discussed in the next section, the ultimate end bearing values for
each soil type are divided by 3 in order to estimate a "mobilized" end bearing
value. An additional soil type (soil #5) has been added so that a void in the soil
profile can be modeled. Soil voids are common occurrences in Florida. The end
bearing of a pile tipped in soil type 5 is set equal to zero regardless of the soil
strengths above or below the tip. The friction value of a soil type 5 is also set
equal to zero.
Another reason for adding the void soil type is to allow the user to more
easily evaluate the pile capacity of a soil profile that is subject to scour. The user
can now simply change the soil type for all SPT values above the scour elevation
to a soil void. Previously, this would have necessitated calculating new depths for
all soils below the scour line prior to inputting any data.

SPT91 Pile Capacity


Several axial pile capacities are calculated by SPT91. These include: ultimate
side friction, mobilized end bearing, estimated Davisson capacity, allowable pile
capacity, and ultimate pile capacity. The program can analyze and calculate
capacities for up to 5 different pile widths and provide detailed output for each.
An analysis of the University of Florida database results has shown that in
order to mobilize the pile capacities in the field that are calculated with the RB-
121-A methodology, excessive pile movement is required. The measured load test
capacity is based on the Davisson failure criteria, Vanikar (1985). This is a
measured value which is determined from a field load test.
For the computed pile capacity to better match the field value, SPT91
calculates a value known as the "mobilized end bearing" which is equal to 1/3 of
the RB-121-A ultimate end bearing value. An "Estimated Davisson Capacity" is
included with the SPT91 output and is equal to ultimate side friction plus the

Caliendo
727
mobilized end bearing. The word "estimated" is used to emphasize that this is
not an actual field value.
The allowable pile capacity is taken as one-half the Estimated Davisson
capacity. This is often called the "design load" and _represents a sustained load
that can typically be applied at the top of the pile without causing adverse
deformations. The ultimate capacity is equal to the ultimate side friction plus 3
times the mobilized end bearing. A very important use for this value is to provide
an estimate of the total static resistance that must be overcome during the pile
installation in order to establish a pile tip at a given elevation. This is often used
as input for wave equation analysis when evaluating driving systems.

SPT91 End Bearing


The basic assumption made in RB-121-A that the soil 8.0B above and 3.5B
below the pile tip contributes to end bearing, has also been programmed into
SPT91. An exception occurs when the bearing layer is weaker than the overlying
layer. In this case, unit end bearing capacity is calculated only for the weaker
bearing layer soil. This is done to preclude a "punching" type failure.
Unit end bearing at a given elevation is based on soil type and the
corresponding blowcount. Figure 1 is a plot of mobilized unit end bearing versus
blowcount for the different soil types recognized by SPT91. The equations for
each soil type curve are shown on the figure. The equations describing the
original RB-121-A curves were divided by 3.0 to establish the "mobilized" value
as discussed above. The unit end bearing value is calculated for each SPT N
value between 8.0B above and 3.5B below the pile tip. If there is no N value at
exactly 8.0B or 3.5B, an interpolation is done on the unit bearing value.
A weighted average algorithm is used for calculating both average unit end
bearing and skin friction values. In order to illustrate the algorithm, Figure 2
shows SPT N values plotted along the length of an 18 inch pile. The weighted
average calculations are shown on Figure 3 for the end bearing contribution from
the soil 8.0B above the pile tip. The figure shows mobilized unit end bearing
plotted as a function of elevation. Trapezoidal areas are formed. The areas are
summed and divided by the total length to yield a weighted average mobilized
unit end bearing value. This same process is completed for the values 8.0B above
and 3.5B below the pile tip. This weighted technique allows the user to include
an occasional high or low N value in the data set without having it dis-
proportionately effect the average value.
Finally, the average of the two averages is computed. If the pile embedment
is equal to or greater than the critical depth of the bearing layer, this final
average value is multiplied by the cross sectional area to yield the mobilized end
bearing. If the pile embedment in the bearing layer is less than the critical depth,
a critical depth correction on the unit end bearing is made. Critical depth is
discussed in a subsequent section.

728 Caliendo
Side Friction
Unit side friction at a given depth is also based on the soil type and the
corresponding SPT N value. Figure 4 shows the ultimate side friction values
plotted as a function of soil type and N value.
A similar weighted average technique is used to calculate the unit skin friction
resistance. The skin friction calculations are done for pile lengths above and
within the bearing layer. A correction is made on the unit skin friction for that
portion of the pile in the bearing layer between the top of the layer and the
critical depth. Critical depth corrections are described in the next section.

Critical Depth Corrections


The concept of critical depth is used in both the end bearing and side friction
calculations in a fashion similar to that originally proposed in RB-121-A The
ultimate end bearing provided by a soil layer cannot be fully developed until the
pile is embedded to a critical depth within that particular soil. This depth is
defined by a critical depth width ratio (DIB). Where D is the depth of
embedment within the bearing layer and B is the pile diameter. Table 2 shows
the critical depth ratios employed by SPT91. These have been modified from
those provided in the RB-121-A study which are also shown on Table 2.

Table 2. Critical Depth Ratios for RB-121-A and SPT91

Soil Description Critical Depth Critical Depth


Type Ratio (DIB) Ratio (DIB)
RB-121-A SPT91

1 Plastic Clay 2 2
Clay silt-sand mixtures,
2 very silty sand, silts and 5 4
marls
Clean Sands
3 N = 12 or less 10 6
N = 30 or less 15 9
N > 30 20 12
4 Soft limestone, very shelly 10 6
sands

(D/B) Corrections, End Bearing


When the actual depth of embedment in the bearing layer is less than the
critical depth and when the bearing layer is stronger than the overlying layer, a

729 Caliendo
correction (reduction) is applied to the unit end bearing. The corrected unit end
bearing is determined by:
1) calculating the unit end bearing capacity at elevation of top of bearing layer
2) calculating the uncorrected end bearing at the pile tip elevation.
3) . interpolating between the two values in accordance with the following
equation:

where:
q = corrected unit end bearing at pile tip
qLC = unit end bearing at layer change
qT = uncorrected unit end bearing at pile tip
D A = embedment length in bearing layer
Dc = critical depth of embedment

A correction to the unit end bearing is not applied when the overlying layer is
stronger than the bearing layer.

(DIB) Corrections, Side friction


If the pile tip embedment in the bearing layer is less than the critical depth and
the overlying layer is weaker than the bearing layer, the side friction in the
bearing layer is corrected (reduced) in accordance with the following equation:

CSFBL = SFBL[q + DA (q -q )]
qT LC 2D T LC
C

Where:

CSFBL = corrected side friction in the bearing layer


SFBL = uncorrected side friction in the bearing layer
q1" qLO D A' and Dc ...previously defined

If the pile tip embedment in the bearing layer is greater than the critical depth
and when the overlying layer is weaker that the bearing layer, the skin friction for
that portion of the soil profile between the top of the bearing layer and the
critical depth is corrected (reduced) in accordance with the following equation.

Where:

730 Caliendo
CSFACD = corrected side friction in the bearing layer between the top of
layer and the critical depth
USFACD = uncorrected side friction between the top of the bearing layer
and the critical depth
unit end bearing at critical depth
unit end bearing at layer change

No corrections are applied to the unit skin friction values for the bearing
layer when the overlying layer is stronger than the bearing layer. The unit skin
friction along the pile length below the critical depth is not corrected.

Post Processor
A post processor allows the SPT91 user to view and obtain graphical output
of SPT N values versus elevation. This affords a quick check of the input values.
The user can additionally obtain plots of pile capacity versus tip elevation for
each pile width analyzed. All 5 different pile capacities curves may be plotted or
the user may choose to select the curve(s) of interest. The scales are
automatically adjusted by the program to ensure that the plots are contained
within an 81/2 by 11 page with suitable margins.

DATABASE RESULTS

As mentioned previously, the University of Florida maintains a large database


of driven pile and drilled shaft static load tests. This effort is supported by the
Florida Department of Transportation. Many of the changes that have been
incorporated into the SPT91 program are a consequence of comparing predicted
pile capacities with measured values from the database.
It is not the intent of this paper to discuss the database results in detail since
this is addressed by others including; Davidson, Townsend, Ruesta, and Caliendo
(1994), Davidson and Townsend (1993) and Sharp, McVay, and Townsend (1987).
However, Figure 5 shows SPT91 predicted pile capacity plotted against the
measured Davisson values for 61 of the first 72 pile load test results from the
database. Data points falling above the 45 line represent a load test site for
which an unconservative SPT91 prediction was made (i.e. predicted capacity was
higher than the measured capacity). Alternatively, points falling below the 450
line indicate conservative SPT91 predictions and points falling on the line show
perfect agreement.

731 Caliendo
Special thanks to Mr. David Werth for his assistance with several of the
figures shown in this paper.

REFERENCES

Bartholomew, M., Caliendo, J.A., Lai, P.W. and Graham, K. (1991). "Static Pile
Bearing Analysis Program - SPT91 User's Manual." Structures Design Office,
Florida Department of Transportation, Tallahassee, Florida.

Davidson, J.L and Townsend, F.C. (1993). "Maintenance of Load Test


Databases."Report No. 99700-7555-010 to Florida Department of
Transportation, Dept. Civil Engineering, University of Florida, Gainesville,
Florida.

Davidson, J.L, Townsend, F.C., Ruesta P.F., and Caliendo, J.A (1994). "Pile
Load Database and an Evaluation of the Program SPT91." Proceedings,
International Conference on Design and Construction of Deep Foundations,
FHWA, Orlando 1994.

Ho, K.H. and Webb, T.B. (1987). "A Computer Program to Estimate Load
Capacity from Standard Penetration Test Results." Florida Department of
Transportation, Research Bulletin 121-C, Gainesville, Florida.

Nottingham, LC. and Renfro, R.H. (1972). "A Computer Program to Estimate
Load Capacity from Standard Penetration Test Results." Florida Department
of Transportation, Research Bulletin 121-B, Gainesville, Florida.

Schmertmann, J.H. (1967). "Guidelines for use in the Soils Investigation and
Design of Foundations for Bridge Structures in the State of Florida." Florida
Department of Transportation, Research Bulletin 121-A, Gainesville, Florida.

Sharp, M.R., McVay. M.C., and Townsend, F.C. (1987). "Development of Design
Procedures for Estimating Capacity and Deformation of Pile Groups, Volume
1: An Evaluation of Axial Pile Capacity from Insitu Tests." Report No. 99700-
7379-119 to Florida Department of Transportation, Dept. Civil Engineering,
University of Florida, Gainesville, Florida.

Vanikar, S.N. (1985). "Manual on Design and Construction of Driven Pile


Foundations." Federal Highway Administration, Technical Report No. FHWA-
DP-66-1, Washington, D.C.

732 Caliendo
80
Type 4, Soft Limestone, Very Shelly Sands
~70 q = 3.6N/3
00
~
'-"
C 60
~ Type 3, Clean Sands
~ 50 q = 3.2N/3
~ Type 2, Sand-Clay Mix,
l:Q
~ 40 Silts & Very Silty Sands
Z \. q = 1.6N/3
...... ~
tAl
tAl
~ 30
~
N
'1!1 - Type 1, Clays
1-4
~
1-4
20 q = 0.7N/3
l:Q
0
~ 10
~
0
0 10 20 30 40 50 60 70
SPT BLOW-COUNT, N

Figure 1. Design Chart for Mobilized End Bearing


SPT Blo'Wcount
o 20 40 60 80 100
o
Prestressed Concrete Pile,
2
I Mobilized Unit End Bearing
for soil type 2, q = 1.6 N/3
~
I
for soil type 3, q = 3.2 N/3
-10 6
tntimate Unit Side Friction
2N(110-N}
6 for soil type 2, f = 4583.3
~or soil type 3, f = 0.019 N ~
-20 8 (EI. -20.5') -,---

-40

-50

-60

Figure 2. Sample Problem

734
Mobilized Unit End Bearing, TSF

0 10 20 30 40 50
-25

8.32 Prestressed Concrete Pile, B= 18"


8.53

-30

-35

E-o
~

I::l
34.56
0
~
-40
ell
> Pile tip (El. -39')
~
~

Segment Trapezoid Sub-Area

#1 0.5 8.32 + 8.53)/2 = 4.21


-45 2.5 8.53 + 11.73)/2 = 25.33
#2
#3 2.5 11.73 + 21.33)/2 = 41.33
#4 2.5 21.33 + 24.53)/2 = 57.33
#5 2.5 24.53 + 19.2)/2 =54.66
#6 1.5 19.2 + 34.56)/2 =40.32
Summation = 223.2
-50
Weighted Average above pile tip = 223.20/12
= 18.60 TSF

-55

Figure 3.
Sample Calculations showing weighted average of unit
end bearing

735
2 I I

~ 1.8
00.
Type I, Clays
~
'-"
~=2N(11O:)/4006~
Type 2, Sand-Clay Mix,
Z 1.6
Silts & Very Silty Sands
o
E: I .4 f= 2N(l10-N)/4583.3
U
~ 1.2
~
~ I ~
-..J
W
Ol
~
t;j 0.8 Type 3, Clean Sands
~
~ f= 0.019N
~ 0.6
-<
~
~~:, ..
~
~~::~::
......
0.4
~
S0.2 Type 4, Soft Limestone & Very
Shelly Sands f= O.OIN
o I I I I I I I I I I I I I I I

o 10 20 30 40 50 60 70
SPT BLOW-COUNT, N

Figure 4. Design Chart for Ultimate Side Friction


COMPARISON OF CAPACITIES
SPT91 PREDICTED Vs. DAVISSON FAILURE
.-...
~ 800
o
-
I-

~ 600
1/
.
1/
t)

n.

t)
o
400 .
W
l-
t)

@ 200
.. .1/
.. .. .
. ..
. .
~.
. .
.
.
.

/
0::::
n.
,.... ::
. .. . .
(1)
l-
n. o ......
CJ)
o 200 400 600 800
DAVISSON FAILURE CAPACITY (TONS)

METHOD: DAVISSON DATABASE" RANGE: 1 72


CRITERION: FAILURE
CAP / SET: CAPACITY

MIN SPT91: 23.3 TONS NUMBER OF EVENTS: 61


MAX SPT91: 676.7 TONS MIN % SPT91 / METHOD: 36.3 %
AVG SPT91 : 180.6 TONS MAX % SPT91/METHOD: 262.1 %
AVG (LOG) % SPT91/METHOD: 82.6 %
MIN METHOD 14.8 TONS STANDARD DEVIATION (LOG): 20.1 %
MAX METHO 698.9 TONS ERROR OF ESTIMATES: 124.0 TONS
AVG METHOD 237.6 TONS

Figure 5. SPT91 Predicted vs. Davisson Failure Capacity


from the University of Florida Database (after
Davidson & Townsend, 1993)

737
NEURAL NETWORK PREDICTIONS OF LOAD-DEFLECTION
CURVES FOR CONCRETE PILES IN FLORIDA

George G. Goble, Ph.D., P.E.\ Jan F. Kreider, Ph.D., P.E. 2 ,


Peter Curtiss, Ph.D. 3 , Jay Berger"

1 Introduction
The prediction of pile capacity from subsurface investigation information has
been a challenge that geotechnical engineers have struggled with since the
beginning of the geotechnical engineering profession. First, the mechanics
of soil-pile behavior is not well defined. Soil properties are also not easily
defined since the information generally used in design is obtained from
Standard Penetration Test (SPT) data or from the Cone Penetration Test
(CPT). In the United States, the SPT is by far the dominant soil investigation
technique. These investigative tests give only very limited information
regarding the mechanical properties of soil and cannot be expected to pro-
duce accurate design information. Furthermore, the soil properties will
change as a result of pile driving. One could even consider pile driving to be
a soil improvement technique. Finally, even with the static load test available
the pile capacity is often not well defined since the evaluation of the load test

1Principal, Goble Rausche Likins and Associates, Inc. 5398 Manhattan Circle, Boulder, CO
80303 USA
'Principal, Energan, Inc. 1455 Oak Circle, Boulder, CO 80304 USA and University of Colorado
at Boulder.
3 Research Professor, JCEM, CEAE Department, University of Colorado, Boulder, CO
80309-0428 USA
'Branch Manager, Goble Rausche Likins and Associates, Inc. 5398 Manhattan Circle, Boulder,
CO 80303 USA .

738 G. Goble et al.


curve is not interpreted in a standard manner. Geotechnical engineers feel
strongly that some judgement should be used in determining pile capacity
from a static load test curve.

Artificial neural networks (NNs) are vastly simplified mathematical models of


some aspects of biological neural networks. This paper reports on several
applications that represent the first use of tools from this modern branch of
artificial intelligence for pile capacity analysis. Instead of coding many rules
and mathematical routines, NNs are able to recall key information patterns
within multidimensional information domains. The ability to "learn" the rela-
tionship between pile capacity, pile physical characteristics, and soil prop-
erties is one of the NN features of particular value. We show that NNs can be
trained to predict the capacity of prestressed concrete piles in Florida soils,
as described below.

This paper willsummarize the algorithmic basis for neural networks. Our goal
is to use this new approach to find relationships that have been unavailable
using standard methods. In addition, our goal is also to develop a method
for assisting designers based on measured data, not theoretical models.
Neural networks are adopted as an alternative to traditional statistical
approaches.

The use of NNs offers a possible solution to the pile capacity prediction
problem. A substantial amount of data of the type used by the pile foundation
designer, together with static load test results are available for training the
networks. In this case, data base information is available for a substantial
number of prestressed concrete piles that were driven in Florida, mostly on
Department of Transportation projects. Using these data a NN was trained
and tested. The results indicate that it is possible to train the net, that the
results agree well with available test results, and that nets can be used to
predict the static load test curve.

2 Background of Pile Capacity Determination

The design and construction of pile foundations precedes recorded history


in many parts of the world. In the earliest times, the pile was probably driven
until it would penetrate no further under the action of the available driving
system. With the rise of civil engineering, attempts were made to predict the
static strength of the pile based on an orderly evaluation of the driving process

739
using energy concepts. The method known as the dynamic formula became
well developed during the last century. Of course, the observation of the
driving operation could be of no help prior to the time that field work began.
When subsurface investigation information became available the attempt was
made to predict static behavior from that information. In fact, the motivation
for the development of the SPT was to assist in the prediction of pile driving
operations (Fletcher, 1965).

Mechanical models of pile soil behavior were developed when the techniques
of modern soil mechanics became available. These models were used
together with a variety of assumptions to predict static capacity. Unfortu-
nately, these tools gave poor agreement with the response measured in static
load tests and a variety of methods were developed based on several different
assumptions. Consider the reasons that the predictive methods may fail:

1. The mechanical properties of the soil assumed from the subsurface


investigation information may be incorrect.

2. The mechanical model may not correctly and realistically represent


the soil behavior during the static load test.

3. The pile capacity may be poorly defined as illustrated in Fig. 1. The


typical behavior that is expected from a pile driven into a clay or silt soil
is shown in Fig. 1(a). The behavior shown in Fig. 1(b) represents the
typical response from a pile driven in sand. In the first case, the deter-
mination of the pile capacity from the load test curve is well defined while
in the second case the capacity is highly dependent on the method used
to define the capacity. In coarse grained soil, there is usually no well-
defined capacity from the load test result. Available methods can provide
a large range of capacities from which to choose.

The above problems are further complicated by the fact that the pile driving
operation may change the soil mechanical properties by changing the soil
density. If, after the pile is driven, the properties are different than when the
subsurface investigation was made, it cannot be expected that the results of

740
1000 .... --.-.---.-------------------------,

~
::: i---~~~==~------=-:-J
700 -1-------,~
I .-/ II .1

-
g
600
500
j-.-T'"--
----~- I
~ 400 I f--- - 7
.iOCl I ~ .------.----..
200 iJ:
II

1 O~ t -,-
o 0.7 0.4
j

0.6 0.8 1
1 1.2
j

1.4 1.6
Top Movemen1 (inches)

(a)

800
1--
I
100 ~-_.-

600

]:: 500
~
~ 400
t
I

/
...... ----- .------------
1
j
1

tl/r
o
. J

g- 300
>-

200
J
100
j
o
o 0.2 0.4 0.6 0.8 1 1.2
/
1.4 1.6 1.8 2
Top Movemenl (inches)

(b)

Figure 1. Illustration of pile capacity definition; (a) clay soil, (b) sandy soil.

741
the subsurface investigation can be used to predict soil behavior after driving.
Methods of pile capacity prediction have been summarized in a state of the
art report by Vesic (1977).

For these reasons, pile capacity predictions based solely on subsurface


investigations agree poorly with static load test results. This conclusion is
well supported by many of the prediction symposia that have been conducted
on pile capacity. At several conferences the delegates have been asked to
predict capacity, given the subsurface investigation information. Then, at
the time of the meeting, a pile load test was performed and the results
compared. In the most common cases, the scatter of the predicted results
was quite large. However, the mean of all of the predictions usually agreed
surprisingly well with the measured result. Perhaps this implies that the
collected wisdom of all of the predictors will agree well with reality. This
represents an ideal application for neural nets.

3 Introduction to Neural Networks


An artificial neural network is a massively parallel, dynamic system of inter-
connected, interacting parts based on some aspects of the brain. Neural
networks are considered to be intuitive because they learn by example rather
than by following programmed rules. The ability to "learn" is one of the key
aspects of neural networks. A neural network consists of several layers of
neurons that are connected to each other. A "connection" is a unique
information transport link from one sending to one receiving neuron. The
structure of part of an NN is schematically shown in Fig. 2. Any number of
input, output and "hidden layer" (only one hidden layer is shown) neurons
can be used. One of the challenges of this technology is to construct a net
with sufficient complexity to learn accurately without imposing a burden of
excessive computational time.

The neuron is the fundamental building block of a network. A set of .inputs


is applied to each. The elements of the input set are each multiplied by a
weight, indicated by the W in the figure, and the products are summed at the
next level neuron. The symbol for the summation of weighted inputs is termed
INPUT and must be calculated for each neuron in the network. In equation
form, this process for one neuron is

742
IN pur = >:, 0, 11'1, + B (1 )

where () i are inputs to a neuron, i.e., outputs of the previous layer


lv', are weights, and
B is the neuron's bias

fIJDDEi' LAYER OUTPUT LAYER


k

TARGET :

TARGET 2

TARGET"

Figure 2. Schematic diagram of a neural network.

After INPUT is calculated, an activation function F is applied to modify it,


thereby producing the neuron's output as described shortly.

Artificial networks have been trained by a wide variety of methods [Wass-


erman, 1989; McClelland and Rumelhart, 1988]. Backpropagation is one
systematic method for training multi-layer neural networks. The weights of
a net are initiated with small random numbers. The objective of training the
network is to adjust the weights iteratively so that application of a set of inputs
produces the desired set of outputs matching a training data set. Usually a

743
network is trained with a training data set that consists of many input-output
pairs; these data are called a training set. Training the net using backpro-
pagation requires the following steps:

1. Select a training input-output pair from the training set and apply the input vector
to the network input layer.

2. Calculate the output of the network, OUTI.

3. Calculate the error ERRORI between the network output and the desired output (the
target vector from the training pair).

4. Adjust the weights of the network in a way that minimizes the error.

5. Repeat steps 1 through 4 for each vector in the training set until the error for the
entire set is lower than the user specified, preset training tolerance.

Steps 1 and 2 are the "forward pass." The following expression describes the
calculation process in which an activation function F is applied to the
weighted sum of inputs INPUT as follows.

OUT = F(I N PUT) = F(LOiLv, + B) (2)

where F is the activation function,


B is the bias of each neuron

The activation function used for the present work was selected to be

1 (3)
F(lNPUT)== -INPUT
(l + e )

This is referred to as a sigmoid function and is shown in Fig. 3. It has a value


of 0.0 when INPUT is a large negative number and a value of 1.0 for large and
positive INPUT, making a smooth transition between these limiting values.
The bias B is the activation threshold for each neuron. The bias avoids the
tendency of a sigmoid function to get "stuck" in the saturated, limiting value
area.

744
Generic Sigmoid Function
output =1 / (1 + exp (-input))

09

0.8

--;- 0.7

~
~ 0.6

1~ 0 ,5 --- -- - - - - - -

,g
. . .. .- 0.4

~
d 0 ..3

0.1

o ~.-=::-:-;::==---,-,------,-------r---l----,-----,--,--,---------j
-5 -4 -3 -2 -I 0 1
Input (indeptlndenl "fQrlable)

Figure 3. Sigmoid function used for "squashing" the sum of the weighted
neuron inputs (abscissa) to produce output (ordinate).

Steps 3 and 4 above comprise the "reverse pass" in which the delta rule
[Rumelhart and McClelland, 1988] is used as follows. For each neuron in the
output layer, the previous weight Wen) is adjusted to a new value W(n+ 1) to
reduce the error by the following rule

Ll(n+ 1)=Ll(n)+(115)OUT (4)

where Wen) is the previous value of a weight


Wen + 1) is the weight after adjusting
11 is the training rate coefficient
o is calculated from

(5)
b= (O~~~~T)(T ARGET - OUT) = OUT( 1 - OUT)( TARGET - OUT)

in which the derivative has been calculated from Eqs. (2) and (3) and TARGET
(see Fig. 2) is the training set target value. This method of correcting weights
bases the magnitude of the correction on the error itself.

745
Of course, hidden layers have no target vector: therefore, backpropagation
trains these layers by propagating the output error back through the network
layer by layer, adjusting weights at each layer. The delta rule adjustment,
B , is calculated from

(6)

where aj and a
J+ 1 belong to the jth and 0+ 1)th hidden layers, respectively
(being numbered with increasing values from left to right in Fig. 2). This
overall method of adjusting weights belongs to the general class of steepest
descent algorithms.

A network containing values of weights and biases after training contains


meaningful system information; before training, the initial, random biases and
random weights have no physical meaning.

4 Available Network Training Data

The development of a NN requires information on system performance to


train the net. A large data base of pile performance has been assembled by
Goble Rausche Likins and Associates, Inc. and the FHWA (GRL-FHWA). It is
ideal for this application. First consider the variables that determine pile static
load performance. These factors are the soil strength, soil type, pile type,
length of time after driving, and the load test procedure. In the GRL-FHWA
data base, the practice used in the United States dictates what information
is available. Subsurface investigation in the United States is usually limited
to SPT testing. Thus, soil type and N-value are known. In cohesive soils
especially, but also in sands, the pile capacity will depend on the elapsed
time between driving and static load testing. The data base contains a large
volume of data on prestressed concrete piles driven in Florida so the NN was
trained for that pile type. Of course, that implies that soils characteristic of
Florida are emphasized. In other soils or for other pile types, the present
neural net would not be expected to apply.

The NN that was finally selected used the following data fortraining purposes:

1. soil type along the shaft


2. soil type at the pile toe

746
3. sum of the SPT N-values along the pile shaft
4. SPT N-value at the pile toe
5. pile length, area, and circumference
6. pile elastic modulus
7. static load test curve

Since the net was limited to prestressed concrete piles, pile type was not
included as an independent variable. The solution assumed that, in appli-
cation, the information listed in items 1 - 6 above would be entered by the
pile designer and the NN would produce a predicted static load test curve.
The engineer would then evaluate the curve and predicted pile capacity.

5 Neural Net Design


For each site over 20 pieces of data related to the pile and site characteristics
have been recorded and are contained in the GRL-FHWA data base. In
addition, the SPT blow count and soil type have been measured at a variety
of depths. It is our belief, however, that not all of the data are necessary, and
indeed may introduce unwanted complexity to the problem. The pile data
used in the present analysis are only those listed above.

The SPT data presented an interesting problem since the number of mea-
surements and the depths will be unique for each site. Simply treating each
SPT depth and blow count as a separate input would lead to a very large (and
slow to train) network. For this work we have condensed the SPT data, relating
to pile shaft resistance, into one value -- the area beneath the blow count vs.
depth plot. Furthermore, the dominant soil type along the pile shaft was
included as an input. Soil resistance at the pile toe was described by the SPT
N-value and the soil type reported by the SPT.

6 Results

There is no easy way to determi ne the proper configuration of a neural network


by virtue of the data alone. Often, trial and error attempts along with con-
siderable experience are necessary to arrive at an architecture that is both
sufficiently accurate and that can be trained with a reasonable computational
effort.

747
It is desirable to have a network with enough complexity to allow the modeling
of the system but simple enough to prevent it from merely memorizing the
training set (this is analogous, in traditional regression analysis, to having a
very high order polynomial which can be made to fit any data set). Thirty
different network architectures were systematically tested to create a map of
the RMS error versus the number of network parameters. The number of
parameters in any given network is the sum of the number of connections
between cells (Le., the number of weights) and the number of cell bias nodes.
Table 1 lists the key results from the sensitivity tests performed on a data set
of 71 sites selected from a large pile driving data base with sites both in Florida
and elsewhere. The training rates and annealing factors were chosen based
on previous and ongoing research with pile data5

The limit of 2 million back propagation iterations was determined by com-


paring the accuracy of predictions for networks trained with between 200,000
and 5 million passes. Figure 4 shows results of the network size study
tabulated in Table 1, where the RMS error is plotted as a function of the total
number of network parameters, the principal measure of network complexity.
The error follows a clear pattern: larger for the simplest networks and
decreasing until a plateau is achieved for the more complex architectures.
This I~veling off begins at around 80 parameters; it was decided to use a
7:6:4:1 network (81 parameters) for the predictions described next.

Figure 5 is a typical comparison of the net-predicted and measured pile load


vs displacement curve. This figure shows that the pile data sets are amenable
to the NN approach - the agreement between measurement and prediction
is within 10%. However, in NN applications, it is customary to conduct a more
stringent test. Data not used in training are used to test the fully trained
network. Figure 6 is a comparison of the measured and predicted load-
deflection curve for a data set not included in the network training. Again
the agreement is seen to be very good.

5The data used for this aspect of the study did not include all soil information listed in the
previous section. The toe soil types and the SPT count were added later for physical and
prediction accuracy reasons.

748
Table 1. Data from network architecture investigation

Number of Training
Parameters Rates Network Rates Error

Cl
E
.5 .2
~
:c co4l c:

i
.r::. 4l
Kl Cl co CD
E
:g i .~
c:
75
~ c: 0
Architecture data set iii ~ I- m c.. ::::!: FM5 Average
No hidden 7:2:1 14-Jun 3 16 19 5.0 2.5 0.995 5E+06 0.1 151.8 -0.5
layers 7:3:1 14Jun 4 24 28 5.0 2.5 0.995 5E+06 0.1 142.5 -0.1
7:4:1 14-Jun 5 32 37 5.0 2.5 0.995 5E+06 0.1 152.0 0.4
7:5:1 14-Jun 6 40 46 5.0 2.5 0.995 5E+06 0.1 153.1 0.4
7:6:1 14-Jun 7 48 55 5.0 2.5 0.995 5E+06 0.1 123.1 -0.2
7:7:1 14-Jun 8 56 64 5.0 2.5 0.995 5E+06 0.1 116.1 -1.8
7:8:1 14-Jun 9 64 73 5.0 2.5 0.995 5E+06 0.1 118.5 -1.7
7:9:1 14-Jun 10 72 82 5.0 2.5 0.995 3E+06 0.1 117.9 -2.6
7:10:1 14-Jun 11 80 91 5.0 2.5 0.995 5E+06 0.1 121.2 -1.1
7:15:1 14-Jun 16 120 136 5.0 2.5 0.995 5E+06 0.1 113.6 6.6
7 nodes In 7:7:7:1 14-Jun 15 105 120 5.0 2.5 0.995 5E+06 0.3 108.6 1.8
first hidden 7:7:6:1 14-Jun 14 97 111 5.0 2.5 0.995 5E+06 0.3 107.7 2.8
layer 7:7:5:1 14-Jun 13 89 102 5.0 2.5 0.995 5E+06 0.3 114.2 2.5
7:7:4:1 14-Jun 12 81 93 5.0 2.5 0.995 5E+06 0.3 113.9 2.3
7:7:3:1 14-Jun 11 73 84 5.0 2.5 0.995 5E+06 0.3 116.3 0.0
7:7:2:1 14-Jun 11 65 76 5.0 2.5 0.995 5E+06 0.3 119.3 1.8
6 nodes in 7:6:6:1 14-Jun 13 84 97 5.0 2.5 0.995 5E+06 0.3 112.0 -0.6
first hidden 7:6:5:1 14-Jun 12 77 89 5.0 2.5 0.995 6E+06 0.3 104.4 5.8
layer 7:6:4:1 14-Jun 11 70 81 5.0 2.5 0.995 5E+06 0.3 108.2 -0.4
7:6:3:1 14-Jun 10 63 73 5.0 2.5 0.995 5E+06 0.3 114.1 1.9
7:6:2:1 14-Jun 10 56 66 5.0 2.5 0.995 5E+06 0.3 121.2 0.2
5 nodes in 7:5:5:1 14-Jun 11 65 76 5.0 2.5 0.995 5E+06 0.3 120.1 -1.9
first hidden 7:5:4:1 14-Jun 11 59 70 5.0 2.5 0.995 5E+06 0.3 120.5 11.4
layer 7:5:3:1 14-Jun 9 53 62 5.0 2.5 0.995 5E+06 0.3 120.8 -0.5
7:5:2:1 14-Jun 9 47 56 5.0 2.5 0.995 5E+06 0.3 123.1 -0.8
4 nodes in 7:4:4:1 14-Jun 11 48 59 5.0 2.5 0.995 5E+06 0.3 122.1 -2.7
first hidden 7:4:3:1 14-Jun 11 43 54 5.0 2.5 0.995 5E+06 0.3 130.4 -7.1
laver 7:4:2:1 14-Jun 11 38 49 5.0 2.5 0.995 5E+06 0.3 133.2 -3.5
3 nodes In 7:3:3:1 14-Jun 11 33 44 5.0 2.5 0.995 5E+06 0.3 138.3 -4.0
firsl H.L. 7:3:2:1 14-Jun 11 29 40 5.0 2.5 0.995 5E+06 0.3 138.2 -8.5

749
RMS ERROR VAS NETWORK ARCKTECTlJRE

160 '--~--~

..,..=.......... 'J,
It
- ... ~..... ,.'" .-. '1'"" ... --1;
150
_l... ~ _
.. ! . i.
_._..
'i
:
~ ..-. -.-. ,-
:
r
,:'
--i--"-'--"-i
.. _.1. - "'1
140
..........'i i..-, . . . :_.~~....
' GO I j,
o
. -.r",:--------~
' - . -....-..
.-
.. ~ t .
I
. '''i
130 i ..: -t-----~

120

110
,------Ji-------:~~~=:;i~~=:;f~;::-~;-~:-i
100 =--=-:---i-=--:-:t~~t~:-~:::~-~--J
~
2
r
90 ~ ~~D~::FUmA I.; ~ - :~~~~~=~~.:==.~

---=-~=-~-.~.:!.,:=-~F~.slNGl.E=:-LA:~= ...-.-.-
~ 80
15
G)
70
i
60
- -::.,
50
....--.. __ .-. __ .----.. ~--

!
-.-......'"

.............i......................
.
..
--+_
.....---...........:-.........

,.......... - - - 0- _.
,I~I"~I~

3 NODES IN LAYER 1
-J
;

40
------t
~
. j~--=~~~~:=~
j

t-------~--i--
---0--- 4NOOESINLAYERl

---Co-- 5 NODES INLAYER 1


,
- ....1
--~~~:J
30
"-"1'
--t--
-1-
- -- ..- - - 6NODESINLAYERI
20 +-----l.-----.-+-.... -------+-
, -- -1 -_.--- .
- - 0- - - 7 NODES IN LAYER 1 :~~!
10 +-----,'------1----------~---"L--__,----.,__-J--._~

0
j "'r
o 25 50 75 100 125 150
NUMBER OF NETWORK PARAMETERS

Figure 4. RMS error in load prediction vs. network complexity as measured


by the number of network parameters (weights and biases).

750
Load Deflection Cur ve
Site - #3
900
----
~
800 -------- -----------"so-::::--li=-:::iF--
-- ~ data
-

700
________________ ~- I
~ NN prediction
3. cOO i
~
'0
:'00
400
r----
1- -----
--.-~
..3 300 - ---- -

200f,-Jj----~
100 l~- -------------------
a E--
o
-------.----,-----,-------,------,----------1
0_2 0.4 0.6 0.8 1.2
Displacemenl - Inches

Figure 5. Comparison of measured and predicted load-deflection curves


for site 3 selected from the Florida training data set.

Finally, Fig. 7 shows the accuracy of the trained network in predicting the 13
sites used in training plus an additional 10 testing sites, for a total of 508 load
vs. displacement points. Most of the data lie along the line that indicates
perfect agreement. The root mean square error of the full data set is less
than 10%. For the testing set alone, the RMSE is 11%.

However, there is a group of points for which the prediction (ordinate) is much
higher than the measured data. These points above the perfect agreement
line are all from one load test (site 49). This site was not included in the
training set demonstrating that a NN cannot be expected to predict well for
situations that are outside of its training space. Furthermore, the poor per-
formance of the NN on site 49 can be traced to the relatively short time
between the driving of the test pile and the static load test. At site 49 the
waiting period was only six days; whereas, for adjacent sites 47A and 48A

751
Load Deflection Curve
Site - #48 A
1800 - - , - - - - - - - - - - - - - - - - - - ,----------,

1600
1 40 0
I-------------;;;:=:::::::IF=-~-'_______iTest

+- ----=~~----:=-----=='!'~-3=~!__.j ---El-
-- data

NN prediction
III 1200 +-----.i'-~=-----------____I
a.
~ 1 000 +---~l------------____I
I
11 800 + - - - - - - I ! - - - - - - - - - - - - - - - - 1
o
-' 6 00 +-----lIl~------------____I

400 + - - - ; ; ; r < - - - - - - - - - - - - - - - j

200 . . . , . - - - - - - - - - - - - - - - - - 1

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2


Displacement - Inches

Figure 6. Comparison of measured and predicted load-deflection curves


for site 48A selected from the Florida testing data set.

the times were 35 and 29 days, respectively. The network trained for much
longer durations was simply unable to predict the capacity for a much shorter
waiting period.

Appendices 1 and 2 contain all of the load test curves in the data base. The
predicted and actual curves for the training set appear in Appendix 1. Not
surprisingly, the agreement is good in almost all cases. This demonstrates
the "trainability" of the network. The same comparisons for the testing data
set are given in Appendix 2. Here again, the network predictions were
accurate with an overall RMS error of 11% on a data set that the neural net
had never before been exposed to. This difficult test was passed successfully
except for site 49 discussed above. That case demonstrated that an additional
input to the network is needed - the elapsed time between driving and testing.

752
Load Deflection Curve Data
All Sites
1800 - ---------------

1600 --------------------:".-c-----1

~
... ""-:._--.._----~_....~-__1
1400 - - - - - - - - - - - . -
.~ .:. A.
I 1200 ....
-g~ 1 00 a + - - - - - - - - ' ;..c.-.---..c=t~W_,6,--~""------___1
A.

"0 800 .. :.:."


2
OJ
i1 600
'"' -
CL 4 a0 t------,-

200 [,MIIIIII
o ~ ---, " " "
o 200 400 600 800 1000 1200 1400 1600 1800
Measured Load - kips

Figure 7. Predicted vs. actual load data for Florida


training and testing data sets.

7 Conclusions

1. Neural networks offer a significant step forward in simplifying and


improving the analysis of pile capacity data. It is expected that a pile design
tool could result from applying NN ideas.

2. A neural network learns best from fact patterns that include all physically
possible combinations of the independent and dependent variables. Neural
networks cannot predict events not included in the training set; nor can any
other prediction scheme.

3. With a larger data base it can be expected that better predictions of


load-deflection curves and pile capacity will be obtained.

753
8 References

[1] D. B. Fogel, 1991: "An Information Criterion for Optimal Neural Network
Selection." IEEE Trans. on Neural Networks, Vol. 2, No.5, pp 490-497, Sep.

[2] P. D. Wasserman, 1989: NEURAL COMPUTING, THEORY AND PRACTICE,


Van Nostrand Reinhold, New York.

[3] J. L. McClelland and D. E. Rumelhart, 1988: EXPLORATION IN PARALLEL


DISTRIBUTED PROCESSING, MIT Press, Cambridge.

[4] A. S. Vesic, 1977: "Design of Pile Foundations," National Cooperative


Highway Research Program Synthesis of Highway Practice, Transportation
Research Board, National Research Council, Washington, D.C.

[5] G. F. A Fletcher, 1965: "Standard Penetration Test: Its Uses and Abuses,"
Journal of the Soil Mechanics and Foundation Engineering Division, ASCE,
Vol. 105, No. GT8, August.

754
Appendix 1 - Comparison of Test and Predicted
Load-Deflection Curves for the Training Data Set

Load Deflection Curve Load Deflection Curve


Sit. - # 1 Sit. - q2A
1200 1000
~
900
1000
800
~
~
700
.~ ll. 600
/.1"
...
:>
.",-;3
I sao //
1l '00
./
~
300
or..?
laO rI'/
2~O
100
.:If'
0 a
a 0.2 0.' 0.6 0.6 I 1.l '-' a 0.1 0.2 Q,J 0.4 O.S 0.6
[)i:.plcc8I"T'I.,1 - klc:r.5 DispIoc..-nMn1 - fthu

Load Deflection Curve Load Deflection Curve


Sit. - #3 Sit. - B4
9DO 100
~
8DO ~
7DO
I ~ 600
../'
~,,~- ..-?"
500
6DO

'DO
c-ii--- ::f .400 //
Ji'" I
,//
oY' ! 300
300
I

.f 200
.F
zoe
~?" 100 ./
a
0; a
D 0.2 0.' 0.6 O,B 1.l 0.1 0.2 0.3 0.' 0.5 0,6 0.7 0,8 0.9
DI~I-Ird'.u Ci~I-lnch..

Load Deflection Curve Load Deflection Curve


Sit. - #6 Sit. - #11
900
800
~
----- 1400

'200 ----'
700
.# '000
~
600

'DO
->tf!Y i BOO
/
'00

I

] 600
d
300
of
1/
200
IJ
.00

200
;r
'00
I:'
o a
o 0.2 0.4 0.6 O.B 1 1.2 1.4 a 0,2 0.' 0.6 O.B I 1.l 1.4
OI~~I-lnCf,.s Oi5plocemll"ll -.-.chIs

755
Load Deflection Cur ve Load Deflection Curve
Sit. - #34 Sit. - #46
900

-- 900
...-'"

-------
SOO 800
/
Ol
a
:i 500
700
600 , ~
!t 800
:5 ~oo
700

/""
/
I
/ I
/.-
roo
/ i '00 ~
~ '00 ... 300
200
,;f ~
100
.,;; 200
100
F',f
!' !'
o o
o 0.2 0.' 0.8 0.8 1.2 o 0.05 0.1 0.15 0.2 0.25 0.3 0.35 a.' a.'5 0.5
Di~I-ncn.1 ~I-ntuo.

Load Deflection Cur ve Load Deflection Curve


Sit. - #47b 511. - #48b
1800 1800

14CO 1800
/::.---- 1400
~ ~

a. ,coo
1200
.~ X 1700
f
~
~
j:f ~ 1000
I 800 I
Jl
~ :ti " aoo
'i"
600
; ~ 600
'00 .00
.J'
700
if 200
IF
o o
a 0.2 0." 0.6 O.S 1 1.2 1.4 1.& 1.8 2 a 0.2 D.' 0.8 O.S I 1.2 I.' 1.6 \.8 2
Displ.ol:ima'lt - nc'-s Q;spIoc........, - ......

Load Deflection Curve Load Deflection Curve


Sit. - #53 511. - #62
'100 aoo
7DO
1000
600
SOO
R !t 500
~ ~
I 600 I .. 00

3" '00 ~ 300


200
200
100

0
0.1 0.2 0.3 0.' 0.5 0.6 0.7 0.8 0.9 0 0.\ 0.2 0.3 a.' O.~ 0.6 0.7 D.S
Oiliiploc:..".,-t - h:hes Displacwnll"ll - Inches

Load Deflection (;;urve


511. - #62

sao
R
:i
I 600

!" '00

200

0.1 0.2 O.J 0." o.s, 0.5 0.7


Orsptoc.-nn - n:r.s

756
Appendix 2 - Comparison of Test and Predicted
Load-Deflection Curves for the Testing Data Set

Load Deflection Curve Load Deflection Curve


1000
SII. - #28
, 900
Sit. - #5

900
~-
800
...... -alii
800
700
600
500 ./
~ ....,....-- R 600
3i1 500
I
700

,;/
, ~

'00
/ 8 <DO .,or
~ ..J .loa
.300
JF
laO
'00
I J
If-
o
o
r
0.1 0.2 0.3 0.'" 0.5 0.6
ZOO
100
o
o
, ;'

D. I 0.2 0.3 D.' 0.5 0.6 0.7 0.8 0.9


DisploclW11Wl! -,~ Ois.p!ot=:.,..,.,,1 - Incheli

Load Deflection Curve Load Deflection Cur ve


Sit. - #470 SII. - j48A

---
t 600 1500
! 1600
1 .. 00
.,-/::::::------ 1",,00
~

i ,..,...?/ .-~
ll. '200
\ 000
..' ~ '000
V
:
'00
jf
I
" BOO
f
600

'00
..,.., 3 eoo
l'
.P
<00
lOa
I 200
;
~
o a
o 0.2 0_4 0.6 0.8 , 1.2 1.4 1.6 1.8 2 a 0.2 0.' 0.6 0.8 I 1.2 I.' 1.6 1.8 2
Oisplocwn." - tlclws DispIoc..,...,t - """"'

Load Deflection Curve Load Deflection Curve


511. - j49A Sit. - #498

I
\600 \ 600

I.~
~ -<J
~
"00 1<00 Ta:sl data

, 200
~ 1200
~
;;-prldiclion

R '000
? 1000
,/
I'
.------ /

-----
3i
I 800 800

] 800
! '00
/
1/ //
'00

200 lr
'00
200 r
~ ~
a
a 0.2 0.4 0.6 0_8 1
Oisplocrnll'rl - k"lchu
1.2 1,4 1.6 1.6 a 0.2 0_4 0.6 0.8 1
Oisplac:.-nllnt - Inch...
1.2 1.4 1.6 UI

757
Load Deflection Curve Load Deflection Curve
Sit. - #50 Sit. - #51

I
, 500 1600

a. 1000
:;

"00

, 200

BOO
vi' /___
' ----- I~
1MI da1'o

;::-prediCl!on

1I
l:
I
1'00

1200

1000

800
f
~/' ------
] 600
V ~ 600
f
'00
.t '00
.F
}
.00
~
.00

a
r a
~
a 0.2 0." 0.6 0.8 1 1.2 1.4 1.6 1.8 o 0.2 0.' 0.6 0.8 1 1.2 I. ..
OLsplocltTlrl' - ftt-.s IJlspIoca-nonI - .......

Load Deflection Curve Load Deflection Curve


Sit. - #52 Sit. - #63
1500 700

, "o0 600
~
1.00
/ ....., ~
/.----- ~oo

X 1000
:;
~
l:
~ 400
/"
I

"3 600
600
ff
I
~ 300
/
' 200
/
'00

100
if ,00
/
r a
if
a
o 0.2 0." 0.5 0.6 , 1.2 1.4 o 0.1 0.. 0.3 0.' 0.5 0.6 0.7 0.8 0.9
Oisploc...,.,." - nc:t-.s: Oicploc.....,1 - ncNS

758
Pile Load Test Database and an Evaluation of the Program SPT91

John L. Davidson 1, Frank C. Townsend 1, Pedro F. Ruesta2 and Joseph A. Caliend0 3

ABSTRACT

A pile database has been developed at the University of Florida as part of an


ongoing research program with the Florida Department of Transportation. LOTUS 1-2-3
Release 3 was chosen as the database software because ofthe 3-D spreadsheet stacking and
the flexibility provided by the macros. This paper describes the database set-up, the menu-
driven system and how the database can be used to evaluate a pile prediction program.
Macros are used to generate Davisson, DeBeer, FDOT and Fuller-Hoy failure capacities
from the pile field load test results. Macros are also used, with the prediction program, to
calculate the predicted capacities and to provide comparison plots and statistics.

The program SPT91, written at the University of Florida and based on a method
originally proposed by Schmertmann in FDOT Bulletin 121, is used to illustrate how the
database has been structured to evaluate such a code. Comparisons between SPT91
capacities and those of the four load test methods can be made on different groupings of
piles from the database, e.g., based on pile diameter, length to diameter ratio and soil type.

INTRODUCTION

Deep foundation databases can serve two purposes. First, they are an organized,
detailed and valuable record of particular foundations. They contain infonnation on the
pile or shaft, e.g., diameter, length and method of installation; infonnation on the site,
usually an SPT profile; and the results of the field load test. The databases can therefore
be used, with suitable caution, to evaluate new foundations which have characteristics
similar to those in the database. The second use of the database is in the evaluation and
modification of prediction methods. If the necessary foundation and site information is

lprofessor, Civil Engineering Department, University of Florida, Gainesville, FL 32611


2Research Assistant, Civil Engineering Dept, University of Florida, Gainesville, FL 32611
3 Assoc. Professor, Civil and Environmental Eng., Utah State Univ., Logan, Utah 84322

759 Davidson et aI. (79)


available in a record, the predictive method, usually a computer code, can be used to predict
behavior under load. Then, since the actual field load test results are available, a compari-
son can be made and the particular method statistically evaluated.

This paper describes a Pile database and how it has been used to evaluate one
particular prediction code, SPT91. The SPT91 program was written at the University of
Florida and is based on original work reported by Schmertmann in FDOT Bulletin 121 in
1967. The latest program modification was in 1991, hence the designation, SPT91. Details
of the program are provided to this conference in the paper "Static Pile Capacity Predictions
with SPT91 " by Caliendo et al, The pile database was used to evaluate the predicted results
from SPT91 and compare them with the Davisson, DeBeer, FDOT and Fuller - Hoy failure
criteria, defined from the field load test.

THE PILE DATABASE

The database was created using LOTUS 1-2-3 R 3. 1 in order to take advantage of
the software's macro and three-dimensional capabilities, Individual pile records are stored
on successive sheets in the database. Menu driven macros are used to manipulate the data,
making it a simple matter to update the database as new data become available and to
perform statistical analyses on the records. Each LOTUS database file is limited to 120
records for diskette storage purposes. The database currently contains 72 data sheet
records containing all the parameters necessary for running the program SPT91.

A database file consists of four major parts; the Database Directory (Sheet A), the
120 Database Records (Sheets B - DQ), the Database Macros (Sheet DR) and the Database
Template (Sheet DS). Each pile record is listed in the Database Directory (Sheet A), which
serves both as a Table of Contents and as a Summary Table. The directory contains all
relevant data on the pile, the site, the predicted capacities and the load-test determined
capacities.

When a new pile record becomes available, it is added into the database with the
help ofa template which is created using the menu system. The template is placed after the
last database record sheet and has yellow text where data are to be entered, Red text,
which is later removed, provides guidance information. A record contains detailed
information about the pile, the site and the field load test. Once these data are added a
menu option is chosen to automatically create an input file for the computer program
SPT91. Predicted values of skin, tip and total capacities from the SPT91 output file are
then copied back into the record sheet. The record is then complete with regard to user
data entry and a summary of the data is copied to the Database Directory,

The Pile Database Macro sheet contains the 181 macros used in the database. The
vast majority ofthese are used by the menu system and are not individually available to the
user. The last sheet in the database is the Database Template sheet. This contains the pile
template used in the creation of new records and copies of all the screens used in the menu
system.
760 Davidson et at. (79)
The Database Menu System

The database is menu-driven. Local menus operate on a single database record


while global menus operate on a selected range of database sheets. When the database is
retrieved into LOTUS 1-2-3, an introductory screen, Figure 1, is presented for
approximately six seconds, after which the Main Menu, Figure 2, appears.

AXIAL LOAD

SIDE

RESISTANCE
11f ":

1
1..\ SPT91

Q - Q. + Qb
11
BASE RESISTANCE

Figure 1 Pile Introductory Screen

The Main Menu lists seven options. The first option allows the user to exit the menu
system to the current data sheet, the number of which is shown on the menu. Options 1
and 2 allow the user to select a database sheet number to work with by either scrolling
through the data sheets or by entering a specific database sheet number. Option 3 accesses
the Local Menu operations and Option 4 the Global Menu operations. Option 5 places the
data sheet template after the last recorded data sheet and makes this template the current
data sheet. The last option provides an exit from the database.

Local Menu Functions

The Pile Local Menu is shown in Figure 3. The menu functions operate only on the
current data sheet, i.e., on a single pile - in this case Pile Number 1. Option 1 provides a
checklist of available data on the current data sheet. Option 2 creates an input file for
SPT91. Option 3 creates an SPT input file for a program called PL-AID while Option 4
imports the results from PL-AID back into the Database. Option 5 creates the different
capacities from the pile load test load-settlement plot. Option 6 updates the database Pile

761 Davidson et al. (79)


o EXIT MAIN MENU TO CURRENT DATA SHEET.
1 SELECT SPECIFIC DATA SHEET BY SCROLLING.
2 SELECT SPECIFIC DATA SHEET BY DATABASE NUMBER.
3 GO TO LOCAL DATA SHEET MENU.
4 GO TO GLOBAL MItNU.
5 CREATES TEMPLATIt FOR NEW PILE DATA SHEET.
/I EXITS DATABASE.

CURRENT DATA SHEET NUMBER 1

Figure 2 Pile Main Menu

: LOCAL MENU FOR DATA SHEET 1

o EXIT TO CURRENT DATA SHEET.


I vaw CHECKLIST OF CURRENT DATA SHEET.
2 CREATE INPUT FILE FOR SPT!>1 PROGRAM,
3 CREATE SPT INPUT FILE FOR PL-AID PROGRAM.
4 IMPORT RESULTS OF PLAID PROGRAM.
S CREATE LOAD SETTLEMENT CAPACITIES.
6 UPDATE SHAFT DATABASE DIRECTORY,
7 GO TO LOCAL PRINT MENU.
a GO TO LOAD SETTLEMENT PLOT MENU.
!> GO TO COMPARISON SELECTION MENU.
10 RETURN TO MAIN MENU.

SELECT ).\'IENU OPTION: 5

Figure 3 Pile Local Menu

762 Davidson et al. (79)


Directory Sheet "With information from the current data sheet. Options 7, 8 and 9 are used
for accessing the Local Print, Load-Settlement Plot and Comparison of Capacities menus,
respectively. The last option on the Local Menu is for returning to the Main Menu.

The Local Print Menu is shown in Figure 4. Option I prints all five pages of data
sheet pile information with report quality. Options 2, 3, 4, 5 and 6 print specific pages,
with the information as stated on the menu. Option 7 prints all data sheet information with
draft quality (3 pages).

LOCAL .. PRINT MENU

o EXIT TO CURRENT DATA SHEET.


1 PRINTS ALL DATA SHEET INFORMATION.
2 PRINTS ONLY PILE DATA INFORMATION..
J PRINTS ONLY LOAD SETTLEMENT DATA.
4 PRINTS ONLY INSITU TEST IUl:SULTS.
S PRINTS ONLY SPT91 INPUT & OUTPUT DATA.
6 PRINTS ONLY PL.AID LOAD-SETTLEMENT DATA.
7 PRINTS ALL DATA INFORMATION (DRAFT QUALITY).
8 RETURN TO LOCAL MENU.

, ,',',

~ELECT'~ENtr OPTION ~ , ...-"'""":-':'-=-,

Figure 4 Pile Local Print Menu

The Load - Settlement Plot Menu, Figure 5 allows six different plots. Option 1
plots only the load-settlement data while Options 2 through 5 add the failure criteria of
choice and Option 6 the predicted PL-AID load-settlement. Option 7 prints a hard copy
of the last viewed plot. . Figure 6 is an example, a load-settlement plot showing the .
Davisson capacity construction, i.e., Option 2.

The Comparison of Capacities Menu, Figure 7, provides options of plotting SPT91


results versus the methods of failure criteria in different formats. Figure 8' is an example.

Global Menu Functions

The main function of the Pile Global Menu, Figure 9, is to perform statistical
analysis on all the piles (or on a selected group) in the database. Option 1 is for plotting
and evaluating data without any of the restrictions that are imposed in Options 2, 3 and 4.

763 Davidson et al. (79)


o EXIT TO CURRENT DATA SHEET.
1 LOAD-SETTLEMENT PLOT PROM LOAD TEST.
2 LOAD-SETTLEMENT PLOT WITH DAVISSON CRITERION.
3 LOAD-SETTLEMENT PLOT WITH FDOT CRITERION.
4 LOAD-8ETTLEMENT PLOT WITH PHWA CRITERION.
S LOAD-SETTLEMENT PLOT WITH FULLER-HOY CRITERION.
6 LOAD-8ETTLEMENT PLOT WITH PL-Am PREDICTION.
7 PRINTS CURRENT PLOT.
8 RETURN TO LOCAL MENU.

$ELECT . ME.NU OPTION: 5

Figure 5 Pile Load-Settlement Plot Menu

500

400

'"' 300
III
..
----- - - - - .
Z
o
t 200
CI
0(
o X=0.15 + DI120
"" 100

o
o 0.5 1 1.5 2.0

SETTLEMENT (INCHES)

Figure 6 Load-Settlement Plot with Davisson Construction

764 Davidson et al. (79)


COMPARISON OF CAPACITIES MENU

o EXIT TO CURRENT DATA SHEET.


1 PLOT SPT91 VS. DAVISSON.
2 PLOT SPT91 VS. FDOT.
J PLOT SPT91 VS. FHWA.
4 PLOT SPT91 VS. FULLER HOY.
~ PLOT ALL FAILURE CAPACITIES (BAR GRAPH).
6 PLOT ALL DESIGN CAPACITIES (BAR GRAPH).
7 PRINT CURRENT PLOT.
a RETURN TO COMPARISON SELECTION MENU.

SELECT MENU OPTION: 51

Figure 7 Pile Comparison of Capacities Menu

LOAD (TONS)
o 30 60 90 120

SPT91 PRE

Figure 8 Bar Graph for Comparison of All Failure Capacities

765 Davidson et al. (79)


o EXIT TO DATABASE DIRECTORY.
1 PLOT BY DATABASE SaEET NUMBER RANGE.
2 PLOT PO-E DIAMETER RANGE.
3 PLOT BY LID RATIO RANGE.
~ PLOT BY SOn. TYPE.
S GO TO GLOBAL PRINT MENU.
6 RETURN TO MAIN MENU.

SELECT MENU OPTION: 5

Figure 9 Pile Global Menu

When Option 1 is selected, a menu appears and the user enters information concerning the
characteristics of the data to be analyzed, Figure 10. The database sheet number range,
method of comparison, criterion (failure or design values) and selection of either analysis
of capacity or settlement are prompted for.

,DATABASE, SHEET,'NUMBER: RANGE

AVAll..ABLE DB SHEET NUMBER RANGE: I-no


METHOD CRITERION CAP {SET OPTION
1 DAVISSON 1 FAILURE 1 CAPACITY
2 DEBEER :z DESIGN :z SETTLEMENT
3 FDOT
4 FULLER-HOY

ENTER MIN DB #I RANGE: ~~~iii


!! ENTER METHOD, k'-

ENTER CAP/SET: rc- -::t:=!l

ACCEPT ENTIUES (YIN) : ~~

Figure 10 Pile Database Sheet Number Range Menu

766 Davidson et al. (79)


The data that fit the parameters are collected from the Database Directory sheet and a plot
of SPT91 predicted capacity versus the chosen criterion capacity shown: Pressing Enter
exits the plot and provides a table of characteristics and statistics. Finally a menu prompts
the viewer to either continue without a print or print the statistics and plot on a single page
with or without data-labels. Figure 11 is an example print out.

COMPARISON OF CAPACITIES
SPT91 PREDICTED Vs. DAVISSON FAILURE
900

...-..
CI)
Z 800
0
E-<
'-"
700
/
>-
-
E-<
600
1/
~/
U r
<C
lJ,.,.
<C SOl

~
U
Cl 100
l.Ll
E-<
~~/ ~

-
U
Cl
300
~I!l ~ ~

~~
-\!Sl
l.Ll ~~ ~ ~
~
~ ~ c;;:
lJ,.,. 200 .."

......
~
lJ,.,.
,100 ~ II""
~ ~
~ ~ ~ 181

CI)

iJ ~
181

200 ~ m ~ 600 ~ ~ ~

DAVISSON FAILURE CAPACITY (TONS)'

METHOD~ DAVISSON DATABASE # RANGE: 1- 72


CRITERION : FAD..URE
CAP/SET: CAPACITY

, MIN 'SPl91: 23.3 IONS NUMBER, OF EVENTS: 61


MAX SPT91: 516.7 TONS, ,MIN 'lit SPT91/ METHOD :' " " 36.3 'lit'
AVG SPT91 :, 180.6 TONS, ' MAX 'lit SPT91 / ..ElliOD : 262.1 'lit'
AVG (L.OG) 'lit SPT91/METHOD: 82.5 'lit
M1N METHOD: 14.8 TONS 'SfANDARD DEVIATION (lOG): ' 20.1 'lit
MAX METHOD': , ,698.9' TONS, ERROR OF ESTIMATES: 124.0 . TONS'
AVG MElliOD: 237.5 ' TONS

'Figure1! Typical Comparison Plot and Statistics Table,

767 Davidson et aI. (79)'


Option 2 on the Global Menu is for plotting and evaluating data in a particular
diameter range. A menu, Figure 12, prompts the user to enter the desired diameter range.
The minimum and maximum diameters available are shown on the menu. The data that
fit the diameter range are collected and prepared for viewing. Again a capacity
comparison plot and statistics table can be printed. Option 3 is similar to Option 2 except
that length-to-diameter ratios rather than diameters are chosen.

AV AILABLE DIAMETERS IN RANGE OF DB # RANGE

MINIMUM DIAMETER MAXIMUM DIAMETER

10.00. INCHES 30.00 INCHES

ENTER MIN DIAMETER FOR RANGE' !!jf!t~l:'2i~r;ij

ENTER MAX DIAMETI!:R FOR RANGE: ih1!iE2'4'liOI

Figure 12 Pile Diameter Range Menu

Option 4 is for evaluating data by soil type. The procedure is similar to that for
Options 2 and 3. Four soil types are considered -- clay, silt, sand and rock -- for both
primary side soil and soil at the base of the pile. Clay has been given the designation "1".
Therefore, if a pile were embedded in clay only (side and tip) it would have the designation
"11", indicating that the primary side and base soils were clay. Silt has been designated
as "2", sand as "3" and rock as "4". Although there are sixteen available soil type
designation options (see Figure 13), a maximum of six can be analyzed at one time. The
first prompt on the menu is to enter the number of different soil types to be analyzed. The
cursor then moves to the Soil Type Choice position, where the soil type designations are
entered. If the user enters "2" different soil types to analyze, then the cursor will move to
the Accept Entries position after selecting the second soil type designation. The available
soil type designations are shown in red while the unavailable soil types are shown in blue.
A count of soil type designations is shown to offer assistance. After entering each soil
type, the color ofthe number designation on the menu changes from red to blue, indicating
that the soil type is no longer available for selection.

768 Davidson et al. (79)


PRIMARY SIDE SOIL f BASE SOIL
11 CLAY/CLAY 21 SILT/CLAY 31 SAPID/CLAY 41 ROCK'CLAY
12 CLAY/SILT 22 SILT/SILT 32 SAHD/SD..T 42 ROCIUSn..T
13 CLA Y/SA:MD 23 SILT/8AN"D 3.5 SAND/SAND 43 1l0CKJ'SAMD
14 CLAYIR.OCK 24 SILTIROCK 34 SAN"J)IROCK 44 ROCIUROCX

RED"'I SOIL TYPE J$ AVAn..ABLE IN DB'" IlANGE~


BLUE _. SOIL TV.:': 18 NOT AVAD..ABLZ IN' 'DB. RANGE OR
BAS _ .. EN PREVIOUSLY SELECTED

ENTER N OP!L-~~ri@: '"


. DIFFERENT SOIL"'" .... ~~~:~::g;~E. .t4f ..
TYPES TO. VXE"W: .. N;UMBER:;;.. 1 .
(6 MAX) ,

ACCEPT ENTRIES (YfN): ~5~~~

Figure 13 Soil Type Range Menu

Option 5 accesses the Global Print Menu which regulates printing of the Database
Directory. A complete printout of this directory totals 21 pages. The menu allows the user
the option of printing all or only certain ranges of the directory. Another Option prints the
macros names accessible to the user, the key strokes required to run each macro and a brief
description. The final Option prints all database macro names and the location in the
database where each macro can be found.

EVALUATION OF SPT91 PREDICTIONS

It is not the intent of this paper to specifically evaluate the program SPT9I but
rather to illustrate, using SPT91, how the database has been structured in order to evaluate
any such code. As already noted, database macros are employed to generate, from the pile
field load test results, the Davisson, DeBeer, FDOT and Fuller-Hoy failure capacities.
Macros are also used with the SPT9I program to calculate the predicted capacities. The
Global Menu system then allows comparison plots and their statistics to be generated for
different groupings of piles, e.g., by diameter, lengthto diameter ratio and soil type.

Figure II illustrated a typical output. In this case the SPT9I predicted capacity was
being compared to the Davisson failure capacity. Data points located above the 45 degree
line represent unconservative SPT9I predictions versus the failure criterion used in the com-
parison. Points falling below the 45 degree line represent conservative results and points
falling on the line display perfect agreement. The statistics table shows that in this case there
were 61 piles available for comparison. The table also provides a number of minimum and
maximum values, averages, a standard deviation and an error of estimates. The program
SPT91 on average predicted a capacity equal to 82.5% of the capacity detennined from the
field load tests and using the Davisson criterion.

769 Davidson et aI. (79)


The ratio of SPT91 predicted capacity to the capacity from the method being
analyzed, designated JR, is not normally distributed but is skewed. If underprediction
occurs, JR has a value between.zero and one. If overprecliction occurs, JR can have a value
between one and infinity. Ratio JR is more closely log normally distributed. Average and
Standard Deviation values were therefore calculated in the database using the equations:

Avg JR = L log R
10 n

STD L (log JR - log Avg JR)2


1/2

[ ]
n - 1

Table 1 is a summary of how SPT91 predictions compared with the four failure criteria.

Table 1 Statistics for SPT91 Predicted vs. Measured Failure Criteria for Database
Range 1 to 72

Error
Number MinJR MaxJR AvgJR Stand.
of
Method of Dev.
Estimates
Events
(tons) .
% % % %

Davisson 61 36.3 262.1 82.5 20.1 124.0


FDOT 67 42.6 274.2 110.4 18.2 76.3
DeBeer 62 31.9 210.1 82.1 19.2 133.2
.Fuller-Hoy 71 30.0 259.1 73.1 19.3 153.9

Comparisons by Diameter Range

The smallest diameter of the 72 piles examined was 10.0 inches and the largest was
30.0 inches. This 10 to 30 inch range was divided into five diameter ranges. Statistics for
each range were collected to see if any trends in the data were evident. Table.2 is a
tabulation of the results for all five diameter ranges for the Davisson comparison. Figure
14 is a plot of average JR values versus diameter of piles for the Davisson criterion. Based
on the records available in the database it would appear that there is a slight trend towards
more conservative SPT91 predictions (i.e., lower average JR values) with increasing pile
diameter. Similar comparisons and plots can be made based on length to diameter ratio.

770 Davidson et aI. (79)


Table 2 Statistics for SPT91 Predicted vs. Measured Davisson Criterion by
Diameter Range

Error
Diameter Number Min lR MaxlR Avg lR Stand.
of
Range of Dev.
Estimates
Events
(tons)
(in) % % % %

10 - 14 23 41.7 262.1 97.0 21.5 64.6

14 - 18 21 41.4 226.6 79.8 20.8 72.7

18 - 22 17 41.4 210.5 93.6 17.5 52.6

22 - 26 11 38.4 88.4 56.9 10.8 165.2

26 - 30 9 36.3 111.4 67.4 18.4 234.2

100
.~
e
,...,
80
/ \
l;I.:l
~
....:l
-<
>
~
60
\ ~
/
l;I.:l
c:;l
40
;2
l;I.:l
>
-< 20

0
o 10 20 30

AVERAGE DIAMETER IN RANGE (INCHES)

Figure 14 Plot of Average lR Values vs. Diameter Range for Davisson Comparisons

771 Davidson et al. (79)


Comparisons by Soil Type Range

There are a total of sixteen possible soil combinations using the Soil Type option on
the Pile Database Global Menu. Currently, there are no data in the database that fit the five
categories Clay/Rock, Rock/Clay, Rock/Silt, Rock/Sand and RockIRock. Therefore, eleven
soil type combinations can be evaluated. Table 3 is a tabulation of the results for the
Davisson criterion and Figure 15 a bar graph comparison. It is also possible to group the
piles having the same side friction soil but different bearing soils, or having the same bearing
soil and different friction soils. It would appear that SPT91 on average predicts conserva-
tively low in clayey soils and approximately correctly in sandy soils.

Table 3 Statistics for SPT91 Predicted vs. Measured Pile Capacity using the
Davisson Failure Criterion by Soil Type Range

Error
Soil Number Min lR Max lR Avg lR Stand.
of
Type of Dev.
Estimates
Range Events
(tons)
% % % %

Clay/Clay 2 42.3 50.5 46.2 5.4 257.8

Clay/Silt 3 66.8 83.7 77.3 5.5 100.7

Clay/Sand 3 53.1 70.0 61.1 6.0 103.6

Silt/Clay 2 70.9. 91.6 80.6 7.9 92.9

Silt/Silt 7 36.3 140.6 66.9 20.9 216.7

Silt/Sand 7 37.8 132.6 66.5 20.5 173.1

SiltIRock 3 54.0 210.5 106.9 29.5 104.4

Sand/Clay 2 110.4 114.1 112.3 1.0 17.5

Sand/Silt 3 71.0 110.6 84.9 10.1 71.5

Sand/Sand 24 41.4 262.1 97.1 22.2 70.5

SandIRock 5 54.7 106.8 80.8 11.6 60.7

772 Davidson et al. (79)


125

106.9
If - - - - - - - - - - - - - - - - - 1""
'<I---.
100 -
-
~ 77.3
80.6
r--: <:"
OIl
75
...
~

~
>
PC 46.2
SO
'"
I
1:.:1
~ .. ::-1.
~ ~.
OIl 2S I-::l,
>
<I: .::r.';"~:
~
~ ;;:'
~' -..'?':-

- ...-~
-+
~'?'.':'
0
;.. l::l ;.. l::l :lIlI
:z: ;.. :lIlI
-<
...l
(.)
""
::l
~
-<
~
-<
...l
!-<
::l
l::l
z
-<
(.)
0
...-< !-<
t:l
:z:
-<
(.)
0
~ ;.. ;.. S::! ~ ~ !!! S::! ~ ~ !!!
!-< !-< l::l l::l l::l l::l
-< -< <I: !-< !-< :z: :z: :z: z
...l
(.)
...l
(.)
...l
(.)
...l
<ii ::l
'" =
'" =
'"
<I:
l
-<
l
-<
'"
<I:
'"

Figure 15 Average IR Values vs. Soil Type Range for Davisson Comparisons

CONCLUSIONS

A pile database has been developed. It forms a record of foundation members, their
sites and field load tests. Menu-driven macros are used to manipulate the data, making it
a simple matter to update the database as new data become available and to perfonn
statistical analyses on the records.

Computer prediction codes, such as SPT91, can be easily evaluated by comparing


predicted capacities with capacities detennined from the field load tests. Comparisons can
be made on selected groupings from the database, e.g. by diameter, length to diameter ratio
or soil type. This allows the developer of the code to identify specific shortcomings and
modify his program for particular situations.

REFERENCES

Ruesta, P. F., Pile Load Test Database and an Evaluation of the SPT91 Program, Master
of Engineering Thesis, University of Florida, Gainesville, Florida, 1993.

Davidson, 1. L. and Townsend, F. C., Maintenance of Load Test Data Bases, Final Report
to Florida Department of Transportation, Gainesville, Florida, 1994.

Caliendo, 1. A., Bartholomew, M., Lai, P.W., Townsend F. e. and McVay, M.e., Static
Pile Capacity Predictions with SPT91, International Conference on Design and
Construction of Deep Foundations, Orlando, 1994.

773 Davidson et al. (79)


Design Parameters for Steel Pipe Piles Driven in a Soft Rock
Yuji Michi 1, Makoto Tsuzuki 2 and Tatsunori Matsumoto 3

Abstract : Standard Penetration Test is widely accepted as a soil parameter for


foundation pile design in Japan. The SPT's may be useful for certain soil types but
often are blindly applied to any soil type without reference to the soil parameters
obtained from laboratory tests and other in-situ test results. This paper discusses the
reliability of the soil parameters obtained from SPT, CPT and laboratory tests, and the
methods for foundation pile design specified by the several Japanese organizations and
well-known methods world wide such as AASHTO and API. It also verifies it through
the static load testing of an open-ended steel pipe pile in the diatomaceous mudstone in
Noto Peninsula. The cast-in-situ concrete pile in the alternative layers of sand and clay
in Osaka is also evaluated.

INTRODUCTION

A new highway route is under planning in Noto Peninsula, Japan. Most of the
route area in Noto Peninsula is covered by a diatomaceous mudstone having relatively
high compression strength, which is classified as stiff clay or soft rock. A pile
foundation will be applied for the new highway. Hence, the test piling was performed
at Nanao district on Noto Peninsula in 1991 in order to gather the reliable soil
parametric data and to evaluate the design methods of the pile in the mudstone aiming
to provide the high quality foundation. The test program consisted of various soil
investigations, static loading tests and dynamic loading tests.
Such a prudent test program was desirable due to troubles experienced during
the installation of foundation piles in the mudstone during construction of the
Notojima Bridge (Nishida et al. 1985). The pile driving refused far earlier than the
design penetration depth which was derived from an empirical equation based on the
N-value from the Standard Penetration Tests (SPT's). The equation is proposed in the
Specifications for Highway Bridges in Japan

I Yuji Michi, Yoshimitsu-gumi Co., Nagasaki-cho, Komatsu, Japan


2 Makoto Tsuzuki, Managing Director, Fugro McClelland Japan Corp.,
1-21-2 Jingu-mae, Shibuya-ku, Tokyo, Japan.
3 Tatsunori Matsumoto, Department of Civil Engineering, Kanazawa University,
2-40-20 Kodatsuno, Kanazawa, Japan.

774 Michi, Tsuzuki & Matsumoto


This paper firstly describes the result of soil investigations and the static loading tests.
Then, the ultimate and allowable pile capacities predicted by the different design
methods commonly used in Japan and other countries are compared with the test result.
For a cast-in-situ concrete pile installed in the alternative layers of sand, silt and clay,
the above mentioned test and evaluation procedure were also applied. Finally, a
discussion is made on an appropriate combination of the soil parameters and design
methods based on the series of tests.

CODES IN JAPAN AND OTHER COUNTRIES

Several codes and standards for foundation pile design are available in Japan
Japan Road Association(1990) (called JHA code for abbreviation in this paper) ;
Architectural Institute of Japan (1988) (lAC code); Japan Port and Harbor Association
(1991) (JPH code) ; The Japanese National Railways (1986) (JR code). The empirical
equations are proposed in these codes, in which unconfined compression strength qu
or N-value are used for the soil parameters.
The empirical equations used in the Japanese codes are summarized in Table 1

Table 1 Summary of Japanese codes for pile design (1 tf/m 2 = 9.8 kN/m 2 )

Pile type Soil JHA code JR code JPH code JAC code
type
Unit Disp. pile Sand 30N 25N 30N 30N
Ultimate (for N-;;:;'40) (limit=800)
Point
Capacity Clay 30N 4qu or 8N 8cu 6cu
qp (tflm 2 ) (for N>20) (limit=1000)
Non-disp. Sand 300 7N 015N
pile (for N'6 30) (limit=350) 0=0.5
Clay 3qu 3qu or 6N 6cu
(limit=900)
Unit Disp. pile Sand 0.2N 0.2N 0.2N Nf3
Ultimate (limit =10) (limit =10)
Shaft Clay Cu or N qu(2 or N Cu ~(qu(2)
Capacity (limit =15) (limit =10) (limit =10)
qs (tflm 2 ) Non-disp. Sand 0.5N 0.2N Nf3
pile (limit =20) (limit =10)
Clay Cu or N qu(2 or N qu{2
(limit =15) (limit= 8)
Safety Disp. Pile End bearing qp=0.3 Fs '62.5 Fs =3(total)
pile: Fs=3 qs=0.3 Fqp =3
or FQs=3
Non-disp. Shaft bearing qp=O.6 Fs ;SZ.5 F s =3(total)
Resistance pile pile: Fs=4 qJqs=0.3 F qp =O.ldl3;S::'
Factor FQs='X~
I
- .-
Rp=Apqp Rp : Ultimate Point Capacity A p : Closed Area d : pile diameter
R s = Asqs Rs : Ultimate Shaft Capacity As: Shaft Area

775 Michi, Tsuzuki & Matsumoto


The undrained shear strength, Cu, is usually taken as cu=qul2. Allowable bearing
capacity, Ra, is estimated using total safety factor, Fs , or partial safety factors, Fqp and
F qs , as follows in Japan excluding JR code:

Ra=RlFs , ,.(1)
or
R a = Rp/Fqp + RslFqs (2)

where R is the total bearing capacity consisted of the end and shaft capacities, Rp and
Rs are the end and shaft capacities, respectively. For convenience in the following
discussion, the total resistance factor, <j), and the partial resistance factors, <j)qp and <j)qs,
are defined as <j)=l!Fs, <j)qp=I/Fqp , <j)qs=l/Fqs for these Japanese codes.
Some representative codes in other countries were also examined: Baker et aL
(1991) (AASHTO); Canadian Geotechnical Society(1985) (CFEM); Det Norske
Veritas (1977) (DNY); ISSMFE(1977) (CPT method); American Petroleum Institute
(1980) (API method). In contrast with the Japanese codes, they utilize various soil
parameters such as cone resistance qc and sleeve friction is from CPT, effective
overburden pressure ov', internal friction angle cjJ, bearing capacity factors, as well as
qu and N-value. A distinctive feature of these codes is flexiblity to apply plural design
methods and equations associated with different resistance factors.

R a = <j)qp Rp + cjJqs R s , , (3)

Here, cjJqp and cjJqs are the resistance factors for the end and shaft capacities respectively.
Rp and R s are the estimated ultimate end and shaft capacities

STATIC LOADING TESTS IN NANAO SITE

Test Piles and Ground Conditions

The layout of the piles and the soil investigation points are shown in Fig. l. The
open-ended steel pipe piles were driven by a diesel hammer having a piston weight of
4.2kN. The test piles designated as TI, T2 and T3 each had a penetration length of
8.2m. The pile geometry and properties are listed in Table 2 Each test pile was
instrumented with strain gages at 10 levels in total. A steel channel was used for
protection of the strain gages which increased the cross-sectional area of the piles to
0,041m2
Several investigations were conducted in 8 boreholes Note that the borehole BI
through Bg were drilled in the virgin ground prior to installation of the piles. The SPT's
were conducted in boreholes Bs and B6. A number of unconfined compression tests
were carried out using the soil samples from boreholes. The ] I CPT's were performed
at the site, The soil profile and the results of SPT's, CPT at point C2 and the
unconfined compression tests are shown in Fig.2 The test site was characterized as a
thick deposit.

776 Michi, Tsuzuki & Matsumoto


Table 2 Geometrical and mechanical ro erties of iles
Pro ert Value
Length L (m) 11.0 (Piles Tl and T2)
11.5 (Pile T3)
Outer radius ro (nun) 400.0
Inner radius Ti (nun) 387.9
Wall thickness t w (nun) 12.1
Cross-sectional area A (m )2 0.041
Youn 's modulus E MPa 2.06 x 105

oC 4
Legend
e8 5
Borehole
e8 4
o Piezocone
R1 T3 R3
e8 3
0 Ci) oCg @ cfl1
T -8 2 Ca T2 C 10
Plate ~ ~B1
loading
test
~
C5 0
- Ba 0 C 7

Cs
0R 0 0
C2 0R
2 R4 5

o 1 23m

Fig. 1 Layout of piles and soil investigation points at Nanao site

(b) (c) (d) (e) (f) Pile T,


(a)
SPT N-value qc (MPa) f5 (MPa) qu (MPa) Esc (MPa)
o 5 10 15 20 25 0i'-::::i=::i""":;"-'T---; 0.05 0.1 0.15 02 0r---,.......::,0~4..,......:.;0..::....8.,...-,1.2 0 20 40 60 80

o~
= lh o
I -2 til
DJ'O
N :J lJ) o
C -4 0 c
.2 80
rn- o
~ -6 E-2
lJ) 0 :J
W -a.~ E
o
-10 o o
-12

Fig.2 Ground conditions at Nanao test site (after Matsutmoto et at, 1993)

777 Michi, Tsuzuki & Matsumoto


of the saturated diatomaceous mudstone. The unconfined compression test results
expressed the inherent properties of the test site (Matsumoto et aI., 1993). It is seen
from Fig.2(e) and (f) that the variations of qu and the secant modulus 50 with depth
are relatively unifonn up to the depth ofT.P.-12m (T.P: Tokyo Datum). The variation
of N-values and the cone resistance, qe, are also uniformto the same depth.
The variation of sleeve mction, /S, had a shape of a saw. The high mction was
measured at the beginning of continuous penetration after stoppage for connecting the
rods at 1m interval for a couple of minutes then decreased with the progress of
penetration. The pore pressure, pw, profile indicated the opposite of the fs profile. Such
soil behaviors influence to the pile set-up phenomena.

Results of Static Loading Tests

The design penetration depth of a steel pipe pile in Japan is usually defined by the
. seismic load. Hence, the shaft resistance to the tension force due to the seismic load
should be calculated at the comer pile of the footing. For this purpose, the static load
test (SLT's) were carried out in different manners depending on the test pile type. The
ordinary vertical cyclic loading test was perfonned on piles II and T2 as well. However,
the inside soil of the latter pile was removed in advance so that the outer shaft
resistance only was mobilized. The pull-out test was performed on pile T3 in order to
know the behavior of the shaft resistance under compression and tension forces.
The load Po vs. displacement Uo curve at the head of pile TI is shown in Fig.3
The ultimate capacity P u lt{=4.7MN) was attained at the 4th cycle of the loading steps.
The displacement of the soil plug, Uplug, was also measured at the surface of pluged soil.
The change of Uplug, was identical with Uo throughout the test. This indicated that the
pile TI reached the ultimate state with the perfect plugging mode.
The force distributions along the pile generated by the compression loads are
shown in Fig.4. Most of the load was sustained by the shaft friction at each loading
steps When the compression load of 4.7MN was applied at the pile head, 1.3MN was
transmitted to the pile base and remaining 3.4MN taken in shaft resistance.
The Po vs. uo curve of the pile T2 is shown in FigS The test results shows that
pile T2 had an ultimate capacity P ulr.=3.8MN which was 09MN smaller than that of pile
TI The Po vs. uo curves of piles TI and T2 are compared in Fig.6. It is seen that both
curves are identical until Po reached 3.MN. This may imply that only outer shaft
resistance motivated in pile TI until Po reached 3MN and the inner shaft resistance,
which was always equal to the soil resistance below the soil plug, was then motivated
for further increase in Po. The axial force distributions of pile T2 (Fig.7) were also
identical those of the pile TI until Po was increased to 3MN.
The relationship between the pull-out load Po and displacement 110 of the pile T)
is shown in Fig.8. The ultimate capacity was 3.7MN which was only O.IMN smaller
than that of pile T2 under the compression load. This may suggest that the shaft
resistance under tension and compression loads are nearly identical. The axial force
distributions of pile T3 (Fig.9) were similar to those of the piles T] and I2, although the
axial forces of pile T3 were tensile. This means that the distribution of shear stresses
along the pile surface under tension and compression loads is similar.

778 Michl, Tsuzuki & Matsumoto


Pile head force, Po (MN)
00 1 2 3 4 5
- - - ---
.--.
E 20 - -- ----
- - -- -
--- -
--
E
0
:l
40

....c- 60
OJ
E 80
OJ
0
co 100 -- - - -- - -
D-
C/)
120
"'0
"'0
loading
co 140
------ unloading I
I

OJ
.c I
I

OJ 160 I
I
0.... \
180 I
I
- - - - - - ____ J

200

Fig. 3 Load-displacement curve of pile Tl at the head in compression test

Axial force, Px (MN)


1 2 3 4 PileT 1
00 o (EI.+4.14m)

1 1
.--.

--Ex- 2 ~.1m
3 ., (EI.+1Am)
"'0
co
OJ
.J:: 4 4
Q)

D- 5 5

E 6 6

-a'--
Q)
0
c
7

....ro 8
.~ 9
0
10

11 11

Fig. 4 Force distributions along pile Tl

779 Michi, Tsuzuki & Matsumoto


Load on pile head, Po (MN)
---E ___
E
Load on pile head, Po (MN)
o 1 2 3 4 5
-E
o
:::J
00 1 2 3 4 5.S-
o
:::J
oG"'-==--.-----r----,r---,----,
~ 10 ~
c 10
CD
E 20 E
CD CD ~,
()
--loading () ,
eu 30 ctl 20 "
c. - - - - - unloading 0.. '(j)
(J)
I "~ I

:a 40
,I
I "0
"0
I

I
~ I eu 30 -<> - - pi"Ie
T l 'I
1? 50 I
I

~
Q) - -

- 6 - pile T2 i
CD -----",
0:: 60 ' - - - _ - L - _ . . . . l . - _ . . . . . . L . . _ - - - I . _ - - - J Cl: 40 .........._-'--_--'--_--L..._--'-_...J..'....J

Fig. 5 Load-displacement curve of pile Fig. 6 Load-displacement curves of


T2 at the head in compression test piles Tl and T2

Axial force, Px (MN)


o 1 2 3 4 Pile T 2
o ';----,-----;....----r-------::I=---r--......,.;;-~---, 0 (E I. 4.01 m)

1 1

Ix 2 2
(EI. 1.18m)
"0 3 3
co
~ 4 4
Q)
"0. 5 5

E
o....
6 6
'+-
Q) 7 7
()
c 8
.....eu 8
.~ 9
o 9
10 10
11 L.--'----'----'-------'_.......-~--'-____' 11

Fig. 7 Force distributions along pile T2

780 Michi, Tsuzuki & Matsumoto


140...--....,..---r-----,.----,..------,
E
-E
~-c
o
_ c::
120

~
(D
E100
(1)
...c.u
(D Ci3 80
=0.
Q.U)
a '0 60
ci.12
0-r.n a.co~ 40
....
1:)
:::J _~ __-

o 20\..---=-=---~--
I
::J
a... o~--~-~:'-.L...-.-.....l-.-~
o 1 234 5
Pull-out force at pile head, Po (MN)
Qn tension)
Fig. 8 Load-displacement carve of pile TJ .at the bead in tension test

Axial force, PlC (MN)


(in tension) 4 Pile T
oo
3
1 2 3
i--..--....,..,.---r-----F.-r-~Tr-__...,~ 0 (8. 4 .18m)
,
\

1 \
, I

-E
X
"D
2
3 +-+-4-t--t--lr-~
I

.-\..
J


2~f
3 1.01m
(8. 1.01m)
ltl
(I) , ,
4 4
~
<D 5

5
'Q.
E 6 6
,g .
7 7

g
(I)
8

,, 8
~ ,
U) 9 ," 9
o at the ultimate
10 state '0

11
"
Fig. 9 Force distributions along pile T3

781 Michi. Tsuzuki & Matsumoto


STATIC LOADING TEST IN TENNOUJI SITE
Test Pile and Ground

Tennouji is located in the central area of Osaka. A soil investigation including


SPT's, unconfined compression tests, triaxial undrained shear tests (UU tests),
oedometer tests was carried out. The soil types with N-value and the laboratory test
result are shown in Fig. 10. The test site consisted of alternative layers of gravel, clay
and sand, which is typical of the Osaka area. A total of 5 SPT's were conducted in the
site. The variation of N-values at the test pile location is plotted by the solid circles in
Fig.1 O(b). The value of undrained shear strength from UU test is comparable with qu
values. It is seen from the effective overburden pressure, av', and the pre-consolidation
pressure, pc, that the test site is over-consolidated with OCR's in a range of 5 to 7. The
pile was a cast-in-situ concretepiIe, having a diameter of 1m, a length of 26. 5m and a
penetration length of 20m, installed by the reverse circulation drilling method.
(a) (b) (c) (d)

n
(e) (f)
SPT N-vaJue w (%) su,q.)2(MPa) E so (MPa) PC' a v'(MPa) Pile T,
o 0r--..--:2::,:::0~----:;4O,,----..--.::,60. 0 20 40 60 80 0 01 02 o 20 40 o 0.4 08 12

I~
grav61
.". bed
'----'
-10 o. 0 o
g
N
C
0-20 ". 0

~
.~

>
<ll
i:U o Quf2
-30 .0 0 o
CIt 0 o

o
00 0 0 0
\
o

Fig.10 Ground conditions in Tennouji test site (after Ueda et aI., 1993)

Results of Static Loading Test

The Po YS. Uo curve from the static test is shown in Fig.II. The uo increased
abruptly when Po increased to 5.9MN. It may be adequate to adopt this load for the
ultimate capacity which is called P u /t2 hereafter. On the other hand, the Japanese
Architecture code prescribes that the ultimate load, Putt, is defined at the load
corresponding to a displacement 10% of the pile diameter. Thus estimated Pull, called
PultI, is 9MN which is 1.5 times as large as P u/t2.
The force distributions along the pile in the static loading test are shown in
Fig.12. When PO=P utt2=5.9r-vrN was applied, only I.Ir-vrN was transmitted to the pile
base and the shaft resistance was 48r-vrN. When Po=P u/tI=9.8r-vrN was applied, the
load transmitted to the pile base increased to 4. 6r-vrN while the shaft resistance was
increased to 5.2r-vrN which was only O.4r-vrN larger than that at Po=5.9r-vrN. Since a
safety factor F s is usaually taken as 3 in Japan, then it is clearly seen that the working
load ranging from 2r-vrN to 3r-vrN is generated by the shaft resistance only.

782 Michi, Tsuzuki & Matsumoto


Pile head force, Po (MN)
00 2 4 6 8 10 12
..,----,-,--

-E 20

-E 0
40

--
::J

c
Q)
60
80 Pu1t 1
E
Q)
(.)
100
D!sp"_o-.!..!0%_ di~m. ____
eel
a.
C/)
120
:0 loading
"0
a:l 140 ------ unloading
Q)
..c
Q) 160 ~-------------------
a:: 180
200

Fig. 11 Load-displacement curve at the pile head


(after Takamori et al. 1993)

Axial force, Px(MN)


00 2 4 6 8 10 12 Pile T,

gravel

---E
~
4
bed

x
diluvia
-g
Q)
8 clay
..c
Q)
0. 12
sand
E
,g 16
day
Q)
()
c
i9 20
(/)

sand
o
24 ca
sand

Fig. 12 Force distributions along the pile


(after Takamori et al. 1993)

783 Michi, Tsuzuki & Matsumoto


COMPARISONS BETWEEN MEASURED AND ESTIMATED CAPACITIES
ON NANAO AND TENNOUJI TEST SITES

The upper bars in Figs.I3 and 14 indicate the shaft and end capacities calculated
by the different codes. The lower bars indicate. allowable shaft and end capacities
multiplied by the resistance factors. The qu and N affix to the code names indicate the
soil parameter used in the calculations. The numbers in parentheses are the values of
resistance factors, <Pqp and <Pqs. For The Tennouji site (Fig.I4), a set of measured
values of the end and shaft capacities, Rpm and Rfm, at Po=5.9MN (PuIt2) was derived
from the yielding load and 9MN (Putt 1) was derived at displacement 10% of the pile
diameter.

Ultimate Capacities in Nanao Site

The rnA N code prescribes that the end capacity should be zero when SPT
results provide N-values below 20. The values of the total capacity, R, calculated from
Japanese codes, except JHA Nand JR N codes, are comparable with measured total
capacity and the portion of Rp and R s derived from JAC qu code shows good fit
among Japanese codes. The R calculated by CFEM N code might be compared with
the measured R, but the proportion of the Rp and R s are incomparable with measured
Rp and R s . CPT method, API method, A-method in AASHTO overestimate R,
although the percentages of Rp and Rs are comparable with the measured ones. The Rp
and R s calculated by CFEM plasticity code and by AASHTO CPT code are in
remarkably good agreement with the measured ones.

Ultimate Capacities in Tennouji Site

For the cast-in-situ concrete pile, the Rp calculated by the Japanese codes give
similar magnitude. The calculated R are comparable with the measured R derived from
Po, called Putt I, at displacement 10% of the pile diameter. This may attributed to the
fact that the ultimate capacities of cast-in-situ concrete piles were been defined by
Pull I and that the empirical equations have been calibrated against thus defined
capacities
The R calculated by the codes in other countries are comparable with the
measured R, called P u 112, defined in Fig.II, excluding CFEM N code

Allowable Capacities in Nanao and Tennouji Sites

At the Nanao site, the allowable capacities of the driven steel pipe pile derived
from Japanese codes tended to be smaller than those derived from the codes in other
countries. On the other hand, the allowable capacities of the cast-in-situ concrete pile
in Tennouji site derived from the all codes may be equal in spite of the fact that the
ultimate capacity shows a large discrepancy between Japanese and other codes. This is
due to the different definitions of the ultimate capacity as indicated by Putt! and Putf2
on the load displacement curve in Fig 11.

784 Michi, Tsuzuki & Matsumoto


R u = R p + R s (MN)
Ra = R pa + R sa (MN)
o 2 3 4 5 6 7

Measured
Cepqp, qs)
JHA N I
CO. 25'1 O. 25)1
.' . :" . :'.-'~I'i;::':' '.' ..-' . : , -'. "- ",:'-, ': "...-.

1I1l11l111''''W-e ,;,''C.'''' ':": 'I CO. 33" 0.33)1


JR N C

~ CO. 30. 0.30)


...,.".".,.-,. . I
CO. 30, 0.30)
I
(0.40'1 0.40)1
'"
CO. 33, 0.33)

CPT method
I
(0. 5~, O. 5~)
API method . '

(0.50. 0.50)
CFEM N
=nmmnnnr~(0.30,I 0.30)I
CFEM plasticity I
0.00, 0.60)
I
CFEM CPT
(0.50, 0.50)
I
AASHTO O!
CO. 70, 0.70)
AASHTO /3 I
CO.70, 0.50)
AASHTO A
I I i

CO.70, 0.55)
AASHTO CPT I
(0.55, 0.55

0m Rp [J Rs
1

Fig. 13 Comparison of measured end and shaft capacities of Nanao test pile
with those derived from the desgin codes in Japan and other countries

785 Michi, Tsuzuki & Matsumoto


Ru =Rp + R s (MN)
R a =Rpa + R sa (MN)
o 2 4 6 8 10 12

Measured 1
L .... , ... :..."':::.,:: .. :'.:1
. ,,,,','::1
Measured 2 (<!>qp, <!>qs)
I
..........,.: ..
'::.:>}:> ..... : 1
JHA N
(0.33 0.33)
j
-c..... . . . . ' . j
JHA qu
(0.33, O. 33)

JR N .. ':' I
I
CO. 60. 0.30)
I
JR qu ~
I
CO. 60, O. 30)
JPH qu
I I
1
JAG qu
CO. 30, 0.50)

I
CFEM N I
CO. 30. 0.30)
I
CFEM plasticity I
I
0.00, O. 60)

DNV ]
1
CO. 83, O. 83)
I

AASHTO N 1111111111111111
I I I I
CO.45, 0.45)

~ R._p D R_s

Fig. 14 Comparison of measured end and shaft capacities of Tennouji test pile
with those derived from the desgin codes in Japan and other countries

786 Michi, Tsuzuki & Matsumoto


It is interesting to note that the allowable capacities derived from the empirical
equations using N-values in the all codes are smaller than those derived using other soil
parameters. This result suggests unreliability of the N-value comparison with other soil
parameters that might have world wide recognition among professional geotechnical
engineers. In current practice in Japan, the majority of site investigations are carried
out by SPT's for the foundation pile design purposes. This situation is being changed
using adequate soil parameters on specific sites solely, or in parallel with N-values, to
improve the quality assurance of the foundation piles.

CONCLUDING REMARKS

The bearing capacities of the foundation piles from static loading tests performed
on two different sites are compared with the calculated capacities by several
representative design codes in Japan and other countries, aiming at finding out a
proper design method associated with reliable soil parameters. These test and
evaluation results conclusively suggest that;
1) The empirical equations using qu and qc and is of the CPT's provide the most
reliable pile capacity particularly in the diatomaceous mudstone.
2) The pile capacity derived from N-values by the SPT's is not reliable enough, in
comparison with CPT's, to fulfill the needs of high quality foundation design.
3) In cases where the yield point is clearly obtained in the load displacement curve as
in Fig.ll, the ultimate capacity defined at the pile displacement reached to the 10%
of the pile diameter should not be used as a design criterion.
4) It is questionable to apply the same magnitude of safety factor for the ultimate
capacities defined by the different criteria.
5) Definition of the ultimate capacity by the displacement 10% of pile diameter is not
realistic to apply on a large diameter foundation pile particularly for the specific
type of the super structure which has narrow allowance to the excess settlement

ACKNOWLEDGMENT

The authors would like to thank Japan Railway West for their permission to use
the Tennouji site data for this paper.

REFERENCES

American Petroleum Institute (1980) : Recommended practice for planning, designing


and constructing fixed offshore platforms, API RP 2A, 11 th Edition..
Architectural Institute of Japan (1988) Recommendations for Design of building
Foundations (in Japanese)
Baker et al (1991) : Manuals for the design of bridge foundations, NCH, Report 343,
Transportation Research Board.
Canadian Geotechnical Society (1985) : Canadian Foundation Engineering Manual
(2nd Edition)

787 Michi, Tsuzuki & Matsumoto


Det Norske Veritas (1977) : Rules for design construction and inspection of offshore
structures.
ISS:MFE (1977) : Report of subcommittee on standardization of penetration testing in
Europe. Proc. 9th Conf., Tokyo, YoU, 95-152.
Japan Road Association (1990) : Specifications for highway bridges, Part IV :
substructures (1990). The Japan Road Association, Tokyo (in Japanese).
Matsumoto, T, Kusakabe, 0., Suzuki, M. and Shogaki, T (1993) : Soil parameter
selection for serviceability limit design of a pile foundation in a soft rock. Proc.
Int. Symp. on Limit State Design of Geotech. Eng., Copenhagen, VoUI3, 141-
151.
Nishida, Y., Sekiguchi, H., Matsumoto, T, Hosokawa, S. and Hirose, T (1985) :
"Drivability of steel pipe piles driven into diatomaceous mudstone in the
construction Notojima Bridge." Proc. Int. Symp. Penetrability and Drivability of
Piles, San Francisco, VoU, 187-190.
Takamori, H., Koshino, H., Matsuo, M., Hasuda, T., Tsuji, E. and Katori, H. Y.
(1993) : "Field load tests of pile for Tennouji tenninal project (Part-2) Results of
vertical loading test and estimation of bearing capacity." Proc, Annual Meeting
of Architectural Institute of Japan, 1755-1756 (in Japanese).
The Japanese National Railways (1986) : Foundation Design Standard for Railway
Structures. Japan Society of Civil Engineers.
The Overseas Coastal Area Development Institute of Japan (1991) : Port and Harbour
Facilities in Japan. The Japan Port and Harbor Association
Veda, S., Hasuda, T, Tsuchida, K., Machida, S., Koshino, Hand Kawabata, Y
(1993) : "Field load tests of pile for Tennouji terminal project (Part-I) Outlines
of the project and plan of the tests." Proc, Annual Meeting of Proc. Annual
Meeting of Architectural Institute of Japan, 1753-1754 (in Japanese)

788 Michi, Tsuzuki & Matsumoto


Aspects of pile design for cyclic and dynamic loading
with reference to API conditions for offshore structures

J. P. SUlly1, M. Paga 2 , R. Bea 3 , E. Gajardo 2 , R. Gonzalez 2


and A. F. Fernandez 2

Abstract
The recently revised American Petroleum Institute guide
to recommended practise (API RP2A Recommended Practise
for the Design, Construction and Installation of Offshore
Platforms) includes suggested methods to be used for the
design of piled foundations under dynamic and cyclic
loading. The different approaches to the evaluation of
axial and lateral load response of offshore piles under
both cyclic and dynamic loading is considered in terms of
the loading rate. In both the t-z and p-y cases,
adjustment factors to account for loading rate and
degradation are included in the pseudo-static analysis.

Introduction
The API RP2A (1991) code ("Recommended Practise for
Planning, Designing and Constructing Fixed Offshore
Platforms") is widely used in the USA for the design of
offshore steel platforms. The code contains
recommendations regarding the modelling, analysis and
design of structure and foundation for the various
distinct loading conditions to which the platform may be
subj ect. In the case of seismic loading, the philosophy
of the code consists of designing the platform to behave
elastically with respect to the forces corresponding to
the expected level of shaking during the design life, as
well as satisfying a series of requirements in order to
provide a certain level of ductility so that the
structure would withstand, without collapse, the loading
associated with an extreme or low probability seismic
event. In the API code, the extreme or rare seismic
event is considered to produce a loading condition equal
to twice. the level assumed for the design condition.
However, the relationship between the design and extreme
level events is dependent on a series of complicated
inter-relationships involving the structure itself, the
soil conditions at the site being considered and the

1 GEOHIDRA, C.A., Apartado 47851, Caracas 1041, Venezuela.


3 Professor, University of California,Berkeley, California, USA
2 INTEVEP, S.A., Apartado 76343, Caracas 1010A, Venezuela

789
J. P. Sully et ale
seismotectonic characteristics of the region.
Furthermore, the relationship does not depend solely on
the expected levels of maximum bedrock acceleration, but
also on the time history records in terms of the effect
of the dynamic soi I properties, structural response and
the soil-structure interaction. This aspect is taken
into account in the code by recommending that a nonlinear
structural analysis be performed for the extreme event
loading in which the associated forces are taken to be
twice, or more, those corresponding to the design level
event.

Herein, a simple methodology is proposed for adapting the


API code to situations where the relation between the
design and extreme level events is different from that
assumed by the code, which incidentally is based on
information relevant to the California seismotectonic
environment. The probabilistic approach used includes
the uncertainties involved in both the seismic event and
platform strength determinations and considers not only
the factors themselves, but also the errors associated in
the modelling of such parameters.
Reliability Formulation

The capacity of an offshore platform soil-structure


system as a function of the lateral load-displacement (F-
.1) relationship shown in Fig. 1. The reserve strength
ratio, RSR, is defined as the ratio between the ultimate
lateral capacity, Ru , and the design load, So:

RSR = Ru/S o (1)

The RSR can be interpreted as the reserve capacity that


the system has to withstand, without col laps ing, dur ing
the extreme event and is a fundamental parameter to the
method employed. Following the reliability approach
proposed by Sea (1991), the RSR can be expressed as:

(2)

where ~ is the design annual safety index, defined as:

(3)

IlR = In R so (4 )

Ils = ln Sso (5)

790
( 6)

R SO and Sso are the medians of the distributions of the


structural capacity and maximum annual load,
respecti vely. a is the measure of uncertainty that the
resistance, R, is greater the load, S, and is determined
by means of the standard deviations a R and as of these
variables.

ULS CAPACIT~
Ru

ULT1MATE
LIMIT STATES
Rr
u..
RESIDUAL
w
u CAPACITY
a:
ou..

DESIGN SERVICEABI L1TY


LOADING LIMIT STATES

6e

DISPLACEMENT 6

Figure 1 Load response curve for soil-platform system


(modified after Bea, 1991)

The probability that loss of serviceability or failure,


P fQ , occurs as a result of seismic loading, SQ' is given
by Eq. (7) and is related with the annual safety index as
given in Eq. (8):

(7)

(8)

791
where <1> (~) is the accumulati ve normal probability
function of the safety index ~.

In Eq(2}, FR is called the force ratio, which is


determined according to:

(9)

(IO)

where T is. the average return period for the seismic


design event.

The uncertainty associated with the seismic action, as,


can be considered as being comprised of two main
components:
the natural or inherent uncertainty of the
phenomenon itself (Type I)
- the uncertainty of the model used to represent the
seismic action (Type II)

Type I uncertainty is unaffected and insensitive to the


level of information available. However, the Type II
uncertainty can be reduced~ according to the amount and
reliabili ty of data regarding the regional seismicity,
which determines how well the conditions can be
represented by the model. Hence, as can be calculated
from (as}r and (as}rr as suggested by Eq. (II), if the
corresponding standard deviations are small:

(II)

The reserve strength ratio, RSR, defined earlier can be


obtained from a static, monotonic or incremental
(pushover) analysis. In this case, it is termed the
nominal value, RSR. In order to incorporate the effects
of ductility and degradation, the dynamic RSR can be
obtained from the nominal value according to:

(12)

where 11 is the ductility factor, a is the degradation


index and BRU is the bias in the estimated strength.

Also, from Figure 1:

11 = tiejlJ.p (13)
792
(14)

~e and ~p are the displacements corresponding to the


limi t of elastic behavior and to the point of maximum
plastic deformation, or failure, whereas Ru and Rr are
the ultimate and residual load capacities, respectively.

The bias, BRU ' is the factor that relates the best
estimate of the ultimate resistance, Ru , to the nominal
value, Run' as determined according to standardized
procedures included in the design codes.

(15)

Run can be determined also from a static, monotonic or


incremental analysis and Ru evaluated using the BRU
value.
Conditions Inherent in the API Code

The conditions used for development of the API RP2A code,


in relation to seismic considerations, are described
briefly, since the idea here is to consider how these can
be modified to account for the actual conditions at the
site of interest.
The nominal reserve strength ratio, RSR N , is taken equal
to 2. For cross-braced jacket structures, the ductility
factor, ~, is equal to 2. The residual strength, Rr , is
equal to the ultimate strength, Ru , (no strength loss
occurs) and a is 1.0. The strength bias, BRU = 1.2 and
the action bias is assumed to be unity.

In terms of the uncertainty values, the following factors


are used:

<iR = 0.25; (<ish = 1.00; (<ishr = 0.40

Furthermore, the return period, T, for the seismic design


event is stipulated as 200 years. Based on the above
information, using the equations given above, the
following parameters can be determined as assumed in the
API code:

<is = 1.08; <i = 1.11

793
FR = 0.1316; RSR = 4.8
From Eq. (2), the value of ~ can be determined as 3.24
and consequently, by means of Eq. (8), the annual
4
probability of failure is 5.4 x 10-
Modification to API Code

One option for considering the actual site seismicity in


the platform design would be the modification of the
provisions in the API code, which would essentially
involve the development of a new code. Alternatively, a
viable and practical approach would be to modify the
design acceleration spectrum by means of an adjustment
factor. This approach would maintain the same safety
index adopted in the API code. The modification factor
can be considered as a bias that relates the nominal
spectral acceleration, Sn, that would be used in order to
satisfy the API safety index requirements:

(16)

The bias, Bs ' can be estimated as the ratio between the


reserve strength ratio for the site under study, RSR site '
and that necessary in order to maintain the safety index
assumed in the API code, RSR API .

(17)

Application of Proposed Method

The use of the methodology discussed requires that a


seismic hazard evaluation be performed and that seismic
response analyses be carried out for the soil profile at
the particular site under study. Obviously, detailed
geotechnical investigations are also required to define
static and dynamic soil engineering parameters. The
application of the method to an offshore site in
Venezuela is presented by Paga et al (1994). For that
particular site, the resulting bias, Bs, was 0.52. This
is the factor which is used to adjust the spectral
acceleration ordinates and results from the fact that the
uncertainties in the site specific seismic data are much
lower than those assumed in the API code formulation. In
this case, the factor reduces the design acceleration
while at the same time maintaining the same safety index
recommended by this widely used code.
For the case in question, it was considered convenient to
use a factor of 0.7 rather than 0.52, with the option of
further reducing the value based on confirmation from the
results of an inelastic structural analysis of the

794
platform-soil system. This analysis is considered
desirable to verify the level of ductility assumed in the
procedure and can be used also to fine tune the
structural detailing and confirm that the API conditions
are being satisfied. The potential reduction in platform
costs more than justify the additional time and expense
involved in the final inelastic analysis.

The technique can also be applied to aspects of the


foundation design for both axial and lateral loading.
This is discussed in the following sections.

Pseudo-dynamic axial pile capacity

The first step in the evaluation of axial and lateral


pile response is the calculation of static axial
capacity.

The approach employed for determining the required pile


length for the dynamic loading is that presented by Bea
(1992) based on considerable field results and research.
In effect, the dynamic load is converted into a static
component and the standard plot of ultimate pile capacity
with depth can be used to determine the required
penetration.

The total load to be sustained by the pile is obtained


from:

(18)

(19)

where:

P t is the total static and dynamic axial load to be


considered for determining the pile length
P s is. the static axial component
P d ~s the pseudo-static component
P e is the total dynamic load
RSR N is a factor to guarantee resistance requirements,- as
described previously
Pc and Pd are factors to consider loading effects in
terms of soil resistance
P e is determined using the design response spectrum
produced for the site.

795
The two factors Pc and P are determined by consideration
d
of:

- the effect of the number of cycles of loading,p c


- the effect of the loading rate, P
d

These factors will thus be different for the cases of


earthquake loading (low number of cycles, high rate of
loading) and wave loading (high number of cycles, lower
loading rate). The effect of these two conditions are
determined from the results of laboratory tests on
recovered samples. As an example of the importance of
loading rate on soil behavior, both the modulus and
strength increase as the loading rate increases,
typically a 15% increase occurs for each log cycle
increase in loading rate. For a particular offshore site
in Venezuela, an increase of between 15% and 21% per log
cycle was measured during laboratory direct shear tests
at different loading rates.

The P values for determining the component of dynamic


load via Eqs. (18) and (19) are given in Table I.
As a preliminary value, RSR can be taken as equal to 1.5
N
for both earthquake and wave loading. Subsequent to
performing the initial dynamic analysis, a revised value
can be defined.

Table I P factors for different dynamic loading


conditions

Loading Condition P Factors

Earthquake
Pc
0.92
I Pd
1.6
Waves 0.8 1.4

The factors listed in Table I have been determined for


the case of axial loading. Similar factors can be
determined for the lateral loading condition.

The above method applies the factors RSR, P and P to


N c d
the estimated dynamic load. No factor is applied to the
static load component. This equation can be compared to

796
the more traditional approach where a factor of safety is
applied to the load capacity of a single pile. The
factor may be applied as a global factor to the total
pile capacity or as partial factors to the axial and tip
components of the total capacity. For example:

(20)

or
(21 )

where:
Q is the allowable pile capacity
a
Q is the ultimate pile capacity
u
Q is the contribution of the pile sleeve friction to the
f
total capacity
Q is the pile tip bearing capacity contribution to the
t
total capacity
F is the global factor of safety
F 1 is the partial factor of safety on the pile sleeve
friction
F is the partial factor of safety on the pile tip
2
bearing capacity
The API RP2A (1991) code considers a level of allowable
pile capacity whereby a global factor of safety is
applied to the ultimate pile capacity as given by Eg.
(20). For the condition of working and environmental
loads, the code recommends a load factor of 1.5 which is
the same as applying a resistance factor F = 1.5 in Eg.
(20) .
The only factor applied in the reliability approach is
the product of RSR N , Pc and Pd and this is applied only
to the dynamic component of the total load. The value of
RSR N is determined from a probabilistic study of the
loading which takes into account the probability of
failure under the loading conditions and the uncertainty
in the determined load and the structural strength or
resistance. In the deterministic sense, RSR could be
N
considered as an applied factor of safety on the dynamic
load component.

In the more traditional deterministic analysis, this


factor would be applied without modification. However,

797
two additional factors are considered in the approach to
consider the effect of both loading rate and number of
cycles of loading on the operational soil strength.

P is a factor which considers the effect of the number


c
of cycles on the operational soil strength. It
represents the degradation of soil strength under cyclic
loading and can be determined from the results of cyclic
laboratory tests. The factor has been evaluated for both
wave (cyclic) and earthquake (dynamic) loading.

P is a factor which considers the effect of rate of


d
loading on the undrained soil strength and can be
evaluated by evaluating the difference between the
laboratory and field loading rates. Again results from
cyclic DSS, or similar, tests can be used toevaluate Pd'
A strength increase of 15% per log cycle of loading rate
was obtained.

The magnitude of both Pc and P will depend on the type


d
of loading and hence will be different for wave loading
(large number of cycles, lower loading rate) and
earthquake loading (small number of cycles, higher
loading rate). Consequently, the pseudo-static component
of the dynamic load can be determined once the total
dynamic load is known.

For a pile which has been dimensioned according to the


combined static and dynamic load, P t' the operational
factor of safety under static loading conditions is given
by:

(22)

If the dynamic load is twice the static, then for this


cond i tion F = 3. For the case of the offshore site
studied, the static and dynamic loads were estimated to
be 3000 kips and 7000 kips, respectively. Consequently,
the static factor of safety is 3.3 (P = 10000 kips).
t
The approach advocated by the reliabil ty approach
presented by Bea is that of ultimate (limit) state
design, which requires that at the estimated maximum
loads all the available soil resistance has been
mobilized, i. e. an extreme-condition operational factor
of safety of unity is effected. Hence, the static load
or capacity should not be factorized in order to achieve
this condition.
798
The reliabilty with which one can determine the axial and
lateral pile capacities is thus of great interest. It is
generally accepted that the axial capacity of steel pipe
pi les can be estimated with reasonable accuracy (Bea,
1991) and so the application of factors to account for
uncertainties is not necessary. (This uncertainty is,
however, considered in the formulation of RSR .) In the
N
case of lateral capacity, the uncertainties can also be
considered in RSR N Furthermore, it is generally
accepted that estimates of lateral load capacity based on
API procedures are somewhat conservative.
Based on the above, it would then appear reasonable to
use the formulation of Eqs. (18) and (19) with the
factors as described previously to determine the total
equivalent static axial load and subsequently, the
required pile penetration by means of a static capacity
curve ca lculated according to API procedures. Var ious
offshore platforms have been designed in this way and so
certification is not a problem, especially since API
permits variations in design philosophies.

Recent information (Bea, 1992), indicates that a new API


RP2A code has been approved and is in preparation for
release. In this new code, the recommended load
reduction factors are given as:

(23)

with the proviso that, when P and P are approximately


s d
equal, Eq (23) should be modified to:

(24)

In the standard API equations, however, .no account is


taken of the p factors discussed earlier.
using equations (18), (19), (23) and (24), the different
magnitudes of P can be evaluated for the offshore
t
platform and loads given above. For the standard API
approach, P t is equal to between 9600 kips (Eq. 23) and
12500 kips (Eq. 24), whereas the reliability approach
gives 10150 kips. Hence, the results of the analysis of
the axial pile capacity appear to be consistent with the
new API RP2A code in preparation (Eqs. 23 and 24), which
obviously are based on reliability studies.

Three types of data are required for the analysis of


axial and lateral pile capacity under static and dynamic
(and cyclic) loading, namely:

799
* t-z curves which describe the axial load transfer
as a function of pile head displacement versus
depth
* Q-z curve used to describe the mobilization of tip
load with tip displacement
* p-y curves as a funtion of depth which describe
the change of lateral soil resistance with varying
degrees of pile deflection
All of the above data have to be generated for both the
cyclic and dynamic loading conditions.
P-y curve determination for lateral loading

To account for rate of loading effects and degradation


due to the numbers of cycles of loading, correction
factors can be defined based on the results of laboratory
tests performed on undisturbed samples recovered during
the site investigation phase. For the lateral load case,
the rate of loading factor ~d is the same as for the
axial load case for both earthquake (~d=l. 6) and wave
(P d =1.4) loading given in Table 1.

The ~d factors were determined from the relationship:

Pd = 1 + r log(ts{tr) (25 )

where r is a soil dependent load rate factor, ts is the


reference loading rate and tr is the field loading rate.
The DSS data indicated a representative value of r = 0.15
for the offshore soils. The ratio of reference
(laboratory) to field loading rates was calculated to be
4 log cycles for earthquake loading and 2.6 log cycles
for wave loading.

The Pd factors so determined can be applied to both the


axial and lateral components. However, the degradation
factor which considers the number of cycles of loading,
will be different for lateral loading than for axial
loading. This results from the fact that the lateral
displacements diminish with depth, so that at the surface
the soil will suffer larger degrees of degradation than
the soil at depth, where the lateral displacements are
much smaller. consequently, the percentage degradation
will vary with depth.

800
The ~c factors for quake and wave loading were determined
based on the results of cyclic direct simple shear tests
(OSS). The degradation index, 8, where:

us: = N- t = Gn /G 1 (26)

(N is the number of cycles, t is the degradation


parameter and G is the shear modulus at N=1 and n cycles)
was determined from the change in modulus with number of
cycles of fully reversing harmonic straining. Hence, ~
c
is given by (Bea, 1992):

(27)

where w is the ratio of percentage change in soil


strength to percentage change in soil stiffness.

The t parameter values from the OSS results on the


offshore clay were compared with data obtained from tests
with Orinoco clay and also San Francisco Bay Mud. The
comparison is presented in Fig. 2 where the three sets of
data are in good agreement. The generalized curve in
Fig. 2 based on results from similar Venezuelan clays was
used to extrapolate the laboratory test data to higher
strains when t values were required at those levels of
strain.

T-z and Q-z curves

The axial capacity can also be determined by defining the


soil parameters in terms of shaft and tip load transfer
functions, commonly known as t-z and Q-z curves. The t-z
curves are generated at varying depths along the length
of the pile while the Q-z curve relates only to the tip
response. The curves provide a relationship between pile
axial displacement and load transfer. These curves are
generated using basic soil parameters such as the
undrained shear strength, strain at 50% strength (for
clays) , etc. In this case, the ~ factors are
incorporated directly into the t-z and Q-z curves.

As for the API calculation, the safety factor on the


dynamic load component (RSR ) is calculated from
N
probability considerations. For the Rio Caribe platform,
RSR is taken to be equal to 1.5.
N

The t-z and Q-z curves for the offshore platform


foundation soils can be generated using any of the many

801
available computer codes. The load transfer relations
calculated by the code are based on both theoretical
developments and field measurements. Data can be
produced for the static, cyclic (wave loading) and
dynamic (earthquake loading) conditions.
In addition to the t-z and Q-z relations, the codes also
provide the variation of pile head load as a function of
pile head displacement. This is in effect the load
displacement curve from which the ultimate pile capacity
can be determined and the axial stiffness evaluated.
The difference in the load displacement curves for the
wind and quake conditions is due to the different p
factors for each case.
0.30

0.25 ........-
-~
~/'/

...-
0.20 UPPER RANGE
FOR S.F. BAY MUD /f/ /~

Q)

Q)

E
...
c
0.15

ORINOCO
//
///~VERAGE FOR S.F. BAY MUD
c
a.. CLAY ",/ /
o.ro-
1jj~(- RIO CARIBE DATA
0.05
....,.:.....: ..-.

o+ - - - r - - - - - . - - - , - - - - - , - - - , - - - - - r - - - . - - - - - i
o 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2
Cyclic Axial Strain,tc(%)

Fig. 2 Degradation parameter t for offshore soils.


References

Bea, R.G. (1992) Pile capacity for cyclic axial loading.


Proc. ASCE, Jour. Geot. Engng., 118:1:34-50.
Sully, J.P., Gajardo, E. and Paga, M. (1992) Pseudo-
dynamic axial pile capacity. Technical Note to Lagoven,
S.A.
Sully, J.P., Gonzalez, R., Paga, M. and Gajardo, E.
(1992) Dynamic and cyclic pile analysis parameters for
axial and lateral loading. Technical Note to Lagoven,
S.A.
802
Integrity Testing of Drilled Shafts:
A Computer Vision Approach

Mahmod M. Samman, Ph.D. 1


Mrinmay Biswas, Ph.D., P.E.2

Abstract
In this paper, we investigate a nondestructive testing approach that can
potentially detect and characterize defects in drilled shafts without complicating the
construction process and without using any theoretical models. A number of
computer vision (pattern recognition) techniques are used to interrogate stress wave
signals from laboratory size models which contain simulated defects of different
depths, widths, and locations. We present the results of the techniques as defects vary
in size and location. Our work indicates that this technology is potentially capable of
providing size and shape information about defects in drilled shafts.

INTRODUCTION

Drilled shafts are common foundation elements used to support a variety of


structures; most notably bridges. Shafts are long concrete piles constructed by
drilling a hole in the ground, placing- rebars in place, and casting concrete to fill the
hole. Drilled shafts are of vital importance to the structural integrity because they
provide the necessary firm support for structures. Since they are buried and cast-in-
place, defects may form during construction and go unnoticed.

1 Associate, Stress Engineering Services Inc., 13800 Westfair E.Dr. Houston, TX


77041-1101
2 Associate Professor, Department of Civil & Environmental Engineering, Duke
University, Durham, NC 27708-0287.

803 Samman and Biswas


To inspect these shafts, partially-destructive methods as well as nondestructive testing
(NDT) methods are being used. Partially-destructive testing (such as taking bore
samples from the shafts) is very costly and time consuming. On the other hand,
commercially available NDT techniques do not provide sufficient information about
the shape or size of defects. Also, except for the cross-hole testing, none of the NDT
methods known to us can provide any information beyond the first full-section defect.

The cost and uncertainty involved in inspecting drilled shafts are of concern to both
contractors and public agencies. There is a need for NDT techniques that can achieve
the following:

1. Test new and existing shafts.


2. Pose no threat to the structural integrity of the shafts.
3. Readily detect large and small defects.
4. Provide information about the size and shape of defects.
5. Minimize interference with the design and construction processes.
6. Perform the testing in a timely and cost-efficient manner.

Almost all nondestructive testing techniques were developed for homogeneous


materials. Early attempts to use such techniques on concrete faltered because of the
inadequacy of the techniques and also because of the poor performance of the
available equipment. As techniques developed and equipment improved, several
other techniques became available to the civil engineering community. (Malhotra and
Carino, 1991)

Problems that face NDT of drilled shafts fall in one or more of the following
categories:

1. Limited accessibility to the structure.


2. The large-size of the structure.
3. The complexity of the soil-structure interaction.
4. Heterogeneity of the shaft material.

Currently, shafts are being inspected using stress wave testing, either by time-domain
or frequency-domain analyses. The theory behind this technique is as follows: When
a stress wave is introduced at the top of a shaft, it propagates down along the length of
the shaft until it reaches a discontinuity and reflects back up. In a solid shaft, without
defects, the first discontinuity is the end of the shaft. In a defective shaft, the first
discontinuity is an intermediate defect.

By recording the vibration at the top of the shaft using sensitive sensors, the departure
and arrival times of the stress wave can be identified. Since the speed of stress waves
in concrete can be estimated, the location of the discontinuity can be identified with
reasonable accuracy. Modal hammers are used as exciters to impart the impulse that

804 Samman and Biswas


generates the stress wave. Geophones or accelerometers are the sensors used to
measure the motion caused by the departure and return of the stress wave at the top of
shafts. (Hearne, Stokoe, and Reese, 1981)

The basic time domain analysis of the stress-wave signal is called the sonic echo
method. It can only provide the location of defects that extend across the full cross
section of shafts. The method, however, does not provide any details about the size or
shape of defects.

Converting stress-wave signals to the frequency domain, called the sonic mobility
method, provides more information. The mobility plot is the frequency response
function of the system which is the ratio of the geophone spectrum to the hammer
spectrum. This plot can be used, as in the sonic echo method, to locate full-section
defects. In addition, it can be used to estimate the low-strain stiffness of the shaft
from the slope of the initial part of the mobility plot.

The most recent processing technique for stress-wave signals is the impedance log
method. (Pacquet, 1991) In this technique, the experimental mobility curve is
compared to a theoretical curve for a shaft with no defects. Then, the difference
between the two curves is used to back calculate the impedance along the shaft.
'rhen, using the mass density of the shaft and the velocity of compression waves in it,
impedance values are translated into changes in the cross sectional area of the shaft.
If a shaft has no total separation along its span, this technique can provide the most
details about its defects. (Rix, Laurence, and Reichert, 1993)

An alternative technique that offers substantial improvement over basic stress-wave


testing is the cross-hole test. (Olson and Wright, 1990) The technique involves
casting tubes inside shafts during construction for providing access along their length.
Then, by transmitting a stress wave from one hole and receiving it from the other, the
existance of defects can be verified and characterized. This technique offered the
following benefits:

1. The test provides valuable information about the axial extent (depth) of
defects.
2. It is the only test that can go beyond the first full-section defect.

But the technique has some drawbacks:

1. The technique may compromise the structural integrity of small-diameter


shafts.
2. It involves considerable complications and time delays during construction.
3. It can only be used on new shafts. Existing shafts cannot be inspected
unless ducts are drilled inside the shafts to provide the required access

805 Samman and Biswas


which would be prohibitively expensive and also could threaten the
structural integrity of the shaft.

The above description of stress wave testing techniques was not meant to be a survey
of available NDT methods. It was rather a brief review of some widely used
approaches. For more details, the reader is referred to the cited references as well as a
wealth of literature available on the topic.

PATTERN RECOGNITION TECHNOLOGY

Pattern Recognition (PR) is the science and art of mimicking human identification of
images, graphs, and waveforms using computers. A branch of computer vision, PR
relies on digital signal & image processing, and methods of mathematics. This
technology is being successfully used for diagnostics in many other fields such as
medicine and speech analysis. The technology is relatively inexpensive, can be fully
automated, and produces results in a timely manner.

In the field of PR, single-valued functions, similar to time varying signals, are referred
to as waveforms to distinguish them from planar curves and multi-valued functions.
The following is a brief description of the waveform recognition techniques we used
for analyzing signals. The description is not sufficient for reproducing the results.
Details on the techniques can be found in Samman and Biswas (1994).

Adaptive Template Matching

Template Matching is a method of image recognition and segmentation based on a


point-by-point check. Adaptive Template Matching (ATM) refers to a technique that
performs a point-by-point magnitude check for detection of differences between two
digital signals.

The technique identifies the similarity (or difference) between two signals as follows:
Given two digital signals and a threshold as a percentage of the maximum magnitude;
the technique finds the minimum tolerance value using which one signal falls within
the tolerance space of the other for all magnitudes above the given threshold.
Therefore, using that minimum tolerance and threshold, the technique considers the
two signals to be "similar".

For the recognition process, the minimum tolerance value is the parameter or feature
used for quantifying the similarity. As the difference between the signals increases,
the value of the minimum required tolerance increases.

806 Samman and Biswas


Waveform Chain Code

In PR, chain codes are used for detection of boundaries in digitized images. The
Waveform Chain Code (WCC) is a dedicated chain code for waveform analysis. The
WCC characterizes a waveform by its scaled relative slope and / or curvature. Hence,
the technique is very sensitive to minor changes in peak locations.

First, slopes of the two signals are computed by forward differencing. Then, to
establish the relative nature of the code, slope values are normalized with respect to
the maximum slope in the signals. These relative slopes are then scaled to 50 so as to
present the final result in a percentage form. The scaled relative slopes are then used
to find the scaled relative curvature also using forward differencing. The resulting
scaled relative slopes and curvatures take values in the range from +50 to -50. At
least one value of the scaled relative slope is equal to either +50 or -50.

The scaled relative values of slope and curvature are referred to as the "coded" slopes
and curvatures respectively. To compare two signals, we calculate the absolute
difference between their coded slopes or coded curvatures at every point in the signal.
These differences are called slope differential values (SDV) and curvature differential
values (CDV), respectively. A large SDV or CDV at a point indicates a significant
difference between the two signals in the neighborhood of that point. For the
recognition process, two measures are used as the features of recognition:

1. the maximum values in SDV or CDV curves, and


2. the area under SDV or CDV curves, which is equivalent to the summation of SDV
or CDV values, respectively.

Signature Assurance Criterion

The Signature Assurance Criterion (SAC) is a measure of similarity between two


signals. It draws on the Modal Assurance Criterion (MAC) used in experimental
modal analysis of structures. The SAC compares the magnitudes of the two signals at
each abscissa point, and returns an ensemble value for the two signals. The ensemble
value provides the measure of similarity between the signals. The SAC can range
from zero to one. A value of one indicates identical signals.

EXPERIMENTAL PROCEDURE

Since the purpose of the experiment was to examine the potential of the PR
techniques, we used simple laboratory-size models in an ideal setting with no soil-
structure interaction. We built nine specimens made of a single batch of cement
grout. All cylindrical specimens were 10.16 em (4 inches) in diameter and 121.9 em
(4 feet) in length. After 33 days of curing, we set the models in an upright position on
elastic supports as shown in the schematic diagram of Figure 1. The elastic supports
Samman and Biswas
807
were one-inch-thick six-inch-diameter smooth-finish natural rubber disks of 40
Durometer hardness. To keep specimens in an upright position, we tied them to a
sufficiently rigid steel frame at the top using plastic straps.

After we tied the specimens to the setup frame, we cleaned the top surfaces, and
brushed off small grout clusters. To provide maximum bonding between specimens
and the accelerometers for vibration measurements, we mounted accelerometers on
the specimens using threaded mounting studs. These mounting studs are threaded at
both ends, one screws into aluminum mounting disks attached to the top of specimens
using epoxy resin, and the other goes inside the accelerometers.

From all nine models, we obtained time-domain, spectrum, coherence, and frequency
response function signals. As expected, signals were relatively consistent in the
frequency domain. The dominant frequency was around 1.5 kHz, the frequency of
stress wave vibration. From each model we obtained composite frequency response
function signals each of which was an average of five individual signals.

After performing stress wave tests on the intact models, we started simulating defects
by cutting through samples using a hacksaw with Tungsten-Carbide-coated 0.16 cm
(1116 in) thick blades. To study the effect of the location of defects, we made these
blade-deep cuts at 30.5 cm (1'-0"),60.96 cm (2'-0"), and 91.44 cm (3'-0") from the
top of samples number 1 through 3, 4 through 6, and 7 through 9, respectively. We
will refer to these three locations as top, middle, and bottom locations respectively.

To study the effect of the width of defects, we increased the width of these blade-deep
cuts gradually to 0.635 cm (0.25"), 1.27 cm (OS'), 1.905 cm (0.75"),2.54 cm (1.0"),
3081 cm (105"), and 5.08 cm (2.0"). Then, to study the effect of depth, we increased
the depth of the 5.08 cm-wide blade-deep cuts to 2.54 cm (1 ") and 5.08 cm (2")
respectively. Figure 2 illustrates what we mean by location, width, and depth of
simulated defects. For each size of defect, we obtained composite frequency response
function signals.

RESULTS

Three typical frequency response functions are shown in Figure 3. All three signals
were obtained from the same specimen. They were obtained from the original intact
specimen, after introducing the minimum simulated defect, and after introducing the
maximum simulated defect 30.48 cm from the top of the specimen. The minimum
defect was 0.635 cm wide and 0.16 cm deep, and the maximum was 5.08 cm wide
and 5.08 cm deep. As shown in Figure 3, a part of the signal was associated with low
coherence which indicates a lack of consistency. This appeared to be caused by very
small magnitudes in the spectrum of the response signal.

808 Samman and Biswas


Waveform recognition techniques work by comparing two individual signals. The
result of a single comparison is a measure of similarity between the two signals. As
stated earlier, such measures of similarity are often called features. To establish an
"average" base line, we operated candidate recognition techniques on each two-shaft
combinations from the nine intact specimens. For any particular feature, the base-line
was the average of the corresponding values from the thirty six possible
combinations. Using this procedure, we established average base lines for all
investigated features. After introducing defects, for each size and location of the
defect, we compared one signal from the defective model to all nine base-line signals.

To investigate candidate recognition techniques, we analyzed signals in two scales:


linear scale and logarithmic scale. The intent was to vary the emphasis on low-
magnitude parts of signals. The following is a summary of the results.

1. Linear Scale Analysis

We observed a good correlation between three features and the size of simulated
defects. The three features were the area under the Slope Differential Value (SDV)
curve, the value of the Signature Assurance Criterion (SAC), and the minimum
tolerance of the Adaptive Template Matching (ATM) technique. Using ATM,
threshold values of 30%, 40% and 50% yielded the exact same answers. This is due
to the dominance of the stress wave frequency in linear scale signals. Figures 4
through 6 illustrate the change in these features in all models. In the three figures,
features changed consistently as the size of defects increased. This indicates that the
three features can be used to assess the size of defects.

In addition to the size of defects, to accurately evaluate the integrity of a shaft, shape
information is needed. One would like to find a feature that is only sensitive to the
width of the defect, another one only sensitive to the depth, and a third only sensitive
to the location. After examining the data, we concluded the following:

1. The SAC is sensitive to the width of defects and is relatively insensitive to


their depth and location. Figures 7 through 9 illustrate how SAC values
change with width, depth, and location respectively. This makes the SAC an
excellent indicator of the width of defects.

2. Neither of the two remaining features appeared to have an exclusive


correlation with the width, depth, or location of the defects. Therefore, SDV
and ATM can only be used to provide size not shape information.

In short, all of the three features above were good indicators of the size of defects.
The SAC was an excellent measure of the width of defects, and none of the features
correlated exclusively with the depth or location of defects.

809 Samman and Biswas


II. Logarithmic Scale Analysis

After analyzing signals in the logarithmic scale, we concluded that results were
generally inferior to those in the linear scale. This finding confirmed that changes in
low-magnitude parts of the signals are not as relevant as those in large-magnitude
parts.

SUMMARY AND CONCLUSIONS

We investigated the feasibility of utilizing a set of pattern recognition techniques to


characterize defects in drilled shafts. Using laboratory-size models made of cement
grout, we simulated defects by cutting through the models at different locations and in
different sizes. We obtained representative stress-wave signals by hitting each model
at the top with an instrumented hammer and recording the consequent vibration using
a sensitive sensor. We obtained these signals for each defect size. Then, we
examined the signals using the pattern recognition techniques.

Our work indicated that some of the pattern recognition techniques are potentially
capable of providing needed information about defects. We found that three
techniques can be used to indicate the size of defects, namely the Waveform Chain
Code (WCC), the Signature Assurance Criterion (SAC), and Adaptive Template
Matching (ATM). Also, we observed an excellent correlation between SAC values
and the width of simulated defects which indicates that shape information may also be
obtainable.

Our proposed approach to nondestructive testing of drilled shafts differs from existing
stress-wave based techniques in the following aspects: (1) We interrogate the entire
frequency response signal to find relevant clues that are not known a priori to
characterize a defect if one exists. (2) Instead of using the wave propagation theory
and the physical characteristics of concrete to set a theoretical baseline signal, we use
real signals obtained from previously tested full-size piles to learn what a shaft signal
should look like.

To prove the validity of the techniques described above, more testing is underway.
Factors that will be examined include soil-structure interaction, size, and the effect of
steel bars. At this stage, we view this technology as complimentary to existing
methods.

810 Samman and Biswas


ACKNOWLEDGMENTS

This work is a part of a research project sponsored by the U.S. Department of


Transportation, the Federal Highway Administration. The support is appreciated.
Mr. Carl Ealy is the technical project manager. Mr. 1. Randy Long of Stress
Engineering Services, Inc. helped in planning the experiments. The authors are solely
responsible for the facts and the accuracy of the data.

REFERENCES

Hearne, T. M., Stokoe, K. H., II, and Reese, L. C. (1981). "Drilled Shaft Integrity by
Wave Propagation Method", Journal of Geotechnical Engineering, ASCE,
107(GTlO), 1327-1344.
Malhotra, V. M., and Carino, N. J. (ed.) (1991). Handbook ofNondestructive Testing
of Concrete, CRC Press.
Olson, L. D., and Wright, C. C. (1990). "Nondestructive Testing for Repair and
Rehabilitation", Concrete International, ACI, 12(3),58-64.
Pacquet,1. (1991). "A New Method for Testing Integrity of Piles by Dynamic
Impulse: The Impedance Log" (In French), International Colloquium on Deep
Foundations, Paris, 1-10.
Rix, G. J., Laurence, J. J., and Reichert, C. D. (1993). "Evaluation of Nondestructive
Test Methods for Length, Diameter, and Stiffness Measurements on Drilled
Shafts", paper no.930620, 72nd Annual Meeting of the Transportation Research
Board, Washington, D.C.
Sarnrnan, M. M. and Biswas, M. (1994) "Dynamic Testing for Nondestructive
Evaluation of Bridges. I: Theory", Journal of Structural Engineering, ASCE, 120
(1) 269-289.

811 Sarnrnan and Biswas


~--+----- accelerometer
J------ modal hammer

2-channel
setup frame
signal analyzer

tight strap
computer
specImen
floor
data
elastic support
storage

Figure 1: Sketch of the experimental setup

top of specimen

top

r
width (w)

location '------. 4depth (d)

1
bottom

Figure 2: Characteristics of simulated defects

812 Samman and Biswas


.
,,'.

," .,
.
,
",

-- max defect

- .. - - - min defect

solid
low coherence

o 2000
Frequency (Hz)

Figure 3: Typical frequency response functions from the models

1700
1600
1500
1400
1300 area
1200 SDV
1 1100
1000
sample no

(depth is .16 em unless noted)

Figure 4: The area under Slope Differential Value (SDV) curves for different
size defects

813 Samrnan and Biswas


1
0.8
0.6
SAC
0.4
0.2 1
o
Olf"lr-
~C"J:g'o::t" ..... 00 sample no
...... 0\V"l 00 0 'o::t"oo
o 'C'l'
..... If"lo
Mlf"lNari
II II
"'O~ "'0
width of defect 00 ~

~~(depth is .16 em unless noted)


(em) If"lari

Figure 5: Signature Assurance Criterion (SAC) for different size defects

100
90
80 ATM
70 Ill1mmurn
60 tolerance (%)
1 50
40
sample
no

(depth is .16 em unless noted)

Figure 6: Minimum tolerance of Adaptive Template Matching (ATM) for


different size defects

814 Sarnrnan and Biswas


------~
0,8

U 0,6
<
rJ).
0.4
~
~
I'----
---r-
0.2

o
o 234 5 6
width of defect (cm)

Figure 7: .signature Assurance Criterion (SAC) versus width of defect

0.8 1J----+----+----+------+---+--------1

U 0,6
<
rJ).
0.4 ++----+----+----+------+---+--------1

0.2 +-+.iL---+----+----t-------+---+--------1

O+-----+-----+----t------+------+-------1
o 234 5 6
depth of defect (cm)

Figure 8: Signature Assurance Criterion (SAC) versus depth of defect

815
Samman and Biswas
0.9

0.8 v,...
"
.~
"q
<>- . TOP

0.7
"'~9. r\
~
, u, - - -t:r- MIDDLE

0.6
,,
v, ,
,

"--
,,
'\
'~ ~ ---0-- BOTTOM
0.5
\"
'' ~
SAC
0.4
"
" ~
z:,.
"
-, ,
, ......... "--'
~
0.3

'.' " "'~


" v,

0.2 , ,

0.1 d=.16 em unless otherwise stated


' ..... , , ~
r.._ . _ g.......
'1

I I I
- .... --
0
o 2 3 4 5 5 5
d=2.54 d=5.08
width of defect (em)

Figure 9: The effect of location on Signature Assurance Criterion (SAC)

816 Sarnman and Biswas


BEARING CAPACITY OF EXPANDED-BASE PILES IN TILL

By William J. Neely; Member, ASCE

ABSTRACT
The results of loading tests are used to develop empirical correlations for
estimating the bearing capacity and load settlement behavior of expanded base piles
in glacial till.

INTRODUCTION
Prediction of the bearing capacity of driven piles in glacial tills is complicated by
the inherent variability of such deposits and by the very considerable problems in
obtaining representative and relatively undisturbed samples for testing. For these
reasons, pile design in such materials is based more on local experience than on the
results of theoretical bearing capacity analyses. Consequently, the design and
performance of driven piles in glacial till have not been widely reported, and there
has been very little basis for improving existing design procedures or for comparing
pile behavior from different till sites over a wide geographic region.
This paper summarizes important findings from analyses of loading tests on
expanded-base piles with cased shafts founded in glacial till. The loading tests are
from sites throughout the New England states and Southern Ontario, Canada, which
are underlain by tills of the Wisconsin glaciation. The results are used to develop
empirical design methods for estimating the ultimate bearing capacity and load
settlement behavior of expanded-base piles.
GENERAL CHARACTERISTICS OF GLACIAL TILL
Terzaghi and Peck (1967) defmed till as "an unstratified glacial deposit of clay,
silt, sand, gravel, and boulders." It has also been stated that the outstanding
characteristic of till is that it is unsorted; it may consist of 99% clay-size particles or

(Vice President, Geotechnical Systems, VSL Corporation, 2840 Plaza Place, Suite 200, Raleigh, NC
27612

817 Neely
99% large boulders, or any combination of these and intennediate sizes (Flint 1957).
Glacial tills are broadly grouped into lodgment and ablation tills. Lodgment till is
an accumulation of glacial debris transported in the base of a glacier and deposited
under pressure on the subglacial floor, lodging there. Lodgment till tends to be
compact and largely unstratified. Ablation till is non-compact material that was
deposited from drift in transport. The nature of lodgment tills tends to be closely
related to the local bedrock conditions (Linell and Shea 1960). Bedrock such as
gneiss, schist, phyllite, and slate give rise to till deposits in which the clay-size
fraction dominates, giving the till the characteristics of clay, while granite bedrock
results in coarse-grained till with characteristics similar to those of a compact rock
fill.
Linell and Shea (1960) compiled an envelope of grain size distribution curves for
till samples from 24 locations in New England, which shows that the fines content
(i.e., particles passing the No. 200 sieve) varies from 18% to 56%. Linell and Shea
(l960) also conducted triaxial tests (unconsolidated undrained, UU tests) and direct
shear tests (consolidated drained, CD) on the minus No.4 sieve fractions of glacial
till samples from five New England locations. The measured shear strength
parameters varied significantly depending on molding water content, activity, clay
mineralogy, and other factors. Drained, or effective stress, strength parameters close
to optimum water content ranged from q> == 32; c == 0, to q> == 38; c == 0.4 tons/sq ft
(38 kN/m 2).

PILES
The data base comprises loading tests on 20 cased, expanded-base piles (or
pressure injected footings) with shaft diameters ranging from 12-in. (305mm) to
about 19-in. (483mm). The driven lengths of the piles averaged 35.5 ft (l0.8m) in the
range 9.5-76 ft (2.9-23.2m).
The installation procedure consisted of advancing the drive tube, making the base,
and forming the shaft. The drive tube was usually advanced by a cylindrical drop
hammer operating inside the drive tube and falling on a dense plug of zero-slump
concrete at the bottom of the drive tube. At some sites, the bottom of the tube was
closed by an expendable steel boot-plate, and the tube was advanced by top-driving
with a diesel hammer. After reaching a predetermined depth, the drive tube was
withdrawn slightly to allow the driving plug to be expelled by repeated blows of the
drop hammer; zero slump concrete was introduced into the tube at the end of driving,
and the boot-plate was then driven off by the drop hammer. Drive-tube diameters
were 16-in. (406mm) aD for shaft diameters of 12 5/8-in. (320mm), 20.5-in.
(520mm) 00 for shaft diameters up to 17 5/8-in. (448mm), and 22-in. (560mm) for
shaft diameters of 19 lI8-in. (486mm). Drop-hammer weights were 2.5 tons (22.3
kN) for the 16-in. (406mm) 00 drive tube and 3.5 tons (31.2 kN) for the larger drive
tubes. At the end of base construction, a corrugated steel shell (casing) was placed in
the drive tube and securely connected to the base and sealed with a plug of zero-

818 Neely
slump concrete. The drive tube was then extracted and the shaft casing filled with 4-
inch to 6-inch (100 - 150mm) slump concrete. This method of pile shaft construction
results in very small lateral pressures between the cased shaft and the surrounding soil
(Nordlund 1982); consequently, expanded-base piles with cased shafts function
essentially as point-bearing units.

LOADING TESTS
Loading procedures varied somewhat from site to site, but typically each loading
increment was maintained for 4 hours, except at design load and double-design load,
where the load was held for a minimum of 24 hours or until the rate of movement was
less than 0.02-in. (0.5mm) in 24 hours. Movements at the head of the pile were
monitored using an engineer's level reading on a scale graduated to O.Ol-in.
(0.25mm) attached to the jack and/or by three dial gauge [O.OOl-in. (0.025mm)
resolution] mounted on an independently supported reference beam.
Dead weight for reaction was supported on two timber mats, 20-ft (6.1m) long
and 5-ft (1.5m) wide, with the inside edge of each mat being about 5-ft (1.5m) from
the center of the test pile.

ULTIMATE LOADS
There are many methods of interpreting the ultimate or failure load of a pile from
the load-movement curve of a test not carried to failure (Hirany and Kulhawy 1989).
In this and previous studies of expanded-base piles (Neely 1990a; 1990b), failure was
interpreted as the load corresponding to a pile head movement equal to 5% of the
diameter of the expanded base, calculated assuming a spherical base and a
compaction factor of 0.8 on the bulk volume of zero-slump concrete used in base
construction.
If the movement at maximum test load was less than 5% of the base diameter, the
ultimate load was interpreted from the available load-movement data using the
stability plot method (Chin and Vail 1973). The stability plot method is based on the
assumption that the relationship between pile load Q and pile head movement ~ is
hyperbolic and that a plot of DJQ versus ~ is linear as described by the expression
DJQ=m~+c (1)
where m = the slope of the line and c = the intercept on the DJQ axis.
Fig. 1 (a) shows the results of a loading test on a 41.5-ft (l2.6m) long expanded-
base pile driven about 2.5-ft (0.8m) into very dense sandy, gravelly till overlain by
finn to stiff sandy clay; the maximum test load was 300 tons (2670 kN). The stability
plot of the load-movement data is shown in Fig. 1(b). Using Eq.l, and the values of
m and c from Fig. 1(b), gives a failure load of 364 tons (3240 kN), which agrees with
a visual inspection of the load-movement data indicating that a failure load around
350 tons (3115 kN) is not unreasonable.

819 Neely
c
Q

~
C1
~
l
.,
2
m = 0.002248

0
...J
Ie

.!!
l:i:

Intercept c = 0.00074
O.S:;-----;:!-:;:----+ ..L--J
OJ 0.6 0.9 o 0.2 0.4 06
Pile Head Movement, .1 inches Pile Head Movement. .1 inches
Fig. I(a) LoadMovement Curve Fig. l(b) Stability Plot of Load Movement Data
for 19 1/ r in. Diameter Pile (I in. = 2S4mm; I ton" 89 kN)
(1 in. 2S.4mm; 1 ton - 8.9 kN)
2401,....---"'T""---.....- - - - - .
Pile Head

S;.....- - - , . . . - - - - . - - - - -
2.02 = 2.02 I[ 2.286 I[ 10
0

] +0.00225
Q-
Q_ = 294 tons
c
o
~
.c
.~
1:.

w =2.021[ 2.285 x 10'] +O.0012S


Q-
Q-" 288 tons

0.6 0.9
O,:-----!'=----+:-----'
0 0.3 0.6 0.9
MovaDeD1, 41 iDches Mowmem, A inches
F"18- 2 Load-Movemeat Curw:s FI8- 3 Stability Plots ofLoad-Mow:mc:at
fer 16 1/rin. Diameter Pile Data (1 in. - 2S.4mm; I too so 8.9 kN)

For most of the piles in the data base a significant portion of the movements
measured at the pile head were due to elastic compression of the pile shaft. It was
therefore important to verify that the load-movement response at the head of the pile
is a reliable means of evaluating the ultimate load of point bearing expanded-base
piles. Fig. 2 shows the load-movement response measured at the head of a pile and
the corresponding curve of the displacements measured by the tell-tale just above the
top of the expanded base. The stability plots of the pile head and pile tip movements
are presented in Fig. 3. Even though the maximum test load was applied in two
cycles, and individual load increments were maintained for between 2 hours and 36
820 Neely
hours, most of the data points lie on two straight lines. The instrumented pile was 28
ft (8.5m) long and the diameter of the expanded base was 3.37 ft (l.Om). Substituting
the values of m and c from Fig. 3 in Eq 1 yields derived ultimate loads of Qum = 294
tons (2620 kN) based on pile head movements, and Qum = 288 tons (2565 kN) using
movements measured just above the top of the expanded base. Such good agreement
tends to support Roscoe's (1983) conclusion that neither elastic compression of the
pile shaft nor application of the test load in load-unload cycles has much infl uence on
the value of the derived ultimate load, Qum' Values of the derived ultimate load for
the 20 expanded-base piles in the data base average about 252 tons (2240 kN) in the
range 98-434 tons (870-3860 kN).
RESULTS
The ultimate point resistance, qp, was obtained by dividing the measured failure
load, or the derived ultimate load, Qum' from the stability plot by the area of the
expanded base.
In examining the results of the loading tests a distinction was made between short
piles and long piles. The point resistance of short piles is considered to increase with
increasing depth in homogeneous soils above a critical depth for the bearing stratum.
Long piles are commonly considered to be founded below the depth where point
resistance is no longer directly proportional to depth. In this study, a long pile is
defined as having a DJBb ratio greater than 10, where Db = embedment depth of the
maximum cross section of the expanded base, taken as the sum of the driven length
and one-half the diameter of the base, B b Of the 20 piles in the study, 80% (16 piles)
were long piles. In addition, 14 of the piles in the data base were driven through
relatively soft, compressible material overlying a thick till stratum; the remaining 6
long piles were installed in relatively homogeneous soils.
The ultimate point resistance, qp is plotted against N, the measured standard
penetration resistance, in blows per foot (blows per 300mm), near the pile point in
Fig. 4. On average, the point resistance oflong piles may be taken as
qp (tons/sq ft) = I.2N (2)
There appears to be very little difference between the point resistance of long
piles driven into relatively homogeneous soils (qpIN = 1.3; mean N = 33 blows/ft) and
long piles that penetrated compressible soils and into a thick bearing stratum (qpIN =
1.1; mean N = 53 blows/ft), provided that the expanded base was at least 1.5 base
diameters below the surface of the bearing stratum. It should also be noted that the
ratio qpIN. tends to decrease with increasing N, confirming a trend that was observed
for expanded-base piles in sand (Neely 1990a).
The point resistance of the four short piles (i.e., DJBb < 10) appears to be in fair
agreement with the following expression
(3)

821
Neely
120r------~,---T'""
.---'.------",-----,
o PilC$ driven into homogeneous soils
PilC$ driven through compressible soil
into thick bearing strarum
~ 8.2 4

/ q, = UN (!Sf)
/" . ., Average
/ 9.4
o 5.3 4
~.. ~
~a SS .............. ~IlL='
~.~ s._
;r~~
l~ I ~.'lO.-:-notod
00 20 40 60 80 100
Standard Penetration Resistance, N, blows/ft.
Fig- 4 Correlation between Ultimate Point Resistance
and Standard Penetration Resistance
(ltsf .. 95.8 kNlm';l ft .. 0.305m)

Values of qp are plotted in Fig. 5 against the ratio DJBb' For a given value of DtlBb'
the ultimate point resistance increases with N. The values of qp range up to about 30
2
tonslsq ft (2.9 MN/m ) for long piles in medium dense till to as much as 90 tons/sq ft
2
(8.6 MN/m ) in very dense till. The data indicate that a limiting value of qp is not
reached, even for exceptionally long piles; rather the trend is one of a continuing
increase in point resistance with increasing DtlBb ratio, but at a reducing rate. A
similar trend was observed in two previous studies of expanded-base piles in sand
(Neely 1990a; 1990b) and, on the basis of an elegant theoretical analysis, Kulhawy
(1984) was able to show that the point resistance (and shaft resistance) of long piles
do not reach a limit at a critical depth.
From bearing capacity theory, the ultimate point resistance is given approximately
as
qp = PoNq (4)
where Po = the effective overburden pressure at the level of the expanded base; N q = a
theoretical bearing capacity factor that varies with the angle of friction of the soil and
the assumed failure mechanism.
Values of N q were detennined for seven of the piles in the data base for which
sufficient infonnation was available to calculate the effective overburden pressure at
the level of the maximum cross section of the expanded base. The values of Nq from
the present study are plotted against the corresponding effective overburden pressure
in Fig. 6. It can be seen that the N q values decrease from about 100-120 for Po = 0.5
tsf (50 kN/m 2 ) to about 30 for Po > 1.5 tsf (150 kN/m\ It is interesting to note that
822 Neely
the N q values for glacial till lie within the range of values reported previously for
piles in relatively clean sand (Neely 1990a).
The curves in Fig. 5 suggest that, for a given relative density, ultimate point
resistance increases with depth (or Po) but at a decreasing rate, implying that the
bearing capacity factor N q in Eq 4 also decreases with increasing Dt/Bb ratio. This
so-called scale effect arises because the strength envelope is not straight, but curved;
in other words, the angle of friction--and N q which is a function of the angle of
friction--decreases with increasing confining pressure.
Ultimate Point Raistance, q, (tsf)
90

11
Very Dense
(N) SO)
A

OTaS A
C

jJ A

2
o Medium Dense
C Dense
A Very Dense

)OL...-_ _.....;L-_ _...... I.---L_~

C
F18- 5 Corre1atioIl between Ultimate
Poinllledmce mel IN'B. Ratio
(Itsi'''' 95.8 kNIm~

The dependence of the angle of friction on pressure is most pronounced for soils
that are initially dense, are initially of relatively uniform grain size, and have been
heavily overconsolidated. Curvature of the strength envelope is associated with
crushing of particles and, in dense soils, with a greatly reduced rate of volume
increase (dilation) at failure (Bishop 1966). Bishop also noted that particle
breakdown results in a grading tending to approximate to that found in naturally
occurring glacial tills, for which the angle of friction has proved to be relatively
insensitive to stress (Insley and Hillis 1965). However, the stress range represented
by the data in Fig. 6 is close to the origin of the extensive envelope of Insley and
Hillis where there is some curvature which is not evident at elevated stress levels.

823 Neely
Although the limited evidence from the present study tends to support the view that
scale effect may be important for piles in glacial till, it should also be kept in mind
that variations in grain size characteristics and fines content could also account for
variations in the angle of friction. For example, in their investigation of the strength
of New England tills, Linell and Shea (1960) found a variation of 6 in the drained
friction angle of tills having fines contents between 18% and 56%. A reduction in the
friction angle from 38 to 32 would correspond to Nq values of 130 and 46
respectively, based on the bearing capacity theory of Berezantsev et al. (1961), i.e.,
very similar to the range of backcalculated N q values. In tills containing appreciable
fines, it would certainly be prudent to use a bearing capacity factor corresponding to
an angle of friction near the low end of the range given by LineH and Shea (1960).

Depth of Bue. 14 shown 13 e. g (12. 1)


Bue Diameter B.
60
(5.3) 0
SPT NValue shown 13 e.g. 58

'58
(12.1)0 075
(9.4)
30 47
0 0 (14.6) 62
(11.0)
o 50
(28.8) 0 (30.9)

10L.----:l":---~::----~:__--":"':---':'
o 0.5 1.0 1.5 2.0 2.5
E.ft'ectiw Owrllurden Pressure, P~ t~
fiB. 6 ComIatioD between BeariDI Cap8city FlCtor and
E6ective ()verburdeIl PresaIre (1 tal - 95.1 kNlm1

Three of the piles in the data base included a tell-tale just above the top of the
expanded base, permitting direct measurement of pile tip movements. The difference
between the movement at the head of the pile and the movement of the tell-tale
represents the elastic compression of the pile shaft. Values of Ec, the elastic modulus
2 2
of concrete, were found to decrease from about 3370 kips/in (ksi) (23200 MN/m ) at
2 2
an axial stress of 1.0 ksi (6.9 MN/m ) to 3130 ksi (21560 MN/m ) for axial stresses of
up to 2.0 ksi (13.8 MN/m2); at the maximum test load, corresponding to an axial
2
stress of about 2.2 ksi (15.2 MN/m ), the value of Ec decreased to 2940 ksi (20255
2 2
MNI m2 ). The value of Ec for nominal 4000 Ib/in (27.6 MNI m ) concrete,
calculated in accordance with ACI 318-83, is approximately 3600 ksi (24800 MNI
2
m ).
Analysis of the tell-tale data always yielded values of Ec less than that calculated
using the ACI 318-83 formula, implying that no shaft resistance was acting on the

824 Neely
piles during the loading tests. Tip movements of piles that were not instrumented
were computed as the difference between the measured movement at the head of the
pile and the calculated shaft compression, assuming Ec = 3600 ksi (24800 MN/ m 2)
and zero shaft resistance. Although data from the instrumented piles show that E c
decreases slightly with increasing stress level, this does not significantly affect the
calculated stiffness of the till.
An estimate of the settlement of an expanded base may be obtained from
Pb = PIp (1- v2)/O.785B bEv (5)
where P = pile load; B b = diameter of expanded base; I p = influence factor dependent
on depth (Burland et aI., 1966); v = Poisson's ratio (taken as 0.3); Ev = vertical elastic
modulus of the glacial till.

80 20

0 WBb = 88; N = 47
6 WBb = 9.4; N = 75 -0

60 ..
::
0\
c
9
".::: .5
4
0\
c lot!
wi
='
.5 "'5
"8
~ 2
u
'j
iii o Medium Dense
o Dense
6 Very Dense

O......_~:--_~_~.L-_'"""'="~_~
o 0.2 0.4 0.6 1.0 0.1 Q2 Q4 Q6
Degree ofLoadio& qlq, SettJemem Ratio, ~ iD %
F"\I- 7 Variatioa ofE.IN with Degree F"18. 8 CorreIItioD betwea EIutic Modulus
ofLoldiDa qlq, (ItIt'- 9H kNhn~ aDd Scttlcmerc of'Exp&Qded Rue
(l tIf' - 9S.S kNhn~

Values ofEy for two of the instrumented piles (the tell-tale in the third pile did not
function properly) are plotted against degree of loading, q/qp in Fig. 7, where the
actual bearing pressure, q, is expressed as a proportion of the ultimate point
resistance, qp' corresponding to the derived ultimate load. The pattern of decreasing
E/N with increasing q/qp shown in Fig. 7 is very similar to that developed by Stroud
and Butler (1975) from a study of the settlement records of large structures supported
on glacial soils. Fig. 8 shows the correlation between the vertical elastic modulus, E y ,
calculated from Eq 5 at an axial load about 50% of the derived ultimate load, Qum'
and the relative settlement of the pile base. Although Ey increases with increasing
standard penetration resistance, it can be seen that the stiffness of the till is also

825 Neely
strongly influenced by the amount of base settlement; in other words, the stress-strain
characteristics, at stress levels equivalent to 0.5qp, are non-linear. The assumption of
linear elastic behavior under working load conditions appears to be approached for
factors of safety of about 3.0. Fig. 9 shows the correlation between Ev and SPT N-
value at bearing pressures of about 0.30qp' With the exception of two piles where the
bases were formed just below the surface of the till, there appears to be a potentially
useful correlation between the elastic modulus of the till and the N-value near the tip
of an expanded-base pile.

CONCLUSIONS
The ultimate point resistance, qp, of an expanded-base pile at depths of more than
10 base diameters below ground level can be calculated using the empirical
correlation, qp = 1.2N tsf (95.8 kN/ m2), where N is the measured standard
penetration resistance near the pile base. For piles at depths less than 10 base
diameters, the ultimate point resistance may be taken as qp = 1.2N DtlBb in tsf (95.8
kN/ m\ There appears to be no significant difference between the point resistance of
piles driven into relatively homogeneous soils and piles driven through soft,
compressible soils and into a thick stratum of till.
For long piles, i.e., DtlBb>lO, a limiting value of point resistance is not reached; the
results show that qp continues to increase with increasing depth, but at a decreasing
rate. The point resistance of expanded-base piles in glacial till is significantly less
than that reported previously for similar piles in sand, primarily because of the greater

826 Neely
fines contents and lower friction angle of the tills. The vertical elastic modul us of the
till correlates reasonably well with standard penetration resistance, provided the
factor of safety against bearing capacity failure is at least 3.

APPENDIX. REFERENCES
ACI 318-83. "Building code requirements for reinforced concrete." American
Concrete Institute, Detroit, MI, Nov., 1983.
Berezantsev, V.c., Khrisoforov, V.S., and Golubkov, V.N. (1961). "Load bearing
capacity and deformation of piled foundations." Proc. 5th Int. Conf. on Soil
Mech. and Foundation Engrg., 2,11-15.
Bishop, AW. (1966). "The strength of soils as engineering materials."
Geotechnique, London, England, 16 (2), 91-128.
Burland, J.B., Butler, F.G., and Dunican, P. (1966) "The behaviour and design of
large diameter bored piles in stiff clay." Proc. Conf. on Large Bored Piles,
Institution of Civil Engineers, London, 51-71.
Chin, F.K., and Vail, A.J. (1973). "Behavior of piles in alluvium." Proc. 8th Int.
Conf. on Soil Mech. and Foundation Engrg., 2.1, 47-52.
Flint, R.F. (1957) Glacial and Pleistocene Geology. John Wiley & Sons, Inc., New
York, N.Y.
Hirany, A, and Kulhawy, F.B. (1989). "Interpretation of load tests on drilled shafts.
Part 1: Axial compression." Proc., Foundation Engrg. Congress, ASCE, 2, 1132-
1149.
Insley, A.E., and Hillis, S.F. (1965). "Triaxial shear characteristics of a compacted
glacial till under unusually high confining pressures." Proc. 6th Int. Conf. on Soil
Mech. and Foundation Engrg., 1,244-248.
Kulhawy, F.H. (1984). "Limiting tip and side resistance: Fact or fallacy?" Proc.
Symp. on Analysis and Design of Pile Foundations, ASCE, 80-98.
Linell, K.A, and Shea, H.F. (1960). "Strength and deformation characteristics of
various glacial tills in New England." Proc. Res. Conf. on Shear Strength of
Cohesive Soils, ASCE, 275-314.
Mitchell, J.K. (1968. "In-place treatment of foundation soils." Proc. Special Conf.
on Placement and Improvement of Soil to Support Structures, ASCE, 93-130.
Neely, W.J. (1989). "Bearing pressure-SPT correlations for expanded base piles in
sand. "Proc., Foundation Engrg. Congress, ASCE, 2, 979-990.
Neely, W.J. (1990a). "Bearing capacity of expanded-base piles in sand." J. Geotech.
Engrg., ASCE, 116 (1), 73-87.
Neely, W.J. (1990b) "Bearing capacity of expanded-base piles with compacted
concrete shafts." 1. Geotech. Engrg., ASCE, 116 (9), 1309-1324.
Nordlund, R.L. (1982). "Dynamic formula for pressure injected footings." J.
Geotech. Engrg., ASCE, 108, (3), 419-437.
Roscoe, G.H. (1983). "The behavior of flight auger bored piles in sand." Proc. Conf.
on Piling and Ground Treatment, Institution of Civil Engineers, London,
England, 241-250.

827
Neely
Stroud, M.A., and Butler, F.G. (1975). "The standard penetration test and the
engineering properties of glacial materials." Proc. Symp. on the Engrg. Behavior
of Glacial Materials, Birmingham, England, 12pp.
Terzaghi, K., and Peck, R.B. (1967). Soil Mechanics in Engineering Practice. 2nd
Ed., John Wiley & Sons, Inc., New York, N.Y.

828 Neely
Elasto-plastic analysis of laterally
loaded piles

Shunsaku Tanaka 1, Yoshinobu Sawan0 2, Fuminao Okumura 3 ,


Akihiko Nishimura4, Tadatomo Watanabe 5

Abstract

Limit state design introduction to Foundation Design Standard


for Railway Structures has been under way in Japan. Defining ultimate
limit state of pile foundations is one of the major tasks to introduce limit
state design. In Japan, scales of structures are mainly determined by
earthquake resistant design. In this paper the authors analyzed lateral
loading test results of small size pile foundations conducted from elastic
range to yield range. The tests simulated foundation behavior in large
displacement and failure modes by large earthquake. The test results are
analyzed by the method based on elasto-plastic principle both for pile
and soil. Analyzed behavior of single pile and pile group are coincident
with the test results such as bending moment distribution and load-
displacement curves. The applicability of the analyzing method is
demonstra ted.

Introduction

In accordance with three limit states defined in Nishimura et


al.(1993), such as serviceability, ultimate I and ultimate II limit states, in
this paper the authors focus on pile foundation design by limit state
design. Behavior of pile foundation at the ultimate limit state is quite
complicated, because there are many critical states for pile foundation
support, structural members and structural functions. During strong
earthquake, the pile foundation may suffer strong lateral movement.
Lateral movement causes pile fracture by bending, intolerable lateral
deformation by supporting soil yielding, unrestorable settlement by
eccentric force and combination of those stated above.

1 Researcher, Railway Technical Research Institute (RTRI), 2-8-38


Hikari-cho, Kokubunji-shi, Tokyo 185, Japan,2 Chuou Fukken Consult-
ants Co., 3 Senior Researcher, RTRI, 4 Chief, RTRI, 5 Researcher, RTRI

829 Tanaka et al.


Lateral loading tests of small size pile foundations were conducted
from elastic range to yield range and the tests results were utilized for
predicting foundation behavior at large displacements. The authors
analyzed these test results by the method based on elasto-plastic princi-
ple both for piles and soils.

Model pile test

Sets of steel pipe were used for pile models. Four tests of two
series were conducted for modeling effects of number of piles in pile
group. The thickness and diameter of piles were 3.2mm and 101.6mm
respectively. The depth of pile in soil layer was 2000mm. SOmm of pile
top was free above the ground surface. Test types are shown in Table 1.
Table 1. Test types

Type # of pile Distance between


in pile group piles (D:pile diameter)

GH4 4 2x2 3Dx3D


GH6 6 2x3 3Dx3D
GH9 9 3x3 3Dx3D
TH1 1

Series 1 is GH4 to GH9. Series 2 is TH1.

Artificial test ground was constructed in Maglev test pit of Rail-


way Technical Research Institute(RTRI). Construction method of the
artificial test ground was the same for two lateral loading test series.
Sand(Inagi sand) and gravel were used for making 180cm of the upper
and SOcm of the lower layer, respectively. Each dumped sand layer
thickness was 30cm and was compacted four times by a vibration
roller(weight 6S0kg). Each compacted thickness was 2Scm. Main char-
acteristics of the upper artificial ground layer of series 1 are shown in
Table 2. Figure 1 shows the pile configuration for series 1.
Table 2. Characteristics of artificial ground
Water Content (%) 16.6
Wet Density (g/cm 3) 1.68
Void Ratio 0.884
Degree of Saturation (%) 51.2
Cohesion (kg/cm 2) 0.13
Angle of Internal Friction (deg) 34.0

830 Tanaka et at.


GH4
I
I
~
. II I ,I
~
-,

li, IlJ'o

Ii
I,
:IE. ~ ~ I =18"
l2QQJ
I I 13001

----+-------;-----1---
1SOO 2000 2000
T 1500
,
7000

Fig.l Pile configuration for series 1 (unitmm)

7000,

1500 2000: 20001 1500

Fig. 2 Configuration of artificial test ground and piles (unitmm)

1500

Fig. 3 Testing rig for Lateral loading tests (unit:mm)

831 Tanaka et at.


Piles were driven into ground by 2000kg hammer. The depth of
penetration into the lower layer was 200mm. A concrete footing was
made at the pile head. Testing rig for lateral loading tests is shown in
Figure 3. Surcharge weight was used for simulating weight of super-
structures. Axial load to piles caused by surcharge weight affects bend-
ing moment distribution along piles.

L:iteral loading was repeated three times to the same displacement


in two directions for the purpose of simulating earthquake force. Loading
height was set for simulating superstructure-pile behavior during earth-
quake. Load was maintained for short time for data recording. Loadings
were conducted by deformation control at the loading point. During
these tests, loads and deformation at loading points and strains along
piles were measured and bending moments and shear forces were calcu-
lated from the test results.

Analysis

Non-linear behaviors of piles and soils were shown from lateral-


loading test results. This fact indicate the plastic deformation of the piles
and the surrounding soils. The authors simulated these test results by two
dimensional analysis assuming non-linear behavior both for piles and
soils. Hysteresis behavior of single pile and pile groups are analyzed by
an elasto-plastic simulation program called FAMC, or Frame Analysis
in Consideration of Moment-Curvature Relationship. This program was
originally developed for analyzing elasto-plastic behavior of concrete
structures.

MCkNom)

M p = 10.75 -_._._._---------~-~----------------

M y = 7.22 _. - -
E I
_ _..
z:=
I
,
99. OCkN-m Z
)

,,
,
,,
1= 210:., 6CkNom Z )
L.......J--'- --'- ~ rP C1/ m)
0.0343 0.07
=rP y =rP p

Fig.4 Assumed correlation of moment and curvature ofpiles

832 Tanaka et at.


p

P o. P -------------..,--------

k 0, k v
o

Here.
P.b: Limit of the lateral effective resistance earth pressureCk~/m~)

P eh (z) = P ehl (z) (sand)+ Peh2 (z) (cohesive soil)

Pehl(Z)=2Te,z'Kp (sand)
z : DepthCm)
~ Average effective weight per unit volumeCkN/m )
J
K, : Coefficient of passive earth pressure

Kp=.tan [45+ ~ J
2
{(degree): Friction angle of sand at depth of z}

JeT: Z + 2 c) ~
Z
Peh2 (z) =[ 1 + 2 D 6 c (cohesive soil)
o Pile diameter(m)
c CohesionCkN/m~)

k b : Lateral sub grade reaction effectCkN/m J )


_-l.
k h =O.2aE o D
a Modification coefficient according to calculation method of Eo
Ca = 2 : Standard penetration test)
Eo: Elastic modulus of ground

k. : Vertical subgrade reaction effect


P : The limit of the vertical effective resistance earth pressure
k. and P were assumed from the test results.

z P .h P k hCXIO J k .(X]OJ
(m) I Ck~/m~) I CkN/m 2 ) kN/m J ) I kN/m J )
O. I I 52. I 34.4
O. 3 110. 3 5 I. 6
O. 5 134. 7 51.6
0.7 158. 0 68.9
O. 9 181.2 5589.9 68.9 1148.6
1.1 204. 5 68.9
1.3 227.8 86. I
I. 5 251. I 86. I
1.7 274. 4 86. I
1.9 2212. 1 103. 3
Fig.S Assumed correlation oj load and displacement oj ground

833 Tanaka et al.


Assumed tri-linear correlation between moment and curvature of
piles is shown in Fig.4. Axial force effect to moment-curvature relation-
ship of piles was not considered in the analysis. Plastic behavior of
surrounding soils was considered in lateral ground reaction. The bi-
linear correlation between lateral resistant force and displacement is
shown in Fig.5. Subgrade reaction and effective resistant earth pressure
were calculated from the soil characteristics of the test ground. The
effective resistant earth pressure was the limit of subgrade reaction force.
Coefficients of vertical ground reaction were introduced from load-
displacement behavior during surcharge loading.

In the analysis, lateral load was increased at the loading point in


one direction. Load-displacement curve and bending moment distribu-
tion along piles were calculated.

Analysis of single pile

The result of single pile analysis is compared with the test result of
THl as shown in Fig. 6. On the condition of subgrade reaction coeffi-
cient and the limit of subgrade reaction stated above, the analysis was
not coincident with test results. Subgrade reaction calculation formulae
was defined from the lateral loading test results of real piles. In those
loading tests, footing were loaded laterally and the pile heads were dis-
placed as the fixed head condition. But in this loading test of the single
pile, pile head was rotated freely and the condition of subgrade reaction
was considerably different from the assumption of calculation. The
analysis condition was modified as the parameters of subgrade reaction
and the limit of effective resistance earth pressure for fitting the test
results and the analysis. The modified condition was that subgrade reac-
tion was multiplied by three and the limit of the effective resistant earth
pressure was multiplied by two. The modified analysis condition results
in good simulation of test result.

The bending moment distribution along the single pile while


loading is shown in Fig.7. By the modified analysis condition, analysis
and test results of load and maximum bending moment along the pile
was compared and shown in Fig. 8. On the same analysis condition
above, the relationship between the load and the depth of maximum
moment along pile is shown in Fig.9. The values of maximum bending
moment along the pile and the depth of the maximum bending moment
were simulated well by this analyzing method. The depth of the maxi-
mum bending moment became large along with the lateral loading in-
crease.

From these analysis, the authors conclude the estimation formulae

834 Tanaka et al.


for the subgrade reaction and the limit of the effective resistant earth
pressure for single pile is not accurate enough in the large displacement
range because of the difference of the loading conditions between this
test and the pile group loading tests.

24 5 '

19.6

~
14.7,

9.8 /AVd
V/ .///
/
la

4.9 J ~~'

; l 0 If#~:""
"fr';~"""-" i
,
~'1fJf1
,.....
..' !.
10
I
20
I
30 40 50

. p /ftJ
/ " ---LL-
li~ I I
f l/Y--_ :Test result I
/. / ..

:-/ ... :Analysis,- condition: 3k, 2Pe

I
:Analysis, condition: k, Pe
I I I I I

Lateral displacement (mm)

Fig. 6 Experimental and analytical load-displacement curves(TH1)

tl -Of) 147-1176-8 82 1 -~ 8:;~~ 0 \\2: ~ ,5 8~


!Ioment(kNm)

8 82 \116 14

U' \,,:/ ~'


m-n ,,\ \"
or
I

~~l
O. 7
0.8
1 \

"'"
\

\
\

\
\

\ \
\
I

!! i
,;!.
,i /
!,'

'! /.
;,/ //
i
!
i

,,.../
\

./

/ ~
No I LoadCk:'i)

3I
1 -:

4.9
i
.J3-fi- I. 3 4 - 5. 5

:ft: -+_--+_\_4\\~!/_i-j_--'I---j--;;~+----;-:-;;-~:o~-1
I' .

I I "" Iii,,""" ~ I _~: ~


:5-+-_-+-_-I__
1
9 16.3
Dc p t h(m) '--_"--_-'--_-'--_-'-__-'--_-'-__-'--_,-'1-'-0....:..._'-,-l7.;....-,-2---,

Fig. 7 Bending moment distribution along the single pile (TH1)

835 Tanaka et a!.


14.7,
,,-..,

Z
~
E
'-'"

6 ............
:Test result
:Analysis
~
'5-
bJ) 9.8, P.
l::
.9 1J.'
ca ,
..... ,,
l::
II.J
, t:s.
E
0
E 4. 9
E
='
E
~
ca
::E
,
O. 0 4. 9 9.8 14. 7 19. 6
Lateral load (k~

Fig.8 Load v.s. maximum moment along pile

o. 0 .---~--,----r----'----,---'----"-~

- O. 5


6.
:Test result
:Analysis
- 1. 0 1....,-:---'-'-'.-!"---'-----'-----'---l..----L--'----'---'---'---L--",-,---,-,---1
'
o. a 4. 9 9. 8 14. 7 19. 6
Lateral load (kN)

Fig.9 Load v.s. depth of maximum moment along pile

836 Tanaka et al.


Analyses of pile groups

Analysis model of pile group is shown in Fig.10. The analysis


conditions were two cases such as with and without pile group effect
consideration. Pile group effect is calculated from the chart in Founda-
tion Design Standard for Railway Structures (1986, FDS86). This chart
was introduced from the results of small pile model tests. The lateral
subgrade reaction to all piles were reduced by the coefficient of pile
group effect (eg).

Analyses and test results of pile groups (GH4, GH6, GH9) are
shown in Fig.11, 12 and 13. Non-linear behavior of test results were
simulated well. As the number of piles in the pile group increased, The
pile group effect became large. In these cases, the analysis with pile
group effect simulate test results better than the analysis without pile
group effect consideration.

In the small displacement ranges, the analysis without the consid-


eration of pile group effect predicted satisfactory the envelop of hystere-
sis curve of pile group. But in large displacement ranges the analysis
with the consideration of pile group effect predicted the test results of
repeated load-displacement behavior better than the analysis without
consideration.

The bending moment distribution along piles were simulated well


by this method. Analysis and test results of maximum bending moment
of the front pile to the loading direction were compared in Fig.I4 and 15.
In Fig.14, the test results and the analyses agrees well. Pile group effect
consideration does not affect the analyses results.

In Fig.I5 of nine-pile-case, in the small loading range, the test


results and the analyses agree fairly well. But over the displacement
range of 30% of the pile diameter, the test results and the analyses differ
very much. In the large displacement range, the difference of load distri-
bution to the front, middle, back piles to the loading direction became
large. but in the analyses, load distribution was not considered, in other
words, surcharge weight ware supported by each pile equally. Lateral
subgrade reaction to the piles were also treated as the same value for all
piles. But in the loading test subgrade reactions to the front piles were
bigger than those of the back piles. These assumption of the analysis
results in the difference of the test results and the analyses.

837 Tanaka et al.


Lateral load (N)

>

Back pile to loading Front pile to loading

kb:Lateral spring constant


of ground

EI:Rigidity of pile

ky:Vertical spring constant of ground

Fig.10 Non-linear analysis model for piles

13'
58.8

_... [~~'
d;--
I -1"""'-:::::::
f-.-.-.

/:/
~/i-" ~ ' I :
. /
,~:~.
?t~Y :'----,:-
'

29.4 ''-;' i;
-14.7 ! t '.. ' /-',
:/,/.~ 'I

.-il-,!IJl-
,', ,"
-. "
''?~
7"
","
." . I

~
.:.,' .... ~~~O~E~O
~,.
' -~.xm0
'/1' /, .
:-
~ ~:tiJl

:,', / /, . ! V
--;lI,; . ,- -
JL -
:...V.
:Testre~ult .
:Analysls, pIle group effect not conside
..... :Analysis, pile group effect considered

Lateral displacement (mm)


Fig.ll Experimental and analytical load-displacement curves(GH4)

838 Tanaka et a!.


"'0
ro ! - - - - I - - - t - - - t - - + 19.6 HJ ~ I " .",.'
o
...... , ~q7"'20 ...--~ ~O"'.,., 60" .-.... 80 10
......
ro ~~ , . ,J" 11''1:~~'I-'
\-; .".... ,C, ~~~
....
Q.) .... ~v/.-p,?f

j
'V/ I :II fiji I" I
:Test result --1--1--1---4
f----t--<:.<:t'/"--- - -
~L----" :Analysis, pile group effect not considered
I---f---f--
~_.:.-I_--l-'_----l..,A_:Analysis, pile group effect considered (e g =0.54)

Lateral displacement (mm)


Fig.12 Experimental and analytical load-displacement curves(GH6)

122.5 -

98.0

73.5
j~ V~
0 '. V IV
.--- - ,

49. 0,
_i/ll/
' ,.!j.,1/ ./!.J II .:

ri~~~'V/
" ..~!~
./-.
..<,- "
:
"'0 24.5
ro ~'I'j ~.
. -.
o
..- ~ ..'/.~.
" I .... 1 . . . . . '".1' .... -
20 40""'60
.......
80 10
..-
ro ...., "
,.
,.
... .. ',' ,. ,':~
\-;
.-7'--'"
,.,~,
~'
~'~l
')

.....Q.)
j r~; /.'
bh~f/; rtfj1!
! !
; ;)/ j III ~JJ
i - :Test result
:-{/ .. I

~ ./ :Analysis, pile group effect not consider

f A :Analysis, pile proup effect considered (


,
-- -
Lateral displacement (mm)

Fig.13 Experimental and analytical load-displacement curves(GH9)

839 Tanaka et al.


:Test result
~ 4.90 0
's..
gn 3.92 6.
- o
ca
.-;: .2. 9.4

7
CI)

~ 1.96
5 O. 98 ~
~ ~~!.--- I

~ 0 9.8 19.6 29.4 39.2


Lateral load (kN)

Fig.14 Load V.s. maximum moment along pile(GH4)

,-.....
E
Z
7. 84 :Test result . 1
~
'-'"
6. 86 o :Analysis, pile group effect not considered

-'0..
CI)

5. 88
6 :Analysis, pile group effect considered
f

c.o (e g=O.50) /
c::
0 4. 90
ro
.....
c::
CI)

E
0
E
E
3. 92

2. 94
~
///0 l~
96
~.D\.~Cl
:l
1.
E
>:: ~~
c;::
~ O. 98 ....--::;::::.~.
Y'.~
._~r"
~_l I I 1_

o 19.6 39.2 58.8 78.4 98.0 117.6

Lateral load (kN)

Fig.IS Load v.s. maximum moment along pile (GH9)

840 Tanaka et al.


Conclusion

For the sake of limit state design introduction to pile foundation


design, deformation characteristics of pile foundations and lateral soil
support change in the large displacement ranges should be evaluated.
The authors analyzed the test results of lateral loading tests of small pile
foundations by the method based on elasto-plastic principle both for
piles and soils. Applicability of the analyzing method is discussed. The
analyzing method based on elasto-plastic assumption both for piles and
soils can simulate the nonlinear behavior of pile groups.

The authors are doing the same analysis for the conventional
bridge foundation designs. They are trying to find the simplified way
of designing pile foundations in the plastic ranges.

References

Japanese Society of Concrete Engineers(1986) Standard specification for


design and construction of concrete structures, Part l(design)

The Japanese National Railways(1986):Foundation Design Standard for


Railway Structures. The Japan Society of Civil Engineers

Nishimura, A. et a1.(1993) Limit state design application to railway


foundation design standard, lnt. Sympo. on LSD in Geotechnical Eng.,
Vol 2/3, Danish Geotechnical Society

Okumura, F. et a1.(1993) On deep foundation design for railway struc-


tures by limit state design, lnt. Sympo. on LSD in Geotechnical Eng.,
Vol 2/3, Danish Geotechnical Society

Railway Technical Research Institute (1992) Design Standard for Rail-


way Structures (Concrete Structures), Maruzen

841 Tanaka et at.


ANALYSIS OF PILES UNDER LATERAL LOADING WITH
NONLINEAR FLEXURAL RIGIDITY

Lymon C. Reese l and Shin Tower Wang2

ABSTRACT
The p-y method for the analysis of piles under lateral loading is described briefly,
and the procedures to consider the nonlinear flexural rigidity of reinforced-concrete piles
are discussed. A numerical iteration technique is used for solutions to satisfy the
nonlinear soil resistance as a function of the pile deflection and for crackedluncracked
flexural rigidity as a function of applied moment. Several examples are given to illustrate
the effect of nonlinear flexural rigidity on pile response.

INTRODUCTION
Laterally loaded piles are found in many structures, both onshore and offshore.
As a foundation problem, the analysis of a pile under lateral loading is complicated by the
fact that the soil reaction (resistance) is dependent on the pile movement, and the pile
movement, on the other hand, is dependent on soil response as well as the rigidity of the
structure. Thus, the problem is one of soil-structure interaction.

The p-y method is being used extensively in analysis of piles under lateral
loading. The pile response is obtained by iterative solution of a fourth-order differential
equation, using difference-equation techniques. The soil response is described by a
family of nonlinear curves (p-y curves) that give soil resistance p as a function of pile
deflection y.

The behavior of a pile subjected to bending is dependent on its flexural rigidity


EI. The value ofEI is found from the product of the modulus of elasticity of the pile and
the moment of inertia of the cross section about the axis of bending. The value of EI is
essentially constant for the level of loading to which a structural-steel member is
subjected, but both E and I vary as the stress conditions change for a reinforced-concrete
member. For concrete, the value of E varies because of nonlinearity in stress-strain
relationships, and the value of I is reduced because the concrete in the tensile zone below
the neutral axis becomes ineffective due to cracking. The tensile weakness of concrete
and the ensuing cracking are the major facts contributing to the nonlinear behavior of
reinforced concrete.

1 Professor of Civil Engineering and Principal. Lymon C. Reese & Associates. Austin, Texas.
2 Project Manager, Lymon C. Reese & Associates, Austin, Texas.

842 Reese and Wang


In most cases, the EI value of a reinforced-concrete pile is assumed to be
constant, for simplicity in the analysis. The assumption is not rational but little
infonnation is available in the technical literature on the magnitude of the error it
introduces. This paper deals with the influence of variation of EI on pile behavior. The
analysis of a field-loading test and some recommendations for design are presented.
CRACK MECHANISM
Cracks fonn when the flexural stress due to bending exceeds the tensile strength
of concrete. Immediately after formation of the first crack, the stresses in the concrete
near the cracking zone are redistributed. As loading continues, additional cracks open,
and, in general, the initial cracks penetrate more deeply with increase in load.
Many variables affect the development and characteristics of cracks. The major
ones are percentage of reinforcement, bond characteristics, and tensile strength of
concrete. Because concrete is a heterogeneous material, cracks occur at random and the
location and spacing of cracks are subjected to considerable variation. Studies have
shown that the crack spacing and crack width follow a normal distribution and are
influenced by each other. The crack patterns of a typical beam after testing are shown in
Fig. 1 (Mathey and Watstein, 1960). Statistical studies, based on the accumulation of
such experimental data, are necessary in order to develop prediction methods.

'r \ '7 -
\"\ '7 \ I 'I 1-" I I
'f
,
,
'
) , /
( I /

'" ' - '= \, (I t, ( \---.., ' J \ r r ; ,-


~ " ': ' . ' ',", '~,: I / /
L ]Y'21 _ ~._ _.... '/_' _

[.....;:.~:JZ:._r9-_\_\.'_>_.. ! ! !I~,- l_)_/_' ~_:__I \_;'_~_1_'1_'_/_-__

f/",'
I "lI- Ie

Figure 1. The General Crack Patters of Beams after Testing


(after Mathey and Watstein, 1960)
RELATIONSHIP BETWEEN FLEXURAL RIGIDITY AND BENDING
MOMENT
The variation of EI with the magnitude of the bending moment was studied in a
general manner by Eppes (1959) and his results are presented in Fig. 2. Three distinct
stages of behavior can be distinguished from these measured relationships. They are: (1)
Uncracked stage: the concrete is uncracked and the full uncracked section is available to
carry stress and provide rigidity. The flexural rigidity is more-or-less constant and is
equal to the computed EI for the gross section. (2) Crack-propagated stage: in this region
the EI value is considerably reduced due to the formation of the flexural cracks. The rate
of decrease depends mainly on the amount of reinforcement in the section, which
controls the rate of opening of cracks towards the neutral axis. Propagation is faster for
beams with lower steel ratio. The moment at the beginning of this stage is called the
cracked moment. (3) Fully-cracked stage: a further decrease in the EI value takes place
843 Reese and Wang
as the depths of the cracks continue propagating, and some additional cracks are formed,
thus reducing the EI to a value which is close to that of a fully cracked section. The
calculated EI by the cracked-section method is close to results that have been measured in
the laboratory.
800
700 1/j'IID\1
T T
co
0
......
x
C\I
c:
600
alculated
500 F-----'==..:=.-=::: Vu I (At T 1

T 400
.D
300
u.i 200
100

0
0 10 20 30 40 50 60 70 80
M, kip-in.
Figure 2. Measured Relationship between the Flexural Rigidity and Bending Moment
(after Eppes, 1959) For conversion factors see Appendix.

ANALYTICAL METHOD OF COMPUTING FLEXURAL RIGIDITY


The analysis of a cross section of a reinforced-concrete beam proceeds in a
straightforward manner if the nonlinear stress-strain curves for the concrete and steel are
known. A value of the curvature ep is selected and the position of the neutral axis is
estimated. The strains are computed across the section and slices are taken parallel to the
neutral axis. The stress-strain curves are employed and increments of forces are
computed across the section. The net force across the section is computed; if the section
is not in equilibrium, a new value of the neutral axis is selected and a new trial is made.

With the section in equilibrium, the value of moment can be computed by


multiplying the force on each slice by the distance from the neutral axis. The value of EI
can be found by dividing the value of M by the selected value of cI>. A computer code can
be written to do the tedious computations.

The relationships of M versus ep and EI versus M for a beam with no axial load
were computed and are shown in Figs. 3 and 4. The value of EI was initially nearly
constant until excessive curvature caused the section to crack. After cracking occurred,
EI was calculated using the transformed cracked-section, in which no tensile stresses in
the concrete section were taken into account. A large change in EI at the point of
cracking is shown in Fig. 4. Generally, the range of ep for an intact section is very small,
and most concrete beams behave nonlinearly even under service-loading conditions. It is
evident that the use of the crack-transfonned section should be considered for the
computation of the EI at pile sections where the crack occurs.

NONLINEAR EI EMPLOYED IN ANALYSIS OF PILES


In the analysis of piles subjected to lateral load (Reese, 1983), the flexural
rigidity, EI, is one of the parameters occurring in the differential equation for the solution
of deflection as shown in Eq. 1.
844 Reese and Wang
12.0

.D
I
C
fc'-4 ksi
8.0

Of
CD ly-60 ksl
...o
><
....c:
Q) ,2*8
E 4.0
o
:::E

0.0 L..- '-- -'--


I'"
-'-
30' "I -'--_

o 0.0001 0.0002 0.0004


Curvature, cD

Figure 3. Moment-Curvature Diagram for a Circular Cross Section


25
Computed From Moment Curvature Diagram
N
c: 20
T
.D
15
...
0

...
0
10 Large Strain
><
u:i 5

OL.-_ _.l....-._ _....I..-_ _--'-_ _---J. '--_

o 2.0 .0 6.0 8.0 10.0


Moment x 10+ 6 , in.-Ib
Figure 4. Relationship between Bending Moment versus
Flexural Rigidity for a Circular Cross Section

d2 ~ ~
dx 2 (EI clx2 ) + Px ~2 - P - W =0 (1)

where

Px = axial load on the pile, F,


Y = lateral deflection of the pile at point x along the length of the pile L,
p = soil reaction per unit length FIL,
EI = flexural rigidity, F-L2 and
W = distributed load along the length of the pile, FIL.

A physical definition of the soil reaction p is given in Fig. 5. Figure 5a shows a


profile of a pile that has been installed by driving or by other methods. The soil
resistance is examined for a thin slice of soil at some depth Xl below the ground surface,
as shown in Fig. 5a. The assumption is made that the pile has been installed without
bending so that the initial soil stresses at the depth Xl are unifonnly distributed, as shown
in Fig. 5b. If the pile is loaded laterally so that a pile deflection YI occurs at the depth Xl
the soil stresses will become unbalanced as shown in Fig. 5c. Integration of the soil
stresses will yield the soil reaction PI with units of FIL. An expression for p can be
conveniently written, as follows
Reese and Wang
845
p =-EsY (2)

where

Es = a parameter with the units F1L2, relating pile deflection y and soil
reaction p.

1-""
IAlflf"l"I-""--J
(0)

II

Ib) Ie)

Figure 5. Definition of Soil Reaction p and Deflection y (after Reese, 1983)

It is evident that the soil reaction p will reach a limiting value (and perhaps decrease) with
increasing deflection. Furthennore, the soil strength in the general case will vary with
depth. Therefore, only in rare cases will E s , sometimes called the soil modulus, be
constant with depth.

The bending stiffness EI of a metal pile will probably be constant for the range of
loading of principal interest. However, the EI of a reinforced-concrete pile will change
with the bending moment Because the bending moment is larger near the top of the pile,
in many designs it is desirable to reduce the bending stiffness by reducing the wall
thickness of a steel-pipe pile or by reducing the number of bars in a reinforced-concrete
pile for the lower sections. Thus, in the general case the bending stiffness will not be
constant with x or with y.

In view of .the nonlinearities of Eq. 1, numerical methods must be utilized to


obtain a solution. The difference-equation method can be employed with good results.
Equation 3 is the differential equation in difference fonn which simplifies the influence of
bending stiffness EI in the fIrst tenn of Eq. 1.

846 Reese and Wang


Ym - 2R m -l + Ym-l (-2R m -l - 2R m + P xh 2) + Ym (Rm-l + 4Rm + R m +1

- 2P xh 2 + k m h4 ) + Ym+l (-2R m +l + P x h2 ) + Ym+2 Rm+l - W m h4 =0


(3)
where

Ym = [mite deflection at point m along the pile L,


Rm = Eml m, F-L2,
h = increment length L,
k = E sm , FIL2, and
Wm = distributed load at point m along the pile, FIL.

The pile is subdivided into n increments and n + 1 equations can be written in the form of
Eq. 3, yielding n+5 unknown deflections. Two boundary conditions at the bottom of the
pile and two at the top of the pile allow for a solution of the n+5 equations with selected
values of Rand k. The value of n and the number of significant figures in Y can be
selected to give results with appropriate accuracy. The solution of the equations
proceeds readily by Gaussian elimination. The value of n ranges from perhaps 50 to
200; on most computers double-precision arithmetic is necessary.

The solution proceeds as illustrated in Fig. 6. Figure 6a shows a pile subjected


to a lateral load. Figure 6b shows a family of p-y curves where the curves are in the 2nd
and 4th quadrants because soil resistance is opposite in direction to pile deflection. Also
in Fig. 6b is a dashed line showing the deflection of the pile, either assumed or computed
on the basis of an estimated soil response. Figure 6c shows the upper p-y curve
enlarged with the pile deflection at that depth represented by the vertical, dashed line. A
line is drawn to the soil resistance p corresponding to the deflection Ywith the slope of
the line indicated by the symbol E s . Figure 6d shows the values of E s plotted as a
function of depth x. In performing a computation, the computer utilizes the computed
values of E s and iterates until the differences in the deflections for the last two
computations are less than a specified tolerance. The bending moment along the pile is
computed during the iterations, and the new value ofEI is internally selected for each pile
increment based on the magnitude of the computed moment. The new EI will then be
used to compute the pile deflection for the next iteration until the convergence is reached
within the tolerance.

p y
... ...
\
\

\ \
\

x \ \
\ ,
~ ,,
(0 ) (b) (e) (d)

Figure 6. Procedure for Iterations


847 Reese and Wang
The number of iterations, for a tolerance of 0.()()()()1 in. (0.()()()25 mm) between
computed deflections for successive iterations, is usually less than 20. A high-speed
computer can converge to a solution in less than one second of central-processor time.
Thus, if data on nonlinear flexural rigidity and p-y curves are available, a solution to a
given problem can be obtained with little difficulty.

EFFECTS OF USING CRACKED/UNCRACKED EI IN ANALYSIS


In most cases, the EI-value is assumed to be constant in the analysis, but little
information is available in the literature to guide in selecting a reasonable value for this
constant. A parametric study was undertaken for a particular pile and for a particular set
of p-y curves. The percentage of EI of the gross-section was varied. It should be noted
that the gross-section EI defined in this paper is the product of the elastic modulus of
concrete and the moment of inertia of the cross section; the presence of rebars is not
taken into account for simplicity. The results of the parametric study for different
selections of EI values are shown in Figs. 7 and 8. As may be seen, the bending-
moment curves were affected marginally but the deflection curves were affected
markedly.

A study was conducted to find the effect on pile behavior by using a rigorous
procedure which employed the nonlinear variation of EI. The results for a particular pile
and soil are shown in Fig. 9. For comparison, an analysis using the assumption of a
constant gross-section EI was also made and is also shown in Fig. 9. The figure
presents a plot of deflection at the top of the pile versus lateral load. It is apparent that
there are significant differences in the load-deflection curves and these differences
increase with load. On the other hand, it is interesting to note that variation in EI has
little influence on the maximum bending moment. In fact, bending moment along a pile
does not depend strongly on its structural characteristics. If the pile becomes stiffer by
increasing steel ratio by 1%, the cracking zone is limited to a small range. Consequently,
the differences due to the EI selections are small, as shown in Fig. to.

-0.:1 0.0 0.:1 1.0


o .0 rT..JTl
I IT IlrTTT1"..-,;~n::>rr"'[J;7'1'TT"TTI-I-:'"Tl.I:I"TT"T'TT'1r'TT"TTT1
I rTlTl
The curves from left to right were
0.6 f computed using the following
values ofEI:
1.01 I--> ...... ,~l~ ...... ' ........,... ..... ,......... , ... .... ,.......I
100% Gross Section
75% Gross Section
1.8 t- .. ; ........ t .... , .. :.. ,.. .. ,~ .... ,.. ,:.. ",, .., ....,,;...... 1
50% Gross Section
~ 25% Gross Section
01.4 t-I-_ ...., ....... ,., .. " .. , .........:........ " ..... "" ...... <" .. 1
~ 10% Gross Section
t-~_ ...,.. ,.... ,iI..... ,... " ......
] 3.0
I-
~
'O ........ :, ........ , ........ : ...... "I

~
3.6 1-1-_ ..., ....... ,..... ,,,,: .........:......... :..... ,.. ,........ '........ 1
~
4.2 1-~_
I-
.... ,.. ,.... ,'I' ....... ,.........:......... :........ ' ...... "i" .......1
~
4.8 1-1-_ , iI. ; .;."." .. ,: ,,,,, ' 1
I-
~
:1.4 t-I-_ ..., .. + ......".... ,"....",,,....,.... '.. 1
f: '
6. 0 L:..~_'-- '---'-_'---'-_'---J

Figure 7. Comparison of the Pile-Deflection Curves with Different EI-Values


848 Reese and Wang
"Oftent Cinch-Pounds) Cl000000'.)

-1.0 0.0 1.0 :iI.O 3.0


0.0

The curves from left to right were


0.6
computed using the following
1.:iI
values of EI:
100% Gross Section
-; 1.B 75% Gross Section
8... 50% Gross Section
~
:iI.4 25% Gross Section
-; 10% Gross Section
1 3.0
~
3.6
~
~
Q
4.:iI
~
4.B

5.4

6.0

Figure 8. Comparison of the Bending Moment Curves along a Pile


under Different EI-Values

o Constant Gross Section EJ


Nonlinear EI

160
C/) 140
a.
.:.:: 120
1:1 100
ro
.3 80
ro 60

-"-
CD
ro
-'
40
20
OL-..........-.I_...1.---'-_L...---'---'-_.l.-....J
o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 0 2345678
Deflection at Pile Head, in. Max. Bending Moment
x 10 6 , In.-Ib

Steel Ratio = 1.3% . Pile Diameter 30'

Soil Properties: Stiff Clay. r ~ 120 Ib/ft 3 c - 2 kips/ft2

Figure 9. The Deflection at the Pile Top and the Maximum Bending Moment
Considering Nonlinear EI (p = 1.3%)
849 Reese and Wang
o Constant Gross Section 8
Nonlinear EI

160
U) 140
a.
..lr:: 120
"0
- 100
co
0 80
......J

...co 60

-<ll
co
....J
40
20
o L.-....L.......--L....----L_.L---I..---l...----JL--....I..--J
o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 0 1 2 3 4 5 6 7 8
Deflection at Pile Head, in. Max. Bending Moment
x 10, in.-Ib
Steel Ratio ~ 2.3% Pile Diameter 30'

Soil Properties: Stiff Clay, 120 Ib/ft 3 , C 2 kipslft 2

Figme 10. The Deflection at the Pile Top and the Maximum Bending Moment
Considering Nonlinear EI (p =2.3%)

.FIELD MEASUREMENT
As discussed earlier, the deflection and bending moment from the lateral loading
of a pile are dependent not only on response of the soil but also on the flexural rigidity of
the pile. A number of loading tests for reinforced-concrete piles under lateral load have
been conducted in the past. However, in only limited cases has the flexural rigidity been
measured. Direct measurement provides an easy and accurate method to assess the value
of the flexural rigidity. Figure 11 shows a section through a test pile as arranged to
obtain values ofE!. The test pile is a 16-inch diameter augercast pile which was installed
in a 50-ft deposit of soft to stiff silty clay. After the regular test for lateral loading was
completed, an excavation was made and the soil was supported against collapse as
necessary, as shown in Fig. 11. Concrete was cast around the pile as indicated to
restrain it against excessive deflection when it was reloaded. Prior to the reloading,
strain gauges were affIxed to the front and back of the pile on the diameter in the plane of
loading. Loads of 2.5,5.0, 7.5, 10.0 and 15 tons were applied and held for a period of
5 minutes to obtain data from the strain gauges.

Figure 11 b shows an enlarged section from the pile at the location of a pair of
strain gauges with the strain indicated that was measured at the front and back of the pile.
If the section is symmetrical about its neutral axis, the strains at the front and back will be
equal. The EI of the pile at the points in question can be found from the equation shown
in the fIgure:

EI = Mb/(1 + 2) (4)
where

850 Reese and Wang


I" b
.,
UPPER
DIAL
~
-12'
T T
35"

TOP OF PILE TP 1/1


EI -~ ~
(b)

18"

., , , , 22"1 , ,, ,
.
'. ,,
,
.,
32" 75" I---.....I...---l
{
49.25"
52" 60"
,
# , ,, ,
, , ,,
, ,
, ,,, ,
.
9"
/I 8

12"

24"
~:;:;~
. '1
1 I

~
Not to scale (a)

Figure 11. Sketch of Strain-Gauge Arrangement

851 Reese and Wang


M = bending moment at the point in question (to be detennined from
the lateral load that is applied and the distance the point of load
application to the strain gauges),
b = diameter of pile at point in question, and
Q,E2 = strains at the point in question.

The bending moment and flexural rigidity have the following relationship

EI = M / <I> (5)

where

M = the applied moment at the section of interest, and


<I> = the curvature of the section of interest

The experimental values of EI derived from the above equation are presented in
Fig. 12. The considerable amount of scatter in the experimental results is not surprising
in view of the nonlinearity of the system. The selected EI value is, in general, between
the uncracked-EI value (upper bound) and the cracked-EI value (lower bound). Figure
13a shows the soil conditions at the site of the pile tests. Analyses indicated the
advantage to be gained from the use of about 2 m of select fill around the laterally loaded
piles. Figure 13b shows load-deflection data from a field test and a curve by making use
of nonlinear values of EI.

6.0


5.0
...z

4.0
-
N


II

...

Cl
....
><
...w
3.0


.
'

2.0
1.0

Moment (m - kN)

Figure 12. EI versus Moment


852 Reese and Wang
EL. FT

221.0 ---R'~~~~~~~~~;:r---

215.0

Compacted gravelly clay Sort silty clay


cI.O ksr. 50-0.005 w1th silt and
y-115 pcr sand lenses
c - 0.5 kSf
50 0.02
i-58 pcr

178.0 ----------1
Sort to stHf clay
with silt and
clayey sand
lenses
c 0.8 ksf
50 0.0 I
i 58 pcr
156.0 -4

Normally consolidated
stHf clay with silty
sand lenses
c 1.5 ksf
50 0.005
i 58 pcf
C : Undrained shear strength
50: Axial strain at one-halt the maximum prIncipal stress difference
a

20 --
I:J
Results from computer
Results from field test

10

O ...---.-_r--r----.--r--r----.-_r-_._-r-....--,
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Deflection, In

Figure 13. Soil Conditions at Test Site and Comparison of Results


from Field Tests and from Computation

853 Reese and Wang


CASESTUDY
Construction work for pile tests at the Sunshine Skyway Bridge was done in
1981 by Misener Marine Construction, Inc., on behalf of the Florida Department of
Transportation. Results of lateral-load tests are described by Reese and Long (1982).
Two identical test shafts were installed where the water depth was about 15 ft (4.6 m)
with the top of each shaft extending 7ft (2.1 m) above the water. The purpose of the
testing was to develop p-y curves; therefore, the drilled shafts were designed with an
outside casing, as shown in Fig. 14 a and b. The addition of the steel shell provides
more bending resistance and allowed for more deflection than otherwise. Soil
infonnation is shown in Fig. l4a.

The EI of the uncracked section was computed as 1.5 x 10 12 1b-sq in. (4.3 x 1()6
kN-sq m), while the EI of cracked section was about 0.9 x 10 12 lb sq in. (2.6 x 1()6 kN-
sq m). The difference in EI for cracked and uncracked section was reduced because of
the effect of the steel casing. The back-calculated value of EI from strain gauge readings
and from inclinometer data at the higher loading was near 1.0 x 10 12 1b-sq in. (2.9 x 1()6
kN-sq m), which is consistent with the cracked section value.

The two shafts were tested by applying a load to a high-strength steel bar that
passed horizontally through the axis of each shaft. The load was applied at 5 ft (1.5 m)
above the water surface and the load was measured by an electronic load cell. Because
the steel casing gave a drilled shafts a large bending capacity, failure was controlled by
excessive deflection rather than by the development of a plastic hinge. Figure 14c gives
the comparison between the measured and predicted results.

As may be seen in the figure, through most of the range of service loading, good
agreement was found between the measured and predicted results which used the value
for uncracked section. The predicted value becomes unconservative when the uncracked
EI value was used throughout the whole pile length. Above a deflection of the pile head
of about one inch, the computations with the EI of the cracked section agreed fairly well
with the results from the e.xperiment.

CONCLUSIONS
Based on the study presented in this paper, the conclusions concerning the effects
of nonlinear flexural rigidity on the behavior of laterally loaded piles can be summarized
as follows: 1. Selection of a constant flexural rigidity in the analysis of a pile has
negligible influence on the computed maximum bending moment. 2. The pile deflection
can be significantly underestimated if a reduction of flexural rigidity due to tensile
cracking is not considered in the analysis. 3. To evaluate the flexural rigidity in
response to cracks is important not only for controlling the pile deflection, but also for
obtaining the correct fonn of soil resistance from a full-scale loading test. 4. In the
analysis of a pile under lateral loading, only the upper portion of the pile will be
subjected to a large moment. In the analysis of a reinforced-concrete pile, the value of EI
for the cracked section needs to be used only for the upper portion of the pile. The value
of EI for the gross-section can be used with no appreciable error for the remainder of the
length of the pile.

Reese and Wang


854
,--- Casing
71.:- .-/ Q
I.
48"
01
1/2" steel
casing
15'1 " , __
,...-
- _=.:..:..:..:.=---
Mudline 6"

Sand I Shells I
Y =110 Ib/cu ft
<p =37 b
51'
Si Ily-Clayey Sand
Y =112 Ib/cu It
c = 0.6 kips/sq It. 41 =34

Cloy, )'=110 Ib/cu It, c-3.0kips/sq ft. <p-28


Sandy Cloy, Y=\lO Ib/cu fl. c=1.0 kips/sq ft
a ep =33

100

....,;.....
80

60
.!!
!II 40
...
-Go

a
Uncracked EI
Cracked EI
Measured
~
20

c
1 2 3
PUe heed deUectJon, In.

Figure 14. Soil Conditions, Test Shaft, and Results of Testing and
Computations at Site of Case Study

855 Reese and Wang


APPENDIX: CONVERSION FACTORS FROM U.S. CUSTOMARY TO
SI UNITS

To Convert to Multiply by
ft m 0.305
psf Pa 47.880
pef N/m3 157.100
ft-kip kN-m 1.356

REFERENCES
Eppes, B. G., "Comparison of Measured and Calculated Stiffnesses for Beams
Reinforced in Tension Only," ACI Journal, V. 31, No.1, Oct., 1959.

Mathey and Watstein, "Effect of Tensile Properties of Reinforcement on the Flexural


Characteristics of Beams," ACI Journal, V. 31, No. 12, June, 1960.

Reese, L. C. and Long, J. H. (1982), "Analysis of Two Drilled Shafts Subjected to


Lateral Load, Sunshine Skyway Bridge," Geotechnical Engineering Report
GR81-21, The University of Texas at Austin.

Reese, L. C., "Behavior of Piles and Pile Groups Under Lateral Load," Report No.
FHWAIRD-85/106, U.S. Department of Transportation, Federal Highway
Administration, Office of Research, Development and Technology, Washington,
D. C., July, 1983.

856 Reese and Wang


Determination of P-Y Curves in Fractured Rock Using Inclinometer Data

Dan A. Brown 1 and Shu Zhang 2

Abstract

Derivation of p-y curves from lateral load tests on deep foundations is a


tedious and expensive task, requiring large numbers of strain gauges along the
length of the pile to develop bending moment vs depth relationships. A method
is proposed which allows derivation of p-ycurves from simple inclinometer data
using a least squares regression technique. Inclinometer measurements are
simple and inexpensive to obtain during routine lateral load tests of drilled
shafts. The method outlined in this paper provides a means of "calibrating" p-y
curves using data from tests where only inclinometer data are available. The
technique is demonstrated using the results from two field tests of drilled shafts
in partially weathered rock, a material for which conventional empirical p-y
curves are not available for routine design of highway bridges.

Introduction

In the U.S. highway industry, the most common method of analysis used
in the design of deep foundations for lateral loading is the p-y method,
represented by the finite difference model implemented in the computer code
COM624 (Wang and Reese, 1991). This analytical model utilizes nonlinear load
transfer functions of force, p, vs lateral displacement, y, to represent the
resistance of the soil to lateral displacement by the pile.
The analytical p-y curves used for design are based upon those derived
from double differentiation and double integration of the bending moment vs
depth relationships from relatively few, well-instrumented lateral load tests. Data

lAssociate Professor, Department of Civil Engineering, Auburn University,


Alabama 36849

2Graduate Student, Dept. of Mathematics, Auburn University, Alabama

Brown and Zhang


857
derived in this manner are expensive and difficult to obtain and therefore such
data are not available for a wide variety of soil conditions. Because of the
degree of empiricism included in the p-y method of analysis, there is a need for
a simple and cost effective approach to interpreting lateral load tests which does
not require extensive instrumentation.
This paper outlines a method for interpretation of lateral load tests which
allows derivation of p-y curves from simple inclinometer data using a least
squares regression technique. Inclinometer measurements can and have been
routinely made on lateral load tests of piles and drilled shafts. The method
outlined in this paper provides a means of "calibrating" p-y curves which follow
a prescribed analytical function using data from tests where only inclinometer
data are available. The technique is demonstrated on a hypothetical example
for which the "correct" p-ycurves are known and on two examples of actual field
tests of drilled shafts in weathered rock material for which design guidelines for
p-y curves are not available.

P-Y Analysis of Laterally Loaded Piles

The p-y method for analysis of piles subjected to lateral load consists of
a beam-on-elastic-foundation using nonlinear springs to model the soil
resistance to displacement. This approach has been widely accepted because
of its simplicity and ability to capture the essential aspects of pile behavior,
including nonlinear soil resistance, gapping, variable soil and pile properties, and
others.
Where bending moment measurements are made at a number of points
along the pile during load testing, then bending moments and pilehead
deflections can be used to derive p-ycurves by twice differentiating the moment
as a function of x to obtain p, and twice integrating to obtain y. This procedure
has been used successfully to derive p-y curves from load tests on a point-by-
point basis. The greatest difficulties with the use of this procedure are: a) the
large numbers of strain gauges required to define the distribution of bending
moment along the pile, and b) the sensitivity of the double differentiation
procedure to errors in the bending moments derived from the bending strain
measurements (or in the bending stiffness of the pile which would result in
errors in transforming strains to bending moments).
A limited data base of load tests from which p-y curves have been
derived provides the basis for general analytical methods for representation of
soil resistance. These analytical expressions include a considerable degree of
empiricism and have limited reliability in predicting pile deflections and bending
stresses when extended to load test data beyond that used to develop the p-y
criteria. Improvements in reliability of analytical p-ycurves are hindered by the
unavailability of appropriately documented full-scale test data (Gazioglu and
O'Neill, 1984; Murchison and O'Neill, 1984).

Brown and Zhang


858
Although extensive instrumentation is required to derive p-y curves from
lateral load tests in a conventional manner, it is quite a simple matter to obtain
measurements of slope and deflected shape along the pile using a slope
inclinometer and guide casing. The casing can be attached or embedded prior
to pile installation or embedded in a drilled shaft by attaching to the rebar cage
prior to concrete placement. Since the only expendable item is the casing,
these measurements can be obtained at a very nominal expense above that of
the test piles. The integration of these slope measurements provides a fairly
reliable indication of deflections, y, along the pile profile. However, direct
derivation of soil resistance, p, from these slope measurements in the
conventional manner would involve differentiation three times and the
amplification of measurement errors involved in this process result in unreliable
calculations. In order to obtain reliable indications of p-y relationships from
inclinometer measurements-, it is therefore necessary to use a different
approach.

Back Calculation of P-Y Curves from Slope Inclinometer Data

The method outlined below is a means of deriving p-y curves from a


measured deflected shape by making a "best-fit" to the data using a preselected
analytical function for the p-y relationship and using a least squares technique.
The variation in the shape of the p-y curves and the variations with depth are
defined using several variables which are the subject of the fitting process. For
the simple use of existing analytical p-y curves fitted to a specific site, these
variables can simply be taken as the input soil strength and stiffness
parameters. For example, existing p-y criteria for clays utilize cohesion or
undrained shear strength, Su' and the strain at 50% of the failure stress in
uniaxial compression, Eso ' For sands the angle of internal friction, <, and the
initial subgrade modulus, k (in units of F/L3 ) is common. When used to develop
or evaluate a more general form of an analytical p-y curve, other empirical
parameters could be used as the unknown fitted variables. The former might
be most useful for evaluating test data at a specific site so that alternative
boundary conditions, pile or shaft lengths, etc. could be evaluated. The latter
might be more useful in a general sense to develop more reliable analytical p-y
curves for a particular type of geomaterial. In either case, the technique would
be the same but the actual variables to be determined using the technique
would be different. In the method and examples outlined below, soil strength
and stiffness parameters are used as unknowns.

Definition of Problem
Consider a pile which is divided into i-1 intervals, so that there are i
nodes along the length. Additionally, deflections are determined at each of j
sets of boundary conditions (for a load test, j values of pilehead shear). Thus,

Brown and Zhang


859
there are a total of i x j = m nodes for which deflection measurements are
obtained. The horizontal deflections, y, are:

y(u)= (1 )

where:

U= (2)

and: u = soil parameter affecting analytical p-y curve,


= number of soil parameters used to define analytical p-y curves and
n
distribution with depth.
The deflections, y, are computed from the soil parameters, U, using the finite
difference code COM624. This code solves a set of m non-linear equations (i
equations at each of j loads) to get y. The code also uses a predetermined
analytical p-y criterion which depends upon U as well as other known aspects
of the problem such as soil unit weight, pile diameter, etc.
From the load test measurements, a set of known deflections are
determined as b:

b(u)= (3)

It is desired to determine the set of unknowns U by using the measured


deflections b. Because m n, there is not an exact solution for u. The best

Brown and Zhang


860
estimate of u is determined using a least squares "inversion" technique similar
to that used for interpreting geophysical data. Thus, u is found to minimize:

Y(U) = [y(u) -b]T[y(u) -b] (4)

Note that for a long pile, many of the nodes along the pile would have
immeasurably small deflections, and so the set of values in equation 4 need
only include those nodes which are considered relevant to the problem.

Method of Solution
The best solution y(u'j is unknown, but b is known, and b is the goal of
y(u) so that y(u)-bf is minimized. By Taylor series:

t:.U TY(u)t:.U (5)


y(u+ll. u) -Y(u) = Jt:.u + + ,..
2

where: J = the Jacobian matrix which describes the effect of changes in


u on y.
If only the first term is used, then:

y(U+t:.u)-y(u) = Jll.U (6)

Let
(7)

(8)

then
(9)

Equation (9) is considered as a linear least squares problem, for which


the computed changes in the unknowns, ~Uk' can be solved. The solution
involves an iterative process in which the differential at point Uk is used to
estimate the next point, U k +1 . Equation (9) is solved using singular value
decomposition (SVD). SVD is a widely used technique for such problems
because it has the important feature of stability for the inversion of ill-conditioned
matrices.
Because the problem is non-linear, the estimation of U k +1 is not exactly
correct. In order to provide a stable solution which will converge, the estimation
of U k + 1 is made using equation (10):

Brown and Zhang


861
(10)

where: c =a damping coefficient which. reduces the changes in the


unknowns from one iteration to the next to enhance the stability
of the procedure.
As an initial estimate, c is taken as 0.2. Additionally,
(11 )

so that the unknowns are never changed by more than 20% for a single
iteration. If the solution does not "converge", that is, if
(12)

then reduce c to 0.1, 0.05, 0.025, etc. until Uk+1 does converge. The "best fit"
solution is determined when c is reduced to some suitably small value (say,
.00001) without improving the fit as per equation (12).
The Jacobian matrix, J, is obtained using the COM624 code and forward
differences:

(13)

where oU k is chosen as a small proportion of Uk which takes the same direction


as ~Uk.1' Then the Jacobian matrix is:

5Y1
-
5Y1
-5Y1-
5U1 5u2 5un

5Y2 5Y2 5Y2


-
J= 5u1 5u2 5un (14)

5Ym 5Ym 5Ym


-- --
5u1 5u2 5un

The particular algorithm used to describe the form of the p-y curve will
define the relationship between the unknowns, u, and the computed deflections,
y. The resulting fitted solution will only be as good as the algorithm used for the
p-y curves is at describing the soil response at a particular test location.
Obviously, the solution is also sensitive to errors in the inclinometer

Brown and Zhang


862
measurements, although the least squares fitting process will tend to minimize
the effects of small errors. Note that it might be possible to improve the solution
by first "fitting" the measured data in some manner so as to smooth the data
and minimize the effect of errors. However, any such smoothing would also
tend to reduce the sensitivity of the solution to actual measured changes. The
solution technique described above can be used with any general form of the
analytical p-y curve which is implemented through COM624 or a similar code.
Note also that the solution will be affected by the EI of the pile or shaft, as
would be any technique used to derive p-y curves from field measurements,
including strain measurements in the pile. In principle, the technique described
above could also be used to optimize the bending stiffness of, for example, a
reinforced concrete drilled shaft, provided that other parameters were known
with some confidence.

Examples

The use of the technique described above is illustrated using several


examples, including a hypothetical example for which an exact solution is
known, and two actual examples from field load tests of drilled shafts in
fractured rock. In these cases the analytical function used for the p-y curves is
the polynomial function commonly used for stiff clays and implemented in
COM624 (Reese and Welch, 1975).

Hypothetical Example
A hypothetical example is presented to demonstrate that the technique
can converge to a solution for a problem for which the exact solution is known.
This example consists of a 36 inch (0.91 m) diameter by 40 ft. (12.2m) long
drilled shaft embedded into a formation consisting of either clay in which the
strength parameters of that formation vary with depth. This example utilizes the
analytical p-y criteria for clay which is included within the code COM624; the
regression algorithm works toward a best fit solution to the "measured"
deflection data by adjusting the strength and stiffness parameters specified as
an initial guess for the soil formation. The adjusted parameters for clay are
cohesion, SUI at the top and bottom of the formation and Eso ' the strain at 50%
of failure in a compression test. The parameters which are used to adjust the
analytical p-ycurves could just as easily take some form other than modification
of soil strength and stiffness properties; however, the approach used is quite
simple to implement and allows a user to easily produce site specific p-ycurves
for a project without changing the existing COM624 code. It must be noted that
the soil strength and stiffness parameters which are derived from this analysis
are simply parameters used to formulate site specific p-ycurves and should not
be taken as an indication of soil strength properties for any other purpose.

Brown and Zhang


863
For the hypothetical example in clay, the "measured" deflections were
computed using COM624 and the p-y criterion for static loading in stiff clay of
Reese and Welch (1975) with values of Su=3psi (20.7kN/m 2) at the top of the
shaft, Su=6psi (41.4kN/m 2 ) at the bottom oUhe shaft, and 50=.015. The back-
analysis technique is implemented in a code named "INCPILAU" which seeks
values of Su at the top and bottom of the shaft and strain parameter, 50' which
provide the best fit to the measured deflection data.
Note that one could easily determine a best fit to measured data at a
sand site using, for example, a p-y criterion for clay soil although the fit would
not be as good since the analytical expression for clay will not fit as well at a
site comprised of sand.
Illustrated on Figure 1 is the progression of iterations for the hypothetical
example from INCPILAU when the code is given starting values which are well
removed from the true solution. The "measured" deflection values were
rounded to the nearest 0.01 inches (0.25mm). Numerous sets of starting values
were tried, and the code was found to readily converge to very near the exact
solution with starting values which were generally within a factor of 2 to 3 of the
correct solution. It is possible to induce the code to converge to a different
solution (another local minimum) using, for example, lower Su values and lower

... 1.0E+1
g 1.0E+0

W
rn::J 1 OE-1

:2 1 OE-2
rn
Q) 10E-3
0:::
10E-4
0 10 20 30 40 50 60

70
0 Su at Top of Shaft

60
0 Su at Base of Shaft

f', eSO (x 10)


0
..- Correct Values
50
~
0
L()
40
...0
Q)
0.

CO
0...
30
~OO
6/;,0000
~
20
::J
66
U)
10

0
0 10 20 30 40 50 GO

Iteration Number

Figure 1 Solution of Hypothetical Example in Clay

Brown and Zhang


864
Deflection mm I

o 10 20 30 40 50 60 70 80

-1

.-
E
-2
<ll
"0
ro
.....
(9
3: -3
0
<ll
I1l
..c
.- Computed, B9kN load
a.. -4
<ll Computed, 17BkN load
0
Computed, 267kN load

Computed, 356kN load


-5
Correct Values

Figure 2 Deflection Response of Hypothetical Example in Clay

Computed, a.8m depth

o Computed, 1.2m depth

6. Computed. 1 8m depth

150
o Computed, 2 4m depth

E Correct Values
""-
Z
- ~

a.. 100
<ll
U
c
ro
......
l/)
'ii)
<ll
0::::
'0
if) o_.I---,----,------,r--,----r----,
o 50 100 150 200 250 300

Displacement, Y (mm)

Figure 3 P- Y Curves for Hypothetical Example in Clay

Brown and Zhang


865
however, the Rnorm values for alternative solutions were observed to be
Eso ;
much larger (generally by several orders of magnitude) than for the "best"
solution. Analyses using several different starting values are thus suggested for
real examples, with the best solution chosen as the one with tl:le lowest Rnorm
value (the measure of the error between the computed and measured deflection
values, = #y(u,J-bf).
Presented on Figures 2 and 3 are plots of the fitted and exact deflection
vs depth profiles and p-y curves for the hypothetical example. As would be
expected from the convergence data shown on Figure 1, the fitted solutions
agree almost exactly; any small differences are due to the use of "measured"
deflection data which was rounded off to the nearest .01 inches (0.25mm).

Actual Load Test Examples in Weathered Rock


A number of actual load tests have been analyzed using this procedure
with excellent results, particularly for piles which are displaced to large lateral
deflections (several inches) so as to define the soil resistance over a broad
range. In rocklike materials, smaller lateral deflections are sometimes all that
can be mobilized during testing. Two load tests cases are illustrated below to
demonstrate the use of the method as well as some of the limitations. These
tests provide case studies for the proposed method which demonstrate the utility
of the procedure when dealing with geomaterials for which there is not a reliable
and established experimental basis for a specific p-y criterion.
Because there is no widely accepted p-y criterion for dense fractured
rock, analyses were performed using the analytical functions available for clay
soils. Both the soft clay criterion of Matlock (1970) and the stiff clay criterion of
Reese et al (1975) were considered, and several analyses were performed with
different starting points. Each of several different p-y criteria may be fit to the
measured data and perhaps obtain a reasonably close fit. For these test
results, the "best" solution (the one with the lowest Rnorm value) was chosen
using the stiff clay criterion.
Note that the bending stiffness, EI (modulus of elasticity, E, times
moment of inertia of the cross section, I), of a reinforced concrete drilled shaft
is less reliably known because of the nonlinearity in the bending stiffness of
such members. Thus, experimental data of this type include an additional
uncertainty with respect to back-analysis to obtain p-y data. Nevertheless,
these cases represent what is likely to be a more typical use of the method.

North Alabama, Test in Fractured Shale


Data for this case was obtained by the first author as part of a load
testing program for a bridge replacement project at 1-65 over the Mulberry Fork
of the Warrier River. The test shaft is a 30 inch (0.76m) diameter reinforced
concrete shaft which is socketed 15 feet (4.6m) into a fractured shale. The
shale near the ground surface is highly fractured, and standard penetration tests

Brown and Zhang


866
at shallow depths (within the upper 1.2m) were 35blowslft. (b/300mm).
Compression tests on 2 inch diameter (50mm dia.) cores of intact rock layers
below a depth of about 1.5m were in the range of 5 to 35 mPa, with one test
on a sandstone layer at 100mPa. However, the rock mass was quite fractured,
with ROD values typically around 50% or less.
The agreement between the experimental measured deflections and the
computed deflections is illustrated on Figure 4 with deflection vs depth plots for
each of the 6 load levels used in the back calculation process. The computed
data are generated using cohesion values of 676 and 8832 kPa at the top and
bottom of the shaft, respectively, with an 50 value of .00651. The iterative back
analysis routine located other "local minima" using larger cohesion at the top of
the shaft and a corresponding larger 50 value.
The p-y curves derived using this process are shown on Figure 5. The
lines shown on these curves represent computed p-y response using the stiff
clay criterion with the cohesion and 50 values derived from the back calculation
procedure. The solid data points represent the p-y values at the computed
deflections along the shaft during the back-calculation analyses. Because the
fitting process was done only up to these deflection values (which vary with
depth), the back calculated p-y curves are valid representations of the
experiment only up to that deflection and should be used accordingly; computed
response for design at this site for larger deflections would represent an
extrapolation beyond the experimental data. Note also that the values of
cohesion (Su) derived from this analysis do not imply that such strengths
actually exist, particularly at depth; the Su values are only used to produce the
p-y curves. In addition, the data which most influence the fitting process are
those nearest the ground surface. The computed response is therefore not very
sensitive to the Su value at the shaft base in this case, and the Su value at the
shaft base really only is used to dictate the rate of change of Su with depth.
The INCPILAU code could be set up differently, so that a rate of change of Su
with depth could be a variable; however, the use of Su at the pile toe is more
easily implemented into a subsequent analysis for design using COM624.

CalTrans Drilled Shaft Load Test, Test Pile A


Data for this case were provided by Speers of the California Dept. of
Transportation (CaITrans) (1992) and represent a load test on a large diameter
drilled shaft at a site near San Francisco in a dense fractured sandstone.
The shaft was 72 inch (1.8m) diameter and had a length of approximately
21 feet below grade. Inclinometer measurements were obtained using a casing
which was tied into the reinforcing cage at the time of construction. The shaft
was loaded laterally in an incremental, monotonic fashion using a hydraulic jack
and load cell. For purposes of analysis, the bending stiffness, EI, was
estimated using the code "STIFF1" (Wang and Reese, 1987) which computes
the EI of a cracked reinforced concrete section as a function of bending

Brown and Zhang


867
Lateral Deflection, mm
o 2 4 6 8 10

-04
E
OJ
U

~
:::J
(f) -0.8
"0
1500 c
:::J
e
(J
/.
/ Experimental
Z :!: Computed
.>t: o -12
-0- 1000 Ui
III CD
0 .r:
-'
m a.
OJ
CD 500 o
~ -16
-'

o 5 10 -20
Displacement at Top of Shaft, mm

Figure 4 Deflection Response of North Alabama Test in Shale

3.00
E
-
Z
E
..lI::
2.00
a..
Q)
U

-c:
ro
( /)
'w
Q)
1.00
. -e-
'--A--
0.3m depth

0.6m depth
0:::
-. . 0.9m depth
'0
C/)
a 00 --+--.,...---...----....,....--......-----,-----,
000 400 800 1200
Deflection, Y, mm

Figure 5 P- Y Curves for North Alabama Test in Shale

Brown and Zhang


868
moment. Bending moments were estimated using preliminary analyses with
COM624.
Presented on Figure 6 are the load vs deflection and deflection vs depth
profiles for the experimental measured data and the computed solution for each
of the 4 load levels used in the analysis. It is noted that a less than perfect
agreement exists, but the analytical solution captures the general trends in the
data. Shown on Figure 7 are p-y curves for the 24 (O.6m) and 48 inch (1.2m)
depths using each of the three solutions mentioned previously. Also shown on
Figure 7 are the deflection limits in the measured data; the extensions of the
analytical p-y curves beyond those deflections are an extrapolation of the fitted
analytical p-y criteria beyond the experimental data and are not justified by this
test. Because of the relatively limited deflections mobilized during this test, a
soil resistance limit was not observed, although the measurements suggest
considerable nonlinearity in the soil resistance. Thus, the solution is not
particularly sensitive to the shear strength, Su' which would most strongly
influence the ultimate soil resistance. The initial stiffness of the p-y curves is
most affected by the 50 value, with Su influencing the point at which the curves
begin to exhibit nonlinear response.

Summary and Conclusions

An analytical method is described which provides a rigorous and reliable


method of interpreting lateral load tests on piles or drilled shafts using
inclinometer data. The method utilizes a least squares regression technique
which will readily converge to a solution with minimum error (as indicated by
Rnorm) between the predicted and measured deflection vs depth profile over a
range of loadings. Although any analytical p-yexpression could be fitted to the
data, the method is demonstrated on both hypothetical and actual load test data
using p-y criterion which are available and accepted. Of course, the ability of
the procedure to provide an accurate representation of the soil resistance is
limited by the reliability of the inclinometer measurements, the correct
representation of the bending stiffness, EI, of the pile or shaft, and the suitability
of the analytical p-y criterion to match the general shape of the soil resistance
at a particular site. The proposed method provides a basis for efficiently
evaluating a representative number of lateral load tests for which only limited
instrumentation are available in a given type of geomaterial and thus provides
a tool to improve the data base of load tests from which the empirical p-ycurves
used in design are based. The method also provides a useful tool for
characterizing the soil resistance for design for lateral loading on a specific site
for which load test data are available. The characterization of soil resistance
obtained in this manner would allow for evaluation of other pile head restraint
conditions and possibly other embedded lengths and diameters (with proper
judgement applied).

Brown and Zhang


869
Deflection, mm
o 2 3 4 5 6 7

6 -0.5

Z X
E 5
E
"0
-10
Q)
co 4 "0
0
-.J ~
(j
"0 3
co :;: -1 5
Q)
.c X Measured Data
0
~ 2 ':. Soft Clay. Su=l 6-2 5mPa. e50= 00013 4i
CO
0::
-B- Soft Clay. 5u=1 7-3 3mPa. e50=.0002
:5
511ft Clay. Su=l 4-2 6mPa. e50= 0001 0.. -2.0
~ aJ
0 Measured Data, 2_67mN load
0
0 2 3 4 5 6 7 Measured Data, 4.00mN load
-25
Measured Data, 5.34mN load
Displacement at Groundline (mm)
Measured Data, 6.45mN load
-30 Computed Values

Figure 6 Deflection Response, CalTrans Test in Sandstone

14 Experimental Deflection Limit


E
--
Z
.::
'-'
12 1.2m depth

10 0.6m depth
a..
Q)
(J
8
c
-en
( /J
(/J
Q)
6

4 ---e--
Soft Clay, Su=1.6-25mPa, e50=.00013

Soft Clay, Su=1.7-3 3mPa, e50=.0002


0:::
2 -..- Stiff Clay, Su=1 4-2.6mPa, e50=0001
0
if) 0

0 2 4 6 8 10 12 14

Deflection, y (mm)

Figure 7 P- Y Curves for CalTrans Test in Sandstone

Brown and Zhang


870
It may be noted that other analytical expressions might be used for
fractured rock and one might be found which would provide a better Rnorm.
Such a case would represent a more faithful reproduction of the measured
deflection data; however, given the uncertainties with respect to bending
stiffness in the shaft and accuracy of inclinometer data, derivation of a "new" p-y
criterion for fractured rock is not warranted on the basis of these two tests.

Acknowledgments

The authors wish to acknowledge the financial support of the Auburn


University Highway Research Center, the Alabama Highway Department, and
the Georgia Department of Transportation. The helpful comments and
suggestions provided by Dr. Glenn Rix of the Georgia Inst. of Technology and
the test data provided by Mr. Dan Speers of CalTrans are gratefully
acknowledged.

ApPENDIX I. - REFERENCES

Gazioglu, S.M. and O'Neill, M.W. (1984). "Evaluation of P- Y Relationships in


Cohesive Soils." Proc., ASCE Symposium on Analysis and Design of Pile
Foundations, San Francisco, pp.192-213.
Matlock, H. (1970). "Correlations for Design of Laterally Loaded Piles in Soft
Clay." Paper No. OTC 1204, Proc., Second Annual Offshore Technology
Conf., Houston, Texas, pp. 577-594.
Murchison, J.M. and O'Neill, M.W. (1984). "Evaluation of P-YRelationships in
Cohesionless Soils." Proc., ASCE Symposium on Analysis and Design of
Pile Foundations, San Francisco, pp.174-191.
Reese, L.C. and Welch, R.C. (1975). "Lateral Loading of Deep Foundations in
Stiff Clay." Journal of the Geotechnical Engineering Division, ASCE, Vol.
101, No. GT7, pp 633-649.
Speers, D. (1992). personal communication, data from internal CalTrans
testing program.
Wang, S-1., and Reese, L.C. (1987). "Documentation of Computer Program
STIFF1," Ensoft, Inc. Report, 69 p.
Wang, S-1., and Reese, L.C. (1991). "Analysis of Piles Under Lateral Load -
Computer Program COM624P for the Microcomputer." U.S. Dept. of
Transportation, Federal Highway Administration Report No. FHWA-SA-91-
002,229 p.

Brown and Zhang


871
Appendix II. - Notation

The following symbols are used in this paper:

b = pile width
b = measured deflection on pile
c = a damping coefficient which reduces the changes in the unknowns
between iterations
EI = flexural rigidity of the pile
j = subscript for number of measured values of deflection at a given lateral
load
j = subscript for number of loads for which deflection measurements are
obtained
J = Jacobian matrix which describes the effect of changes in u on y
k = subscript for iteration number in solution sequence
k = subgrade modulus for sands; kb = initial slope of p-y curve in sand
m = subscript for total number of measured values of deflection on pile =
(i)U)
n = number of soil parameters used to define analytical p-y curves and
distribution with depth
p = unit soil resistance, which is a nonlinear function of y
Rnorm = summation of the square of the error between the computed and
measured deflection values
Su = undrained shear strength of cohesive soil, cohesion for clay soil p-y
criteria
u = generic term used for any soil parameter which affects p-y curve
x = depth
y = lateral deflection
Eso = strain value at a stress = 50% of failure stress in triaxial compression
test on a clay soil
<l> = mohr-coulomb angle of internal friction for sand

Brown and Zhang


872
Modelling of the shaft capacity of grouted driven piles in calcareous soils

Hackmet A. Joer l and Mark F. Randolph l

Abstract:
The frictional capacity of piles driven into compressible sediments is often very
low, particularly in calcareous formations such as cover the seabed in many regions of
the world. These formations can be found in different states (ranging from mud, sand
and silts to very hard rock). In order to increase the friction between the pile and the
soil, grouting techniques may be used. The paper presents results from laboratory tests
performed on model piles driven into calcareous formations. The samples were
prepared with different degrees of cementation in order to simulate formations of
differing strength.
1. Introduction
In calcareous formations, and other compressible soil types, the shaft capacity
of driven piles may be very low. This low capacity is mainly due to the crushing and
rearrangement of the soil and the formation of a zone of low effective stress
immediately around the pile during the driving process. The conditions around the pile
are therefore ideal for improving the frictional capacity of the pile by grouting, since
the grout will tend to flow along the pile surface - through the zone of remoulded
material under low effective stress. This type of foundation has become known as a
'grouted driven pile' and different techniques have been proposed for constructing such
piles. Barthelemy et al (1987) and Rickman and Barthelemy (1988) describe a
successful technique for grouted driven piles, where grout is pumped through valves
pre-installed in the wall of the pile, and discuss the cost advantages of this type of pile.
The paper describes model-scale apparatus that has been developed to study
the performance of grouted driven piles in calcareous soil. Results are presented from
tests performed on a model pile of 50.8 mm diameter driven into a sample of 400 mm
diameter and 300 rrun depth (Joer et al, 1994). The samples were prepared with
different degrees of cementation in order to reproduce the naturally occurring
calcarenite which can be found in different states, ranging from mud to very hard rock.
The grouting technique used is detailed. The measured shaft capacity can be related to

IThe University of Western Australia

873 Joer and Randolph


the strength of the material or its degree of cementation. A design approach for
grouted driven piles based on the experimental results is proposed.
2. Sample preparation
The soil used in this study is calcareous soil from the North-West Shelf of
Australia. Two batches of soil were tested. The first batch of soil was dredged from
the sea bed, and stored (immersed in sea-water) in 200 litre drums lined with a thick
plastic bag. Once in the laboratory and prior to testing, the soil was removed from the
drums, placed on the floor and left to air dry for about two weeks. Then, it was sieved
and all dirt and particles greater than 2.4 mm diameter removed. The soil was then
placed in an oven at 50 for 24 hours.
The second batch of soil was air lifted from the sea bed and dried in an
industrial kiln. It was then sieved over a 2.4 mm diameter sieve to remove any large
shell fragments and dirt. Figure 1 shows the grading curve of the soils, referred to as
Cl (first batch) and C2 (second batch). The two curves show a similar trend, although
three different parts can be distinguished. At the extremes, between sizes of 2.4 rrnn
and 0.12 mm, and for particles smaller than 0.013 mm, the second batch, C2, shows
higher percentages than Cl, while for intermediate sizes, between 0.12 mm and 0.013
mm, the first batch has a higher percentage. The d50 for the two soils is about 0.1 mm.

100 ITTllTmr-TIIiTlTn---rIlrr;:I;':IJ:g;l:J='MrTlrnTn
I.l;
~ 90t--t-t-+++ttH--+-+++++t++---+--:>II"7f<1~+----j--l4+++++J
-- 80 +--t-t-+++ttH--+-+-H++f+I---+,L/;.~:....j-If-+++H---+--+-+-+++-4+l
.5 70 Illlmtttt--r-H-tttttt-i=,...../~/V~-tttf:l::::=:::=!=~tffi
~ 60 +--+-+-+++t+H--+-+-+-+++-f+l-VF-+-+-I-++-I -II- C1
0': 50 +----+---+-+++H+I--+-++-+++I-W--+-+-I~
~ 40 t---I--I-+++++t1f---+--++I.F-tM--+-+-I-+-JH - - 0 - C2
C 30 +---i--i-+-++ttHr--+-?.!j!-~/
-.t+t!+--+-+-f-+-1f-fflFF==;f='Ff'=r=f-+++l
u
20
c.. 10 !;ttJ~~~~~~~l'l-'!Jtttttjjttttttt=tt!!ttljj
~ _
p- .
.t-!li"~:!!!--+-++++H-++---+-+-+-H++++---+-+-+-H-jf.++.j
O+-----L---J.......L-.L...I....U.-4----I----.l....J......J..J...LJL..L+-_L........l-L-LLLL4-_L-L..L.Ll.JLU..j
0.001 0.01 0.1 10
Grain size (nun)
Figure 1 Particles size distributions for the two batches of soils.

Uncemented and cemented samples with 2 %, 4 % and 6 % of cement content


were prepared. For the uncemented samples, the dry soil and water were mixed using a
concrete mixer for 30 minutes. For the cemented samples, the dry soil and cement
were mixed for 30 minutes, then water was added and the total was mixed for a further
30 minutes. Table 1 shows the details of the composition of the samples. Ordinary
(type A) and high early-strength (type B) Portland cements were used to prepare the
cemented samples.
After mixing, the mixture was then placed in a cylindrical container, 400 rrnn
diameter by 400 mm high. The sample was then consolidated at a vertical stress of 100
kPa for 1 hour. In the case of uncemented samples, the test could then begin, whilst

874 Joer and Randolph


the cemented samples were placed in a room with a constant hwnidity and allowed to
cure for a certain period (see Table 1).

State Samples Soil Cement Water content Curing period


No. Names type % (days)
Uncemented 1 HSLU_l Cl - 40 0
2 HSLU_2 Cl - 40 0
3 HSLU_3 Cl - 40 0
4 HSLU_4 Cl - 41 0
5 HSLU_5 Cl - 41 0
6 HPOU 2 C2 - 42 0
2 % cement 7 HSL2_1 Cl A 40 56
content 8 HSL2_2 Cl A 40 48
9 BP02_1 C2 B 44 9
10 BP02_2 C2 B 42 6
11 HP02_1 C2 B 42 6
12 HP022 C2 B 42 6
4 % cement 13 HSL4_1 C2 B 42 9
content 14 BP04_2 C2 B 44 17
15 BP04_3 C2 B 42 10
16 HP04_1 C2 B 42 3
17 HP042 C2 B 42 4
6 % cement 18 HSL6_2 Cl A 40 N/A
content 19 HSL6_3 Cl A 40 N/A
20 BSL6_3 Cl B 40 11
21 BP06_1 Cl B 40 12
22 HP06_1 C2 B 42 7
23 HP06_2 C2 B 42 6
24 HP06_3 C2 B 42 14
25 HP064 C2 B 42 21
Table 1 Sample details

3. Grout types
Modelling grouting operations at small scale is complicated by the tendency for
the cement particles to clog fine orifices and tubes. In the early stages of the study, an
epoxy grout of low viscosity (v = 79 mP) was used, in order to overcome this
problem. However, while this grout gave full coverage of the pile, it was found that it
penneated the surrounding soil to a much greater extend than would a prototype
grout. The use of epoxy grout was therefore abandoned in favour of cementitious
grout.
Initially, grouting mixtures were made up from Portland cements type B and G,
along with a plasticiser (Joer et al., 1994; Joer, 1994). However, more recent tests
were perfonned using a microfine cement (Alofix MC 500) from Japan. This microfine
cement is a mixture of finely ground Portland cement and slag. Figure 2 (Reuben H.

875 Joer and Randolph


Karol, 1990) shows the grading curve of the Alofix MC 500 with the ordinary and high
early-strength Portland cements. The curves show that the d50 of the microfine cement
is about 3.2 Jlm compared with 17 Jlm and 23 Jlm for the type B and A respectively;
thus the microfine cement is finer than the two types of Portland cement by a factor of
about 6. A more detailed study shows that the penetration characteristics of Alofix MC
500 is closer to that of chemical grouts (resin), and it can easily penetrate sand With
particles as small as 0.074 mm, while the Portland cement can only penetrate coarse
sand (2 mm particle size). However, unlike the epoxy grout, the microfine grout did
not appear to penetrate the soil surrounding the model piles to any great extent.

100 r-,----""II$'''FlIiI:=rc~=~~:T1__rrTTT-__r__r___rTTTTn
~
'N~~~ll 11
90 +---+--+-+-+-H-j~""",--,+--l--a...Id:.-I-..-I-.'./I----+----+--J--W-I--l--I-l
I

__ 80 +-----t---t-+-+-HH++----""\H---+--H-~t;::~.t"'---I----I----l----I.....J.--j..+-l-j
.:c~70 +--+--+-+-++++H--.Jjr--\+-+-+-+-+-I--It~~-I---+----I--.j......+..-W-I-l
,,~,
.~ 60 --1.1--- Alofix MC500 11_ h ..
4500 1-+-...",.....I!!-+++++-I-~::I,,\---P.\----I----+--4-l-I-I-I-l
- 0 - - Type B 1\
~ 30 1\ "\ ..

~ 20 - . - Type A ~ '\ \
Co> 10 +====F=F=F=R=t=I=++==::"-+--1---+-U-4"i"I!..l-~--!--l-""r::t:'l..:"'''t..!-WJ
o +---.l...-....L.......L.......L...I-wll..J...It---I----l_L...J..-L..J....l.I.LIl't.;I__---L----l.--Ln...LI......!:d~.~..,.,
0.1 10 100
Grain size (micrometer)
Figure 2 Particles size distributions for the cementetious grouts
(From Reuben H. Karol, 1990).

Sample ton----I

150mm
100 mm
Sample base

(a) 5 pairs of holes (b) 3 pairs of hole (c) 1 pair of hole


Figure 3 Three grouting configurations for the piles.

4. The model pile


The model pile used for the tests consists of a 650 mm long steel pipe, with
diameter of 50.8 mm diameter and wall thickness of 1.6 mm. Various configurations of
the model pile have been tested, as shown in Figure 3. Initially, 5 pairs of diametrically
opposite holes were made over the bottom section of the pile (type a). This

876 Joer and Randolph


arrangement was made in order to place the lowest pair of holes at one pile diameter
from the base of the sample and the other 4 pairs at a vertical spacing of one pile
diameter. Later the number of holes was reduced to 3 pairs diametrically opposed and
at a vertical spacing of one pile diameter, starting from two pile diameters from the
base of the sample (type b). Finally, a configuration with one pair of diametrically
opposite holes was adopted, with the holes situated at three pile diameters from the
base of the sample (type c).

Grout pessure - - - - - - : I .
.....--Stow;:r

Piston--~i--I
Grout JllITIP---
Valve },...--_ _...... _ _Valve 2
<::bnrection groot IXJIDP'grout WI::es _ _~=i'l~-Airpressure
Valve 3

~C----!!E:lk::==r=rY1?2~~
---ffieC3p

_--Pile

11--++&--- Retwn. tutes

650 --,-;H++---ttt.t----Retum roles

__-r--\J1\;1UL roles

I' 50.8 mma-I


Figure 4 Pile details and grouting system

Each hole was connected to a copper tube, 3 mm in diameter, inside the pile.
Each pair of tubes connecting opposing holes was then joined at the top end of the pile
and connected to a grouting system. The fIrst two configurations (type a and b)
allowed the full surface area of the pile to be covered in many cases. However, this led
to accumulation of grout at the upper and lower ends of the pile shaft, which therefore
affected the boundary conditions. The last configuration (type c) gave around 50 %
grout coverage. Later, the diameter of the grout tube was increased to 5 mm and more
grout was pumped into the pile-soil interface. However, since the top and bottom of

877 Joer and Randolph


the pile were sealed, air was trapped in the cavity between the pile and soil during
grouting. In order to cover the pile area, it is necessary to allow the air to be displaced
by the grout. This consideration led to the final configuration of the pile (type d),
whil;h consisted of a pair of grouting holes situated at one pile diameter from the pile
tip and a pair of return holes at one to two pile diameters from the grouting holes, but
at a 90 rotation, as shown in Figure 4.

State Samples Grout Grout Pile


No. Names type Pressure volume Coverage type
kPa cm 3 %
Uncemented 1 HSLU_l G 500 183 100 a
2 HSLU_2 G 500 125 39 b
3 HSLU_3 G 500 67 50.2 b
4 HSLU_4 G 500 80 52.8 c
5 HSLU_5 G 500 N/A 22.1 c
6 HPOU 2 MC 500 70 5 d
2 % cement 7 HSL2_1 G 500 N/A 20 b
content 8 HSL2_2 G 500 170 70 c
9 BP02_1 G 75 286 54.5 d
10 BP02_2 G 50 207 70 d
11 HP02_1 G 200 190 41 d
12 HP022 G 250 167 55 d
4 % cement 13 HSL4_1 G 500 96 59 c
content 14 BP04_2 G 150 96 64 d
15 BP04_3 G 50 156 70.1 d
16 HP04_1 G 500 100 5.4 d
17 HP042 G 200 124 60 d
6 % cement 18 HSL6_2 G 500 207 81 b
content 19 HSL6_3 G 500 183 73 b
20 BSL6_3 G 300 260 90 c
21 BP06_1 G 500 N/A 46 c
22 HP06_1 B 650 110 N/A d
23 HP06_2 G 250 300 33.6 d
24 HP06_3 MC 100 276 61.5 d
25 HP064 MC 50 166 97.8 d
Table 2 Test details.

5. Testing procedure
A 10 mm diameter cone was pushed into each sample before installing the pile
and the total cone resistance was measured. Following the cone penetrometer test, the
pile was installed in the sample with an overburden pressure of 100 kPa applied on top
of the sample. Two types of test were performed: Pull-Out tests Oabelled PO) and
Sleeve tests Oabelled SL). For the Pull-Out tests, the pile was inserted to a depth of
250 mm to 300 mm into the sample (Figure 5 (a)). This was achieved by first driving a

878 Joer and Randolph


coring tube (of the same wall-thickness as the pile) into the sample at the desired
depth, then withdrawing it together with the soil plug. The pile was then carefully
placed into the hole. During the installation, Valve 1, situated between the grout pump
and the connection grout pump-grout tubes, was closed and a pressure of 100 kPa was
applied through Valve 2 to the grouting holes (see Figure 4). lbis was to ensure that
the grouting tubes were clean and to prevent soil from oozing through the grouting
holes. The return holes were left at aunospheric pressure (Valve 3 open). However this
method could not be used for uncemented samples. In such cases, the coring tube was
not used, but the pile itself, with the grouting tubes and the return tubes in place, was
driven into the sample.

(a) Pull-out test (b) Sleeve test


Figure 5 Test arrangments

For the Sleeve tests (Figure 5 (b, the coring tube was first pushed into the
sample; then the pile was connected to the top of the coring tube and both pushed
further into the sample. Once the pile was in position, the coring tube was then
detached from the pile outside the shoe. The valve configurations were the same as for
the previous case during installation of the pile. For both tests, the pile was installed
using a combination of a ramp displacement function at a constant velocity of 0.125
mrn/s and a superimposed cyclic displacement (amplitude of 2 mm at a frequency of
0.0775 Hz). This installation procedure was to simulate the cyclic slip between pile and
soil that occurs during driving.
After installing the pile, the grout was pumped into the pile-soil interface with a
constant pressure at the grout-pump. The volume of grout injected was measured by
noting the movement of the piston in the grout pump (see Figure 4). Thus, both the
grout pressure and the volume of grout injected could be controlled. Using the system
with the return holes, Valve 2 was closed after the pile had reached its final position,
and Valve 1 was opened. The grout was allowed to flow from the grouting holes to the
return holes. When the grout appeared from the tubes connected to the return holes,
Valve 3 is closed and the grout pressure was maintained or increased as required.

879 Joer and Randolph


Table 2 shows the grout pressure and volume of grout measured in each test. After
grouting, the grout was allowed to cure for 48 hours before testing the pile.
The pile was pulled (pull-Out tests) or pushed (Sleeve tests) monotonically at a
constant rate of 0.01 mm/s or 0.1 mm/s respectively. Each sample was then carefully
dismantled, and the pile inspected. The area grouted was defined by using an overhead
transparency wrapped around the pile and tracing the area grouted on the
transparency. A digital planimeter was then used to determine the exact area grouted.
Table 2 shows the grout coverage obtained for the tests, expressed as a percentage of
the initial area of the pile in contact with the sample.
6. Experimental results
6.1 Cone penetration tests
Figure 6 shows the plots of cone resistance versus penetration for an
uncemented sample and samples with 2 %, 4 % and 6 % cement content. As expected,
the cone resistance increased with the degree of cementation of the samples. Although
the shapes of the curves was the same for all the samples tested, there are minor
variations in the cone resistances (qJ for a given degree of cementation, as shown in
Table 3. This is due to the two different batches of soil used, as C2 generally shows
lower values than C1. Indicative values of cone resistance were 1.2 MPa for the
uncemented material, 3.5 MPa for 2 % cement, 7.5 MPa for 4 % cement and 15 MPa
for 6 % cement

Cone resistance (MPa)


0 2 4 6 8 10 12 14 16
0
50
..-
E 100
E
'-"
c:: 150
.-.... 200
0
cu
1::
0
c 250
If
300 --
350

Figure 6 Cone penetrometer tests.

6.2 Driving tests


Figure 7 shows the load-penetration curve obtained for a pile jacked into an
uncemented sample (a) and into a sample with 6 % cement content (b). Values of the
load (F) measured during the driving process are shown in Table 3. These values are
the average force measured over the penetration range 2 to 4 pile diameters (l00 - 200

Joer and Randolph


880
mm). A penetration of two pile diameters was considered sufficient for the load to
reach its maximum, and the penetration of four pile diameters was chosen to avoid
significant base effects. These limits were confmned as appropriate in all the driving
tests.

State Samples Results


No. Names <k F Friction Corrected
Peak Residual Friction
MPa kN kPa kPa kPa
Uncemented 1 HSLU_1 1.3 0.8 200 NID 200
2 HSLU_2 1.5 0.75 140 120 -
3 HSLU_3 1.3 0.85 100 90 199.2
4 HSLU_4 1.4 0.8 110 90 208.3
5 HSLU_5 1.2 0.8 50 NID 226.2
6 HPOU 2 0.6 0.3 37 NID -
2 % cement 7 HSL2_1 3.5 4 55 30 275
content 8 HSL2_2 3.4 4 174 100 248.6
9 BP02_1 3.4 3 130 NID 238.5
10 BP02_2 3.5 4 175 NID 250
11 HP02_1 5 N/A 65 54 158
12 HP022 4.5 6 119 NID 216.4
4 % cement 13 HSL4_1 8.5 5.5 260 160 441
content 14 BP04_2 6 7.5 200 NID 312
15 BP04_3 8.5 6 250 NID 356
16 HP04_1 8 8 85 68 -
17 HP042 7.5 6.5 152 NID 253.3
6 % cement 18 HSL6_2 20 16 430 280 531
content 19 HSL6_3 20 16 350 NID 479.4
20 BSL6_3 20.5 16 550 500 611.1
21 BP06_1 18 N/A 380 200 -
22 HP06_1 12 12 140 NID -
23 HP06_2 14.5 13 275 NID -
24 HP06_3 17 13 150 N/D 244
25 HP064 14.5 13 320 N/D 327
Table 3 Test results.

6.3 Loading tests


6.3.1 Pull-Out tests
Figure 8 shows CUlVes of shaft capacity (in tenns of average mobilised friction)
. versus displacement of the pile normalized by the pile diameter, obtained from Pull-Out
tests performed on an uncemented sample and samples with 2 %, 4 % and 6 % cement
content. The shaft capacity was calculated by dividing the total load measured during
the test by the area of the pile embedded in the sample. This makes no allowance for
the partial grout coverage of the pile shaft. Table 3 shows the values of the shaft

Joer and Randolph


881
capacity obtained for all the Pull-Out tests. The peak values obtained for these tests are
37 kPa for the uncemented sample and 119 kPa, 152 kPa and 320 kPa for the samples
with 2 %, 4 % and 6 % cement content respectively. Strain- softening behaviour is
observed after the peak was reached, although a true residual stress could not be
defined since the average shaft friction continued to decrease even after movements of
100 nun.
Load (leN) Load (leN)
-4 0 4 8 12 16 20 24 28 -5 o 5 10 15 20 25 30
O+---i.--,-,'---+--+--r---+--+~ O-t--+-__+-~--+-~r---+----i

50 50
-. -.
100 5
E 100
'-"
c c
.g 150 o 150
C':l .~
l:2
~ ~
c 200 c 200
~ ~
250 250

300 300
(a) Uncemented (b) 6 % cement content
Figure 7 Load-penetration curves during installation of the pile.

350

300
-.
~ 250
~
'-"

.-....
c 200
0
u
;S 150
<t:
C':l
.c 100
rn

50
ncemented
0
0 0.2 0.4 0.6 0.8 1.2 1.4 1.6 1.8 2
Displacement/Pile Diameter

Figure 8 Pull-Out tests.

882 Joer and Randolph


6.3.2 Sleeve tests
Figure 9 shows curves of shaft capacity versus displacement of the pile
nonnalized with the pile diameter obtained from Sleeve tests perfonned on an
uncemented sample and samples with 2 %, 4 % and 6 % cement content. The peak
values obtained for these tests are 140 kPa for the uncemented sample, and 174 kPa,
260 kPa and 430 kPa for the samples with 2 %, 4 % and 6 % cement content
respectively. For these tests, the residual friction is well defined, with values of 120
kPa for the uncemented sample, and 100 kPa, 150 kPa and 280 kPa for the samples
with 2 %, 4 % and 6% cement content respectively. The test on the sample with 2 %
cement content shows a more brittle behaviour at the peak. This is believed to be due
to a sudden shearing of grout near the base of the pile, where the grout interferes with
the base hole.

450
400
__ 350
~

~ 300
'-"

........uc:0 250

:E 200
.:::
~ 150
.c
CI:l Uncemented
I'
100 ~, ., ,
I
50 , ,
I
0
0 0.2 0.4 0.6 0.8 1
DisplacementlPile Diameter

Figure 9 Sleeve tests.

6.3.3 Grout coverage.


One of the most difficult problems in construction of grouted driven piles is
control of the grout coverage. The model apparatus allows both the grout pressure and
the total grout flow to be controlled within some limits. However, no control of the
grout coverage itself can be realised. The degree of grout coverage is determined by
the soil conditions along the pile-soil interface, and appears to be affected by the
degree of cementation of the soil. The pile-soil interface will also be affected by the
method of installation of the pile, although an identical procedure was followed for all .
tests reported here. The tests indicate that it is more difficult to inject the grout along
the pile-soil interface when the sample is uncemented than when it is cemented. This
can be explained by the fact that the soil in the remoulded zone near the pile wall is
more likely to be pushed against the pile (even to the extent of clogging the grout

883 Joer and Randolph


holes), in the case of uncemented soil. The grout coverage is also influenced by the
pressure applied. Larger coverage is obtained if first the pressure (Table 2) is applied
and then Valve 3 (Figure 4) is opened. rather than applying an increasing pressure from
o up to 2 MPa with the valve open. This second method resulted in little grout
coverage, considered to be due to the clogging of the grout in the tubing system.
Figure 10 shows photographs of the grouted piles after extraction, (a) for an
uncemented sample and (b) for a sample with 6 % cement content. In the cemented
samples, the thickness of the grout was fairly unifonn, at about 0.5 mm, while for the
uncemented samples it varied from zero to 2 mm. Some additional grout bulbs were
noted occasionally for tests in uncemented samples.

(a) Uncemented (b) Cemented


Figure 10 Photograph of the piles after testing.

7. Discussion
The cone resistance (qJ increases with the degree of cementation, although
there is some variability at each cement content. The variability is primarily due to the
two different batches of soil used, which is particularly evident for Sample 6
(uncemented) and Samples 22 - 25 (6 % cement content). Although not shown in the
paper, cone resistance for soil C1 with 4 % cement content was in the range 10 - 12
MPa. Unconfined compression tests were conducted on soil C1, and the cone
resistance can be correlated with the unconfined compression strength (au)' Figure 11
shows the relationship, with both quantities normalised by atmospheric pressure (Pa =
100 kPa). A best-fit cwve is given by the following equation:
1.5
!k=15+10 ( ~ ) (1)
Pa Pa
It may be observed that the cone resistance increases with the unconfined compression
strength to a power greater than unity. This contrasts with the relationship proposed by

884 Joer and Randolph


Houlsby et al (1988), where the cone resistance was found to be proportional to the
square root of the unconfined compression strength. It should be pointed out though,
that the weakest soil considered by Houlsby corresponds to the strongest (6 % cement)
soil considered here.

250

200 I

150
qc
Pa lOO
50

01
.- -- ---
I
--- -- --
0 1 2 3 4 5 6 7
O'u
Pa
Figure 11 Cone resistance - uniaxial compressive strength relationship.

The different shapes of load transfer curves obtained for the two types of test,
shown in Figures 8 and 9, merit some discussion. The Pull-Out tests, although more
consistent than the Sleeve tests, appear to give much greater displacements to peak
shaft friction. However, this is due to lifting of the (stress-controlled) piston at the top
of the sample. Unfortunately, this movement was not recorded in the tests reported
here, although will be in future tests. The other differences lie in the peak values of
friction and the degree of strain-softening. The differences in peak friction are partly
due to different degrees of grout coverage, as discussed further below. However, it
was also clear that in some tests the grout interacted with the exit hole at the base of
the chamber, leading to additional load (thus exaggerating the average friction). 1bis
may also account for the lower degree of strain-softening noticed in the Sleeve tests
compared with the Pull-Out tests.
It is difficult to estimate the load taken in 'end-bearing' in the Sleeve tests,
although inspection of Figures 8 and 9 would suggest a value equivalent to about 100
kPa in average friction. This would correspond to an end-bearing pressure of about 15
MPa on a 2 mm skin of grout at the base of the pile. The apparatus has since been
modified to provide an inner sliding collar at the base, in order to avoid any end-
bearing component of load.
The shaft friction shown in Figures 8 and 9 is an average value from both the
grouted and ungrouted areas of the pile. Tests on ungrouted piles indicated that the
friction from the ungrouted areas was very low (10 - 20 kPa) and may be ignored.
Thus, a corrected friction from the grouted areas may be assessed by dividing the
average shaft friction by the degree of coverage. The values of corrected friction are
shown in the final column of Table 3 (only for tests where the coverage was greater
than 50 %).

885 Joer and Randolph


Figure 12 shows a plot of the corrected (peak:) shaft friction divided by the
cone resistance, qc, against the cone resistance nonnalised by atmospheric pressure.
The curve shows a rapid decrease of the nonnalised shaft friction with increasing cone
resistance, but reaching a lower limit of about 0.02 for cone resistance greater than 10
MPa. A reasonable lower bound fit is shown, which can be expressed as:
~=0.02+0.2e-o04qc/Pa (2)
qc
Where 'rc is the corrected shaft friction, qc is the cone resistance, Pa is the atmospheric
pressure.

0.2
C':I
..eo
c::
....
o
0.18
0.16
0.14
\
,.
] 0.12 \
~ 0.1 \
B
~
o
U
0.08
0.06
0.04
\

--- ............... .
~-----~----~~--
0.02
O+-----+-----+----~----+------i

o 50 100
qc 150 200 250

Pa
Figure 12 Relationhsip between corrected shaft friction and cone resistance

From a design point of view, the ratio of shaft friction to cone resistance
indicated by equation (2) should be reduced to allow for incomplete coverage of the
pile. At model scale, procedures have now been developed to a stage where coverage
can be guaranteed to a level of 80 % or higher. Grouting at field scale does not have
the same problems of clogging of fine orifices, and the degree of coverage may be
expected to be at least as high. It would seem prudent though to include a reduction
factor of 0.7 - 0.8 to allow for incomplete coverage of the pile. This would give a
design estimate of friction, 'rd' of

.!d.",,0.015+0.15e-O04qc!Pa (3)
qc

Conclusions
The paper has described the development of apparatus and techniques for
model-scale testing of grouted driven piles, and presented results from a series of tests.
At the scale of the model tests, grouting is not straightforward, and a number of
modifications had to be made during the course of the tests in order to arrive at a
procedure that yielded high quality load test results. A key step in this development
was the use of superfine grout with a mean particle size of about 3 /lm, which

886 Joer and Randolph


overcame the problem of the grout clogging in the relatively small delivery tubes and
orifices. A further problem was to prevent excess grout from interacting with the
confining plates at top and bottom of the soil sample, leading to spurious overestimates
of shaft friction. Conclusions from the testing programme should be regarded as
preliminary, as some of the earlier results were affected by the problems just described.
Two types of test were perlormed, in compression (Sleeve tests) and tension
(pull-Out tests). The Sleeve tests are considered to offer an accurate guide to the initial
friction development of grouted driven piles, but the peak friction values may be
overestimated due to grout interaction with the sample container. By contrast, the
peak friction values measured in the Pull-Out tests are accurate, but the displacements
to peak are excessive due to (unmeasured) movement of the stress controlled top plate
and sample.
Total grout coverage generally ranged from 50 % up to 100 %, with an
average value of 60 %, apart from a few tests where specific problems occurred. The
tests carried out in the later stages of the work achieved consistent grouting over
upwards of 80 % of the pile shaft.
Allowing for the relative grout coverage, it was found that the peak friction
correlated well with the cone resistance, with a ratio that reduced rapidly to a plateau
value of 0.02 where the cone resistance exceeded 10 MPa.
References
Barthelemy, H.C., Martin, R., Le Tirant, P., Nauroy, J.F. and Cipriano de Medeiros, J.
(1987). Grouted driven piles: an economic and safe alternate for pile foundations.
Proceedings of the 19th offshore Technology Conference, Houston, OTC5409, pp
427-436.
Houlsby G.T., Evans K.M. & Sweeney M.A. (1988). End bearing capacity of model
piles in layered carbonate soils. Proceedings of the International Conference on
Calcareous Sediments, Perth, RJ. Jewell and D.C. Andrews (eds), Vol.l, pp 209-214
Joer, H.A. (1994). Model tests on grouted driven piles in calcareous soil. Proceedingd
of the fIrst Australia - New Zealand Professionals Conference, Sydney, G. Mostyn, N.
Khalili and J. Small (eds). pp 99 - 104.
Joer, H.A., Randolph, M.F. and Gunasena, U. (1994). Grouted driven piles in
calcareous soil. Proceedings of the XIII ICSMFE, New Delhi, VolA, pp 1673-1676.
Reuben H. Karol (1990). Chemical grouting. Second edition, A.C. Ingersoll (eds).
Rickman J.P. and Barthelemy H.C. (1988). Offshore construction of grouted driven
pile foundations. Proceedings of the International Conference on Calcareous
Sediments, Perth, RJ. Jewell and D.C. Andrews (eds), VoLl, 313-319.
Acknowledgment
The work was funded by the Australian Research Council. The authors gratfully
acknowledge this support.

Joer and Randolph


887
A NOVEL FOUNDAnON PILING SYSTEM - THE SPEAR PILE

Ross T. McGillivray! and Mohamad H. Hussein2

ABSTRACT:

This paper describes the design, installation and field testing of a composite
pile system consisting of a 9.I-m long, prestressed concrete bottom section with
variable length steel pipe top sections. The bottom 6-m of each concrete section
tapers from a 356-mm square top to a 203-mm square tip. The steel pipe is
uniform, with an outside diameter of 273 mm. A case history is presented for a 6-
story structure and an over-water helipad founded on Spear Piles. Design
considerations included the selection of working loads, selection of an appropriate
pile driving hammer, and preliminary evaluation of driving procedures. Both static
load tests and dynamic measurements and analyses were performed to evaluate pile
capacities. The dynamic measurements were also used to evaluate hammer
performance and pile stresses during installation. Design loads of up to 890 kN per
pile were used at the site. It is concluded that the Spear Pile is an effective and
economical deep foundation system when subsurface conditions indicate the
potential for large downdrag loads.

INTRODUCTION

Pile foundations are usually employed when shallow foundations will not
provide adequate support or protection against excessive settlement. Piles might be
timber, concrete, steel or a combination of these materials. Pile shapes available
for use include uniform-circular (pipe and concrete), "H" section (steel) or tapered
(timber, concrete and metal shell). Composite piles are typically constructed of two
different materials such as a timber pile with a concrete section build-up or a
concrete pile with a steel "H" or pipe section "stinger". Each of these conventional

! - President, ARMAC Engineers, Inc., Tampa, Florida, USA


2 _ Partner, Goble Rausche Likins (GRL) and Associates, Inc., Orlando, Florida,

USA

888 McGillivray & Hussein


pile types has its advantages and disadvantages (Chellis 1961), and any of them can
be an effective foundation solution if used under appropriate conditions.

Pile capacity is derived from a combination of shaft resistance and toe


bearing. As the pile is forced down into the soil by the applied load, the soil
friction along the sides of the pile is mobilized, as well as the bearing force at the
tip of the pile. In cases where compressible soil above the pile tip will continue
to consolidate after pile installation, the downward movement of the soil relative to
the pile will result in a downdrag load on the pile, reducing its net working load
capacity. The downdrag load is caused by negative skin friction acting on the shaft
of the pile. Studies and field measurements have shown that the downdrag load
magnitudes can be significant (Garlanger and Lambe 1973), and may even be larger
than the applied structural loads in some cases. Modern pile design and
construction methods incorporate provisions that take the effects of negative skin
friction into account (Fellenius 1991). A common construction technique is to
apply a bitumen coating along the pile shaft length where negative skin friction is
expected (Baligh et al. 1978); however, this is an impractical field operation with
questionable effectiveness under many conditions.

This paper presents discussions on design, installation, field testing and


application of a new piling system. This new pile is a composite type pile with a
tapered prestressed concrete tip section and a steel pipe top section.

THE SPEAR PILE

General Description

The pile type discussed in this paper has been named the "SPEAR PILE"
because of the slender, steel pipe top section mated to a tapered concrete bottom
section. A steel transfer plate was used to splice the pipe section to the concrete
section. Although the taper tip section is limited by the available fonns to a length
of 6 m, the straight concrete top of the tip section can be as long as 15 m. The
typical pile details are shown in Figure 1.

The steel transfer plate is cast in the top of the concrete pile section. The
prestress strands taper unifonnly from the tip to the top plate. The piles are cast
in sets, tip to tip and top to top with a spacer shim system aligning the top plate
perpendicular to the pile axis. The taper section tip fonn drops into the standard
356 mm square section precast bed forms.

Structural Design

The tapered pile tip section was cast using 41 MPa concrete. This section
was prestressed with four 9/16 (14.3 mm) low relaxation strands in a square pattern.
Each strand was tensioned to 172 kN per strand. The net prestress was about 4.8
MPa after release of the strands from the anchors. The splice plate had a thickness

889 McGillivray & Hussein


SPLICE PLATE DETAILS
...-_ _ 4 OUTER HOLES FOR CONCRETE PILE
PRESTRESS STRAND DIMENSIONS

~61:m
4 THREADED 16 mm
for ANCHOR RODS '1
14"
356 mm

10-3/4" PIPE
273 mm

TIP OF TAPER

j X
El
E zu
~ 0
lD U "
C'J ""
273 mm ~ "
"

I-"- - 1 0 . 7 5 " - - - - 1 ,:7;i "


"
Ii
----, ~ en
::>
0'
"
""
II

3" .... "


I

.-----'-....,+,-_ _---.-I,.--_ _ .+.--'--~-.l-./ ~:.4 mm

3"~1 A
102 mm 76 mm Ei
El ~
Eo-< "
C"J
tIJ
"" "

.q;
....
tIJ
tIJ , """
""
POl
SPLICE PLATE ANCHOR DETAILS i:tIJ I 'iI
@ I

i, li 6 M
fij ::"
"" p.. ,
I

i "
""
20'

I, """
I I lj
::
"i
i"
II
II
~

1\ "
II
~ ii "
II

,I
"
II
II
PLATE ANCHOR BARS "
"
~
II
"
:z: 1. 2 m LONG "
"
II
0
2 4.4 mm DIA. RE-BAR (f6) ;""1 ii
1=
<J "
""
"
I"I
'"""
THREADED AT TOP
"
"" J:
I:! ,,
~ ,, ,, ""
""
"""
I

:z:
0 ,, ,,, """
"
<J i ,, 1""1 I"I
I
"
-
I
"I
~ I'
-..::.=
--=- ' t--...
SECTION ~~ (tn
LlL-J
203 mm
SPEAR PILE STRUCTURAL DETAILS

FIGURE 1

890 McGillivray & Hussein


of 19 mm, and was fixed to the top of the pile with four #8 reinforcing bars 1.2-m
long.

A 273-mm diameter, 6.35-mm wall pipe was chosen for this project to maximize
the difference between top and bottom section areas. Some 305-mm pipe was used
when the supplier ran low on the smaller pipe to maintain progress, but most of the
piles on the project were the smaller pipe.

CASE HISTORY

Project Description

Cape Canaveral Hospital is located on a dredged-fill island in the Banana


River, west of the cities of Cape Canaveral and Cocoa Beach, and north of SR 520
in Florida, USA. The island is about 212 m long in the N-S direction and 178 m
wide. A 105-m by 64-m wide causeway connects the fill island to the causeway fill
for SR 520. The Ground surface of the island is at about elevation 1.7 meters
above sea level. The site was occupied by two main buildings, the 2-story
administrative wing and the 6-story hospital wing. Although the buildings are
adjacent to each other and one can be easily accessed from the other, they are
structurally separate.

The 6-story building was initially constructed to 4-stories and was founded
on step-taper type piles. The building was to be expanded to 10-stories at a later
date. Prior to the vertical expansion, the Hospital asked that a review be conducted
on the foundation capacity to assure that the vertical expansion was safe. The
review revealed that the full expansion could not be constructed because downdrag
loading resulting from to the settlement of the site fill had reduced the allowable
capacity of the piles. The expansion was limited to two additional stories and the
remaining Hospital expansion was reserved for new site areas.

The 2-story building was founded on tapered, fluted metal-shell piles at about
depth 13 meters. Because no provision was apparently made for downdrag load on
these piles, this building was not scheduled for expansion by the Hospital.

The 1992 Hospital expansion consisted of the addition of in-fill buildings


adjacent to and between existing 6-story and 2-story buildings and the addition of
a Helicopter Landing Pad (Helo-Pad). Because of the highly limited site area, the
decision was made to add the Helo-Pad as an elevated pile-supported platform over
water. The Helo-Pad is located at the northeast corner of the site area. Because of
the environmental sensitivity of the shallow water surrounding the hospital fill area,
the Florida Departments of Natural Resources and Environmental Protection would
not allow the use of a barge for construction. All piles had to be driven using a
crane located on the shoreline. Since the distance from the shore line was up to 24-
m to the pile locations, a very light pile was required.

891 McGillivray & Hussein


Subsurface Conditions

The dredge fill is medium dense silty fine sand with some clay lenses
extending 3.0 to 4.6 meters below grade. The fill is underlain by soft silty and
sandy clay and clayey organic silt to depth 17 m. These strata are underlain by finn
sandy clay and loose to medium dense silty sand to depth 22 m. A stratum of finn
to stiff sandy-gravelly clay with lenses of sand and limerock was found to extend
to a weathered limerock stratum at depth 46 to 52 m. The limerock stratum had
refusal Standard Penetration Test resistances of greater than 100 blows per 0.3-m.
The typical soil conditions are represented in Figure 2. The area around the island
was at about sea level, and exhibited a thin silty sand cap about 1 to 1.5-m in
thickness. It appeared that there had been moderate heaving of the soil around the
fill island relative the to soil levels observed at the same distance from the State
Road fill. Observations around the site structures indicated that at least 600 mm of
settlement had occurred since a side walk had been poured around the buildings.
Constant maintenance has been required to provide a safe ramp from the entry area
to the building due to continued settlement of the fill relative to the buildings.

Foundation Design

The design of the foundation for the proposed site expansion was initiated
by concern by the Hospital about the capacity of the original step-taper piles used
in the original construction. A review of the available project records revealed that
the soil, pile installation and pile load test data were available for the step-taper piles
used in the 6-story building, but no data were available on the fluted-taper shell
piles used in the 2-story building. The initial load test on a step-taper pile was on
a pile installed to depth 27 m. This pile failed at a total load of 800 kN. A second
test step-taper pile was installed to depth 41 m. The deeper pile reached a
maximum load of 1,690 kN. Although no data were available on the fluted-taper
shell pile, it was recalled by the Hospital maintenance staff that the piles were
installed to about depth 26 m. The tapered shell pile was described as having a
total test load of 1,600 kN.

The review of subsurface data revealed that settlement of fill placed over the
thick stratum of soft sediments between depth 4 m and 17 m would result in a
potential for large downdrag loads on the piles (800 kN). However, the review of
performance data for the piles used on the existing buildings indicated that taper-tip
piles could provide a substantial advantage in depth-capacity performance. For this
reason, a taper pile was selected for the Hospital expansion. The pile was designed
to minimize the downdrag load by reducing the pile section above the anticipated
bearing zone. Further, the reduced section above the pile tip would minimize the
potential to damage measures such as a plastic wrap, bitumen coating or PVC pipe
sleeve that would be used to limit downdrag loads on the piles.

Analyses were run using various methods to evaluate pile capacity and
downdrag loads. Pile capacity analyses presented here used the soil strength

892 McGillivray & Hussein


SPEAR STEP SOIL
DEPTH
PILE

--
TAPER DESCRIPTION
I
SPT
N-VALUES

--
~!--- O-r-- 0
37
Silly-Clayey SAND
f-- (SP-SC) 19
4_

-
-- 8_
f-- 20 2
0
7

----
f- Gray-Green 2
Sandy-Clayey 0
12 _ f-- 40 SILT 2
(OL-t.4L)
4

-----
f-- 4
14 _
3
60 Lt Gray Silly SAND 21

---
f-
(SP-t.4L)
20 _ 6
Vl
e::: to- Gray Clayey SAND w/ ShellS
w f- w (Sp-SC)
I- W
W l.L. 7
~

-
24 _ Gray Calcareous CLAY
z

- Z f-- 80 (CL) 7
::I: 27
::I: to-
l- e.. Lt. Gray Gravelly 11
a.. 28 _ f- w Calcareous Sandy CLAY
w Cl
w/ Shell &: LR Fragmenls 14
Cl
(SC-CL)
f- 100 19
Lt. Gray Sandy Calc. CLAY
32 _ (CL) 33

- 14
34
36 _
- 120 Dark Gray CLAY 23
(CH) 23

40 _ - 74
40

- 140 Lt. Gray-Green 22


44 _ CLAY (CH) 64/4"

- 100/1"

48 _
- 160

FIGURE 2 SOIL CONDITIONS AT TEST PILE

893 McGillivray & Hussein


correlation with SPT N-Value recommended in the Florida Department of
Transportation Research Bulletin 121-A (Schmertmann, 1967). A safety factor of
1.5 was used for the negative skin friction, 2.0 on support friction and 3.0 on end
bearing. Figure 3 shows that a straight 357 mm square pile develops 768 kN of
negative skin friction. The Spear Pile was analyzed using plastic wrap for
downdrag protection and an assumption that the SPT N-Values improved by 25
blows in the bearing stratum due to compaction of the loose silty sand by the
wedging action of the Spear Pile's taper tip. If the improvement of penetration
resistance was not used, the pile capacity demonstrated by load tests could not be
duplicated in the analyses. Figure 4 shows that the capacity of the pile for the
assumed increase in the bearing stratum N-Values matches the capacity in the load
test. This figure shows a net Spear Pile capacity of more than 900 kN below depth
26 meters and only 110 kN of downdrag.

Pile Driving

Because the surface soils varied from loose to dense sand, and were known
to contain some debris that could damage the piles, the pile locations were predrilled
with a 305 nun diameter auger to depth 12 m. The drill holes stood open, and when
the test pile was driven, the steel pipe followed the larger hole left by the drilling
and the concrete tip with no observed tendency to cave. The upper 12 m of the
pile was wrapped in 3 turns of 6 mil (.15 nun) polyethylene plastic for downdrag
protection. The plastic was wrapped with duct tape at intervals along the pile with
heavy tape wrap at the bottom to limit the tendency for the soil to peel the plastic
from the pipe. No damage or tearing of the plastic was noted on the production
piles. The plastic wrap around the steel pipe top of the piles was used only on the
heavily loaded piles. Piles loaded to only 267 kN working load, including the off-
shore Helo-Pad piles, were not wrapped.

A 6-m section of pipe was welded to the concrete tip prior to the start of pile
driving. Figure 5 is a photograph showing the weld of the pipe top section to the
steel splice plate at the top of the concrete section. When the pile was driven to
about depth 14 m, a 12-m section of pipe, wrapped with the plastic material, was
welded to the exposed end of the driven pile and pile driving continued until refusal
was achieved. Figure 6 is a photograph ,of the site showing some Spear Piles
stockpiled in preparation for driving. Piles already driven can be seen in the
background, including some with the plastic wrap downdrag protection visible.

Two diesel pile driving hammers were used on the project, an open ended
Kobe K-13 and a double acting LinkBelt 520. The actual pile driving resulted in
resistances of 5 to 20 blows per foot (60 to 2 mmlblow) above depth 18 m. Below
depth 18 m, the penetration resistance slowly rose until the pile reached refusal.
The average pile takeup for 405 piles was 26 m with the shortest pile reaching
refusal at depth 18 m and the deepest pile at 30 m.

894 McGillivray & Hussein


STRAIGHT PILE WITHOUT DOWNDRA(; PHOTECTION
- 356 nun SQUARE PRESTRESS CONCRETE PILE -
1.4,---,-------------------------------------,
Analysis by Method Based on FDOT Bulletin 121 A
1.2
Measured froln Composite Pile Load Test @ 27 m Depth

0.0
z
.:.: 0.6
Z
-~
III
0" 0.4
I::
o '"
...l III 0.2
;:l

"Z.<:
0
-'"'
~~
P::
0

0 -0.2
~

-0.4

-0.6

-0.6

-1
0 10 20 30 40

DEPTH IN METERS BELOW GRADE


Figure 3 0 FRICTION + BEARING <> TOTAL - - B-1

COMPACTION OF BEARING STRATUM BY TAPER TIP


- 203lnmx356mlnx61n, 3561n Square x 31n' CONCRETE, wi 273rnln DIA. PIPE TOP -
1.4 ,---,-----------------------...,..,-------,r-o----------,

1.2 SPT N Ilnproved by 25 in the Bearing Stratum

L ad Test @ 27 m Depth wi SF=2 ----+---'';7'-

0.6
z
.:.: 0.6
Z
-~
~ ~ 0.4
...: I::
o '"
..J , 0.2

"~ 0 15 0 f--+--& """"--+-.+-----------------------\


~~

P::
~ -0.2

-0.4 A alysis by Method Based on FDOT Bulletin 121 A


Pile with Plastic Wrap on Top 40 feet
-0.6

-0.6

_ 1 '--_..l-_ _-'--_ _--'- '---_ _....l..-_ _- - ' - _ ---'------'----'-------'----'

o 10 20 30 40

DEPTH IN METERS BELOW GRADE


o FRICTION + BEARING <> TOTAL - - B-1

Figure 4

895 McGillivray & Hussein


Steel Pipe - Concrete Section Construction
Figure 5

General View of Site and Piles


Figure 6

896 McGillivray & Hussein


The Helo-Pad piles were set by hanging the piles from the crane and hand-
setting the piles through a steel frame template. After the piles were set, the leads
were fixed to the template and the piles were driven using the same techniques that
were developed for the land piles. Predrilling was not required for the Helo-Pad
piles.

PILE TESTING

Static Pile Load Test

The static load test was run on a pile near boring B-1. Figure 2 shows the
pile in relation to the soil profile. The depth of the test pile was 27 m below grade
at a penetration resistance of 29 blows per last inch (1 mm per blow) with a Kobe
K-13 diesel hammer. The pile was loaded to 1,557 kN, and the load was held for
24 hours. The pile was then unloaded, and reloaded until the jack reached its
maximum capacity of 1,931 kN. Figure 7 shows the summary of the load test
plotted as settlement versus load. The maximum settlement at the end of the first
24 hour hold was 21 mm, The net settlement was 7 mm following the rebound.
Evaluation of the static load test data indicates that the pile did not reach failure,
even at 1,931 kN, the limit of the test jack.
PILE LOAD IN TONS
o 60 120 180 240

4
0.2

Vl 8
w
I
u 0.4
z ..... 12
~ Vl
z ..... ::J 0::
\ w
~ 0.6 \
-.""
, >-
!:: ~
w
Z
w
:::i:
\
, ......
......
......
... ......
u

a..

16
:::i:
:J
w '\ u ....J
....J "- ...... ~
:::i:
~
~
w
0.8 '" ......
......
......
......
u
.., 20
,
"
Vl ......
a..
0
"'- ., ..... ......
'-
...... 24
~ ..... .......... I
--- ......'",
......
1.0
W
....J
a.. \ 28
- - a- STEP-TAPER PILE, 1966 \
1.2 134.5' Total Length \
_
42' - 10-3/4" Pipe TIp ExtensIon
ARt.lAC SPEAR PILE, 1992 ; 32
89.5' Total Length
FAILURE

1.4
0 500 1,000 1,500 2,000
PILE LOAD IN kN
FIGURE 7 STATIC LOAD TEST RESULTS

897 McGillivray & Hussein


Figure 7 also shows the results of the 1966 load test on the step-taper pile.
This pile reached its original design load of 1,557 kN with the load held for 24
hours. During that time, significant additional settlement (7.5 mm) occurred. The
pile failed to sustain additional load at 1,690 kN. The step-taper test pile did not
encounter significant driving resistance until depth 36.6 m, about 9 m deeper than
the Spear Pile.

Dynamic Pile Testing

Testing and evaluation of the Spear Piles included dynamic pile monitoring
during the initial installation of one pile (Test Pile A) and restrike of two other piles
(Test Piles B & C). Field testing was performed with a Pile Driving Analyzer'"
(PDA) according to the Case Method (Goble and Hussein 1994). Subsequent data
analyses were done according to the CAPWAP~ Method (Rausche et al. 1994). The
primary objectives of the tests were the evaluation of hammer/driving system
performance, pile driving stresses and structural integrity, and static pile capacity.
The distribution of soil resistance along the pile was also evaluated. This type of
pile testing and evaluation is routine in modem deep foundation practice, and is
recognized by many standards and specifications (ASTM 1989).

Dynamic measurements of strain and acceleration were taken approximately


two feet below the top of each of the three tested piles. Two each strain
transducers and accelerometers were bolted on opposite sides of each pile to monitor
and average effects of non-uniform hammer impacts. The PDA provided signal
conditioning, amplification, filtering, calibration to measured signals and data quality
assessment before applying Case Method equations to measured pile records of force
and velocity under each hammer impact. Figure 8 presents plots of pile top records
of force and velocity obtained under hammer blows during the monitoring of each
test pile. Due to the highly non-uniform nature of this pile type, the PDA was
primarily used to obtain dynamic pile records. Dynamic data analysis with the
CAPWAP program provided much of the information regarding pile and soil
behavior. The CAPWAP analysis is done in an interactive environment using
measured pile data and wave equation type analysis in a system identification
process using signal matching techniques. Results from a CAPWAP analysis
included: static pile capacity, soil resistance distribution, soil damping and stiffness
values along the pile shaft and under its toe, forces (and stresses) along the pile
length at maximum load, and a simulated static loading test showing pile load-
movement relationship under static conditions. Figure 9 presents plotted CAPWAP
analysis results performed with data representing a hammer blow towards the end
of driving of Test Pile A.

All three tested piles were identical with length of 27.43 m and consisting
of a 9.14 long concrete section with an 18.29 m long steel pipe (305 mm outside
diameter and 6 mm wall thickness. Testing was accomplished using the LB 520
hammer. During initial installation, compression pile stresses were approximately

898 McGillivray & Hussein


____ F
____ V*Z

Test Pile A

2000kN

Test Pile B

2000kN

Test Pile C


'.....J

Figure 8: Records of Pile Force and Velocity


Histories Under Hammer Impacts.

899 McGillivray & Hussein


SPEAR PILE. ANALYSIS AT END OF DRIVING
GOI)]e Rauscne Likins & AssocUtes. Inc. CAPWAP IRI

_ _ _ Fa,. Hid 300 _ _ _ Fo,. Hid


_____ Fa,. klPI
CPt Hla

150

10 OIl 10 /-. ml
0
I 3 4 I./e
~, t I.le
\, ,
J

' .. - _ I
-150

10.0
It/l
_ _ _ VII

----- VII
Moo
Cot
c==:=D PIle

Sk ~n I:l:e!] stenee
01Str"' 1Dut Ion
IToe 10.0 k 1001

ml
Pile Forces at ~ut
I./e

400

Leaa In klC1S
300 Mid o I~O 200 JOO 400 PIle Top
klDO

~
17
25 Bottom
.25
150 ~ut 312.6 kIDS

-150
10
,,_,
mo
I./e

1.00
.50

.75
"
TOO Novement In tncn
~

~
'\
RSk

Rte

Ov
302.6 k,OS

10.0 klDS

91 tncn

(1 kip = 4.45 kN, 1 inch = 25.4 mm)

Figure 9: CAPWAP Analysis Results


Test Pile A.

900 McGillivray & Hussein


217 MPa in the steel section, 10 MPa in the uniform concrete section and 2.5 MPa
near the pile toe. Dynamic pile tension stresses were generally negligible.
Towards the end of driving, maximum transferred energy to the pile top averaged
approximately 13.6 KJ which translates to 38% transfer ratio when compared to the
hammer rated energy of 36 KJ. This is generally considered to be an average
hammer performance. Data analysis indicated end of driving mobilized static pile
capacity of 1,393 kN. Computations based on field measured dynamic data
indicated that no soil resistance was present in the upper 10m of pile length and
that only 107 kN of shaft resistance was coming from the rest of the steel section;
92% of pile capacity was contributed by the concrete section and only 3% was in
toe bearing. Capwap analysis with the data representing a hammer blow from the
beginning of restrike of Test Pile B indicated a mobilized static pile capacity of
1,722 kN, most of which (more than 80%) was contributed by shaft resistance along
the concrete section. Due to the high driving resistance (i.e., low pile set under
individual blows) during both initial driving and restrikes, it is believed that the
static pile capacities computed represent only lower bound values due to lack of full
mobilization of soil resistance. Compressive pile stresses were somewhat higher
during restrikes than those encountered during initial driving. Dynamic data did
not indicate structural damage in any of the piles tested.

CONCLUSIONS

The Spear Pile proved to be an efficient pile with respect to load capacity
versus depth. As demonstrated by the comparison of the static load test on the
Spear Pile with the results of the load test on the step-taper pile, the compaction of
the bearing stratum resulted in a higher capacity for the shallow depth of Spear Pile
in comparison with the deeper step-taper pile.

The Spear pile was easy to handle and install because of its light weight and short
segment lengths. The steel pipe top could be easily welded to accommodate the
varying lengths of pile encountered on the site. Only two piles were broken of the
more than 400 piles handled. There was some problem using a steel pipe with only
a 1/4 inch (6.35 nun) wall. Great care had to be used to assure alignment of the
hammer and the pile during driving. Otherwise, buckling of the pile top was
experienced.

Another advantage of the Spear Pile was the ability to easily reduce negative
skin friction. Several options were available, but because of stable ground
conditions only wrapping of the upper 40 feet of pipe section in 6 mil (0.15 nun)
polyethylene sheet was required. The dynamic analyses demonstrated that the
mobilized skin friction above the bearing zone was minimal.

The dynamic testing showed that the pile hammer did not mobilize the full
strength of the pile on re-strike. A larger pile hammer would be required to
achieve that load. Based on the driving and load test record of the step-taper pile

901 McGillivray & Hussein


used in the original construction, it was felt that the use of the smaller hammer
would prevent excessive shearing of the bearing stratum that could cause a loss of
effectiveness in compacting the silty sand bearing stratum.

ACKNOWLEDGEMENTS

The writers wish to express their sincere appreciation to Mr. Larry Garrison,
President of Cape Canaveral Hospital and his staff for their support during the
project and for their permission to publish the data contained herein. In addition,
we wish to thank Mr. Dirk Henderson of Henderson Prestress and Mr. Kenneth
Miller of Miller Brothers Construction for their assistance during the project.

REFERENCES

ASTM D4945-89. "Standard test method for high strain testing of piles," Annual
Book of American Society for Testing and Materials, Volume 4.08, 1018-1024.

Baligh, M.M, Figi, H. and Vivatrat, V. (1978). "Design of Bitumen Coating to


Reduce Downdrag on Piles", Massachusetts Institute of Technology Report No.
R78-5; Department of Public Works, Commonwealth of Massachusetts Contract No.
R21-3, Wellesley Hills, Mass. 144pp.

Chellis, Robert D. (1961) PILE FOUNDATIONS. McGraw-Hill Book Company,


New York, 704 pp.

Fellenius, Bengt H. (1991). "Pile Foundations", Chapter 13 in the FOUNDATION


ENGINEERING HANDBOOK, Second Edition, Hsai- Yang Fang, Editor, Van
Nostrand Reinhold, New York, 511-536 pp

Garlanger, lE. and Lambe, T.W., Editors (1973). "Proceeding of a Symposium on


Downdrag of Piles" M.LT. Research Report R73-56, Soil-331, Massachusetts
Institute of Technology, Cambridge, Mass., 104 pp.

Goble, G.G. and Hussein, M. (1994). "Dynamic Pile Testing in Practice,"


Thirteenth International Conference on Soil Mechanics and Foundation Engineering
(XIII ICSMFE), ISSMFE, New Delhi, India.

Rausche, F., Hussein, M., Likins, G. and Thendean, G. (1994) "Static Load-
Movement from Dynamic Measurements," Settlement'94 Conference, American
Society of Civil Engineers, Texas A&M University, College Station, Texas.

902 McGillivray & Hussein


EFFECT OF CANTILEVER PLATE OF A FOUNDATION PILE
ON PILE DEFLECTION UNDER LATERAL LOAD

Krzysztof Trojnar 11

Abstract
The subject of this paper is a method of increasing the horizontal rigidity of foundation
piles, as used in various fields of civil engineering for the transfer of considerable lateral forces
to subsoil. The design of a pile, with its rigidity increased by cantilever plate, is described, as
well as the results of tests and theoretical analysis confirming the effectiveness of this method
of increasing the lateral load capacity of foundation piles.

1. INTRODUCTION

This study refers to one of essential problems in designing pillars for


viaducts and trestles. Frequently the number of piles results from the need
of assuring adequate lateral rigidity of a support, rather than from required
vertical load capacity [6]. Development of a research basis for designing piles
of increased rigidity creates conditions for economic solutions of pillars to be
loaded by considerable lateral forces.

2. LATERAL LOAD CAPACITY OF PILES

The lateral load capac,ity of piles is determined by both the strength and
rigidity of its shaft and the strength and deformability of top layers of the
ground. In designing the pile foundations, one of the critical factors
determining their usability is that their lateral deflections should be restricted
to acceptable limits. In case of large-diameter piles, the requirements of
limited lateral deflection precede those for the strength of pile material.

1fKrzyszrof Trojnar, Dept. of Bridges Rzesz6w University of Technology,


35-959 Rzeszow, IN. Pola Street 2, Poland

903 K. Trojnar
The layer of ground to a depth of 2-3 meters below the surface has the
greatest effect on the value of the lateral deflection of the horizontally loaded
pile [1], [3]. In this range of depths the bending moment in the pile assumes
the highest values. The surface layer of the ground around the pile practically
offers no resistance to its lateral deflections. If pile bending is considerable,
the ground in front of it is raised up, while a gap is formed behind it.
In practice the lateral load capacity of a pile foundation may be increased by
two methods:
a) Increasing diameters of piles, or their number, in the foundation
b) Improving the ground characteristics around piles.

The Department of Bridges of Rzesz6w University of Technology,


developed an effective method of pile design adjustment to achieve good
interaction between the pile and the top layers of subsoil [4], [5]. Examples
of designs of a pile with higher horizontal rigidity are shown in Figure 1.

SECTION y. Y (alternative)

7. 2. 3.
H
~--- -'I,-
I

~I ~-_._

1I L

"
B
B I
--=--f-

.
,
-0:-,
,
t--
.

, 6.
.;- Q-.I'

Figure 1. Examples of Design Solutions of a Pile


with Cantilever Plate

904
2. DESIGN DESCRIPTION OF HIGHER-RIGIDITY PILE

The specific feature of the proposed pile design consists in the use of
reinforced concrete cantilever plate fixed rigidly to the concrete pile at the
ground level. The plate rests on a previously prepared ground. Its layout in
the area plan and the plate dimensions are adjusted to stress distribution in
the subsoil and are a consequence of the assumed condition of an effective
increase in lateral load capacity of the pile.

2.1. Factors Determining Increase in Pile Rigidity

In order to determine the optimum design of a pile with cantilever plate,


the effect of various design factors on the increase in lateral load capacity of
the pile was analyzed in following respects:
a. Pile embedment in the subsoil
b. Pile shaft diameter
c. Lateral loading of the pile
d. Level of cantilever-plate fixing to the pile
e. Cantilever plate layout in the plan area
f. Cantilever plate dimensions
g. Rigidity of plate joint to pile shaft
h. Deformability of subsoil under cantilever plate

The effect of plate reaction is the reduction of lateral deflection of the


pile and a reduction in the value of bending moment in its shaft. The
potential of cantilever plate utilization for the increase of lateral rigidity of the
pile is in particular a result of plate rotation. Vertical displacements of the
plate end, accompanying such rotation, mobilize subsoil reactions in front of
pile and increase its resistance to lateral deflections.

3. TESTING OF PILES WITH CANTILEVER PLATES

3.1. Test Objectives

The main objectives to be attained by tests was the evaluation of the


effect the cantilever plate had on the deflection of large-diameter pile,
embedded in cohesionless subsoil and loaded with lateral force, as well as
determining the phenomena taking place in the subsoil in the area of plate
pressure. Further objective was to provide research basis for including this
increased lateral load capacity of plate-cantilevered large-diameter piles in any
future design considerations.

905
3.2. Design Parameters of Test Piles with Cantilever Plates

In order to define the optimum design of plate-cantilevered piles


prepared for full scale testing, the effects of individual parameters on the
increase of lateral rigidity of the pile were analysed. The basic principles of
designing piles with plates were determined and then experimentally verified.

Level of Cantilever-Plate Fixing to Pile

For better cantilever-plate effectiveness it is practical to fix it to the pile


at the ground level. The plate prevents the raising of soil in front of the pile
and reduces the lateral deformability of the pile in the adjacent active-layer
area, stiffens the pile head to prevent its turning and reduces effectively the
bending moment in the pile shaft.
The cantilever plate located at the ground level cannot be used in areas where
there is a threat of wash-out, where soil under plate may be disturbed, or
where piles are surrounded by highly compressible layers of soil (fresh
embankments, peats, alluvial deposits), due to their compressibility.

Cantilever-Plate Layout in the Area Plan

The cantilever plate taken for field testing had the shape of a trapezoid fixed
to pile shank at its shorted side - refer to option 6 in Figure 1. The shape of
cantilever plate should correspond to the active area of subsoil in front of the
pile, as well as to the distribution of forces exerted onto soil when the pile is
loaded laterally. Figure 2 shows the layout of ground plasticization zones
around horizontally loaded pile. The part of the plate, which is at the greatest
distance from pile-shaft bending axis, interacts most effectively with the
subsoil due to its turning.

Yt
r

H x

Figure 2. Expansion of Plastic Zones Around Pile - based on phisical testing

906
Cantilever Plate Dimensions

The minimum reach of cantilever plate was determined on the basis of results
obtained in the study [2], which was devoted to analysis of pile interaction
with soil, assuming the soil is a elastic-plastic medium. The model used for
analysis is presented in Figure 3.

~
ACTIVE ZONE \ 'l,~
OF THE GROUND

R1 =e<:'R
< ~'\ ..
E =(3'E
V' = Y \~--_._. .~-
x
H
PILE SECTION

Figure 3. Subsoil Calculation Model [2]

The Model consists of a circle of the diameter ~, representing the ground


around the pile, defined by parameters E, v. There is a non-deformable
cylindrical pile of a radius R in its center.
A ring of ground, of radius R, = a * R and parameters E' = {3* E, v' = v, was
distinguished, which constitutes an active zone of subsoil reacting to lateral
loads on the pile.

Two models of ground were considered


1) Model in which the ground values around the pile remain constant:
E = canst. v = canst.,
2) Model with an active zone of the ground;
An area of dia R" where ground has changed parameters:
E = E', v = v' is created in front of the pile.

907
It was found that the distribution of pile deflections for various values of
ground parameters is contained in a limited zone around the pile a < 2. The
value of the radius, which affects .the amount of deflection in a flat model,
had been determined by analysis of a pile embedded in the isotropic elastic-
plastic' semi-infinite solid and by comparison of pile deflection curves in the
load plane of flat and spatial models. On the basis of the presented analysis
of the ground-a round-pile model the minimum reach of cantilever plate was
determined from the condition of providing cover for the ground active zone:
B.> R,.

Rigidity of Plate Joint to Pile Shah

The cantilever plate joint to pile shaft should be rigid. It should be designed
to withstand the bending moment applied in the location of plate fixing to the
pile. If considerable settlement of the pile under vertical load is anticipated,
the joint should be so dimensioned as to take into account the increased
reaction of subsoil under the plate. The recommended solution for that is of
the Plate to Pile Shaft Joint cast-in-place while concreting the Pile.

Deformability of Subsoil Under Cantilever -Plate

Subsoil used for .testing was a 300 mm-thick layer of rammed sand or,
optionally a low pressure injection of subsoil with cement slurry. The
effectiveness of the increase in lateral rigidity of piles with the use of
cantilever plate depends to a large degree on the strength and deformability
of subsoil under the plate. The subsoil .of higher load capacity can take over
a substantial share of pile load while the lower compressibility of subsoil
ensures .smaller settlement of the plate under load and small losses in the
passive soil pressure under plate in cases .of long-lasting loads.

3.3. Description of Tests

In 1988 the Department of Bridges Rzesz6w University of Technology,


.carried out field tests of large-diameter piles with cantilever plates. The piles
of dia. 1000 mm and 1200 mm,immersed 5m- and 10m-deep in c.ohesionless
ground, were subjected to the test. The effect of plate presence (reaction)
was tested by applying the expanding and pulling-together horizontal forces
to piles at both the ground leveland at some level above ground. The example
of testing the dia.1 200 piles of various lengths is shown in Figure 4. The piles
were loaded to their design load capacity, i.e. by applying the force H of 212
kN - in stage I. Then, after load removal, the force was re-appliedat 340 kN
level - stage II. Then the testing was repeated with subsoil under plate
injected with cement slurry - at .stage III. Piles were loaded by hydraulic
cylinders while using a special test stand. A comparison of the extent of
deflection was made between expanded piles with cantilever plates and piles

908
without such plates being pulled together. The deflection was measured at
three levels above ground with Hugenberger dial indicators.
Test piles with cantilever plates were also endurance tested in 12-month load
cycle. A calibrated set of springs was used for loading the piles. Both total
deflections of piles and cantilever plates were measured. during tests, as well
as permanent deflections of piles, after load was removed.

1'- :
1 : : '.1----.,.--,
I . " .
L..::':': .:.: :.. ~ ,~

a
o H
<D
'N

y-y
~. 1200 L. 800 L.
" '1

Figure 4. Test Pile Testing Technique [4]

909
4. EFFECTS OF PILE RIGIDITY INCREASE

The results of tests of piles with and without cantilever plate confirmed
the effectiveness of combined pile-and-plate systems in the transfer of lateral
loads, 'as well as enabled to determine the degree of effectiveness for the
assumed design parameters/values of test piles. Figure 5 shows the
comparison of deflections of 10m-long pile combined with cantilever plate and
that without plate, as measured in three stages of tests, when short-lasting
horizontal force was applied at a level of 4.2 m above ground.

,....- ------------.../
6. =
t:.. =
4.7 ( stage 3

2.1 stage 2 )
,.
,

250
/,- l..----~' ---.~I
225

200

175
. M/,~'
"
_. -'-.!P-' - . -jO~P--
,/
,
// " 6.

/
1.6

,,1/
(stage 1 )

,/ ~ ,'
9 ,c' /" ,
~ 150 / I
." I I'
"e.
./

o 125
: --
," / , I'"
< " / /' ti.:l:.V sta 9 e 1, AI
o 100 ,
,
~ I
I "- ...!!..! stage 1, 81
...J / /-/" -/
~ ~ stage
._._._~~ stage
2, AI
75 I / 2, 81

50 ,
, "
Y / I,'
t/
/
/
/
AI
~L'__'_: stage
stage
3, C!
3,

pile without cantlever plate


AI

25
I /
81 pile with cantilever plate
C! pile with injection under plate

o
o 2 J 4 5 6 7 8 9 10 11

o E F LEe T ION [MMJ

Figure 5. Comparison of Selected Results of Pile Testing [5J

910
When the pile is loaded for the first time (stage 1) with the force of 180 kN
the effectiveness of plate presence (reaction), determined as the ratio of pile-
head deflection increments at the ground level, is equal to 1.6. On second
loading (stage 2), for the same value of force, the ratio equals 2.1. In both
stages 'of testing, the subsoil under plate was in the form of 0.3m thick layer
of rammed sand. The effectiveness of plate interaction with the subsoil and,
consequently, the increase in lateral rigidity of the pile, rose in the third stage
of testing, i.e. when the ground under plate had been injected with cement
slurry, (stage 3) to the value of 4.7.

5. THEORETICAL ANALYSIS OF PILE-AND-PLATE REACTIONS

The analysis of pile-and-plate reactions has been carried out by


numerical method, with the use of discrete elastic supports, as well as by
finite-element method. Figure 6 shows an example of the distribution of main
stresses in the ground in the plate pressure zone, as obtained by finite-
element method.

-; ::.i _,-
j
!

I ! ! I

!
I ,
I
I i

- I ;

- I I

i -
i
- I
-

Figure 6. Distribution of Main Stresses in the Ground


in Plate Pressure Zone, as obtained by
finite-element method

911
REFERENCES

[11. Baquelin F. - The Pressuremeter and Found. Eng., T. T. Pub!. 1978


[2]. Baguelin F. Frank R. Said Y.- Etude theorique du mecanisme de
reaction latera Ie des pieux, Biull. Liaison Lab. P. et Ch. ,nov-dec.1977
[31. Jarominiak A. i inn. - Pale i fundamenty palowe, Arkady 1976
[41. Trojnar K.- Badania metody zwi~kszania sztywnosci bocznej pali
wielkosrednicowych, temat resortowy RPBR MK-702-04-17
Politechnika Rzeszowska, 1990, internal report.
[5]. Trojnar K.- Weryfikacja metody zwi~kszania sztywnosci bocznej pali
wielkosrednicowych, Polit. Rzeszowska, 1991, internal report.
[6]. Zawriew K.S. Szpiro G.S.- Rasczoty fundamentow mostowych opor
glubokowo zalozenia, Sovtransport, Moskwa 1970.

912
The GEWI-Pile,

A Micropile for Retrofitting, Seismic Upgrading and


Difficult Installation

by

Thomas F. Herbst

913
Introduction
When in 1971 the GEWI-Pile was conceived it was the
aim to create a micropile with the smallest possible
diameter and reasonably high bearing capacity. The con-
sequences of this requirement were manifold. The appli-
cation of new technologies have considerably changed
standard procedures practised during this period and
added new features as high safety standard and quality
control, environmental compatibility, corrosion protec-
tion and installation possibilities for difficult site
conditions as far as ground conditions or access are
concerned.

Description of the GEWI-Pile


The DYWIDAG GEWI-Pile is a pressure grouted pile
with smallest possible diameter (Fig. 2). The load car-
rying element is a core of 1 or up to 3 GEWI-bars, sur-
rounded by a cement grout cover. Compatibility between
steel and grout performance govern in particular the
steel properties as long as pure cement grout is used.
At the pile head the load is transferred to the structu-
re either by bond or by a thread anchorage, to the
ground it is transferred by skin friction. Tip bearing
in the case of compression piles is of low influence.
This can be seen from Fig. 1 where the ratio between
pile circumference and pile cross section is plotted
against pile diameter. The curve shows the increasing
influence of the pile surface over its cross section
with decreasing pile diameter. As the GEWI-pile is posi-
tioned at the far left end of the graph it is evident
that the surface contact with the ground governs its be-
aring behaviour. Grouting technology influences not only
the skin friction for load transfer but also the exten-
sion of the pile-soil interaction, which is of particu-
lar importance if GEWI-piles are used for creating a
reinforced soil body.
Pressure grouting through the casing which is with-
drawn is common practice and yields sufficient in strong
granular soils. In soils with weak mechanical properties
it is usual to use a postgrouting system with the GEWI-
Pile, by this the skin friction can be increased sub-
stantially and within the required zones.
The GEWI Steel is a high yield hot rolled unworked
rebar. A feature of the bar is a continuous coarse
thread which is rolled onto the bar during manufacture.
It is unsensitive against rough site handling and is
specially adapted for foundation work. The thread

914 T. F. Herbst
enables the bar to be cut at any point and screw connec-
ted with a coupler, a definite advantage when transpor-
tation lengths, or more especially, head room for in-
stallation is restricted. The thread additionally has
excellent shear bond characteristics so that in the load
transfer length no additional components are required.
It increases favourably the composite behaviour between
steel and cement grout.

. I I I
I
I
V GEWI- PILE

OJ 30 ~
~~
v c
c 0
~ ~
OJ OJ
2S \
~v ~V1 20
\
~ 0
uU 1S r-..
\
~ ~ 10 "-
Cl. 0::
u III eX S r--
-----
o 0.12 0.30 1.00 2.00
----'\? Region 01 smoll Pile diometer DIm)
bored piles

Fig. 1 Ratio pile circumference over pile cross


section to pile diameter

The loads are first of all determined by the size


and quantity of GEWI-bars. Table 1 gives the characteri-
stic values for the GEWI-bars. The working loads are de-
termined by choosing the adequate safety factor, which
in Germany is 1,75 for tension and 1,70 for compression.
For the design it is common practice to consider for
axial loads the load bearing capacity of the steel core
only. With tension piles, this value has to be observed,
but it is possible with compression piles to include the
bearing capacity of the cement grout to a small extend.
As mentioned, the load at the pile head may be in-
troduced through bond (if sufficient bond length is
available), or through a screw-on anchorage when a con-
centrated load application is necessary. For connections
to steel structures, such as transmission towers, staks
etc. anchor nuts are required. In many cases the anchor
nut is combined with a lock nut to prevent loosening.

Stability against Buckling and Bending


It is assumed in most piling standards that piles
even in soils of very weak nature are not liable to
buckling failure. This has been based on experience
915
partly with piles of large diameter, and partly with
piles in which material strength is in excess to the
loading. For very slender piles a question mark was set.
For this reason in recent times, the buckling behaviour
of micro piles has been investigated. This research has
concentrated on the ultimate bearing capacity of instal-
led in the ground and free standing piles in both field
and laboratory tests and was supported by theoretical
analysis.

Only in very rare cases the GEWI-Pile cannot be


used because of the danger of buckling, due to the soil
conditions. The German standard DIN 4128 gives as a li-
miting value an undrained shear strength of c =
10 kN/m 2 above which no buckling analysis forUthe GEWI-
Pile has to be performed. This value corresponds to one
which is attributed to certain soils given in tables. In
situ test results should be at least 20 or 30 kN/m 2 to
compensate for save assumptions in the tables.

GEWI-Piles. steel properties (ASTM A 615)

Threadbar Nominal Cross


.. .. Weight Maximal
size/grade threadbar section Yield U~imate threadbar
designation diameter area strength strength diameter
,
(inches) (Sq. in.) (Kips) (Kips) (Lbs/ft) (Inches)
- - - --
(mm) (Sq.mm) (kN) (kN) (kg/m) (mm)

Single #14/GR 60 1.693 2.25 135.0 2025 7.65 1.862


,
bar 44 1452 601 901 1139 473
piles #14/GR 75 1.693 2.25 168.8 2250 765 i 1.862
44 1452 751 1001 11.39 47.3
#18/GR 60 2.257 4.00 240.0 360.0 1360 i 2.504
2581 1068 1802 2024 I 636
57
#18/GR 75 2.257 4.00 300.0 400.0 1360 2.504
57 2581 1335 1780 2024 636
#20/GR 75 2.500 4.91 368 491.0 16.70 2.717
63 3167 1637 2184 24.86 690
Mul- 13 x #14/GR 60 3 x 1.693 6.75 4050 607.5 2295 --
tiple 3 x44 4355 1803 2704 34.16 --
bar 3 x #14/GR 75 3 x 1693 6.75 506.4 675.0 22.95 --
piles 3x44 4355 2254 3004 3416 --
(up to 3 x #18/GR 60 3 x 2257 12.00 720.0 1080 40.95 _.
3 bars) 3 x 57 7742 3205 4807 60.96 _.
3x#18/GR75 3 x 2.257 12.00 900 1200 I 40.95 .-
3 x57 7742 4006 5341 60.96 _.
I
3 x #20/GR 75 3 x 2.500 14.73 1104 1473 5010 --
I I
3x 63 I 9501 4911 I 6552 74.58 --
Any combination of bar sizes up to 3 threadbars is possible .
Working loads in tension and compression depend on safety factors used.

Tab. 1 Properties of GEWI-bars


916
Torqued Additional
Anchor
Head
Reinforce-
ment
(Tension)
.\ ./
Torqued Additional
Anchor Reinforce-
Plate ment
(Com-
pression)

i
I
Concrete
I Upper Structure
Concrete Pile
Structure Strengthe-
ning

Elastic
Spacer

GEWI-
Bar Outer
Cement
Grout

Cement
Grout

Fig. 2 Fig. 3

GEWI-Pile with Standard GEWI-Pile with Double


Corrosion Protection Corrosion Protection

917
For compression loads where buckling may be consi-
dered the cement grout cylinder is equally compressed
due to the high ribs of the GEWI-Bar. This compression
enables the GEWI-Pile to resist slight bending stresses
as long as no longitudinal cracks occur. If non-design
bending moments have to be taken into account at the
pile head which is quite frequent a spiral cage reinfor-
cement with longitudinal bars has to be provided
(Fig. 4) .It may even be necessary to increase the shaft
diameter to accomodate it.
Load excentricities have to be taken into account
by proper design. For this reason, in general, minimum 3
piles are chosen for a single load foundation and 2 rows
of piles are placed under a strip foundation.

Fig. 4 Testing of the head of a GEWI-Pile

Installation
The equipment and the experience- collected from
soil and rock anchor applications is put to good use for
the drilling of the borehole, injecting of the cement
grout, and placing of the steel core. (Fig. 5) In parti-
cular, the use of efficient hydraulic rotary drill rigs
enables drilling of boreholes through soils of widely
different characteristics from hard rock-or concrete to
soft silts of clays or soils containing floating boul-
ders.
918
In unstable soils the use of casings is common to
stop borehole collapse, and they are left in the boreho-
le until both, the cement grout and GEWI-bar are placed.
The use of a small rotary drilling machine has other
features of note. Piles can be placed vibration free and
at low noise level. Such rigs enable installation of
piles in applicaions previously considered impossible,
or detrimental. Piles can be placed very close to neig-
hbouring buildings, including masonary walls. They can
be installed in rooms with restricted headroom, such as
basements. The mobility, compactness, and relatively
light weights of rotary rigs enable access to areas bar-
red for large piling rigs. Working from light staging or
even flying scaffolds can be considered.

Grouting and Postgrouting

The use of cement grout for the pile shaft and the
application of pressure grouting creates high skin fric-
tions which means higher safety with GEWI-Piles. The
liquidity of the cement grout, guarantes that under

Drilling a Installation of Primary grouting Posl-


cased hole the GEWI-Pile in and retraction of grouting in
sections the casing cohesive
soils

~ I

U
,-:.

Fig. 5 Installation of GEWI-Piles

919
pressure the total volume of the borehole and adjacent
voids are filled and that unstable soils do not collapse
into the borehole upon casing extraction. Indeed the
pressurized grout may even compress weak soil layers.
It is common practice that the borehole is filled
with cement grout starting from the bottom end so that
all water in the borehole is flushed out. If small bore-
hole diameters or casings prevent placing of a tremi
tube and steel core side by side, the borehole is filled
with grout before installation of the GEWI-bar.
Pressure grouting is applied and maintained during
extraction of the casing. It improves the skin friction
of the soil. The properties of the soil layers govern
the grout demand and pressure. Thus, a continuous con-
trol of the foundation soil is simultaneously obtained
and bedding in weak soils is improved.
In cohesive soils single pressure grouting is not
sufficient to develop the required skin friction between
the pile and the soil. It is possible to attach a post-
grouting system (Fig. 6) to the GEWI-Pile for the injec-
tion of additional cement grout into the soil body after
the first hardening of the primary pressure injected
grout. This consolidates the surrounding soil, improves
the soil properties and hence increases the load bearing
capacity. (Fig. 7) Even several post grouting stages are
possible if the grout hose is properly flushed. It is
also possible to postgrout a GEWI-Pile some time after
the installation when the load bearing capacity turns
out to be inadequate.
Corrugated Sheath

-----
GEWI-Bar

\ Outer
Cement grout

Inner High Pressure


Cement Grout Cement Grout, postgrouted Grout Tubes

Fig. 6 Postgrouting _Fig. 7 Effect of Postgrouting


System
920
It has, however, to be born in mind that the soil
properties themselves determine the load which can be
taken. Grouting is basically a technology for improve
the skin friction and bond to the ground.

Corrosion Protection

Standard Corrosion Protection

The centrically placed GEWI-bar or bundle of GEWI-


bars is surrounded by a cement grout cover of at least
of 20 rom. The high ph-value of the cement grout passiva-
tes the steel surface. Because of the high thread defor-
mations the grout body and steel remain homogeneous. For
compression loads this creates a permanent corrosion
protection. Problems of debonding do not occur. This is
especially favorable for tensile loading where uniform
crack distribution of the cement grout and crack width
control is important. Only hair line cracks occur at
higher load levels due to the low strain of the GEWI-bar
at working load. Furthermore, the small surface area of
the steel bar relative to the cross sectional area gives
little chance for the corrosion of the steel bar.

Double Corrosion Protection


In the case that the GEWI-Pile is to be installed
in soils with very aggressive ground water conditions,
a so called double corrosion protection as used with
the DYWIDAG Soil Anchors can be used. A corrugated PVC
sheathing is placed over the entire length of the pile
under workshop condition (Fig. 3). The annulus is preg-
routed with cement grout. For piling double corrosion
protection is a unique feature of highest quality which
is not available for other types of piles.
The Safety Features of the GEWI-Pile
The safety considerations for a foundation element
are of top priority for the design engineer and the ow-
ner as well as for the contractor. The safety features
of the GEWI-Pile are threefold.

1. The Design
The design incorporates many properties which
create a safe foundation element. Amongst them are:
921
- The GEWI-bar is a finished ready to install element.
- Simplicity and ruggedness is also typical for its
few components.
- Proven installation methods include controlled pres-
sure, quantity, flowability and strength of the
grout.
- The safety criteria for load bearing components and
load transfer are defined in the standards.
- The in the range of micropiles limited load bearing
capacity requires a multitude of individual
foundation elements. A more uniform distribution of
the structure loads in the ground is possible.
- A limited pile load makes its load transfer to the
ground more safe.

2. The Quality Control


The modern concept of quality control is applied
to the GEWI-Pile. All load bearing components are sub-
ject to a systematic quality assurance system.
The quality control includes all components affecting
the internal safety.
- GEWI-bar
- Anchoarages and splice components
-Workshop applied double corrosion protection
- Grout

3. The GEWI-Pile Testing


Load testing of piles is usually associated with
high costs, increasing with pile loads. Often the loa-
ding is applied using a counter weight of reusable con-
crete blocks or steel girders or balasted containers.
At very high test loads soil or rock anchors have to
be placed. For GEWI-Piles much simpler testing is pos-
sible.

a) Compression Load Test


With GEWI-Piles the normal problems of pile testing
are virtually eliminated, as the reaction load can
922
be taken by neighbouring piles through tension pro-
vided that prior due consideration is given to ensu-
re that neighbouring piles are available. Piles used
for test loading are usually tested to failure and
are normally not incorporated into the structure
being supported. The use of neighbouring GEWI-Piles
makes the test loading simple. Traverse beams ancho-
red to neighbouring piles are the only expense
(Fig. 8). To avoid eccentricity of loading it is
common to use four reaction piles and if these were
GEWI piles, they would be loaded to 25 % ultimate,
which in general will be about 50 % of the working
load. Connection of the traverse is simple as the
thread of the GEWI-bar enables the standard threaded
hardware to be used. The reaction piles are extended
by using a short length of bar coupled to the pile
and is connected to the traverse with anchor nuts
etc. Adjustment for height is simply a matter of
screwing up or down as necessary. A center hole hy-
draulic jack is arranged over the bar between the
load beam and an anchor plate screwed to the test
pile. This ensures a centered load application.

For accurate load determination a load cell may be


provided. The bar should also pass through it to
ensure centralization. Pile settlement is measured
using standard dial guages fixed to an independent
reference beam. A rigid testing set up is obtained

GEWI - reaction pile GEWI - reaction pile :

load beam beam

Fig. 8 Easy testing set-up for GEWI-Piles


923
if the reaction piles are slightly prestressed. Such
a method allows testing of piles even for soft top
layers of the ground where placing of balast is not
possible.

b) Tension Load Test


In most cases the reaction forces from a tension
load test can be taken directly by the ground in
distributing the load through a concrete slab or a
light grillage. Both must be arranged in such a way
that the soil stresses arising from the reaction
slab do not affect the bond stress of the pile.
Shall the bond stress be limited to a certain soil
layer the GEWI-Pile is particular appropriate. By
interrupting the bond with a sheathing along the
steel core the load transfer is restricted to the
desired zone. The testing set up is the same as for
ground anchors.
c) Test Results
If for the same soil GEWI-Piles are tested under
compression and tension the results show for com-
pression load a higher bearing capacity at lower
settlements. This indicates the partial load trans~
fer by the surrounding cement grout and depends lar-
gely on ground, grout and diameter (Fig. 9).

Load (kN)

0 200 LOO 000 BOO 1000 '200


I

E
.. L r::: f-.-:,,-: \0'
"=.. --:::-.. ~-
........... .:::~ 20" 1-"";'10":"
'EQl :: ......~
...
::' ~ ;;:.- ...... -.,.........
~15

~
E B
::;:..;
.......... ....... - ~
.
.....-,;;.
.... 2G'
r--....
~OO'
~
en 12 1-....
-, r- __ ~

t-- _::J 20'

Legend: - - Compression Pile


- - - Tension Pile
Pile-Length: 13 m (50 mm III Threadbar - DIN)
Soil Condition: 0- 8 m Fill (Round Gravel)
8-13 m Natural Soil (Solid Gravel)

Fig. 9 Results of pile tests in compression


and tension
924
Fields of Application

The applications of the GEWI-Pile are manifold


(Fig. 11).

1. Due to the small dimensions of the drill rig and the


easy coupling of section of the GEWI-bar it can be
used in areas with restricted head and installation
room as under bridge decks and in basements. for
repair or reinforcement of piers and abutments. Even
in difficult site conditions like steep slopes the
drill rig is easier to install.

2. If obstructions like concrete blocks or bolders have


to be penetrated the small borehole diameter allows
the use of high efficiency drill tools.

3. For foundation works on railway bridges advantage is


taken of the fast drilling and installation. The
drill rig is mounted on rail cars, even short traf-
fic pauses may then allow installation.

4. Tension loads occur with different construction pro-


blems. The steel core, easy screw connection, the
load transfer to the ground by skin friction make
the GEWI-Pile an ideal tension member for all these
tasks.

,', dl ".".", ~

:.- :~'.o _:':~ ~' ......::, ; .0

:,: : :.: ~.' :/:~:".:..... ,.....,.....",--

-'---'-"-'--'--- ------~

Fig. 10 Precompression of an installed GEWI-Pile

925
\\ /1/;,\\

/1).\\ /. Z ..\'\. II

-H-

I
I

.... , /}.,\\ '/.

Fig. 11 Field of favourable application of GEWI-Piles

926
5. The high ductility of the steel core allows even
bending due to some ground movement. The GEWI-Pile
may be used for dowels in case of land slide secu-
ring, slope stabilization and embankment securing.

6. For certain cases of vertical loads it may be inter-


esting to create a reinforced soil by a composite
system of closely placed GEWI-Piles and the ground.
Both materials contribute due to their stiffness to
the bearing, may control settlements and may trans-
fer the loads to lower levels jointly. A battering
of the piles increases the surface of the rein-
forced ground and reduces the effective stresses.

7. In some cases it may be necessary to preload the


piles as in the case of underpinning of old structu-
res where settlements have to be controlled. The
continuous thread makes such an operation simple.
(Fig. 10)

Special Considerations for Strengthening of Bridge


Foundations

GEWI-Piles are preferred for bridge retrofitting when:

- the service and dead load is increased,


- new design rules require a strengthening,
- the bearing capacity of the ground or a foundation
element
has been overestimated in former times,
- ground conditions and ground water level have
changed,
- horizontal forces are not sufficiently balanced.
Strongly inclined GEWI-Piles may be a solution.

An example for the strengthening of a bridge pier


is given in Fig. 12 where the foundation has yielded.
Under the bridge deck GEWI-Piles have been installed
around the pillar and connected to it by a concrete
block.

For the strengthening of bridge foundations it is


often the only feasable solution to drill through the
existing foundation slab. The extremely small diameter
disturbes the foundation to the lowest degree and di-
stributes best the load if an equal pattern of the pi-
les over the slab is selected. The anchorage nut in
connection with the high bond to the cement grout along
the bar requires short bond length inside the foundati-
on. In addition, porous concrete of old foundations is
strengthened by pressure grouting techniques.

927
Fig. 12 Strengthening of a yielding bridge pillar

Chipping-off concrete at the pile heads for


leveling the pile head which is for concrete piles of-
ten a problem, is a work of minutes for the GEWI-Pile.
The versatility of load transfer designs at the
pile head makes the GEWI-Pile in particular adapted to
retrofitting work.

Seismic retrofitting
For seismic events different models of wave propa-
gation, attenution by different soil strata and their
impact on the structures have been developed. Little is
known, however, about the behaviour of piles during
earth quakes. Analogies may help to approach the pro-
blem. Soil reinforcements and foundation elements e. g.
have not only been designed and the design has been
verified by observational method at the structure, hi-
story of ground engineering shows that the technical
justification of the bearing capacity of foundation
elements has only been found after considerable amount
of positive experience. The good results have been the
proof that foundation methods were often "invented" on
the basis of a sound geotechnical feeling.
With the application of micropile for seismic re-
trofitting we face a little bit the same condition.
While different models of pile sollicitation still wait
to be backed-up by the results of deeper lying earth
quake sensors and stress sensors at the pile it is at
least reported that foundations with root piles in Ita-
928
ly have already survived earth quakes. Closest related
are mining induced seismic activities. Structures and
installation which display a certain subtleness and
flexibility perform in ground better than those which
are too rigid. Shock absorbing devices follow the same
principle. Not to resist against but to follow move-
ments may be the proper answer also for foundation ele-
ments. We do not yet know about the required degree of
flexibility, but we know that micropiles, in particular
the GEWI-pile is a very flexible pile due to its slen-
derness and its ductile steel core. It can be assumed
that it follows best the shock induced displacements in
the ground which are perpendicular to its axis and that
it remains integrated with the soil. With a group of
micropiles being parallel or battered a reinforced soil
body is created which performs as composit also in a
flexible way, however, the degree of flexibility may be
different from the one of a few single piles. For pro-
per design still considerable research work has to be
done in this field and it is wise to make at present
conservative assumptions. The above mentioned geotech-
nical feeling indicates that micropiles with high fle-
xibility are a highly appropriate approach for founda-
tions in earth quake affected areas.

of a DYWIDAG 57M (1110) GEWI-Pilc


mm

535 kN 625 kN 710 kN 960 kN


15 - I, L I, I.
(120KI 'I (140KI 1 (160KI 'I 1216K) 1
10 _.. , -- -- - - -- , - .
.<:
C E .... ' -.,
E E
"u
~
c0
'in
Sl=:=:;:...:.;;;.;:;...;. :-:...:=Z F..=
.. --==. .= = """
0.;.;.1 .

.~
o f-
N
01-------------------------
.......... Field Data
-5 _ "-.,....--~\...-,.==
. ....,.,.,., .
E Average Values
E
.<:
C
-c
0
........................................ ..
'E" 'iii -10 , ,

'" ''""
u
"' EC.
C.
<Il
Ci u - 15 L-
0 L- .L- ..J..... -'-- ---'- --'-

o 20 40 60 80 100 120
Number of Cycles

Fig. 13 Cyclic compression and tension loading


of GEWI-Pile 0 57 mm
929
As to a great extend existing structures are in-
volved, the additional advantage of the GEWI-pile that
it can be installed under difficult conditions makes
it a very valuable tool for seismic retrofitting. A
typical example is given by the seismic retrofitting
of a historical school building in Vancouver where
GEWI-Piles 0 57 rom have been installed. Extensive
tests under tension and compression to maximum test
loads and cyclic loading up to 960 kN (216 k) with
100 load cycles were executed in advance (Fig. 13).

Conclusion
By combining techniques taken from foundation en-
gineering as rotary drilling and pressure grouting and
the one given by the DYWIDAG threadbar system, a unique
micropile has evolved that can provide a solution to
many difficult problems. Compression and tension loads
even alternating ones can be transferred. Light weight
versatile drill rigs allow installation under difficult
conditions. Groups of GEWI-Piles create reinforced soil
bodies according to a design philosophy which gains
increasingly importance.
Retrofitting of building and bridge foundations, in
particular where seismic events are expected, is a spe-
cial field where the properties of the GEWI-pile with
its ductile core can be made to good use. Double corro-
sion protection is unique for this pile system and of-
fers special advantage in polluted and aggressive
ground.
A forthcoming aspect is its compatibilty with the envi-
ronment as no harmful components nor polluting instal-
lation procedures are used which includes the almost
non existent amount of excavation material.

930
THE DRILL PILE METHOD
- New Low-Noise, Low-Vibration Piling Method -

Masaharu Hashimoto, Osami Hashimoto,


Shinji Nishizawa, Kojiro Ishihara and Yuya Sakurai

Abstract
"The Dr ill Pile Method" is a new type of low-
noise/low-vibration piling method which installs thin-
walled steel pipe piles by rotary penetration but
wi thout discharging displaced soil. The dr ill pile
method also achieves a high skin friction resistance by
minimizing soil disturbance around the pile. As
penetration continues,. spiral ribs located on the inner
wall of-the drill pile provide the inner ~o~l cylinder
with sufficient" resistance to prevent further soil
intrusion, resulting in pressure on the ground under
the pile toe. Therefore, the drill pile has the
character istics of a displacement pile. The setting
depth into the bearing stratum can be checked by
measuring the penetration data (e.g. penetration
torque, penetration speed) in real time at the
construction site.

Scaled model tests clarified the penetration/


bearing capacity mechanism of the drill pile. The
compaction effect of peripheral soil and the soil
plugging behavior inside the pile was confirmed by
field experiments at various sites with different
bearing stratum. The skin friction and point bearing
capacity of the pile were separately evaluated through
vertical loading tests, and an adequate bearing
capacity formula was derived.

1. Introduction

The installation of piles into soil can be broadly


categorized into displacement and non-displacement
methods.

1 General Manager, Kawasaki Steel Corporation, 2-3,


Uchisaiwaicho 2-chome, Chiyoda-ku, Tokyo 100, Japan

931 Hashimoto
The displacement pile method frequently involves
dr i ving the pile wi th a steady succession of blows on
the top using a pile hammer. While this is a
relatively fast installation method and a high skin
friction and point bearing capacity are achieved, this
method generates considerable noise and local
vibrations which may be forbidden by local regulations
or environmental agencies and, of course, may damage
adjacent property.
Dr illing a hole which is filled wi th concrete to
form the pile after hardening is relatively free from
noise and vibration, and causes virtually no soil
displacemen t. However, the problems associated wi th
this method are:
a) high possibility of insufficient cleaning and
removal of loose soil or slime at the hole bottom
which contributes to large settlement (Yamagata,
1980);
b) disturbance of the peripheral soil which reduces
skin friction resistance (JSSMFE, 1992);
c) uncertainty in controlling the setting depth; and
d) storage and disposal of displaced soil.

The rotary penetration steel pipe pile method,


called the dr ill pile method here, 1S a new type of
low-noise, low-vibration method that attains the high
skin friction and point bearing capacity of the
disp.lacemen t pile but avoids the problems associated
with non-displacement piles. The various technical
aspects of the drill pile method are explained here
including:
physical structure of the drill pile and a brief
description of the pile installation equipment;
analysis' and evaluation of the penetration and
bearing capacity mechanisms using scaled-model
experiments;
evaluation of the skin fr iction and point bear ing
capacity through vertical local tests; and
derivation of a proposed bearing capacity formula
based on these tests and analyses.
The steel pipe pile has spi ral ribs and bi ts on
both the inner and outer surfaces at the toe portion,
and is ins talled by rotary penetration. This method
was developed as a low-noise, low-vibration steel pipe
piling without soil discharge that can be applied to
low- to medium-rise buildings with foundation piles of
up to 508 mm in outside diameter. The foundation
structure by this method was approved by Japan
Architectural Center in June 1990 and by the Ministry
of Construction in July 1990. As of December 1993,
approximately 6,000 tons of steel pipe piles
(approxima tely 3,900 piles) had been installed using
this method.

932 Hashimoto
2. Outline of Piling Method
Figure 1 shows the structure of the drill pile used
by this method. Spiral ribs of 13 nun diameter steel
rod are welded to both the inner (vertical length of
1.5 m) and outer (vertical length of 3.0 m)
circumferential surfaces of the pile toe. To
facilitate penetration into the ground, cutting bits
are provided. The pile head is held with a rotary
device which drills the pile into the ground to
penetrate the bearing stratum.

Stopper
Steel pipe pile

Outer spiral
-:;:===~

o Inner spiral rib


o
o
M o 0
o 0
M tn
.-i
.L..._--L..,p._-~

Cutting (rom)
,(bits)

Fig. 1 Schematic View of Drill Pile


The soil around the pile toe is pushed up by the
spiral ribs during pile rotation and then accumulates
around the -pile at the top end of the rib, increasing
the radial earth pressure around the pile and
generating a high skin fr iction resistance. Since the
ribs are only present at the pile toe, the displaced
soil remains underground contributing to clean
execu t ion of work wi th no discharged soil requi ring
temporary storage and disposal.
The inner spiral ribs promote compaction of the
soil inside the pile at the bearing stratum at the
final penet ra tion stage. This increases the friction
between the inner wall of the pipe and the soil plug,
contributing to a point bearing capacity which is about
that of a closed-end pile. ' In an open-end pile, the
point bearing force is due to the friction between the
inner wall of the pipe and the compacted soil along a
length of 2-3 times of the pile diameter from the pile
toe. The length of the inner rib portion, 1.5 m
corresponds to three times the pile diameter of 500 mm.

933 Hashimoto
General-purpose piling machines such as the tripod
type pile driver fitted with an earth auger are used to
rotate and drill the steel pipe pile into the ground in
this method (Fig. 2). Since the chucking of the pile
to the earth auger can be done remotely using a special
rotary jig, greater safety is assured wi thout working
at heights in pile erection and splicing.

auger

I I
. I I Drill pile
I I
Lo1
c::I..l1
~.j

!-~

Fig. 2 Diagram of Drill Pile Method


Main items of execution control with this method
include material, construction equipment and apparatus,
pile erection, rotation torque, penetration and setting
depth, accuracy and safety, but the setting depth is
the most important of all. "Doctor System" shown in
Figure 3 is used to manage the penetration depth and
time of pile, moni tor ing the load cur rent value and
time of auger motor. The execution management da ta,
~ value for each penetrating depth (geometrical mean
of rotational torque T and penetration time per
penetration length t) is calculated by a built-in
computer, and the depth of the bearing stratum and the
setting length of a pile into the bearing stratum a,r..e..
monitored by checking the ~ value. Because the vTt
value cor rela tes wi th the N-value of the soil a t the
pile toe, as shown in Figure 4, it has been found that
in a soil where the N-value increases rapidly, for
example, the v'Tt value also shows the same tendency
(Hashimoto et al., 1992).

934 Hashimoto
Depth
measuring N-Value VTt
wire -....e 10 30 50
I
1.0.2.0

~
Power unit
Power unit (ECU) .c:: I
i
for ~
0.
I

earth auger QJ
f\
0
10 < Top end of

EJa ['~
bearing

[gl
stratum
................
o 9.
L::.J
o
0

~
Q Q

20
1
'" f-- :--

~~~~~r----Penetration
I' ,

depth senser
Fig. 3 Penetration Control Fig. 4 Typical Example of
System (Doctor system) Penetration Control Data

3. Scaled-Model Experiments
3.1 Outline of Experiment
Drill piles with a length, of 1500 rom, outside
diameter of 114 rom, and wall thickness of 4.5 mm were
used as test piles in experiments with and without
ribs, and using different installation methods (rotary
penetration pile/press-in pile) as test parameters. A
rod of 3.2 mm diameter was welded spirally on both the
inner and outer surface of the pile toe at a pitch of
68 mm, vertical length of 128 mm for the outside and
640 mm for the inside. Cutting tools of the same
thickness as the rib were installed on two points of
the outer periphery of the pile tip.
Kashima quartz sand No. 6 was hot-air dried
(moisture content 0.5% or less) and then used for the
model ground. The characteristics of the soil used are
shown in Table 1. The model ground was formed by
vibration compaction of successive layers of 10 em
thickness in a round soil tank (Hashimoto, et al.,
1989) with an inside diameter of 968 rom and height of
1.8 m. The relative density of the model ground was
between 85% and 94% in all tests. A colored layer of
about 1 mm thickness was placed over each 10 em thick
layer to monitor ground movements during pile
penetration.

935 Hashimoto
Table 1 Characteristics of Soil used
in Scaled-model Experiments
60' diameter of soil D60 0.16 rrun
particle
Uniformity coefficient Uc 1.6
Maximum void ratio . ,t max 92%
Minimum void ratio tmin 61%
Specific gravity of soil GS 2.679
particle

The setting depth was five times the pile diameter


in all tests. For the rotary penetration pile, the
push-in force (max. 25.5 kN) was gradually increased
with the penetration depth by a hydraulic jack while
rotating the pile at 4 rpm using a hydraulic motor.
For the press-in pile, static penetration was made at 1
ern/min measured at the ground surface by a hydraulic
jack. Dur ing rotary pile penetration, a surcharge of
0.098 MFa was applied by air pressure through a 32 rnm
thick steel plate.
3.2 Penetration/Bearing Capacity Mechanism
After rotary penetration of the model pile wi th
inner and outer ribs was completed, the ground around
the pile was carefully removed and the surface
observed. Dense soil particles adhered to the entire
outer surface of the pile, as shown in Figure 5. The
colored layer in the ground was distributed in the
upper adhered layer. This shows that the soil
particles near the pile were displaced upward along the
outer ribs due to the rotary penetration of the pile.
Figure 6 shows the compacted condition of the
ground around the pile after installation compared with
the or ig inal ground. The compact ion was evaluated by
reading the load required to make a 5 rnm penetration
(cone index) using a pocket cone penetrometer with the
end modified into a flat head (12.8 rnm diameter).
Before each measurement at the desired depth, the upper
sand layer overlying the measurement surface was'
removed. The measured value, therefore, includes the
effect of stress relaxation due to the removal of the
upper load soil.
The rotary penetration pile achieved compaction
over the entire pile length especially on the
peripheral surface of pile. The peripheral ground
initially disturbed due to the rotary penetration was
compacted by the soil particles displaced upward from
the outer rib, and the radial earth pressure around the
pile became. higher than the initial value.

936 Hashimoto
Pile outer surface
Original ground
Adhered layer on outer
GL surface of pile
(Ground

Colored soil in
,S1'-----o-+-adhered layer
N
N

-400
Top end of---J.....----;:.,,..1 GL. -460
outer rib ~~~----~

-500 Colored layer in


original ground
-570
Pile tip ~--~
mm)

Fig. 5 Adhering of Soil to Outer Periphery of


Pile Due to Rotary Penetration

114iT1111
Press-in Distance from pile
pile circ umferential r- ,- .
circumferential pile
n
Distance from pile penetration
Rot~ry

sur face (rnrn) surface (nun)


50 30 10 0 o 10 30 50
o I I
o
.. 2.5 5 . ,
2. ,,( Or .l9.ln.a1_ \ I .1. r
1\.,
aJ II aJ
~ 100 100 ~
ground)
~ (Or iginal 1
~ rOUl1d)
...::J \ I \Lj

12.5~ "
9 1\
III

\ 1\ \ 1"'- [\'\
\
"C \
C "C
C

...g 300 \
\ 25 \
300
...g
, \
\ \ \
0' 1
50 \ 0'
E
o
: \
E
...
k:V; L7 ...o
I
I I

/
\Lj

I
/
/ ~ I
I
\Lj

~ 500 ~p 500 ~
c I ." V V
V I

-- .
~ 750 .- ... ~
". c:
~
~ :-;;:
20 - ~
III " I
..... ..... ... "-.... ~; 10I III
.....
Cl

700 I 700
-- Cl

~ The numerical values in the figure correspond to


ratio of cone index of compancted ground to that
of original ground.
Fig. 6 Comparison of the Effect of Pile Installation
Methods on Compaction of Peripheral Ground

937 Hashimoto
The increase in cone index of the ground at the pile
toe was not so great for the rotary penetration pile as
for the press-in pile. The experiments confirmed that
for the rotary penetration piles, penetration occurred
when applying pressure continuously on the ground under
the pile toe (Hashimoto et al., 1989), and compaction
based on the advanced pressure effect can be expected.

Figure 7 shows the comparative effect of inner


spiral ribs on the rise of the soil inside piles
installed by the rotary penetration method. In all
cases, the rate of rise of the soil inside the pile
decreased as the penetration proceeded. The decrease
in the rate of rise was remarkable for the pile with
inner ribs. As seen in previous test results
(Kobayashi, et al., 1982), the inner ribs increase the
shear resistance of the inner surface of the pile and
constrain the upward slippage of the soil plug,
promoting compaction of the soil inside the pile and
facilitating the strong blockade at the pile toe.

(Pile dia: 114 mm)


5
111 111
w I
..-I~
ooi 111
0. ~ 4~--+---+---+---I'-~-1
11l00i
"0'0
ooi
til 111
C:..-I 3 ~--+----h"':-......,J.~~"-.)==-'=.,.J
ooi ooi
0.
..-I ........
ooi 111
o
tIl~
0 21----+--:K~--+---+-__j

U-l11l
0..-1
ooi 1
J.J0.
.s:::
Ole
ooi 0
~ ~ O~_--.L_ _- L_ _L - _ - - L _ - - J
a 1 2 3 4 5
Penetration depth/pile diameter

Fig. 7 Comparison of the Effect of Inner


Spiral Ribs on the Rate of Rise of Soil
and the Height of Soil inside the Pile

4. Field Experiments

Field experiments using actual piles were conducted


to quantitatively evaluate the penetration/bearing
characteristics particular to the drill pile.

4.1 Survey on Peripheral Soil and Soil inside Pile

Figure 8 show the SPT (Standard Penetration Test)


N-values around the settled pile at distances of 30 cm
and 50 cm from the pile surface compared with the
values of the original ground before pile penetration.

Hashimoto
938
using a drill pile with outside diameter 406.4 rom and
wall thickness 16 mm, this test evaluated the
disturbance of the peripheral ground, particularly the
decrease in N-value, due to the pile penetration.
However, there was no decrease in the strength (N-
value) of the per ipheral ground and higher N-values
than the values for the or iginal ground were obtained
in the upper sand layer.
......
a Standard
.Cl penetration
::
"-
Cll test
..:::
.u ......
0; ..... N value
o.=.
Cl.l~
a
(JJ
-\7G.r. .... O.O
I //~II
1.3QlX c;, -

~Q
].60 ::;:::::::; \\
I
..g..1 - Soi 1 survey
pen etration
0---0 Soi 1 survey
before

after
"
...

, I
C)! pen etration (No.1)
.' . ~~ .'0 I
.. e:---c. Soi 1 survey after
~ I pen etration (No.2)
. )C
8.80
.B
.'
y- . Soi'l survey after
penetration Soil survey before
S' ~ llpenetration

llITl
.8 ~ !
12.SD=
14.00
.
~.~
./~.

r'~
:,

.:,.; ..
.. ';;
~
"-
:r.J,..
"-- _0
~
~ sao
300
5""
(Unit:
i rom)
~
16.50 r:<: ~
.. .: ~
\7G. r.. -17 .07

' .. t!
20.00 ~:=" ~

Fig. 8 Standard Penetration Test Results


for Ground Around Drill Pile
To invest iga te the condi tion of the soil plug, a
pile 508 rom in diameter was rotary penetrated into the
gravel bearing stratum then pulled out immediately and
cut in half. The formation of a very dense soil plug
consisting of gravel of the 'bearing stratum inside the
pile was confirmed.
4.2 Vertical Load ~est

Table 2 shows the specifications of 7 vertical load


tests conducted at different sites. The internal and
external rib conditions are same in all test piles.
The bearing stratum was sand (4 sites) or gravel (3
sites). Figure 9 shows the whole Po (pile head load) -
So (pile head settlement) curves.

Hashimoto
939
Table 2 Specification of Vertical Load Test Piles
pile Wall Type-of Ratio of depth
Test Setting into bearing
diameter thickness depth bearing stratum to the
No. (rom) (mm) (rom) stratum pile diameter
T-l 318.5 6.9 22.0 Sand 3.5
T-2 318.5 10.3 10.0 Sand 12.6
T-3 508.0 9.0 18.0 Sand 5.7
T-4 508.0 16.0 40.0 Sand 5.9
T-5 508.5 9.0 27.8 Gravel 4.5
T-6 400.0 16.0 17.1 Gravel 1.7
T-7 318.5 12.7 23.0 Gravel 8.2
pile head load Po (kN)
o 1000 3000 5000
o '(
~k
-..., o T-l (318. 5, sand)

20 " r'-o ~ ~ ~. - .:.' OT-2(310.5,sand)


Q
U)
1\ ~ ~ ~. .c.. T-3(fZ1508,sand)
.u 40 '\ 1 "\ ~ ~ T-4(500,sand)
r::: o T-S (500 ,gravel)
111
E
\ ~

X T-6 (400 ,gravel)


111
.-I 60 1\ 1\ 'c ~ ~

"V T-7 (3l8. 5, gravel)


.u
.u
111 \ 0
\~ 1\ ~
en
80 \\ \
'0
ra
111
\~ i\
~
..c 100 \
111
.-I
.~
\ I
0.
120
k
Fig. 9 Po (pile head load) - So (pile head
settlement) Curves

(1) Skin Friction


To evaluate the skin friction resistance
quantitatively, axial stain gages were attached to the
test piles T-3, T-4, T-5 and T-6. Figures 10 and 11
show the changes in skin friction measured at several
positions along the pile length, against the relative
displacement of the pile and ground, that is the
settlement of each part of the pile in sandy soil and
cohesive soil stratum, respectively. These figures
indicate that the maximum skin friction was
demonstrated at almost all locations when the relative
displacement of the pile and peripheral ground reaches
10 to 30 mm. Figures 12 and 13 show the relationship
between the original ground strength (N-value or qu)
and skin fr iction of the piles in the sandy soil and

Hashimoto
940
cohesive soil stratum, respectively, when the pile toe
settlement reaches 10% of the pile diameter (d) (or at
maximum load for piles which did not reach 10%d
settlement).
N
e
.-
~ 20 0-----.--...--...,.-.-.---.,..--.
~ r ~u
,., 16 01+-1-~-+-~=::::?=-+-- a
t ~
c 12 0 -4-~~--+--+--f---+---! g 12 0
.~ rJ' .... ~J\ l.o-

....t: 801~~ v
-~
~ 8 0 .g.~'----~=+--!--:--+---1
.... r
\.4

~c 4~~'~~~~~~~~*!
\L.4 40.~~
-t-
c::
.... W''' 't-- L...-- i---

~ 0 40 120 ~
en.
a 80 40 80 120
en
Relative displacement Si Relative displacement Si
(mm) (mm)

Fig. 10 Changes in Skin Friction Fig. 11 Changes in Skin


with Relative Displacement Friction with Relative
of Pile and Ground Displace~ent of Pile and
(sandy soil) Ground (cohesive soil)

a ~ 10 a
I
r I
z
.:.: .-
a a ... 80
\. I-

cl f I c
....a 12 a
0- r-
....o
,.;
60
't=49
,.;
u ':J}./ V ....u ~
-
.~ 80 I~ . \.4 40
\L.4
..-- - - ~.~y I.Lj
~I
-- f--V""" /to;
~

c c
.... 40
~
en
.'- ./
1-- ......
./ .... 20
.:.:
en ~
~

a .........r--
a a f7
10 20 30 40 50 a 50 100 150 200 250
N-value Onconfined compressive
strength (kN/m 2 )
Ordinary length
o Spiral rib length Ordinary length

Fig. 12 Relationshi~ between Fig. 13 Relationshi~ between


Skin Friction and Ground Skin Friction and Ground
N-value at 10%d Settle- N-value at lO%d Settle-
ment (sandy soil) ment (cohesive soil)

941 Hashimoto
The skin friction values at lO%d settlement were
distributed around N/O.51 kN/m 2 (N/5 tf/m 2 ) or more in
the sandy stratum for both ordinary and spiral rib
length of the piles, and qu/2kN/m2 of more in the
cohesive soil stratum for the ordinary length of the
piles.

(2) Point Bearing Capacity

In an open-end pile, fr ict ion occurs between the


soil inside the pile and the inner wall of the pile
when loaded, this inner friction affects the vertical
strains at the pile end portion and the friction on the
ou ter sur face. Therefore, it is diff icul t to evaluate
the outer fr iction (skin friction) and inner friction
(part of the point bearing capacity) separately by
strain gages attached to the pile. The dr ill pile is
more complicated because the spiral ribs are attached
to the inner and outer surfaces of the pile toe.
Therefore, a design pile toe at the position of 3d
upward from the real pile toe was assumed to evaluate
the point bear ing capaci ty (Fig. 15 (a) ) In this way,
the outer frictional force occurring under the design
pile toe elevation is involved in the point bearing
capacity.

1he value of 3d corresponds to the minimum


installation length of the inner spiral ribs which
causes the soil plugging.

Figure 14 shows the relationship between the load


(Qp) and settlement (Sp) obtained at. the design pile
toe. The bear ing capaci ty coefficient (a) is obtained
by dividing the design pile toe resistance by the
average N-value and pile toe plug cross section. In
this figure the design pile toe of test pile T-6, in
which the setting depth into the bearing stratum did
not reach 3d, was taken as corresponding to the top of
the bearing stratum. Figure 14 shows that the bearing
capacity coefficient (a) was 245 or more when the
settlement of the design pile toe reached 10%d, and 300
or more at the ultimate bearing capacity stage when the
settlement became very large in comparison with the
increase in load.

4.3 Method of Calculating Vertical Bearing Capacity

Formula (1) was derived from the scaled model tests


and the field load tests for designing the vertical
bea ring capaci ty. Figure 15 schemat ically shows the
design concept of the bearing capaci ty of the dr ill
pile.

Ns qu
Ru = aNAp + (----LS + ---LC) (1 )
0.51 2

Hashimoto
942
where

Ru: -ul t ima te bear ing capaci ty, (kN)


a: bearing capacity coefficient of design pile
toe (end bearing capacity evaluation position)
a=245 when Lb/d~3
245 Lb
= --- (---) when Lb/d<3
3 d

Lb: setting depth into bearing stratum, (m)


d: pile diameter, (m)
N: average value of N values of ground between
ld downward and 4d upward from bottom end of
pile, pr9vided that the maximum N value is
100 and N'5.60
Ap: plug cross section of pile toe, (m 2 )
Ls: length in contact with sandy soil, (m)
Ns: average N-value of sandy soil layer of
peripheral ground over the length Ls, provided
Ns'5.50
Le: length in contact with cohesive soil, (m)
qu: average uniaxial compressive strength of
cohesive soil layer of peripheral ground over
the length Le, (kN/m 2 )
: circumference of pile, (m)

The design pile length is as follows:


La - 3d, when Lb/d~3
La - Lb, when Lb/d<3
where, La: length of pile in contact with ground, (m).
Bearing capacity coefficient of
design pile toe a
o 200 400 600
QJ 0 -...,
0
.u
~ ~~ --J.
QJ 5 ~ -I ~
~ 246.0 ~
~~~
~
.,-l
0..
383.2
v-r-
- ,...--

10
c:: 24W~
0'-
.,-l rIP "\ \ \
til- lS
QJ
'0'0 \
\l-j~

o QJ 20
.u
.uQJ
c:e 25
QJ rQ
8,-l 6 T-3(500,sand)
QJ'O
~
30
X T-4(508,sand)
.uQJ
.u~
QJ.,-l
o T-5(500,gravel)
til 0..
35
X T-6(400,gravel)

Fig. 14 Qp (design pile toe resistance)/


(end plugged area x N) - Sp (design
pile toe settlement)/(pile diameter) Curves

Hashimoto
943
O.L. O.L.
J.l
J.l ,;:: U
U
III ",,,,
QlJ.l III
J.l
I:: ,;::
J.l e I:: 0
I:: e<dal J.l
0
U ...
0"''-;'
III
U

c: 1ll
QI '"
'", e
=
c:
....
III::'
-
<II
't: C:.-;
.... 0...
"" .... g,
.... ..J

III
....
e
o
al
Ill""
...... al
J.l ....

........
... J.ll:: Top end of ...."0 uc: ....
....
"'01 c;C:
~ ~ g,
g,'tI
... I:: e ....
~
<II
L
l:Iearing stratum
.. ::l
0 .... J.l =
~O"
~ .... "'111 III 01
">:>'tI "01 :I ...
0 0
""
tIlQl_
---- ,;:: 1::.-;
<II

~}:;~t~~~ion ~2
,;::01 . J.l .... III
QI
",,;:: ~>"O
J.l,;:: <I1Q1_
01 oJ "0 C:J.l
" III...
~ ....
..J
"~ ~
...."0
'range
.;:; (average
value)
Point: l:Iearing capacity Point bearing capacit:y
, evaluation position evaluation posit:ion
I ..
I de~;qnipile toe) . design pile toe)

I(a) Whee. ~bld-= 3: a 245 (bl When Lb/d < 3: a 24?/J (:~/d)
D~sig~ pile length~a-3d Design. pile lenqth~a-Lb

Fig. 15 Interpretation of Design Bearing Capacity

/
I
k:'
- II
6000
..c
0\ ~I 7
:j
a
I..; CI]
5000
I. /
~
17
..c I1l
;;/
~ e
p., ~CJ ,- /
~
:>t
~
.-1
(J
RJ
0..-
-
:z 4000
~
':'16-'lJ/9~CJ
II"Y /
v,+ /

RJ 'lJr-,~ ~~
(J

C'l I1l
~
CI]
JOOO
~~/1/
x 1/
C ~ /
.-1 6
1
I..; "'0
COlO' :/
RJ n2 ./'oJ
I1la
.c ..... 2000 o T - 1( J 18.5 / sa n'd )
~
I1l .....
",
'I
/ 1/ OT-2( <PJl8.S,sandJ
", (J
e
.-1
.....
1000
~
~ -r; / c. T- J(
&T-tl(
Ip 508, sand)
'p50a / sand)
~
.....
I..;
I1l t/ o
T - 5( ,p 5 0 a ,q t" a v e .1 )
XT-6( ,ptlOO.(Jt"<lvell
0 > .pJlo.S,gravel~
a L/ V'T-7(

o 1000 2000 ')000 ~OOOSOOO 6000


Ultimate design bearing capacity (kN)
using formula (1), Pcal
Fig. 16 Comparison between Measured and Calculated
Values of Bearing Capacity

944 Hashimoto
Figure 16 compares the measured value and
calculated value of the bearing capacity using formula
(1) for the 7 vertical load tests shown in Fig. 9. As
a mean of the 7 tests, the measured value is about 1.35
times the calculated value. Formula ( 1), therefore,
yields conservative bearing capacity values for safety.
Comparing these results with the evaluation of the
bearing capacity for conventional low-noise/low-
vibration piles, such as cast-in-place pile reinforced
with cement milk or bored concrete pile, shows that the
vertical bearing capacity of the drill pile is
equivalent to or higher than that of conventional low-
noise/low-vibration piles.

5. Conclusion

The results of the scaled-model experiments and


actual field tests proved that the drill pile compacts
the peripheral ground during pile rotation, especially
sandy soil. The vertical bearing capacity is
equivalent to or higher than that of conventional low-
noise/low-vibration piles.
The drill pile method has great potential foi
application to installation of pile foundations
particularly in urban preas where low-noise, low-
vibrat'ion piling me~hods and a clean piling environment
are desirable or necessary.

References
Hashimoto, M., Nishizawa, S., Sato, S., Sakurai, Y.,
Hashimoto, O. and Takahashi, C. (1992): "Analysis of
Construction Technique and Penetration Characteristics
of Low-noise and Low-vibration Steel Pipe Pile,
"Kawasak i Steel Technical Report. Vol. 24, No.3, pp.
33-40 (in Japanese).
Hashimoto, 0., Kaneko, T., Tateno, J. and Takahashi, C.
(1989): "Model test on bearing capacity of rotary
penetration steel pipe pile," Proc. of Annual
Conference of AIJ. pp. 1299-1300 (in Japanese).
JSSMFE. (1992): "N-value and C.0," p. 80 (in Japanese).
Kobayashi, Y. and Yamakawa, S. (1982): "Influence
fac tors on penetration resistance of steel pi pe pi Ie, "
Proc. of 37 th Annual Conference of JSCE. I II-250, pp.
497-498 (in Japanese).
Yamagata, K. (1980: "Some problems on the vertical
bearing capacity of the large diameter piles," Tsuchi-
to-Kiso, . JSSMFE. Vol. 28, No. 11, pp. 5-11 (in
Japanese).

945 Hashimoto
RESEARCHES INTO THE BEHAVIOR OF
HIGH CAPACITY PIN PILESSM

Dr. D. A. Bruce 1 , J.R. Woloslck2 , and A.L. Rechenmacher 3

ABSTRACT

Minipiles are known by many names, but are generically small


diameter, cast-in-place bored piles. Although they have been used
in Europe for over 40 years, in the United States, where they are
commonly referred to as Pin Piles sm they have became a popular
I

choice for underpinning only during the last 10 years. The paper
describes fundamental laboratory and field researches recently
conducted to better understand load transfer mechanisms. This
work has led to the development of the Elastic Ratio concept which
is now proving extremely useful in analyzing and predicting pile
performance, and in particular the phenomenon of progressive
debonding with increasing load. Pin Piles are becoming more
popular in applications for seismic rehabilitation of bridges, and
this paper focuses on this aspect via recent case histories.

1. INTRODUCTION

The last decade has seen a significant growth in the use of


Pin Piles sm in the United States. Generically, these piles may be
classified as small diameter, bored, cast-in-place elements, and
they owe their origins to developments by specialty contractors in
Italy over 40 years ago. As a result of the kind of research and
development activities described below, their allowable load range
has been extended from 50-100 kips 220-450 kN to up to 300 kips
(1340 kN) while special test piles have yielded capacities of over
1000 kips (4450 kN) in favorable conditions.

1 Director: Business Development and Technology, Nicholson Construction Company,


P.O. Box 98, Bridgeville, PA 15017 (412) 221-4500
2 Regional Marketing Engineer, Nicholson Construction Company, 4070 Nine
McFarland Drive, Alpharetta, GA 30201 (404) 442-1801
3 Geotechnical Engineer. Nicholson Construction Company, P.O. Box 98, Bridgeville,
PA 15017 (412) 221-4500
D. A. Bruce et al.
946
Initially, these advances were made as a result of the careful
execution and analysis of full scale field test programs, and such
experiences have been widely published (References 1-8).
However, within the last few years it has become apparent that
extra dimensions of research efforts were necessary to explore
and understand fundamental aspects of Pin Pile behavior, and
especially those related to the performance of the component
materials in resisting and transferring high axial loads.

This research was funded by Nicholson Construction


Company and conducted jointly with the University of Pittsburgh.
The laboratory work comprised three major phases:

Phase 1, where single, grout-filled steel casings,


simulating the upper (free) section of a typical high
capacity Pin Pile, were compressed to failure, to establish
their composite strength and elasticity characteristics.

Phase 2, as Phase1 but using connected casing sections with


threaded ends.

Phase 3, where similar tests were conducted on internally


reinforced grout columns simulating the lower (bonded)
section.

In parallel, the opportunity was taken to run full scale field


tests at two major contemporary underpinning projects, one at a
petrochemical facility near Mobile, AL, (Reference 7), the other at
a grain silo complex near Port Vancouver, WA (Reference 9). The
latter case history is summarized in this paper.

2. CONSTRUCTION AND CLASSIFICATION

Pin Piles are most commonly used to underpin existing


structures settling, or liable to settle, as a result of changes in
loading or foundation conditions. Construction methods have
therefore been developed to accommodate the gamut of ground and
structure types, while causing the minimum of damage to either,
or the environment. Also Pin Piles operate principally in side
shear and so these techniques have been honed to enhance bond
capacity at the grout/soil interface.

947 Bruce, Wolosick, Rechenmacher


The successive steps of pile construction are well known and
documented (e.g. References 1-9). They comprise drilling and
casing, placing of reinforcement and tremie grouting, and,
typically, pressure grouting of the bond zone.

In most countries, the temporary casing is fully extracted


(as any auger must always be) during the pressure grouting
process. However, in the United States, it has been proved that by
leaving the casing in place through the zones above the bond length,
the Pin Pile performance is greatly enhanced, both vertically and
laterally. This option also prevents wasteful travel of grout into
often permeable upper horizons, and provides excellent corrosion
protection to the interior of the pile in what is usually the most
vulnerable zone. A useful subclassification of Nicholson Pin Pile
types, based on the geology of the founding zone, and the internal
composition (and the mode of action) of the pile is provided in
Figure 1:

Type S1 - A steel casing is rotated into the soil using water


to externally flush the cuttings up around the outside of the pipe.
A neat cement grout is tremied from the bottom of the hole to
displace the water. The reinforcing element is then placed to the
bottom of the hole. As the casing is withdrawn over the length of
the bond zone, additional grout is pumped under excess pressure.
The casing is then seated back into the grouted bond zone. In
granular soils, a certain amount of permeation and replacement of
loosened soils takes place. In cohesive soils, some lateral
displacement or localized improvement of the soil around the bond
zone is accomplished with the pressure grouting. Postgrouting
may be used later to further enhance soil/grout bond (References
10 and 11).

Type S2 - The pile is installed in the same fashion as the S1


except that:
the centralized reinforcing element is not needed;
the steel casing is installed to the full length of
the bond zone after pressure grouting is completed.
post-grouting is not feasible with this type of
installation.

948 Bruce, Wolosick, Rechenmacher


GROUT~~OPTIONAL
FILLED'" FULL LENGTH
PIPE REINFORCING

REINFORCING GROUT
STEEL UPPER FILLEC
SO ILS PIPE

SUITABLE
LOWER SOILS
w
Z
o
N

o CENTRALIZER ~

~'----I,.J.../ - - - - U
TYPE S1 TYPE S2

PIN PILE TYPES IN SOIL

~ GROUT
GROUT
FILLED
PIPE FILLED
OPTI ONAL PIPE
FULL LENGTH
REINFORCING

REINFORCING
STEEL
SOIL

o
ROCK z
Ow
CD Z
CENTRALIZER ..:0
UN
a::
TYPE R1 TYPE R2

PIN PILE TYPES IN ROCK

Figure 1 Generic Pin Pile Configurations in Soil and Rock

949
Type R1 - This pile uses the same technique for advancing
the steel casing as Type S1, except that the depth of penetration is
limited to the top of rock. Once the pipe is seated into rock, a
smaller diameter drill string is advanced through its center to
drill the rock bond zone. Neat cement grout is then tremied from
the bottom, and a reinforcing element is placed in the rock bond
zone to complete the installation. A minimum transfer length is
required for the reinforcement to develop bond inside the casing
(typically 5 to 10 feet) (1.5 to 3.0m).

Type R2 This type differs from the R1 pile in that it uses a


full length steel casing. The possible need for centralized
reinforcement is dictated by internal pile capacity. In order to
advance through both the overburden and the rock, a permanent
drill bit is used on the end of the casing with a diameter somewhat
greater than that of the casing. There are grout ports in the bit.
At the desired final depth, grout is tremied from the bottom, and
additional grout is pumped to ensure full grouting of the rock bond
zone.

3. LABORATORY RESEARCH
Phase 1
Testing of composite members has been conducted for decades,
worldwide, and the results of 68 tests of axially loaded concrete
filled tubes were addressed in a Steel Structures Research Council
(SSRC) report. (Reference 12). Actual steel yield stress varied
from 38 to 88 ksi, (260 to 610 MPa), and concrete compressive
strengths from 2.9 to 9.6 ksi (20 to 66 MPa). A table of data
comparing these test loads with the theoretical allowable loads,
based on the proposed modifications to the AISC allowable stress
equations, was prepared to give an indication of actual safety
factors. These ratios varied from 1.28 to 3.68, average 2.26,
standard deviation 0.45 and a coefficient of variation of 20%.

The tests in Phase 1 were run with composite tubular


members of uniform, high strength steel (nominal 80 ksi) (550
MPa) and grout (minimum 4 ksi) (28 MPa), such as would comprise
certain sections of Pin Piles. (Reference 13). Specimen lengths
were selected to provide a set of slenderness ratios that would be
consistent with previous experiments. Details are summarized in
Table 1. These data proved consistent with the earlier tabulated
work of SSRC.
950 Bruce, Wolosick, Rechenmacher
Spec 1/ Dia. (in) Length (in) Wall (in) Max Load (kips) Elastic Ratio'
1 7 36 0.502 1181 (Equivalent to 0.30
lor 10' length)

2 7 36 0.502 1242 . (Equivalent to 0.30


lor 10' length)

3 7 120 0.502 969 0.32

4 5.5 36 0.363 685 (Equivalent to 0.70


lor 10' length)

5 5.5 36 0.363 584 (Equivalent to 0.54


for 10' length)

6 5.5 120 0.363 450 0.53

Table 1 Summary 01 Phase I laboratory Tests (Single Casing lengths)


1 in. = 25.4 mm; 1 kip = 4.45 kN; 1 ft. = 0.305 m
Calculated for each specimen a compression (in thousandths of an inch) divided
(0 by load (in kips), in the elastic respon~e field.
(}l

Spec # Dia. (in) length\ (In) Wall (in) Max Load (kips) Elastic Ratio *
1 7 B 36 0.502 1300 0.49

2 7 U 36 0.502 1160 0.44

3 5.5 B 36 0.363 630 1.52

4 5.5 U 36 0.363 545 1.06

~ Summary of Phase 2 Laboratory Tests (Coupled Casing lengths)

B = Banded
U = Unbanded
10 fl. equivalent length
1 in. = 25.4 mm; 1 kip = 4.45 kN
Separate tests on material properties confirmed the
specified minimum yield strengths to be 86 ksi (590 MPa) for the 7
inch (178 mm) dia. casing, and 99 ksi (680 MPa) for the 5.5 inch
(140 mm) dia. casing. Grout strengths (28 day compressive)
averaged about 5.6 ksi (39 MPa). Each column responded similarly
throughout the loading range - initial local yielding at the ends
followed by gradual bending. No evidence of buckling was observed.
The shorter specimens exhibited a linear load/deflection
relationship to about 75% maximum load, while the longer casings
were linear almost to maximum load.

Of particular interest in Table 1 is the newly coined term


Elastic Ratio (ER) for each configuration. ER is calculated by
dividing the resultant displacement (in thousandths of an inch) by
the applied load (in kips), and is therefore a simple indicator of the
effective composite elastic modulus of the grout filled casing.
This directly determined value can then be used to ascertain the
seat of load transfer during the cyclic loading of Pin Piles, as
demonstrated below.

Phase 2

The typical Pin Pile casing joint is formed by mating the


male and female ends of successive lengths of casing. This joint
is typically flush, with little or no resulting space between
sections, and its strength is dependent on many factors including
material yield strength, thread pitCh, root size, length of splice,
shoulder contact and the confining effects of pipe and grout. Tests
were conducted (in tension also, but not detailed herein) on the
typical Nicholson casing thread, with 36 inch (91 em) long samples
with and without external banding reinforcement around the
female end. (Table 2).

A comparison of the data of Table 1 (single casing) and Table


2 (coupled casings) shows that for stub columns (simulating the
fully braced pile configuration), no significant difference exists in
the magnitudes of the ultimate loads or the ultimate failure
modes. However, the ER values recorded for the Phase 2 tests
were higher for two main reasons:
a) "Slop" in joints, creating higher total
displacements, coupled with

952 Bruce, Wolosick, Rechenmacher


b) the relatively short test lengths being more
sensitive to these displacements.

The tension tests showed the joints to have about 60% less
capacity than in compression. Also, the failure mode in tension
was explosive (i.e. the thread experienced sudden failure).

Phase 3

Clearly the structural capacity of Pin Piles with full length


casing (82, R2) can be designed conservatively by using the
composite strength of the grout filled casing, ignoring the
confining contribution of the annular grout. However, if the grout
is neglected in the design of an internally reinforced bond zone
(81, R1), the resultant design would be significantly over-
conservative. A series of tests were therefore run on such
simulated bond zone configurations, as detailed in Table 3.

Each specimen was cast and tested in a tubular plastic


mould, the lateral confining properties of which approximated that
of a medium dense sand. The specimens were all 36 inches (91 cm)
long and 10 3/4 inches (27.3 cm) in diameter. The length was
selected to create a stub column, so allowing slenderness effects
to be ignored, while the diameter reflected a typical effective
pressure grouted bond zone diameter in situ.

Linear behavior was noted over an average of 84% of the


ultimate capacity, and failure was characterized by axial crushing.
The relationships between ultimate load, and ER, and the cross
sectional area of steel in the sample are shown in Figures 2 and 3
respectively. The benefit of the simulated spiral reinforcement is
clearly demonstrated - an improvement in capacity (37%) over the
comparable specimen without a confining cage.

4. FIELD RESEARCH - UNITED GRAIN SITE

Concurrent with this laboratory research, Nicholson


Construction was the design-build contractor on two significant
Pin Pile projects. The larger was at an operational grain export
facility on the Columbia River at Vancouver, Washington, where
certain major structures were threatened by settlement as a
result of deterioration of the original 4050 timber piles driven in
1934-1939. (Reference 14). Prior to installing the 840
953 Bruce, Wolosick, Rechenmacher
Equivalent 10 It.
Spec It Reinforcement Configuration Cross Section Sleel (in 2 ) Max Load (kips) Elastic Ratio

lA None . Plain Grout zero


23 212 2.20
18 None Plain Grout zero 214

31~ 30B
2A 1 " 10 123
1.66
28 1 1/ 10 123 300

3A 1 " 14 240
390J 365 1.53
38 1 " 14 240 340

4A 1 " 18 397
49:) ~ 70 1.45
48 1 II 18 397 450

(0 SA 2 1/ 10 2'45
U1 3 6 ] 393 1.68
.Jlo.
5C 2 " 10 245 418

J
6A 2 " 1'1 "B 1 56
572 0.71
68 2 " 14 4 81 580

7A 1 /I 14 + Simulated Spiral cage 240+


50:} 500 1.53
78 1 " 14 + Simulated Spiral cage 240+ 500

8A 11 ea. 0.6" strand" 239


39} 355 2.20' .
B8 11 ea. 0.6" strand' 2'39 315

~ Summary 01 Phase 3 Laboratory Tests (Bond Zones)

Strand Iy = 270 ksi (1860 MPa); all other rebar 60 ksi (410 MPa).
Strand columns exhibiled similar stiffness as plain grout columns.
Strand stiHness in compression is suspect in these tests.
1 in 2= 645 mm 2 ; 1 kip =
4.45kN; 1 in. =
25.4 mm
AV~HI\CE--iJI1ANCE

IlE:AL! sri C
SSO E:~rE:Il/.IE:NTl\l
[NVE:lOrE:
I -ttl oC

sr InAl
sao

'50

0
'"
0
.J
~OO

'"
....:.
-'
JSO
::>

II S TRANOS
JOO
(270 KSII

2S

I 2 J S
CROSS-SE:CTIONAl ARE:A OF RE:INFORCINC STEEL (IN Z )

Figure 2 Ultimate Compressive Load vs. Reinforcing Steel Area. Phase 3 Tests.
All steel = 60 ksi (410 MPa) except for strand.

PLAIN
x_ STRANO (270 K511

2.0

..."-
u

..... l. S
o
EXPE:RI~NTAI.

...c:
o ENvt:LOPE:

.......c: '.0
...:::
..J
-<
>
::>
...
o

O. S

o -t----,,---,---r---.,..--,
o 2} S
CRosS-sECflONAL AREA 01' AEIN~QACINC STEE:L fIN 2 1

Figure 3 Equivalent ER vs. Reinforcing Steel Area, Phase 3 Tests.


955
replacement Pin Piles of 300 kip (1340 kN) service load, about half
of which were to be located in the cramped basements of the three
silo structures, an extensive test program was conducted,
involving six full-scale special test piles.

The design foresaw each pile to be drilled with 7 inch (178


mm) casing to a depth of approximately 70 feet (21 m) from grade
and so a minimum of 30 feet (9 m) into a very dense gravelly and
cobbly bed. The upper portion was to be reinforced by the casing,
with the lower pressure grouted portion reinforced by a central
reinforcing bar, in a standard S1 type configuration. (Figure 1).
The Specification called for an underpinning system to ensure
additional differential settlements of less than 1.5 inches (38 mm)
in 100 feet (30 m) and additional uniform settlements of less than
6 inches (152 mm).

The test program required the successful loading of three


piles to 200% service load, held for a minimum 12 hour period. The
service load was considered to be 300 kips(1340 kN) for the test
program, although final calculated individual pile loads were
slightly lower, depending on location. These initial three piles
(TP1-3) all reached the 600 kips (2670 kN) maximum but all
exhibited what appeared to be explosive internal (structural)
failure prior to the end of the hold period. As a result, and after
structural adjustments, a second group of three piles successfully
passed the test, and subsequently attained ultimate loads of up to
750 kips (3340 kN). These piles established the production pile
structural detailing and criteria for minimum embedments into the
bearing stratum. Test loads were applied in cycles of increasing
load, so permitting the partition of total pile settlements into
elastic and permanent displacements at each maximum load
attained. The elastic component therefore permitted the
calculation of the Elastic Ratio for each load cycle maximum.

Summaries of the soil strata encountered, and the individual


pile installation details are provided in Tables 4 and 5
respectively. Table 6 summarizes the creep and failure behavior.

Table 7 shows the progressive increases in ER with


increasing load for each pile. It was always recognized that this
progressive increase was indicative of progressive debonding: if
debonding were not occurring (i.e. if the pile were acting as a strut
with fixed ends) then the ER would be constant, since deflection
956 Bruce, Wolosick, Rechenmacher
Urrer Length Soil Strala Bond Length Soil Strata

Tesl Pile Sand r-ill Sill Medium Medium Very Very


/I Dense Dense Dense Dense
Sand Sand Sand Gravels
(fl) ( ft) (fl) (ft) ( f l) (ft)

T P-1 22 1G 5 1 :3 12 5
T P-2 20 15 5 7 9 14
TP-3 10 2 1 8 4 14 12
T P-4 10 2 1 G 6 14 5
T P-5 10 2 1 G 6 14 5
TP-6 10 2 1 6 6 14 5

Table 4 Soil Strata Thickness Encountered. United Grain Project, WA


1 ft = 0.305 m

Test Inslalla- Total Bood Casing Casing Rebar


Pile # lion Pile Length Insertion Length Rebar Length
Order Depth Into Bond From Size
Zone ~ Grade (f t)
( f t) ( f t) ( f t) (ft)
T P-1 2 73 30 5 48 # 14 30
TP-2 1 70 30 5 45 #14 30
TP-3 3 69 30 10 49 It 1 4 32
TP -4 4 62 25 10 47 It 1 8 27
TP-S 5 62 25 10 47 1/ 1 8 27
TP -6 6 62 25 10 47 # 18 27

Table 5 Pile Installation Details, United Grain Project, WA


1 ft =
0.305 m

957
' - - , Creep I! 525 I-Jp Creep @ 600 kip Load Creep @ 675 l'Jp Ha;c Test Hold .,.---- ,
Lood Load Load Duratlon
PUC~ 0-10 m1Jl Io--JO 0-10 mln 10-100 100-240 240-720 0-10 m1Jl 10-100 Altalned@Hax 'a..l.Iua
A mln min min mln min Load OeSC:1;>~o:;
. _~L ~ t!!!.)
.02~----- ._--~----- ~lf.illosl'Je D~o;> I
(In) (In) (In) (In In _
TP-I .021 I .010 .OJI I .010 I 600 I-Jps Approlf.
270 r.I1n to Joo YJ;> s
TP 2'.., .. '.02'2""'- '-.0'i6I" . :.:=---.:- "'1""-:-':':::-- _.-- - - - .. - _ . _ - - - .. -.... -- --1-. _ _ ..... - - - . _ - _ - ...
600 YJps J m1n E:l:ploslve O::op
_

to 355 KIps
Tr:r-\-'-:o2Q' -r-:oi2"-r" .026 -''':Oii'';.-I-- _u.. I ----:::-1--===---1--:::::-/ 600 YJps 4S r.I.In E:l:pioslve 0:-0;>
to H1 Idps
TP-1 , .. .oi.j .. l---:oJO- " ":-oi2"T "':o-ji- ../-. ,oli-/'--:Oil'-'/' '-:025-1"':059 1750k.lpsl-4 mln'- . E::l:ploSi~e Drop
to J72 kJps
TP:S' . .oii Mj' .... -r .",010""'" -:0'7i . .:026-I--:oii-r--":0S4 - r . ,070" r7S0k.lps rii"omin . pi~nqtnq 'D~O?
to 534 kJps
TP:6""I"-:oH-I -:Oi)""i-
U

.032 I .036-,:'1---:010'1-,038 -1.. ~()3SI-:-06i.-1 675 k.lps "iSiTi:iilTi~ploslve D:-op


to 150 k.lps

(0 Table 6 Test Pile Displacement Creep and Failure Behavior, United Grain Project, WA
(fI
OJ
1 kip = 4.45 kN; 1 in. = 25.4 mm

- YoO-k.lpl:'Oa"dr--4SO kip Load r-sis 'kJ'PL,'Oad-' -6oOk:iPS...... T---7~eLoad- .. --


?Ue
~
I Casing
Length
Total
PUe Elastlc Elastlc Elastlc Elastlc Elastlc Elastlc ElastJc Elastlc faJJure Elastlc ~Ias::lc
Length Ratlo Length RaUo Lenqth RaUo Length RaUo Length Load Rado Lenqth
(It) lCll ([t! ((tl (t) {(tl I!<.\p'sl -U.!L
~~~~ _~~ ... _.. _~. __. _.~~~_. _... ~~:~ :.:~._I-~:~I-.!~ -f-"~~'- _.~~~ .. _~::. ~~~.1- .. ~~8 .~.1 ~:s_.
- ~p~.~. - . ~: .. _..... ~:..._-~::~. .:::~_ _!.. ~t-.::~- ~~~ . ~~-. _~::7_. __4::~ _..!~_o-I-!.:~~ 't .. _4_2_.8_

-;~~i' :-. i ~:::-:;t~ :::::E:"~11


.;:~~ .....:~ . . .-..:;_. ~ -~:~:.- . ~~:~..
t:...tiL
.
==~t-::=~~;; -: ~:;;:;~..:~~~ ~:~f;~: ~.- ;F.::;;: . :~;-
.. .. . . .
~.:.~~.- _.~~:~_ ~:~;.- ~:~. ~;~. --:~~.- _.~~~- -l~~:~; .j .. ~::~.
'---

Table 7 Test Pile Elastic Ratio and Length, United Grain Project, WA
1 kip = 4.45 kN; 1 fl = 0.305 m
would be directly proportional to load. However, the question was
the relation between ER and effective pile length, and until the
Phase 1 laboratory tests this had not been satisfactorily resolved.
These tests showed that the ER for a 10 foot (3 m) length of grout
filled 7 inch (178 mm) casing was approximately 0.32. Thus, for a
recorded pile ER of, say 2.0, it can be calculated that the effective
elastic pile length would be (2.0/0.32) x 10 feet = 63 feet (19m).

The calculated free lengths in Table 7 are generous: no


allowance is made for the decreased ER value of the pile in the
casing/rebar overlap section. It is clear that explosive, structural
failure occurred when the load had been fully shed to within a few
feet of the bottom of the cased length. (It also is apparent that at
300 kips (1340 kN), the casing had debonded only about 30 feet
(9m) below the ground surface, highlighting the surprisingly high
load holding capability of the poorer upper soils. The table also
shows that the change from a #14 bar (TP1-3) to a #18 bar (TP 4-
6) gave the (centrally reinforced) bond zone an extra 70 - 150 kips
(310-670 kN) capacity to resist explosive bursting failure. This
compares with the same range identified in the Phase 3 Laboratory
Tests. (Table 3).

5. OVERVIEW OF LABORATORY AND UNITED GRAIN SITE


TESTS

The laboratory tests have provided the key to determining


effective debonding lengths in high capacity Pin Piles: the
breakthrough is the Elastic Ratio concept. The Phase 3 Tests, on
simulated bond zones, clearly demonstrated the range of capacities
which can be expected, and this range was confirmed in field test
programs. Both these field programs also highlighted the fact that
large proportions of load may be shed in the upper reaches of Pin
Piles, into soil which is usually neglected as having load transfer
potential at the design phase. This explains neatly the surprising
stiffness of Pin Pile systems in their lower range of capacity.

The field tests also remind us that in certain favorable


geotechnical conditions, the grout-soil bond which contemporary
drilling and grouting methods promote can be so large that it is the
internal load carrying capacity of the pile, i.e. its structural
strength, which is the determinant of ultimate pile capacity. The
ER approach to field analysis of Pin Pile testing offers a precise
analytical and predictive tool, especially when combined with
959 Bruce, Wolosick, Rechenmacher
creep data: when the extent of apparent casing debonding reaches
to within a few feet of the end of the casing, explosive failure may
be expected shortly. At such times, the creep monitored may be
more a result of grout/steel interfacial phenomena rather than
grout/soil conditions as conventionally assumed.

This analytical method opens the door to Pin Pile acceptance


criteria similar to those used for prestressed ground anchors
where elastic performance and creep patterns are used: this would
be more rigorous than current "geometrical construction" type
methods.

Two related questions remain to be addressed, namely the


puzzle of why a failure load can be recorded in a pile, lower than a
load safely reached in a previous cycle, and why failure can occur
during a creep test at constant load. The first case is simply
explained by reverting to the concept of non-recoverable bond:
once the virgin interface around the casing has been disrupted, it
cannot sustain the same level of bond stresses. Therefore, when
reloaded, the load must pass below the point to be resisted (Figure
a). This means that progressively less of the casing is capable of
resisting load, and so progressively higher proportions of the
applied load must be resisted in the bond zone. This bond zone has
a finite capacity (internal and external), and when this is
exceeded, failure results.

The second riddle has a similar explanation. As less of the


casing surface area becomes capable of resisting load as a result
of progressive debonding, the average bond stresses increase on
the surface area remaining in virgin conditions. This increase
accelerates the rate of interfacial creep, which reflects a
continuing, accelerating progressive debonding. At lower loads,
this creep tendency is low, and is soon stabilized: at higher loads,
this creep rate will be higher and may reflect a rate of debonding
so relatively fast that the underlying bond zone is being reqUired
to accept a substantially and progressively higher proportion of
the load over a time interval within the period of the creep test.
So, when the critical amount of load is transferred to the bond
zone, a sudden explosive failure will occur. This time of transfer
may vary from almost instantaneous to many minutes.

960 Bruce, Wolosick. Rechenmacher


6. SEISMIC RETROFIT APPLICATIONS - GENERAL
INTRODUCTION

The California Department of Transportation (CALTRANS)


recorded numerous bridge failures in the 1971 San Fernando
Earthquake. The failures were linked to separations at bridge deck
expansion joints and a lack of ductility in the supporting columns.
As a consequence, CALTRANS retrofitted 1250 state bridges to
provide deck continuity, from 1971 to 1989, although column
ductility retrofitting was delayed until 1986 due to budget
constraints. Column ductility improvements of course result in an
increase of load demand on both superstructures and foundations.

Investigations into various bridge foundation repairs have


been intensified in recent years as a result of the increased
availability of funds following the 1989 Loma Prieta Earthquake.
(15). One of the most common measures is to add
tension/compression piles around the perimeter of an existing
footing. Driven precast concrete and steel piles are typically used
for foundation support of bridges. However, due to constraints
inclUding noise and vibration level limitations, installation
difficulties presented by low overhead conditions. difficult
drilling and driving conditions due to ground obstructions or high
water tables, limited right-of-way access, the inability to extend
the footings and higher tension capacity requirements, alternates
to standard driven piles are becoming more desirable. Since
varying project conditions may be more practically and
economically favorable to certain construction techniques, there is
no one single "best" solution.

7. CALTRANS TENSION PILE TEST PROGRAM


SAN FRANCISCO, CALIFORNIA

In late 1991, CALTRANS initiated a full scale pile testing


program as part of its seismic structural retrofit program.
Foundation retrofits are the most costly element in the seismic
retrofit program, fully justifying research into alternate
construction techniques. This testing program was proposed as a
joint effort between CALTRANS, the Federal Highway
Administration, and contractors who could offer proprietary piling
systems. CALTRANS tested traditional piling systems such as
drilled shafts and driven steel H piles or pipe piles: proprietary
systems offered alternatives. As part of this program, Nicholson
961 Bruce, Wolosick, Rechenmacher
Construction installed six Pin Piles of three different types at the
test site in San Francisco. (16).

A simplified stratigraphy of the test site was:


o to 20 feet (0 to 6 m) Fill
20 to 110 feet (6 to 34 m) Clay (Soft Bay Mud deposits)
110 to 160 feet (34 to 49 m) Sand

During installation, actual pile lengths were varied in response to


the actual conditions encountered. However, to limit test
variables, many of the pile components and dimensions were held
constant. The three types of piles installed were:

NCA-Pile. A high capacity multistrand tendon was installed


within and below the 7 inch (178 mm) dia. steel casing. This
tendon was stressed and locked off against the top of the
casing prior to the pile test. A 35 foot (11 m) long pressure
grouted bulb was formed in the sand, which included a 10
foot (3 m) embedment of the casing, a 5 foot (1.5 m) buffer
zone, and a 20 foot (6 m) bond length for the tendon. This
pile was similar to a type S-1 except that the prestressing
tendon replaced the reinforcing bar.
Pin Pile. A 60 ksi (410 MPa) yield strength steel reinforcing
bar was installed within and below the 7 inch dia. casing
(Type S-1). No prestress loading was applied to this pile.
The pile had a pressure grouted bulb 30 feet (9 m) long in the
sand which included a 10 foot (3 m) embedment of the pipe,
and 20 feet (6 m) of reinforcement extending below.
NFC-Pile. The 7 inch (178 mm) diameter steel casing was
dri lied full length into the lower sands and a 60 ksi (410
MPa) steel bar placed full length. (Type S-2). For these
piles, the grout was introduced as drilling in the sand
progressed.

Two examples of each pile type were installed at the site:


one deep pile founded in the sand and one shallow pile founded in
the Bay Mud. Deep pile lengths varied from about 140 to 155 feet
(43 to 47 m). Shallow piles were installed to about 105 feet (32
m). The piles were tested to a 200 kip (890 kN) design load and
then loaded to failure. The Nicholson pile load test results are
shown in Table 8.

962 Bruce, Wolosick, Rechenmacher


PILE DESCR PILE CAPACITY ACTUAL ELASTIC ACTUAL TOTAL
f\O. (KIPS) DEFL. @ 200 KIPS DEFL. @ 200 KIPS
Tension Comoression Tension Comoression Tension Comnression

10,A NCA-Deep 407 * .461 " --- .503" ---


11, B NCA-Shallow 243 160 .329" N/A .370" N/A

12,E NFC-Deep 500 >400 .302" -.289" .310" -.348"

13,F NFC-Shallow 195 220 .302" -.270" .414" -.420"

73,A Pin-Deep 455 >385 .530" -.516" .631 " -.581 "

74,B Pin-Shallow 189 373 N/A .351 " N/A .378"

Table 8 Summary of Test Results for Nicholson Piles in CALTRANS Test Program
* Pile damaged during tension test loading. No Compression Test Results
N/A Not Applicable
1 kip =4.45 kN; 1 in.= 25.4 mm
Piles in lieu of the 64 specified CIDH concrete piles. Detailed plans and calculations
were prepared, submitted and approved by the CALTRANS Office of Structures.

8. FIELD RESEARCH - CALTRANS NORTH CONNECTOR


OVERCROSSING - 1-110 SITE. LOS ANGELES.
CALIFORNIA

CALTRANS awarded a construction contract for the North


Connector Overcrossing in Los Angeles in 1991. The original
design involved retrofitting bents 2, 3, 5 and 6 by ~trengthening
the existing footings. The design used sixteen 2'13lnch (61 mm)
diameter cast-in-drilled-hole (CIDH) concrete piles placed around
the existing footing at each single column bent.

An experienced and qualified drilled-shaft subcontractor


attempted to install the specified piles. However, due to difficult
drilling conditions, including concrete obstructions and water
bearing (flowing) sand, and the installation difficulties caused by
low overhead conditions, they were unable to complete the
installation of any CIDH piles. CALTRANS was aware of the
Nicholson Pin Pile through the San Francisco Test Program and
SUbsequently, the general contractor engaged Nicholson to install
64 Pin Piles in lieu of the 64 specified CIDH concrete piles.
Detailed plans and calculations were prepared, submitted and
approved by the CALTRANS Office of Structures.
963 Bruce, Wolosick, Rechenmacher
The project site was located in Los Angeles near Figueroa
Street and the southbound on-ramp to 1-5. The soils underlying the
site consisted of loose to slightly compact fill in the upper 25 feet
(8.0 m) and dense to very dense sands and gravels below. The
ground-water table was approximately 25 feet (8 m) below grade.

The project site had been a dump location for a ready-mix


concrete plant, and the upper fill zone contained large chunks of
concrete and rubble. Three of the retrofitted footings were
located adjacent to the Aroyo Seco drainage channel, and were
accessible only by a graded road or from the edge of the Pasadena
Freeway. The fourth footing was located in the middle of the
Pasadena Freeway, creating very difficult access conditions.
Overhead clearance under the freeway superstructure was
approximately 20 feet (6 m).

The Type S-1 Piles for this project were required to support
an ultimate compressive load of 500 kips (2225 kN) with a
maximum pile head total deflection of less than 0.60 inches (15
mm). Each pile comprised:

An upper pile length extending to 30 feet (9 m) below


the bottom of the existing footing, consisting of a 7 inch (178 mm)
o.d. 1/2" (12.7 mm) wall thickness steel casing, reinforced full
length with two 1-3/8 inch (35 mm) diameter grade 150
threadbars and filled with neat cement grout.

A pile bond length extending from 30 feet (9 m) to 60


feet (18 m) below the bottom of the existing footing, consisting of
a pressure grouted bond zone, reinforced with the two 1 3/8 inch
(35 mm) diameter threadbars, extending to the pile tip, and the 7
inch (178 mm) diameter steel pipe, extending 5 feet (1.5 m) into
the top of the bond length.

A specially designed connection between the pile and


the cast-in-place extension to the structure footing.

The production test pile (Bent No.3, Pile No.3, selected by


CALTRANS) was drilled with a high-torque. low-headroom drill rig.
It was installed from existing grade to a depth of approximately
66.5' feet, (20 m) allowing testing to be performed before footing
excavation. The casing was placed in 10 foot (3 m) lengths, and
964 Bruce, Wolosick, Rechenmacher
the threadbars were placed in 10 foot (3 m) and 20 foot (6 m)
coupled lengths, centralized in the pile with plastic spacers.
Maximum grout pressure attained during grouting of the pile bond
length ranged from 100 to 140 psi (0.7 to 1.0 MPa) measured at the
drill rig.

The pile test was conducted by representatives from the


CALTRANS Office of Structures. The tension test was completed
to the required 300 kip (1340 kN) load and the compression test to
the required 500 kips (2225 kN). The pile was loaded in 100 kip
(445 kN) cycles, with the load applied in 20 kip (89 kN)
increments, and reduced in 20 to 100 kip increments. Each
increasing load increment was held for 5 minutes the first time at
that load, and for 2 minutes thereafter. Each decreasing load was
held for one minute.

Figure 4 summarizes the load test data. The pile


successfully resisted the required maximum tensile load of 300
kips (1340 kN) with a total displacement at maximum load of
0.304 inches (7.72 mm) and a permanent displacement of 0.050
inches (1.27 mm) at zero load after loading. Creep movement
during the 5 minute hold at 300 kips was 0.006 inches (0.15 mm).
The pin pile then successfully supported the required maximum
compressive load of 500 kips (2225 kN) with a total displacement
at maximum load of 0.392 inches (9.96 mm) and a permanent
displacement of 0.068 inches (1.73 mm) at zero load after loading.
Creep during the 5 minute hold at 500 kips was 0.007 inches (0.18
mm).

9. OVERVIEW OF CALTRANS TESTS

The Nicholson Pin Pile proved to be an excellent system for


meeting the design load capacity and displacement requirements.
The Pin Piles were also installed with relative ease at difficult
access sites and ground conditions which prohibited the
installation of conventional pile types. Pressure grouting
techniques in the dense sands and gravels resulted in very high
grout/soil bonds and small displacements. Even in the soft Bay
Muds, surprisingly high skin friction values were mobilized. The
response of the Pin Pile to test loads was essentially elastic, with
very small permanent displacements. These observations offer
real hope that the special demands imposed on pile performance by

965 Bruce, Wolosick, Rechenmacher


the particular demands of California can be adequately met by the
appropriate proprietary option.

COMPRESSION
COMPRESSION
450

400

350
Vl
0.324 IN.
0..
- 300
::.c:

TENSION
150
0.254 IN.
100

50

0.'::; 0.3 0.2 O. 1 O. 1 0.2 0.3 0.4

PERMANENT MOVEMENT ELASTIC MOVEMENT

Figure 4 Permanent and Elastic Displacement Analysis


North Connector Overcrossing 1-110, Los Angeles, CA

966 Bruce, Wolosick, Reche.nmacher


REFERENCES

1. Bruce, D.A., Ingle, J.L. and Jones, M.R. (1985). "Recent


Examples of Underpinning Using Minipiles." 2nd International
Conference on Structural Faults and Repairs, London , April
30 - May 2, pp. 13-28.

2. Bruce, D.A. (1988, 1989). "Aspects of Minipiling Practice in


the United States." Ground Engineering 21 (8) pp. 20-33 and
22 (1) pp. 35-39.

3. Bruce, D.A. (1989). "American Developments in the Use of


Small Diameter Inserts as Piles and In Situ Reinforcement."
International Conference on Piling and Deep Foundations,
London, May 15-18, pp. 11-22.

4. Bruce, D.A. (1992). I' Recent Progress in American Pinpile


Technology." Proc. ASCE Conference, "Grouting, Soil
Improvement and Geosynthetics", New Orleans, LA, Feb. 25-
28, pp. 765-777.

5. Pearlman, S.L. and Wolosick, J.R. (1992). "Pin Piles for Bridge
Foundations". 9th Annual International Bridge Conference,
Pittsburgh, PA, June 15-17, 8 pp.

6. Bruce, D.A. and Gemme, R. (1992). "Current Practice in


Structural Underpinning Using Pinpiles." Proc. NY Met.
Section ASCE Seminar, New York, April 21-22, 46 pp.

7. Bruce, D.A., Hall, C.H. and Triplett, R.E. (1992). "Structural


Underpinning by Pinpiles." Proc. DFI Annual Meeting, New
Orleans, LA. October 21-23, 30 pp.

8. Pearlman, S.L., Wolosick, J.R, Groneck, P.B. (1993). "Pin Piles


for Seismic Rehabilitation of Bridges". 10th Annual
International Bridge Conference, Pittsburgh, PA June 14-16,
1993, 12 pp.

9. Bruce, D.A., Bjorhovde, R., Kenny, J. (1993). "Fundamental


Tests on the Performance of High Capacity Pin Piles". Proc.
DFI Annual Meeting, Pittsburgh, PA, October 18-20, 33 pp ..
967 Bruce, Wolosick, Rechenmacher
10. Jones, D.A. and Turner, M.J. (1980). IIPost-grouted Micro
ll
Piles Ground Engineering, II (4), pp. 14-20.

11. Herbst, T.F. (1982). liThe GEWI Pile - A Solution for Difficult
Foundation Problems n Symposium on Soil and Rock
Improvement Techniques Including Geotextiles, Reinforced
Earth and Modem Piling Methods, Bangkok, December,
Paper D1-10.

12. SSRC Report. Task Group 20, Structural Stability Research


Council. IIA Specification for the Design of Steel-Concrete
Composite Columns," Engineering Journal. American Institute
of Steel Construction, Vol. 16, No.4 (Fourth Quarter, 1979),
pp. 101-115.

13. Kenny, J., Bruce, D.A., and Bjorhovde, R. (1992). "Behavior and
Strength of Composite Tubular Columns in High Strength
Steel". Research Report No. ST-13, April 1992. Department
of Civil Engineering, University of Pittsburgh, Pittsburgh, PA.

14. Groneck, P.B., Bruce, D.A., Greenman, J., and Gingham, G.,
"Foundation Underpinning at an Operating Grain Export
Facility," Civil Engineering Magazine , September, pp. 66-68.

15. Zelinski, R. (1992). "Bridge Foundation Retrofits. University


II

of Wisconsin Short Course on Specialty Geotechnical


Construction Techniques, San francisco, CA, November 17.

16. Mason, J.A. (1992). "Tension Pile Test." Proc. 3rd NSF
Workshop on Bridge Engineering Research in Progress, La
Jolla, CA., November 16, 17, pp. 67-70.

968 Bruce, Wolosick, Rechenmacher


TESTING OF GEOJET UNITS UNDER LATERAL LOADING

Dan Spear,l Lymon C. Reese,2 Gordon T. Reavis,3 and


Shin-Tower Wang4

ABSTRACT
A new foundation system named GeoJet was installed in Oakland, California for
testing and evaluation. The GeoJet foundation is constructed by advancing a rapidly
rotating soil processor bit into the soil at a controlled rate. Simultaneously, slurry of
Portland cement under high pressure impacts the cuttings through jets in the processor.
A steel pipe or H-pile is placed into the cylindrical zone of the fresh soil-cement as
structure members. The paper describes the pile loading tests for the GeoJet in Bay Mud
under lateral loading. The test result indicates that the new system has remarkable lateral
capacity created by the soil-cement grout.
INTRODUCTION

CALTRANS proposes to rebuild the Cypress Expressway, parts of which were


made unusable by the 1989 Lorna Prieta earthquake. The new design for the
foundations requires sufficient ductility for earthquake resistance. The Brown & Root
GeoJet Foundation System has unique features over the conventional piles in the Bay
Area. Therefore, a program of loading tests of the GeoJet Units, sponsored by
CALTRANS and Brown & Root, was conducted at a site on the Cypress Expressway,
Oakland, California.

The test foundations were installed in Bay Mud on November 23 to 26, 1992. A
total of seven GeoJet units with 20-in. O.D. (0.51 m) and different inserts for
reinforcement were installed. The unit lengths were 55 ft (17 m) for the laterally-loaded
foundations and 65 to 85 ft (20 m to 26 m) for the axially-loaded foundations.

The aims of the testing were, firstly, to provide data for computing the capacity
of this unique foundation under axial loading; secondly, to gain information to allow

1 Professional Engineer, California Department of Transponation (CALTRANS),


Sacramento, California
2 Professor of Civil Engineering, The University of Texas at Austin, Austin, Texas
3 Manager-GeoJet Foundation System, Brown and Root Civil, Houston, Texas
4 Project Engineer, Lymon C. Reese & Associates, Austin, Texas

Spear et ai,
969
GeoJet Units to be designed to resist lateral loading in Bay Mud. This report presents
the test results conducted for GeoJet foundations under lateral loading.

SUBSURFACE CONDITIONS AT SITE OF TEST


The stratigraphy and data on soil properties come from borings conducted at and
near the test site. A surfacial-filllayer a few feet in thickness is underlain by New Bay
Mud. The New Bay Mud extends to the depth of 43 ft (13 m) and consists of
predominantly very soft to soft clay with one or two interbedded silty-sand layers. The
soil from depths of 43 ft to 62 ft (43 m to 19 m) is Old Bay Mud with greater shear
strength than the upper clay. A layer of dense, silty sand was found at depths of 62 to
64 feet (19 m to 19.5 meters). Stiff clay with interbedded thin silt seams was logged at
depths of 64 to 120 feet (19.5 m to 36.5 meters). The composite soil profile at the site is
shown in Fig. 1, with ranges of N-values for sand and undrained shear strength Cu for
clay. These values were interpreted from results of the testing, as described below.

SOIL SHEAR
PROFILE STRENG1H
c [tsf)
0.:, 1.0 1.5 2.0 2.5 3p _
o-
4 -%-
10 - 10 .
Very soft dark gray clay
20- 20

Very dense
yellow brown silty sand
Very soft dark gray clay ~
30 -

40-
-,- -
N = 36 -70
- ----

!~-====[==
Tan sandy silty clay
sllff, Interbedded with
clayey silly sand

Dense silty sand 60 - N = 30 -70

70 - .' 70 -

SlIff gray clay Interbedded


wIth clayey silty sand

:~-
80 -'

90 -

Figure 1. Soil Profile and Interpreted Shear Strength at Site of Test


(1 ft = 0.304 m, 1 tsf = 23.94 kPa)

The following data were available and were employed in arriving at the values
shown in Fig. 1: blow counts from the Standard Penetration Test, results from
piezocone tests, and results from a series of laboratory tests of undisturbed specimens.
All of the testing was done by personnel from CALTRANS.

As may have been expected, there was a considerable amount of scatter in the
results from both the in situ tests and the laboratory tests. Therefore, it was necessary to
exercise judgment in arriving at the values shown in Figure 1.

970 Spear et al.


ARRANGEMENT FOR TESTING THE GEOJET
FOUNDATION UNITS
The site plan for tests of GeoJet units is shown in Figure 2. Seven members
were installed to the depths indicated. Units TPI to TP4 were tested under axial loading.
I
I

TP59: @ ~, )1
Lateral-test pile Rection piles
(uninstrumented) (H-sections)
I
I
I
@ TP4 @TP3 @TP2 0TP7
Lateral-test pile
(instrumented)

TP6@ ~
I

II Insert
40'x 3/s"
Insert
65' Insert
Insert
4O'X3/S"

HPl2x53 85'
HPl2x53

Insert
25'x 1/2" exterior
55'x3/s" internal

Figure 2. Plan for Tests of GeoJet Piles at the Cypress Site


(l ft = 0.304 m, 1 in = 0.0254 m)

Units TP5 and TP6 were installed for testing under lateral load. Unit TP7 was a
driven pipe which was installed to compare the axial capacity for GeoJet Units and a
steel-pipe pile.

Units TP5 and TP6, identical in structural design, were installed so that the point
of application of the loading was just above the surface of the Bay Mud, which underlies
fill. The loading and measurement of movements of the head of the GeoJet Unit were
perfonned in a shallow excavation in the fill. The water table was near the ground
surface, and water was maintained in the excavation during the loading. The presence of
water at the soil surface can have an important influence on the response of deep
foundations to cyclic lateral loading.

The two GeoJet units were pulled toward each other by use of the arrangement
shown in Fig. 3, with the center-to-center spacing of the units at about 6 diameters. The
entire system was designed to apply sufficient loading to develop the full capacity of the
GeoJet Units in bending.

Desilm of GeoJet Foundation Units


The usual GeoJet Unit is constructed with a structural insert, which is usually a
pipe or a structural shape. The insert, imbedded in soil-cement of appropriate strength,
is capable of sustaining lateral load of a significant magnitude. However, the adaptability

971 Spear et al.


of the GeoJet System allowed for employing a second structural shape to increase the
bending resistance. The design chosen is shown in Fig. 4. That particular figure is for
TP5 because only one of the units was instrumented with strain gauges. Unit TP6 is
identical except for the internal instrumentation.

Bearing with spherical or


cylindrical surfaces Bearing with spherical or
cylindrical surfaces
High strength bar I Hydraulic ram Anchor
Anchor

"
\
---
for tensile loading
-

Bay Mud
Load cell

Figure 3. Arrangement for Application of Lateral Loading to GeoJet Units

Extension of pile Support for extensometers to


be fastened to wall of pit

Depth, rt
Bottom of pit
0

Depth, rt l...!.'!!!. Gauges


Slm. 0 1 2
10 front
2 2 2
Ind
blck 4 3 2
6 4 2
8 5 2
20 10 6 2
12 7 2
14 8 2
16 9 2
18 10 2
30
20 11 2
23 12 2

40

so

60

Figure 4. Arrangement for Performing Test of GeoJet Piles Under Lateral Loading
(1 ft = 0.304 m, 1 in =0.0254 m)

972 Spear et al.


A 20-in. (0.51 m) diameter steel pipe with a wall thickness of 0.5 in. (13 mm)
and a penetration of 25 ft (7.6 m) below the ground line was added to the design. The
outside of this pipe was bearing against the natural soil. The centralized insert was a 12-
in. (0.30 m) diameter pipe with a wall thickness of 0.375 in. (9.5 mm) and a length of
55 feet (17 meters).
The principal reason for the design was to give the two GeoJet Units a sufficient
capacity to resist bending that the deflection could be carried to a significant amount in
order to develop the soil resistance over a wide range. The ability to perform properly
with a large deflection is especially important in Northern California because
considerable ductility is a highly desirable feature in seismic regions.

Figure 4 also shows the positioning of electrical resistance strain gauges for the
measurement of bending moment. The number of gauge points was selected for the
accurate determination of the bending-moment curve for each of the increments of load.
The strain gauges were T-rosettes for greater sensitivity, and the cables to each gauge
were of sufficient length to allow the data-acquisition system to be positioned well away
from the loading system.

Three electronic extensometers were employed for measurement of deflection


along the unloaded extension of each of the GeoJet Units. The lowest transducer was
placed just above the point of application of load and the positions of each extensometer
was carefully measured. Thus, the deflection and rotation at the point of the application
of the load could be readily found. The redundancy was desirable in case some
malfunction occurred in one of the electronic extensometers.
Application of Loadin2. The design of a deep foundation under lateral
loading is normally controlled by one of two factors: pile-head deflection and maximum
bending moment. In the case of testing, deflection was of no particular concern except
that the units should be made to deflect as much as possible so that experimental p-y
curves could be produced to as great a depth as possible. Bending moment, therefore,
should control the maximum loading. Computer Program STIFF was used to analyze
the composite section shown in Fig. 4, and the computed value of ultimate bending
moment in the upper 25 ft (7.6 m) was 14,000 in-kips (1,580 kN-meters).

Calibration of the strain gauges, to convert the gauge readings to bending


moment, was done at the conclusion of testing for soil response. However, a
preliminary calibration factor needed to be found during initial testing that would allow
the bending moment to be computed for strain-gauge readings. This factor could then be
used so that the maximum bending moment would not exceed the ultimate bending
moment (14,000 in-kips/1,580 kN-meters).

For foundations that support a bridge or many other structures, the lateral loading
will be repeated. Previous experiments have shown that repeated or cyclic loading
causes a reduction in soil resistance. However, it is desirable to develop p-y curves for
monotonic-static loading (backbone curves) because some portions of such curves can be
correlated with the mechanics of soils. The most desirable approach is to install one set
of units for static loading and an identical second set in identical soil for cyclic loading.
In order to reduce the cost of testing, an alternate concept was used here as has been
done elsewhere.

The assumption was made that the cycling of lesser load does not affect the
response of the soil at a larger load. Plainly, for this assumption to be valid, the
difference in the successive loads must be significant. The data-acquisition system that
was employed was capable of taking rapidly a full set of readings. Therefore, readings

973 Spear et al.


of all gauges were recommended after the following loading cycles: 1, 3, 6, 10, and 20.
The logic for stopping at 20 cycles was that for most designs only a small number of the
maximum loads are applied. The loading was applied at a rate slow enough that
acceleration was small and dynamic effects were unimportant

To be consistent with the concept of testing, as given above, the recommended


loads in kips were: 2, 5, 10, 15, 20 and so on, in 5-kip increments. The bending
moment in the instrumented GeoJet Unit was monitored carefully by evaluating strain-
gauge readings, and loading was to be discontinued prior to damage. However, even
under a deflection of the head of the unit of several inches, the bending moment did not
reach a critical level.

Figures 5 and 6 show the measured deflections at the point of application of load
for GeoJet Units TP5 and TP6. As may be seen, the repetition of loading had little effect
on the deflection except for the higher loads.
60..---..,.----.,..-----.,.----,-----,------,----,

~
2
j 20 ..... ..:.... i......I.. t.... .... . t.. . .. . . . t..........

1or-l-l-;;-;
ot--+---t---+---+----r---t---1
o 123 456 7
Deflection at loading point, in

Figure 5. Measured Load-Deflection Curves, Static and Cyclic, for 1P5


(1 in = 0.0254 m, 1 kip = 4.448 kN)

Calibration to Obtain EX values. The interpretation of the data from strain gauges
requires the value of the EI of the composite pile. While the value of EI for the GooJet
Units can be computed by the use of mechanics, the more direct way is to calibrate
experimentally. An excavation was made around the pile so that in the zone of the
removed soil the value of the applied moment could be known precisely. The
assumption was made that the soil-cement produced by the GeoJet System would have
identical characteristics with depth. Thus, the calibration curves for the upper levels of
gauges could be used for all of the gauges. During calibration, the strain gauges were
read as half bridges for more flexibility in interpretation.

974 Spear et al.


60 -.----.,.------,---------,---,-------,------,

50 !. +
,
+,
...... , ~ : .


(J)
Q.
40 :
, . " .
~ + ~ + .
:.si:
-ti
ro i,
, .
o 30 l:. t. j..; ) .
~
2ro .:
.....J 20 ..... :1.. ;.. ~ ...... T..


10 .. r-....;.. .. j.. .... r....; .

Ot---;---;---;---t---t------i
o 1 2 3 456
Deflection at loading point, in
Figure 6. Measured Load-Deflection Curve for TP6
(1 in = 0.0254 m, 1 kip = 4.448 kN)

With the value of M for a particular point and with readings from the strain
gauges on opposite sides of the unit, EI can be found from the following equation: lip =
MlEI, where p is the radius of curvature as determined from the strain.
DERIVATION OF p. Y CURVES
The four boundary conditions, Pt, Mt, Yt, and St, were measured, as well as the
bending moment along the length of the pile. An analytical expression was fitted through
the bending moments, using least-squares fitting. Double integration of this expression
and double differentiation, using appropriate boundary conditions in each case, yielded Y
and p, respectively, as a function of depth. Employing this procedure for each of the
loadings, a family of p-y curves was developed by cross plotting y and p.

The derivation of the p-y curves began by plotting strain-gauge readings against
load for all gauges. The plots of load versus gauge readings were fitted by an analytical
expression. The curves for the shallow depths were nearly linear, but curvature of the
plots increased with depth.

After all the strain gauge data had been plotted, the gauge readings were
converted to bending moment by use of the calibration curves. Graphs of moment
versus depth were plotted for each of the applied loads and then fitted by an analytical
expression. Figure 7 shows the bending moments as a function of depth for the various
static loadings that were applied.

975 Spear et al.


0

6-

---
c: 8
.c: 10
Q.
Q)
0 12

14

18

2O+----,!ic----;----Ef---;---;---;---------M"'::r----;--------;--------------!
o 100 200 300 400 500 600 700 800
Bending Moment (k-ft)

Figure 7. Bending Moment Curves as Found from Strain-Gauge Readings for


GeoJet Unit TP5 under Static Loading
(1 ft = 0.304 m, 1 k-ft = 1.356 kN-m)

Integration and differentiation of the bending-moment curves, as noted earlier,


led to the experimental p-y curves. The curves for the case of static loading are shown in
Figure 8. As may be seen, the final curve was for a depth of 15 ft (4.6 m) because
deflection was quite small below that depth.

O-F---r-----,---,.---r---.,----1
o 2 3
y, In

Figure 8. Experimental p-y Curves for Static Loading


(l in = 0.0254 m, lIb/in =0.175 kN/m)

976 Spear et al.


ANALYSIS OF THE RESULTS OF THE EXPERIMENTS
The p-y curves obtained by experiment at the Cypress site for the Bay Mud can
be compared with methods of prediction of p-y curves for soft clay (Matlock, 1970).
Figure 9 shows the comparison between predicted and experimental p-y curves for a
depth of 10 ft (3.0 m). As is apparent, the experiments at Cypress reflect a much stiffer
behavior than the method of prediction. This improved behavior may be due to some
conservatism in the prediction method or possibly due to an under-estimation of the shear
strength at Cypress. As noted earlier, a considerable scatter was found in the values
from the various methods of obtaining the shear strength. As a means of investigating
the influence of the shear strength on deflection, Computer Program LPILE was used to
compute the deflection at the point of load application under static loading as a function
of lateral load and shear strength. The results are shown in Figure 10. The figure shows
that the shear strength can be increased to a value within the range of the values found in
the subsurface investigation to obtain an excellent correspondence between the observed
deflections and the computed ones. The exercise is repeated in Fig. 11 for the case of
cyclic loading.
8OO-r---.,.--------,----:------.,..----,-------.,

700 .. .... + .... j. ...... .. .. .. i.... ..... i .. ..........;......

600 .~\~.s~~~~ j j.
:
I i
!----l---_-
;

Pr9dicted !
.. .1 ..._.._ , __..__ j- -"-i"--"
c
-.
a 400'" .......: j -+ --.
ci
~ 1_-+--;_-j____f __.-.--.. ---.-..

o-t---j------;----;----+-----;----J
o 2 4 6 8 10 12
y, in

Figure 9. Comparison of Measured and Predicted p-y Curves at 10 feet


(1 in = 0.0254 m, 1 lb/in =0.175 leN/m)
CONCLUSIONS
1. The lateral-load experiments at the Cypress site confmned, within reasonable
limits, the previous methods of prediction of behavior of deep foundations under lateral
loading. This conclusion is valid for both static and cyclic loading.

2. The current methods of analysis of deep foundations under lateral loading can
be employed for the analysis of Goolet Units. In this regard, the computation of the
values of bending stiffness, EI, can be computed by taking into account the compressive
strength of the soil-cement with the soil-cement being treated as a weak concrete.

977 Spear et al.


80 -r----------------.....,

60
iii
III
C.
:;:
-ri

...
CIl
.2 40 iii Measured
e
G
Predicted

:
20

O. .-....,.--"'T""-...,...-..,...-.....,..---.----.---!
o 2 4 6 8

Deflection at the loading poInt, In

Figure 10. Comparison of Measured and Predicted Load-Versus-Deflection Cmves for


Static Loading at the Loading Point
(1 in = 0.0254 m, 1 kip = 4.448 leN)
80.....-----------------,

60

III iii
C.
:;:
-ri
~
e
40
...
EI Measured
Predicted

-
~
II

20

O...........-.--_---r-.--....,.----r-r---r--'"T-..--.,
o 2 3 4 5 6

Deflection at the loading point, In


Figure 11. Comparison of Measured and Predicted Load-Versus-Deflection Curves for
Cyclic Loading at the Loading Point
(1 in = 0.0254 m, 1 kip = 4.448 leN)

978 Spear et al.


3. The GeoJet Units behaved. flexibly in sustaining the lateral load at Cypress
with pile-head deflections in the order of 6 inches. This ductile response is a favorable
feature in the design of earthquake-resistant structures.
REFERENCES
Matlock, H. (1970), "Correlations for Design of Laterally Loaded Piles in Soft Clay,"
Proceedings, Second Annual Offshore Technology Conference, Houston, Texas,
Vol. 1, Paper No. OTC 1204, pp. 577-594.

979 Spear et al.


HELICAL PLATE BEARING MEMBERS, APRACTICAL SOLUTION TO DEEP
FOUNDATIONS
ST AN RUPIPER; M A.S.C.E.
ABSTRACT
~echnology h:ls sufficiently been improved to allow the use of helical plate
bearing me mbers to be used in ne:lrly all c:lses Ill' here any other kind of
deep foundation type is used. Industry has developed hydraUlic driven
rOlaling equipment lhal can turn helical plale compression and/or tension
members into soil with resistances up to 100 "Standard Penetration Test" soil
to allowable JO:lds that approach 445 kN. (100 kips) bearing per helical plate
be:lring assembly.

Helical members are versatile, can be installed in all kinds of weather and
nearly all types of soil. to support large and small loads, as individual or
group installations. The installation is self monitoring in that the torque
required Lo install the member is Jogged and the capacity is determined by
the torque. Reference 1, Load lest can be made to correlate the instaJling
torque to load bearing capacity.

INTRODUCT JON
Helic:l( plate be:lring members h:lve been used in the design of new and
retrofitted foundation systems by the author since 1986. AJI installations
have been successful. They were usuaJly inst:lJled in locations where other
types of deep found:ltions were not practical or too costly.

Installations include new and retrofit for residences, commercial buildings.


and olher structures. They have been used Where slab, strip and concrete
caisson foundations have failed. For loads from 18 kN (4 lCips) to 534 kN
(120 Kips). They have been used in sand, clay, gravel and cobbles, elpansive
and collapsing soils; underwater. on levelland and on steep hillsides. They
have been used for retaining walls, retrofit and new.

I. Consulting Engineer, 1033 Villa Maria Court. San Jose, CA 95125


408-287-8759

980 S. Rupiper
They are self testing in that they are turned in with a recorded torque,
determined by a split coupling shear pin torque indicator. or electronic
torque indicator or pressure gages correlated to one or both of the above
torque indicating devices, Many tests have proven that the load carrying
capability of the member is a direct relationship to the amount of the
installing torque. This gives them a predictable bearing capacity even in
unpredictable soils.

Recent experience Wilh the Lorna Prieta and Northridge Earthquakes has
provided us with in-place evidence that they help to limil the seismic force
damage.

GENERAL

There are lWO types of helical plale bearing members available lhal have
been used as deep foundations. The solid square shaft as shown on Figure I
and the pipe shaft as shown on Figure 2.

Presently manufactured square shafl members have shafts thal vary from
3amm (I 1/2"') square to 57 mOl (2 1/4"), using 152 mm (6") to 356mm
1.14"") helices. The maximum allowable torque for the largest shaft is 14.75
kN-m (20 ft.K:ips) which will produce a load carrying capacity of 890 kN
(200 kips), if conditions are right. For minimum selllemenL the design load
is usually considered at l/Z the member's capacity. In most soils the
bearing capacity increases with lime. Reference 3.

The helical plate bearing member shown in Figure 2 consists of pipe sizes
from 38 mm (1 l/Z") diameter to 254 moo (10") diameter. The circular
section has complete circumferential friction. the unit requires more
installing torque lo obtain the same capacity as a square shaft member. The
bearing capacity/torque ratio for square shaft members is 10 to I, for round
shaft members it is 6 to 1. These ratios are very conservative and are based
on not developing a fixed end moment at connection to the structure.

General installaLion techniques are discussed in Reference 2. Helical plate


bearing member manufacturers' qualify the installers to provide quality
installations, They have published minimum standards for installation
techniques. Although the installation is relatively simple and relalively easy
lo obtain a known capacity, it is essential to provide knOWledgeable
engineering, a quality product and the inslaller sbould be qualified and
experienced.

981
SPECIFIC ADV ANT AGES IN USING HELICAL PLATE BEARING MEMBERS
Waler is no problem. they can be installed in water or under water using a
kelly bar. They can be installed in water using small barges with the light
installing equipment. Templates may be utilized to provide accurate
placement. Where Ihere is a requiremenlthat the member extend Ihrough
the waler into the soil below the waler 10 act as a column the member as
shown in Figure 1 can be instaJled deep into the soil as required for load
bearing. Ihen Ihe pipe shart member, Figure 2. or an ordinary pipe can be
instaJled over the solid shaft member and concrele placed in the space. This
aJlows aH the work. 10 be done without divers except where cross-bracing is
required.

Cobbles become a problem, but only in a very dense matrix, where cobbles
are in sand, the member moves around the cobbles. The position of the shaft
may vary from Ihe planned location. but usuaJly Ihe shaft can be shifted to
the desired or tolerable location.

Helical plale bearing members can be a practical uLilizalion Ihrough land


fiJls. The solid shaft members are nearly sealed along the shafts. or can be,
with a bentonite slurry at Ihe lap. lherefore toxic gases wiH probably be or
lillJe concern during Ihe instaJlation. On the olher hand Ihe pipe shafl
members can be used to allow Ihe loxic gases 10 escape. The shaflS should
be sized 10 aHow for more corrosion than in normal soils.
Due 10 Ihe use of sectional shafts they c:ln be installed in areas th:lt :Ire
nearly inaccessible or less costly than using other types of deep foundations.

The inslallation is usuaJly by hydraulic rOlating devices. This produces a


small amount of noise. Jf the members are being installed with hand held
equipment or wilh a hydrauJicaJly powered vehicle Ihe hydraulic pump can
be in a remOle location and the noise of the workmen's voices would be the
loudest noise heard.

When used as Lie back.s. they can be post-tensioned for proof loading. The
lens ion stress is very low in the member because the installation torque is
Ihe governing design parameter. There is little chance that corrosion would
ever become a problem in thaI the shafl could have as much as 1/8"
(}.125mm) corrosion and be able to take Ihe maximum applied tension
before Ihe shaft size required for torque installation would be exceeded.

982
The tie b<lcks c<ln be de~igned for <lny 10<lds <lnd <lny safety f<lctor th<lt will
~<lti~fy the corrosiveness of the soil. or sacrifici<ll <lnodes m<lY be used. II is
the author's opinion that the addition of a few more tie b<lcks is the most
prudent method. <IS there is no dependence on future monitoring of
sacrificial <lnodes. The lower the stress in the tie back the less chance for
f<liJure, The tie back can be e<lsily proof loaded prior to embedding in the
concrete bec<luse it is a stiff member <lnd wiJJ not significantly relax the
tensioning. Also a sleeve m<lY be inserted in the concrete or a hole in the
soldier piles to <lllow tensioning foJJowing the placement of concrete,
drainage <lnd backfill on retaining waJJs. There is little danger of f<lilure with
helic<ll plate bearing on the soil due to future ch<lnges of consistency of the
soil. Reference 3. The helic<l! plate bearing me mber does not depend on the
skin friction of the shaft.

The members displace the soil on instaJJation. but only a smaJJ amount due
to the slenderness of the member. During installation there is little or no
ch<lnce for loc<ll soils subsidence. upheavel or lateral displ<lcement. The
installation is nearly vibrationless. Therefore members c<ln be installed
adjacent to other buildings, <ldjacent to sensitive equipment. in loose sand. in
<lny direction th<lt s<ltisfies the resultant loads.
EXISTING INSTALLATIONS

The <luthor h<ls designed many foundation systems th<lt has used helical
pl<lte be<lring members. The predomin<lnt use h<ls been for residential
structures where the lengths of the shaft varied from 2.13 meters. (7 feet),
to 16.76 meters. (55 feet). These instaJJations were both on the interior and
at exterior perimeters.

Refer to Foundation Comparison Figures S. 6 &. 7 for rating helical plate


bearing members with regard to other deep foundation systems. The rating
was for a row of 22 townhouses (three separate buildings). The units are
occupied on a daily basis by vacationers and the noise and other nuisance
factors are very important. This comparison was made for a project on a
coastal consolidated s<lnd deposit. where it was required to provide a
foundation that reached below the water table. At other sites and conditions
other consider<ltions may be applicable. therefore each site and project
requires its' own rating.

983
Other inst:J.lI:J.tions were ror industri:J.1 buildings. sm:J.J1 bridges. r<ldio tower.
ret:J.ining w<llls. (J:J.rge &. sm:J.lll. tilt-up buildings. swimming pools. hillside
structures. corn mercial buildings. concrete elevated decks. and other
miscellaneous structures and repair of structures. Reference 1.

Inst:J.lI<ltions where the author was not involved include footing :J.nd sl<lb
hold downs. roundations for w<llkways in wet lands. tie downs for fabric
shell buildings. pipe line foundations and hold downs. elevated building
foundations. tempor:J.ry support for lilt-up j<lcking pads. frost heave reaction
piles. lie downs for shear walls. tank supports and tie downs. guy ties for
derricks. transmission lowers and many other uses.

Some of the above installations exceeded 61 meters. (200 feet) with


capacities Qf a single member of mQre than 534 kN (120 kips).

PROPOSED INST ALLATIONS

Because of the ease, economy and predictable :J.spects of installations it is


probable that helical plate bearing members will be widely used in the
future. They can be used for deep foundations. for bridge supports. large
buildings. whatever and wherever any other type of deep foundation is
applicable, and of course for any use mentioned in this paper. The Leaning
Tower of Pisa could be readily stabilized using helical plate bearing
members.

INSTALLING

Helical plate bearing members are easily installed with inexpensive light
equipment, I.e., hydraulically driven rotating drivers held by hand can be
used to drive a Z22 kN (50 Kips) capacity members. A hydraulically driven
rotating drive mounted on a small bobcat can drive a member to 445 kN
(100 Kips) capacity. Reference 1. Larger equipment, but still small
compared to drilling and pile driving equipment. C:ln be used to install
greater capacity members.

Down drag due to unconsolidated soils is Jess of a problem than with larger
circumference deep foundations. Down drag for a 305 mm (I 1/2") square
shaft is 6.3 times less than on a 305 mm (I Z") diameter shaft and 18.8 times
less than on a 610 mm (24") dimeter shaft. Elimination or limiting of down
drag is very important as it is time dependent and can cause future
problems.

984
CONNECTIONS

The connections to the super structure are generally similar to what would
be used for any mini-piles. pipe piles, etc.. Manufacturers have patented
sleel brackets for loads for individual olembers to rated loads of 178 kN
(40 Kips) Reference 4. These brackets can be used with hydraulic jacks to
provide an easy method for leveling small structures. See Figure 3.

Concrete caps at the foundation as shown in Figure 4 and adaptions thereof


have been widely used for retro-fit projects. The advantage is that the load
carrying capacity is at the discretion of the structural engineer and the shaft
is fixed to the foundation in aJJ six directions. The shaft is projected weJl into
the concrete, with the hairpin (s) developing the load in tension.
When used with new construction it is preferable to design for a filed
connection with sufficient reinforced concrete to develop the bending
capacity of the shaft without break out.
The fixed connection embedded in concrete has four important features 1. [t
reduces the possibliJity of corrosion in the area most exposed to the
elements, 2. It reduces the slenderness ratio of the shaft. 3. It provides
adequate resistance to lateral forces breaking the member away during
seismic and or hurricane wind forces. 4. It is good in tension to resist uplift
forces. Tie back connections can be similar, but usually there is less space
available so threaded attachments with heavy washers or pins are required.

EASE OF ANCILLARY CONSTRUCTION

Once the helical plate bearing members are placed the concrete
foundations/walls can be immediately placed or the concrete placement can
be delayed until all me mbers are in place. thus allowing an integral pour.
eliminating the need for shuttling concrete trucks to the site. This also
reduces inspection and testing time. The author has specified use of steel
fiber in the concrete slabs and walls with great success. Shotcreted walls
have proven to be economical and are compatible with solid shaft Lie backs.

985
SEISMIC RESIST ANT
More than thirty installations of helical plate bearing members were in place
in the San Francisco Bay area when the Loma Prieta (7.1) earthquake
occurred in 1989. There was no damage (except for a few hairline cracks) in
the structure with the helical plate bearing members. There was a 73 m
(240') long by 6.1 m (20') high retaining wall in place tied back and bearing
on those members with no damage. At the time we dismissed this no failure
condition as the disorderly sporadic, (tic lack loe), as results of seismic
action.

During the Northridge earthquake, (6.4), in 1994. we had similar non-failure


conditions. except for two known cases. There was a single story home that
had one-half underpinned with helical plate bearing members and the other
half nol. The underpinned one-half came through the earthquake unscathed.
The other one-half was severely damaged with over 25 mOl (1") wide cracks
in the slabs and the slab was elevated from iLs original position. The
damaged one-half will require slab removal. lowering of foundation and
underpinnin8. A shopping center that was underpinned on both ends had a
liltle damage, but the center section that was not underpinned had
significanl damage.

It is the author's opinion lhat the vertical forces of the earthquake are
resolved al the deep bearing plate which serves as an uplift tension and
vertical compression member.
The horizonlal component of the progressive seismic waves pass lhrough the
flexible shafts much like a comb through a hair brush. (Rupiper theory), with
the strongest waves flexing the shafts and the remaining shafts providing
stability. This is the essence of a Jot of fUlure research and additional
research papers, But for the present time iL appears like we are getting a
bonus of seismic dampening with the use of deep helical plate bearing
members as deep foundations.

SUMMARY

Sufficient helical plate bearing members have been used with great success
for many different types of structures for a sufficient length of time to
convince the author that they perform well. Engineers and contractors
should recognize that the helical plate bearing members are not just a deep
foundation, "when everything else fails or is impractical to install", but can
be used as a primary deep foundation as well wilh as good, or better, results
than other deep foundations.

986
REFERENCES

l. Stan Rupiper, 'Helix Piers Are Solution For Column Reactions", ASCE Eighth
Structures Congress Abstracts, Balli more, Maryland- April 10,1990
Pages 535-5.36.
2. Stan Ruplper & William G. Edwards, "Helical Bearing Plate Foundation For
Underpinning", foundation Engineering Proceedings Congress/SCE/CO Div.
EV:lI1slOn, lL - Pages 221-230.
3. S.P. Clemence. Profesor, Syracuse University. Syracuse, N.Y. "Uplift
Behavior Of Anchor Foundations In Soil", ASCE Convention, Detroit. MI - Oct.
24. 1<)85.
4. G. Seider. P.E.. "Eccentrically Loaded Helical Piers Systems", A.B. Chance
Co., Centralia MO - BulJetin 01-9303-
). D,E. Bobbitt, PE. 'Theory And Application or Helical Anchors For
Underpinning & Tieback'. A,B, Chance Co. Bulletin 01-9004.

987
Helical couplers
boll together.

Foundallons:

Extension ~
Sec lion
Lead ~ Top coupler
Sec lion
may be cut
011 In 1I0id
10 design
olevallon.

FIGURE 1 FIGURE 2

E:(JSTTIIG r01JlroATTOIl

RAIRPnl

.il

~~
M> {

FIGURE 3 FIGURE 4

988
EXISTIHG COli CRETE SLAB

:XIIT'I'C COlIC. FIEF.

~ ~1'O:'lrsoLIDATE
Dill' SAIl'[.

!lEW FIEF.

WPICAL SECTION

FIGURE)

989
~ VI
VI
,..!!:
....l a.. VI '"....lCi:
'"'" ..,....l
'"..,'"Ci:
VI ....l....l

'"~Ci-'"
..:'" ;;:
""0 U'" 0", ..,
a ....
;:::1;:;
a gl;j ....
VIa.. '"
..,'"
..'" '"
a..Z
..,Ci:
....., '"
:l1:.., c'"
FOUND AnON COMPARISON ""
:;0 u =>u z~
..,z =z "'..:
Z....l 0 <0
NO. I LEASTx-:t "'0
au a.. a.. ::u ....l::
NO. ~. MOST
I. NOISE J .~
, EQUIP~IENT DISTURBANCE la) , I la) ~
J. STRUCTURAL DISTURBANCE Ib) .'> I lei ,
1. CXCAI'ATlmJ IdH (.J .~ J
'i, HAULING SOIL FRO~1 SITE J I

,.
6. SOIL DISTURBANCE UNDER BUILDING
AREA FOR CONSTRUCTION EOUIP~IENT
UI'i
.
Ig)
I
.'>
.I
5. LENGTII OF INSTALLATION TIME
~. NUMBER OF WORKERS ,Ihl , .)
,)
~
3
10. LARGE EQUIIP~tENT PRESENT Iii 5 lei ,
II. REBAR REINFORCEMENT 3
I~. CONCRETE QUANTITIES Ii! 3 Iii 3
1.1, INSPECTION 3 IiI ~
H. REBAR INSPECTION .'>
I 'i. CUTTING STRUCTURAL MEMBERS It) I
16. PA \'E~lENT DAMAGE FROM LG. EQUIP.
I:. LONG TER~1 SmLEMENT , .i
I~. PROBLE~IS WITH ALL-WEATHER INST. , I
I~. PROBLEMS WITH GROUND WATER II), III .'>
~O. PROBLEMS WITH SAND CAVING .'>
~ I. LEAST HAZARDOUS .'>
~~. DOWN DRAG
~3. FUTURE AD.IUSniENTS DUE TO DOWN DRAG
~". FUTURE AD,IUSnIENTS DUE TO SmLEMENT 3
~~, NON-SEISMIC COmA TlBILITY. E. STRUCT. I 1m! , 1m) , 3
~6. DEFECTIVE INSTALLATION POSSIBLE I Inl.'> (nl , J
~~, PROBLEM WITH L.~TERAL RESISTANCE Ip) I (0/ .'> 101 ,
~S. 1I0N-PREDICnBLE BE.~RING CAPACITY
2~. PROBLEM WITH OBSTRUCTIONS
I
I Iql.'> ,
.'>
J
.10. LATERAL CnEEP irll ~ (r J .~ (r I'

26 110 6~ &3 "



BEST METHOD

FIGllRE 6

990
A. EQUIPMNET ~IA VBE !.ARGE CRANES. ROOF PENETRATION REQUIRED.
B. FLOOR ./OIST IN ROOF &. GARAGE FLOOR MAV REQUIRE CUTTING.
C. VIBRATIONS &. NOISEOF DRIVING EXCESSIVE. lETTING NOT ALLOWED.
D. PROBABLE THAT STEEL CASING REQ'D .. DEWATERING AND/OR DRILLING MUD REQD.
E. IF WATER TABLE HIGH. EXCAVATION BY HAND NOT RECOMMENDED.
F. SOIL TO BE SPILLED FROM DRILL THROUGHOUT UNDERSIDE OF STRUCTURE. WILL
NEED CLEAN-UP FOR EROSION CONTROL.
G A !.ARGE AMOUNT OF SOIL TO BE REMOVED. 9 CVJPIER. OR APPROX. I; C.V. (LOOSE)
TO REMOVE &. HAUL AWAV.
H. IF OBSTRUCTIONS ARE PRESENT IT MAV TAKE LONGER THAN ONE DAY TO INSTALL A
SINGLE PIER EXCAVA TlON. PLUS REINFORCEMENT &. CONCRETING INSPECTION.
I. EQUIPMENT SO LARGE THAT DAMAGE TO STREET, UTILITIES &. TO BUILDING STRUCTURE
(FRONT RETAINING WALL) MAY OCCUR.
.I. LARGE CONCRETE QUANTITIES MEAN PUMPERS. READY MIX TRUCKS &. ADDITIONAL
ACCO~IPANVING NOISE ON SITE.
K. LARGE AUGER. PERHAPS AS MUCH AS 48' DIAMETER REQD. THRU ROOF &: FLOORS OF
GARAGE,
L. RELIEF OF STRESSES WHEN BOTTOM SOIL IS UNCOVERED &. LOOSE MATERIAL ALLOWS
SO~IE SETTLEMENT. SA Y2" PRIOR TO LOAD BEING CARRIED.
M LARGE CONCRETE PIERS ARE INFLEXIBLE. ARE STRUCTURES ON THEIR OWN &. WILL
NOT FLEX WITH REMAINDER OF STRUCTURE IN SEISMIC ACTIVITY.
N, DEFECTI'iE INSTALLATION MAY NOT BE DISCOVERED UNTIL YEARS !.ATER.
0, THESE MEMBERS WILL REQUIRE TIE BACKS. CABLE. RODS OR HELIX.
P. HELIX PIERS ARE THE ONLl' TRULl' MEASUREABLE CAPACITY UNITS.
Q. BOULDERS WILL REQUIRE TO BE BROKEN (NOISE. MORE EQUIP~IENTl.
R. TIE BACKS REQUIRED. SEE (0)

PROJECT GEOTECHN ICAL &. STRUCTURAL ENGINEERS RULED OUT DRILLED CONCRETE PIERS. HAND
DllG CONCRETE PIERS & LARGE STEEL H OR PIPE PILES.

STEEL .lACK PILES NOT CONSIDERED DUE TO BUILDING REACTION LOADS NOT AVAILABLE.

FIGURE 7

991
RECENT DEVELOPMENTS IN THE DESIGN OF
PILES LOADED BY LATERAL SOIL MOVEMENTS

D.P. Stewart l , M.F. Randolph 2 and R.J. JewelP

ABSTRACT

Piles supporting bridge abutments on soft clay may be loaded laterally from
horizontal soil movements generated by the approach embankment. However,
prediction of pile group behaviour under these conditions is difficult. The paper
describes new developments in the understanding of pile group response to lateral
soil movement. The observed response of pile groups is illustrated with data from
recent centrifuge tests. Empirical design charts based on field and centrifuge test data
are presented. Other methods of analysis are also described and are compared with
centrifuge test results.

1. INTRODUCTION

When an embankment on soft clay forms an approach to a piled bridge


abutment, time dependent movements within the clay may produce significant lateral
loading of the piles, Figure 1. This loading will induce bending moments and
deflections in the piles, and in severe cases may lead to structural distress or failure
of the piles or bridge structure. A number of case histories are reported in the
literature, for example Nicu et aI. (1971) and Stermac et aI. (1968).

Lateral loadings predicted by the various semi-empirical or theoretical analyses


available have generally shown poor correlation with the limited data obtained from

1 Research Associate, Department of Civil Engineering, The University of Western Australia,


Nedlands, W A 6009, Australia.
2 Professor, Department of Civil Engineering, The University of Western Australia, Nedlands, WA
6009, Australia.
3 Associate Professor, Department of Civil Engineering, The University of Western Australia,
Nedlands, W A 6009, Australia.

992 Stewart, Randolph and Jewell


full scale field trials. As would be expected, these data arise from situations with
variable geometry and soil conditions and are often of poor quality. Thus there have
been major uncertainties in the estimation of bending moments and deflections
induced in piles under these conditions. Because of this, bridge designers have often
adopted a conservative design, incorporating caissons to shield piles from lateral soil
displacements, or delaying pile installation until virtually full settlement of the
embankment has occurred. However, if the bending moments and deflections
induced in the piles can be estimated accurately, then more cost effective design and
construction procedures may be implemented.
earth pressure from
retained embankment

embankment
'"
lateral loading from - .
soil movements ----.
soft clay

\\ stiff substratum

Figure 1. Pile group loaded by lateral soil movement.

To enable the behaviour of pile groups subjected to loading from lateral soil
movements to be understood more fully, and to provide data to assess the accuracy of
design methods, a series of centrifuge model tests have been performed recently at
The University of Western Australia (Stewart 1992). In this paper, a brief overview
of the observed behaviour of piled bridge abutments on soft ground is given, and is
illustrated with centrifuge test data. Empirical design charts that were produced using
the centrifuge results and supplemented with data from a number of field sites, are
presented. A relatively simple method of analysis, and a numerical technique are also
described and compared favourably with centrifuge test data.

2. EXPERIMENTAL DATA

A limited amount of field data has been published regarding the behaviour of
piles subjected to loading from lateral soil displacements. However, the available
data are largely qualitative in nature or of limited value, as noted previously.
Recently, Springman (1989) conducted a series of centrifuge model tests to examine
pile loadings generated by a nearby surface load. That study examined the response
of relatively stiff piles (equivalent to I m diameter reinforced concrete) installed in a
relatively thin layer of soft clay (6 to 8 m thickness prototype scale).

993 Stewart, Randolph and Jewell


Stewart (1992) conducted further centrifuge model tests to clarify the behaviour
of pile groups adjacent to embankments on soft clay. The response of more flexible
piles (steel H piles - 310UC158 sections) in deeper soft clay deposits (8 to 18 m
thickness prototype scale) was investigated, with particular emphasis on bridge
abutment configurations. In these tests, the model pile groups consisted of 14 vertical
piles (two rows of seven) held in a rigid cap located 2 m above the soil surface and
able to deflect freely. The piles were installed through a soft clay layer and into an
underlying dense sand stratum (providing fixity), before construction of a sand
embankment in six stages adjacent to the group. The embankment had a front slope
of I.5H to 1V, and was constructed to a maximum height of around 8.5 m. Four
instrumented piles were included in the pile group, with the response monitored in
terms of pile bending moment distributions and pile cap deflection. The work has
been described by Stewart (1992) and Stewart et al (l994b) and so only a brief
overview of the observed behaviour will be given here.

Pile bending moment distributions (scaled up to prototype units) from a typical


centrifuge test performed by Stewart (1992) are shown in Figure 2, illustrating
similar responses for different piles in the same row of the group. In this figure, the
term "front row piles" refers to the row of piles closest to the embankment. The
maximum bending moments in the tests occurred close to either the pile cap level or
the interface between the soft and stiff layers. The precise location of the maximum
bending moment was dependent on the variation in stiffness of the soft stratum with
depth and on the presence of raking piles in the group.

Bending moment (kNm)

-800 o 400 800


5
pile cap
tree ., r
o

so~ clay :B piles
~
, front row

..... ... ".'


piles
5

I .~
/
:: 10
...
~.~
Co
Q)
C

15 \1
dense sand ~!>

0
20
<

25

Figure 2. Typical pile bending moment distributions from a centrifuge test.

994 Stewart, Randolph and Jewell


The response of a pile group with increasing embankment height is illustrated
with typical data on Figure 3, as the maximum bending moment (M max ) in each pile
and the pile cap deflection (y) plotted against the average vertical stress applied by
the embankment (q). Two generally linear sections are identifiable in the data,
showing that bending moments and deflections induced in the piles are initially
relatively small. A substantial increase then occurs after a certain threshold loading is
passed. The threshold (at the intersection of the two lines) was consistently observed
to occur at an embankment load of about three times a representative undrained shear
strength of the soft stratum, SUo This load corresponds roughly to a factor of safety
against failure of the embankment of about 1.7 (5.14/3). The increased gradients at
higher load levels are caused by an increase in soil displacement due to the initiation
of significant plastic deformation in the soft stratum. Thus the threshold loading
could be effectively considered as a limit to elastic behaviour. This observation
compares well with field data on lateral soil displacements and pile loadings, as
detailed by Stewart et al (1991) and Stewart (1992). When plotted to logarithmic
scales, the data follow relatively linear trends, indicating that relationships of the
form : y (or M max ) = a qb may be appropriate, where a and b are constants.
Therefore, the threshold loading is not a sharp break but occurs somewhat gradually.
Recently, Kimura et al. (1994) described a series of centrifuge model tests with piles
either fixed or pinned at the base of the soft stratum, and mainly examining the
stability of the pile group. The data shows similar trends to those described above.

800 -r---------~-----, 600 -r-------------:I[~


E
z E
"'" 600
:::
cQl .s
E
, 400
o
E 400 ~
Cl
c
~
"C
'5 go 200
c
1l 200
(,)

~
x ii:
'"
::E
0+-=::::&-,....----,.----.----.--...,.--.-1 O~=*=~=__.-___,_-__._-_._~

o 20 40 60 80 100 120 o 20 40 60 80 100 120


Errbankrrent load (kPa) Errbankrrenlload (kPa)

Figure 3 : Typical pile group response to adjacent embankment construction.

3. EXISTING DESIGN TECHNIQUES

A number of methods have been presented for the analysis of piles and pile
groups subjected to lateral loading from horizontal soil movements, as described by
Stewart et al. (l994a). These methods encompass the full range of geotechnical
analysis categories proposed by Poulos (1989) and may be further classified into a
broad grouping describing the basic approach to the problem:

995 Stewart, Randolph and Jewell


(i) Empirical methods - the pile response is estimated in terms of maximum
bending moment and pile cap deflection on the basis of charts developed from
field or laboratory data;
(ii) Pressure based methods - a pressure distribution acting against the piles is
estimated in a relatively simple manner, and is generally only used to
calculate the maximum bending moment in the piles;
(iii) Displacement based methods - the distribution of lateral soil displacement is
input and the resulting pile deflection and bending moment distribution
calculated; and
(iv) Finite element analyses - the piles are represented in the mesh and the overall
soil-pile-surcharge response calculated.

The empirically based and pressure based design methods are attractive from a
design perspective, since estimates of maximum bending moment and pile cap
deflection can be obtained relatively quickly and easily. However, the majority of
these approaches are very simple and limited in applicability. The displacement
based methods allow more accurate representation of soil stratigraphy and loading
conditions, although accurate estimation of the free soil displacement is notoriously
difficult, even if vertical displacements can be well predicted (Poulos, 1971). Finite
element analysis provides a good opportunity for accurate representation of the entire
problem, although is obviously more complex and time consuming and possibly
unattractive for general design use.

4. NEW EMPIRICAL DESIGN CHARTS

Several empirical relationships have been proposed on the basis of field data,
although these have been found to exhibit a great deal of scatter, or are plotted
inappropriately (Stewart, 1992). New empirical design charts that were developed on
the basis of centrifuge test data and field data (Stewart et al. 1994a) are described
here.

Dimensionless groups were examined in order to compare experimental data


from various sources. The data are plotted on Figure 4 as the non-dimensional
change in maximum bending moment (M q) and in pile cap deflection (Yq) against the
relative soil-pile stiffness (K R). The first two non-dimensional groups were chosen by
reference to the solutions for moment and deflection of a uniformly loaded beam, so
that the data were examined within a consistent framework:

M = ~max
q ~qdL2
eq

where q = embankment load, d = pile diameter or width, L eq = equivalent length of


piles between points of fixity (see below), E p =
Young's modulus of pile, I p =
moment of inertia of pile, E s = representative stiffness of soft clay layer, h s =
thickness of soft clay layer.

996 Stewart, Randolph and Jewell


1.000 , , - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

" c:OJ
.-

~ ~ 0.100 Fbstthreshold
~ E
'5 g>
m~

"o " OJ
"iii .c
g E
E ::l
'9 .~ 0.010
" '"
~ E
Pre-threshold
range

0.001
lE5 lE-4 lE-3 1E-2 1&1 lE+O
Relative stiffness. (K )
R

0.1

Fbst-threshold 0
0.01

0.001 Pre-threshold A
range
A A
A

0.0001
1&5 1&4 1&3 1&2 1&1 lE+O
Relative stiffness. (K )
R

LEGEND

o Centrifuge data - Stewart (1992) 0 Centrifuge data - Watabe (1994)

o Centrifuge data - Springman (1989) D. ... Field data

solid symbols =pre-threshold response; open symbols =post-threshold response


Figure 4. Empirical design charts for maximum pile bending moment and pile cap
deflection.

997 Stewart, Randolph and Jewell


Data are shown for pre-threshold load levels (q < 3s u ) with ranges corresponding to
the gradient of bounding lines through the origin. For post-threshold load levels (q >
3s u) a best fit to the data was judged. Data are shown from centrifuge tests conducted
independently by Springman (1989), Stewart (1992) and Watabe (1994) (described
by Kimura et al. 1994). Field data presented by Heyman (1965), Marche and Lacroix
(1971), Bigot et al (1977), Ingold (1977), Sirawardane et al. (1984) and Hull and
McDonald (1992) were utilised by simplifying the soil conditions at each site. These
charts are an update of those presented by Stewart et al. (1994a) with a great deal of
additional data included. The trend lines shown on Figure 4, are unchanged from the
original charts, and still appear to be reasonable. Most of the available field data are
for pile cap displacements, as these are most easily measured, although some bending
moment data have also been collected.

The pile configurations for the data shown on Figure 4 include piles pinned at
the head, freeheaded piles, and groups connected by a rigid cap. To account for
different head fixity conditions, the case of a rigid pile cap preventing rotation but
not deflection was chosen as the reference configuration. Thus the reference piles can
be thought of as single beams under the action of a distributed load, fixed at the base
of the soft stratum and with a moment loading at the head to prevent rotation.
Comparing the solutions for head deflection and maximum bending moment for this
reference pile with those for a cantilever (free head) and a propped cantilever (pinned
head), an equivalent pile length may be defined as the length of a reference pile to
produce similar maximum moment and head deflection as the pile being considered.
Therefore, for each head fixity condition:

rotation prevented L eq =L
pinned Leq = 0.6 L
free L eq = 1.3 L
where L is the length of the piles from the head to the base of the soft stratum. For
piles prevented from rotating at the head, but pinned at the base, L eq = 0.2 L.
Obviously a pinned head pile will have no head deflection, and so for this case
maximum bending moment only is considered.

The available data suggest that the post-threshold peak bending moment is
relatively independent of relative soil-pile stiffness for K R > 10- 2. For pre-threshold
loadings, the peak bending moment is more dependent on the value of K R . For very
stiff piles, the pre- and post-threshold values of M q converge, and thus a distinct
threshold would be less obvious. Inspection of Figure 4 suggests that M q is roughly
proportional to KRO.O? to K RO.3, while yq is roughly proportional to K RO.5. Therefore,
for a given site, ignoring any change in pile width for a change in I, and thus
considering q, d, L eq , h s and Es as constants, these relationships reduce to M max oc
EI 0.07 to 0.3, and y oc EI -0.5. Thus, a doubling in pile stiffness at a given site would
be expected to lead to an increase in maximum bending moment of between 5 and
25%, and a reduction in pile cap deflection of about 30%.

998 Stewart, Randolph and Jewell


The data on Figure 4 may be used to construct design envelopes for maximum
bending moment and deflection as shown on Figure 5. The effect of such factors as
pile spacing, group size and configuration, and embankment shape are not fully
accounted for. However, it should be noted that pile cap deflections may be reduced
substantially by piles with a large rake away from the embankment, or by the
existence of a thick and relatively stiff surface layer. Also, the pile spacing across the
face of the embankment will have an influence on the pile group response.
Nevertheless, the charts provide a means to obtain a rapid estimate of likely pile
group behaviour. Since the charts are derived from real data, there is high confidence
in the predictions obtained from this method provided that the situation being
analysed is compatible with those from which the data were obtained. Stewart et al.
(1994a) provide recommendations on the use of these charts when piles are installed
after embankment construction is completed, and also describe an alternative
empirical technique based on curve fitting to field and laboratory data.

c:
.Q
13
III
';
'C
<5
E
III
E
o
::::E

Embankment load, Q

Figure 5. Construction of design envelopes using the empirical design charts.

5. NEW DISPLACEl\1ENT BASED DESIGN l\1ETHOD

A new design approach was developed (Stewart et al. 1994a) on the basis of a
relatively simple method proposed by Springman (1989). The method will not be
described in detail here, but revolves around using a simple soil displacement
mechanism to approximate the lateral soil displacement, and then relating this
displacement to that of the pile. Soil displacements are derived from a simple elastic
triangular displacement mechanism similar to that described by Bolton et al. (1991).
The lateral soil displacement is then described in tenns of the average mobilised
shear stress in the soft layer, approximated by a lower bound plasticity solution for
collapse of an infinitely wide strip footing. This comprises a two-zone stress system,
with the piles represented as a frictionless sheet-pile wall located at the stress
discontinuity below the edge of the strip footing. The piles are assumed to carry a
uniform pressure loading across this discontinuity, and the pressure is used to
calculate the pile cap deflection and maximum pile bending moment.

999 Stewart, Randolph and Jewell


This type of approach is particularly attractive for relatively complex soil-
structure interaction problems, since the response of each individual element of the
problem can be incorporated approximately and the overall response calculated. The
equations resulting from the approach are relatively simple, although cumbersome
due to their length. The parameters required for the analysis are pile bending
stiffness, pile width and spacing, depth and undrained shear strength of the soft layer,
stiffness of the soft layer and underlying stiffer stratum, and the embankment load.

Application of the method is illustrated in Figure 6, comparing calculated results


with data from a typical centrifuge test. The calculations are extended up to an
embankment load of 3s u (an approximate elastic limit), to be consistent with
observations from the centrifuge data discussed earlier. The calculated results
provide a reasonable approximation to the test data, although the initial response is
overestimated. However, the stress-strain behaviour of the soft clay is known to be
highly non-linear, and therefore the selection of a representative elastic modulus for
the soft layer is problematic. Nevertheless, the proposed method approximately
models the significant features of the embankment-sail-pile interaction, and has been
shown to yield consistent results for configurations of different relative soil-pile
stiffness.

E 600 300
z
e. E
.sc:
'E
CIl !
E 400 .9 200
0 U
E CIl
Cl ';
c:

"t:l
'6
c: 200 Co 100
co
: u

.
Q)
.D
~
X
<lI
. ~ ii: ~
~ 0
0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Errtlankment load (kPa) Errtlankment load (kPa)

Figure 6. Results from the displacement based design method (lines) compared with
centrifuge test data (symbols).

Non-linear stress-strain behaviour is relatively important in the behaviour of


embankments on soft clay, since the factor of safety against failure is generally low
(= 1.5). For most other soil-structure interaction problems, much higher factors of
safety are generally chosen in order to limit deformations and to ensure an essentially
linear response. Another non-linear effect is the changing geometry of an
embankment as it is built up in height. If a linear elastic deformation analysis was
performed, incremental subsoil displacements at the embankment toe would be found
to decrease with increasing embankment height since the loading effectively moves
further from the toe. However, this trend is inconsistent with field and centrifuge
model data of embankments on soft clay, that indicate displacements increasing at a

1000 Stewart, Randolph and Jewell


gradually greater rate as the embankment height is raised. This further underlines the
importance of non-linear stress-strain response of the soft stratum.

Non-linear stress-strain response can be incorporated into the analysis simply,


by using incremental values of secant stiffness relevant to the average mobilised
shear stress in the soft clay layer. The effect of changing embankment geometry can
also be included by applying a correction factor, as detailed by Stewart et al. (1994a).

Calculations were performed for the same centrifuge test as shown in Figure 6,
but using a non-linear stress-strain curve for the soft clay, and applying corrections
for the embankment geometry. A modified hyperbolic curve was fitted to laboratory
simple shear test data for the kaolin clay used in the experiments. The results are
shown in Figure 7, and can be seen to compare extremely well with the experimental
data, indicating the correct trends and providing a good match to the magnitude of
the results. The method shows a great deal of promise, provided that the stratigraphy
can be represented simply.

600 300
E
z
=.
cCD
E
..
c
E 400 .Q 200
0 t;
E Q>

Cl
C
~
'0
'5 200 a. 100
c ctI
CD U
.Q
~
x
ctI ii:
:::E 0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Errtlankrrenlload (kPa) Errtlankrrenlload (kPa)

Figure 7. Results from the displacement based design method with non-linear effects
incorporated (lines), compared with centrifuge test data (symbols).

6 PLANE STRAIN FINITE ELEMENT ANALYSIS

A number of alternative finite element representations have been proposed for


piles loaded by lateral soil movements. These include axisymmetric analysis with
non symmetric loading (Carter 1982), plane strain analysis (Sirawardane et al. 1984
and Stewart et al. 1993) and fully three dimensional analysis (Springman 1989).
While each of these approaches has merits, plane strain analysis is probably the most
convenient means of approximating the obviously three dimensional nature of the
problem. The application of this technique will be described here and compared with
centrifuge test data.

1001 Stewart, Randolph and Jewell


=

Plan on pile group Wall stiffness per pile Equivalent sheet-pile wall
stiffness per metre width

Figure 8. Equivalent sheet-pile wall representation of a pile group.

The soil stratigraphy and embankment loading of a typical problem of this


nature can often be depicted adequately by a cross-section parallel to the direction of
soil movement imposed on the piles. Representation of this section with a finite
element mesh is then relatively straightforward, and the piles could be modelled with
elements similar to those used for the soil but with higher stiffness (Sirawardane et
a1. 1984). Alternatively, beam elements could be incorporated into the mesh (Stewart
et al. 1993) and thus probably model the bending behaviour of the piles more
accurately. Both approaches model the pile rows as equivalent sheet-pile walls
(Figure 8) and appropriate properties should be chosen. However, with the piles
explicitly incorporated in the finite element mesh in this way, the soil can not deform
around and past the piles, and the results may be significantly in error. If the piles are
modelled with beam elements, the beam nodes can be defined separately from those
describing the soil layers. The soil and the piles can then be joined with joint or
spring elements to allow relative movement, thus allowing the soil to flow past the
piles and providing a more accurate representation of the real three dimensional
behaviour. This relative movement is of minor importance for very flexible piles, but
becomes critical as the pile stiffness increases.

The centrifuge tests conducted by Stewart (1992) were modelled using a plane
strain finite element analysis, with the piles modelled with beam-column elements. A
typical mesh is illustrated in Figure 9. The pile nodes were defined separately from
the soil nodes, and the two were connected by nodal joint elements having shear and
normal stiffness and a maximum normal force. This allowed an approximate
representation of the development of lateral resistance with relative soil-pile
movement and ultimately the full limiting soil pressure acting on the piles, similar to
a p-y curve that is used in other forms of analysis for laterally loaded piles. The soil
strata were modelled with Mohr-Coulomb materials for the sand layers, and a Tresca
material for the clay layer.

1002 Stewart, Randolph and Jewell


~ 0-----0 Beam-Column Elements
.............
.............
!'"&
"""

Figure 9. Typical finite element mesh used for the analysis.

Bending, shear and axial stiffness of the piles were input as the average of the
soil and pile properties over an equivalent 1 m thickness of the mesh (Figure 8). Thus
the wall stiffness was specified per metre width, and the bending moments and shear
forces resulting from the analysis were factored up by the pile spacing to obtain the
moments and forces per pile. Results from analysis of a typical centrifuge test are
presented for comparison with the centrifuge data on Figures 10 and 11. The results
are shown as plots of maximum pile bending moment and pile cap deflection against
average load applied by the embankment, and as pile bending moment distributions
for the embankment stage where the closest agreement between maximum bending
moments was obtained.

The maximum bending moment and pile cap deflection predicted from the
analysis generally compare well with the centrifuge data over a range of embankment
loads, although the moments and deflections are overestimated initially. The
predictions then fall below the test data as the embankment load increases, although
the finite element results increase at a rapid rate during the last stage as limiting
equilibrium is approached. The maximum calculated bending moments were slightly
larger for the front row piles than for the rear row, similar to the test data. The pile
bending moment distributions were reproduced quite well by the analysis, indicating
that the general pattern of soil displacements is relatively accurate. However, it
should be noted that not all analyses of the centrifuge tests were as close to the test
data as the example shown here.

A threshold load at which moments and deflections begin to increase rapidly is


readily identifiable in the finite element results, although this occurs well in excess of
the value (about 3su ) observed in the centrifuge test data. The discrepancy is
presumably a result of the linear elastic behaviour assumed for the soil before yield,
as opposed to the known non-linearity of the kaolin used in the tests. This non-
linearity is thought to be very important, and further work is in progress to
incorporate non-linear stress-strain response.

1003 Stewart, Randolph and Jewell


1200 1000
E FE results
Z
C E 800
900 ..
E
Q) c:
0
E
0
E 600
ti
~
600
c entrif uge data
~.
C> Qj
c: "C 400
'5 Q.
c:
CD
300 ~
J:l
2! 200
>C
ell c::
:::iE
o.
20 40 60 80 100 120 0 20 40 60 80 100 120
ErrtJankrrent load (!<Pa) ErrtJankrrent load (kPa)

Figure 10. Results of finite element analysis compared with bending moment and
deflection data from centrifuge tests.

Bending moment (kNm) Bending moment (kNm)


-300 -150 0 150 300 -300 -150 0 150 300
-5 -5

o
,. o I-

5
1\ 5
tJ.,
I I _I
==a. 10 ==a. 10

~ l1
OJ OJ
Cl Cl
15 15

20
~
..........
~
20
..... ~
.;'

I-....
.~ ~
25 25
(a) front row piles (b) rear row piles

Figure 11. Bending moment distributions from finite element analysis (lines)
compared with centrifuge test data (symbols).

7 CONCLUSIONS

Existing design techniques for piles subjected to loading from lateral soil
movements have generally been found to be inconsistent, or show poor correlation
with the available field data. A recent and comprehensive review of design methods
(Stewart, 1992) has further emphasised these shortcomings by comparison with high
quality centrifuge model test data.

1004 Stewart, Randolph and Jewell


This paper has described some recent developments in the understanding and
analysis of pile groups loaded by lateral soil movements. New empirical design
charts have been developed to facilitate rapid assessment of pile group response. A
new analytical method has been developed based upon using a simple soil
deformation mechanism to estimate the approximate relative soil-pile displacement.
Extension of the method to account for non-linear stress-strain behaviour of the soft
stratum, and applying corrections for the embankment geometry was very successful,
and illustrated the attractiveness of this relatively simple design approach. Plane
strain finite element analysis may also be of some benefit, and work described here
has shown good correlation with centrifuge test data, provided that care is taken to
model the flow of soil past the piles.

ACKNOWLEDGEMENTS

The work described in this paper was funded by the Main Roads Department of
Western Australia, and the support of the Commissioner of Main Roads is gratefully
acknowledged. The first author was supported by a research scholarship from the
University of Western Australia. Thanks to Professor T. Kimura and Y. Watabe for
generously providing details of their centrifuge test results.

APPENDIX. REFERENCES
Bigot G., F. Bourges, R. Frank and Y. Guegan (1977) Action du deplacement lateral
du sol sur un pieu, Proc. 9th ICSMFE, Tokyo, 1,407-410.
Bolton M.D., H.W. Sun and S.M. Springman (1991) Foundation displacement
mechanisms, Ground Engineering, April, 26-29.
Carter J.P. (1982) A numerical method for pile deformations due to nearby surface
loads, Proc. 4th ICONMIG, Edmonton, 2, pp 811-817.
Heyman L. (1965) Measurement of the influence of lateral earth pressure on pile
foundations, Proc. 6th ICSMFE, Montreal, 2, 257-260.
Hull T.S. and P. McDonald (1992) Lateral soil movement loading on bridge
foundation piles, Proc. 6th Aust. New Zealand Conf. Geomechanics, Christchurch,
146-150.
Ingold T.S. (1977) A field study of laterally loaded piles, Proc. 9th ICSNlFE, Spec.
Sess. 10, Tokyo, 77-80.
Kimura T., 1. Takemura, Y. Watabe, N.Suemasa and A. Hiro-oka (1994) Stability of
piled bridge abutments on soft clay deposits, Proc. 13th ICSMFE, New Dehli, 2,
721-724.
Marche R. and Y. Lacroix (1972) Stabilite des culees de ponts etablies sur des pieux
traversant une couche molle, Canadian Geotechnical Journal, 9(1), 1-24.

1005 Stewart, Randolph and Jewell


Nicu N.D., D.R. Antes and RS. Kessler (1971) Field measurements on instrumented
piles under an overpass abutment, Highway Research Record, 354, 90-99.
Poulos H.G. (1971) Difficulties in prediction of horizontal deformations of
foundations, JSMFE, ASCE, 98(SM 8), 843-848.
Poulos H.G. (1989) Pile behaviour - theory and application, 1989 Rankine Lecture,
Geotechnique, 39(3), 365-415.
Sirawardane H.I., L.K. Moulton and RJ. Chen (1984) Prediction of lateral
movement of bridge abutments on piles, Transportation Research Record, 998,
TRB, Washington, 14-24.
Springman S.M. (1989) Lateral loading of piles due to simulated embankment
construction, PhD Thesis, Engineering Department, Cambridge University.
Stermac A.G., M. Devata and K.G. Selby (1968) Unusual movements of abutments
supported on end-bearing piles, Canadian Geotechnical Journal, 5(2), 69-79.
Stewart D.P., R.I. Jewell and M.F. Randolph (1991) Embankment loading of piled
bridge abutments on soft clay, Proc. Int. Conf. Geotech. Engg. for Coastal
Development, Yokohama, JSSMFE, 741-746.
Stewart D.P., RJ. Jewell and M.F. Randolph (1993) Numerical modelling of piled
bridge abutments on soft ground, Computers and Geotechnics, 15(1), 21-46.
Stewart D.P., R.I. Jewell and M.F. Randolph (1994a) Design of piled bridge
abutments for loading from lateral soil movements, Geotechnique, 44(2),277-296.
Stewart D.P., R.I. Jewell and M.P. Randolph (1994b) Centrifuge modelling of piled
bridge abutments on soft ground, Soils and Foundations, 34(1),39-49.
Stewart D.P. (1992) Lateral loading of piled bridge abutments due to embankment
construction, PhD thesis, The University of Western Australia.
Watabe Y. (1994) Personal communication.

1006 Stewart. Randolph and Jewell


Deep Timber Pile Foundations Using Geotextile Reinforced Embankments

Joseph H. Byrne, P.E.l

Abstract

An island homeowner's association was faced with replacing a collapsing bridge. The
bridge provided the only crossing to the island development and was located on the
narrows of a spring-fed lake. Inspection of the bridge revealed that the substructure
units had undergone severe damage due to settlement and translation. Failure of the
bridge piers had occurred from the large displacements. Soil borings revealed the
presence of a three meter thick marl layer with medium-dense sands present beneath
the marl layer. Due to the higWy compressible nature of the marl layer, and the pile
depths required, solving the problem via the use of a heavy concrete structure was not
feasible. Instead, a timber bridge structure was selected. The replacement bridge was
a 9 meter span, 3.7 meter clear width roadway structure designed for HS20-44
loading. In order to achieve sufficient end bearing, 36 cm diameter timber piles, driven
to depths ranging from 13.4 to 14.6 meters, were used. Light-weight fill material was
used on the bridge approaches to reduce overburden pressures, possible negative skin
friction, and settlement. To minimize lateral loads on the piles and compressive
stresses in the bridge deck, woven geotextile fabric layers were placed in the
embankments in the vicinity of the bridge abutments to reinforce and confine the soils.

Introduction

Geotextiles have proven very cost effective alternatives to conventional construction


materials in situations where stabilization of embankments over soft compressible soils
are concerned. Because geotextiles are lightweight and have high tensile strengths,
their application as reinforcing layers in soil embankments can result in an embankment
having a higher shear strength. This allows embankments to be constructed to greater

1 Project Manager, JCK and Associates, Inc., 2525 East Paris, S.E., Suite 160, Grand Rapids, MI
49546

1007 Byrne
heights because lateral displacements are reduced, thus improving the overall stability.
This is not a naw concept. Engineers have known for many years that reinforced
embankments can withstand much higher loads when placed in properly backfilled
soils [1], but only recently have successful pioneer projects and affordable
geosynthetics led to increased use of geotextiles in embankments.

Conventional grav1ty and cantilever wall systems made from concrete resist lateral
pressure by virtue of their large mass, In applications where large settlements are
possible due to soft subgrade soils, heavy structures may not be feasible, Ocotextile
walls are light-weight and derive their str~ngth from their flexibility. They must yield
or deform to mobilize their strength. In combination with light-weight soils and deep
piles, this system can minimize settlements and increase embankment stability while
offering considerable cost savings.

This paper presents a case study where soft compressible soils caused a bridge to fail,
The replacement bridge was a timber structure that used deep timber piles to penetrate
a soft compressible marl layer to achieve bearing capacity. A geotextile reinforced
wall was constructed in the approach embankments to minimize lateral loads on bridge
piles and lateral spreading of the embankment. A discussion of the findings from the
bridge inspection through post construction monitoring follows.

Inspection of Existing Bridge


The existing bridge was a three span structure, 7.3 meters long, 3.6 meters wide, with
a rise of 46 centimeters above the lake surface. The superstructure consisted of four
steel stringers and three timber stringers with a 6.4 cm thick wood plank deck, The
piers and abutments were concrete on concrete spread footings, The wood plank
bridge deck was new and had been replaced approximately 5 years prior to the
inspection.
The bridge superstructure was in fair condition, Analysis of the superstructure
revealed that the bridge deck alone could safely support a 3,6 metric ton load, This
was adequate for small passenger vehicles that frequented the bridge, However, the
bridge substructure was in terrible condition and was not safe for heavier vehicles such
as fire trucks. The concrete had undergone considerable deterioration and had failed
structurally in several locations as evident by severe cracking, The bridge piers were
in the worst condition. The cracked piers also revealed that there was insufficient
steel reinforcement.

A considerable amount of differential settlement and translation of the bridge


abutments had occurred. This caused the bridge piers to experience lateral forces
transferred via the bridge railing and stringers from the abutments to the piers. The
differential settlement of thtl abutments was apparent in the bridge railings. A
photograph of the old bridge is shown in Figure 1,

1008 Byrne
Based on the bridge inspection findings and structural analysis, it was recommended
that the bridge be closed. It was proposed that the bridge be replaced with a new
bridge rather than be rehabilitated because the substructure units were inadequately
designed and had experienced a large amount of settlement. Corrective rehabilitation
would have been too costly to be considered an economic alternative.

The proposed bridge was a light-weight timber structure with a span of 9.1 meters and
a clear roadway width of 3.7 meters. The bridge deck was supported by six, 36 cm
diameter piles. The abutments consisted of7.6 x 30.5 cm planks to a depth of2.13
meters below grade. Some of the advantages of this type of structure are lower cost,
ease of construction, and natural appearance. The timber bridge was prefabricated
and the materials were delivered to the site for assembly.

Figure 1. Photograph of Existing Bridge

Geotechnical Investigation

Prior to design and construction of the new bridge system, a geotechnical investigation
was performed. Two soil borings were drilled to 15.2 meter depths on both sides of

1009 Byrne
the bridge and Standard Penetration Testing was performed in accordance with ASTM
Standard D1586. Split spoon samples revealed that the first 3 meters were compact
fill soils. Below this layer was 4.3 meters of very loose, light brown, fine to coarse
sands with blow counts of only two. Below this stratum was a gray-white marl layer
with blow counts of zero. Beneath the marl layer was medium-dense, light brown,
fine to medium sands with blow counts ranging from 7 to 13. Soil properties from the
geotechnical investigation and laboratory analysis are summarized in Table 1. Soil
borings are shown in Figure 2.

Table 1. Soil Properties

Soil Depth Blows Density Friction Void Porosity Relative


Description Interval per (Kg/m3) Angle Ratio % Density
(meters) 0.3 m (deg.) %
Fine to
coarse sand o t07.3 2 1602 27 0.7 41 10
Gray-white 7.3
marl to 0 1602 0 3.0 75 -
10.4
Fine to 10.4
medium to 6 1762 33 0.6 37 50
sand 15.2

It was evident from the geotechnical investigation that the existing bridge had failed
primarily from excessive settlement. Observed differential settlement of the existing
bridge was approximately 20.3 centimeters. Settlement of the proposed embankments
was calculated from elastic settlement analysis and consolidation theory using
Equations 1 and 2 shown below. A total settlement of 19.3 centimeters was
theoretically possible. Differential settlement was estimated to be 10.2 centimeters.
Because large settlements were possible, it was decided that a pile foundation driven
to the medium-dense sand stratum were necessary to achieve bearing capacity and
minimize settlement of the bridge.

The ultimate bearing capacity of a 17.8 cm diameter timber pile tip embedded in the
medium-dense sand layer was determined to be 96.5 kN using Equation 3 shown
below. Negative skin friction from the overlying soils was possible due to
consolidation of the compressible marl layer and could be transferred to the pilings
resulting in an additional 38.4 kN per pile. The total working load per pile, including
negative skin friction, was 84.2 kN. Using 15.2 meter long piles driven into the
medium dense sand layer, a factor of safety of only 1.2 was possible. Test piles 19.8
meters long were specified in the event that sufficient bearing capacity was not
obtained at 15.2 meter depths.

1010 Byrne
RORING NO 1 RORING NO 2
STA, 1+35, STA. 1+85,
BORING DATE : 10- 7-91 BORING DATE: 10-7-91

DEPTH DEPTH
IN IN
METERS METERS
o o
WATER at ,61 .61
L,:<:}t::::::::::j TOPSOIL - MIXED WITH
SANDY GRAVEL

CD-- 3 --+C. ~S4 CD-- 3


I

~ LIGHT BROWN
FINE TO COARSE FINE TO COARSE
SAND
~
SAND

~l! MARL
GRAY-WHITE MARL
wi LAYER OF WOOD
CD------i I I I I
!"
,
.
'8" 10~ ; , . '. I
~::"I
I" , ',:',,', : I FINE TO COARSE
0---i' :'.,. 'Ii ~~NED T~ ~6~I~M SAND

I : ' ,. ORGANICS-
~ I ' ':' '. TRACE OF MARL
~,,'I
I,'
I . "' "1
15 I' . ' E I EV. 167.4 1 BROWN FINE SILT
EOB

Numbers in circles represenllhe numberofblowB to drive


a splilspoon sampler.3 meters.

Figure 2. Soil Borings at Bridge Abutments

L1Hl=HlX(~) Equation, 1
1

Where;
Mil - Immediate Settlement in the sand
layer,
HI - Thickness of sand layer.
E 1 - Soil modulas,
i1p - Pressure increase due to dead load,

1011 Byrne
Equation. 2

Where;
MI2 - Consolidation settlement of marl layer.
H2 - Thickness of marl layer.
PI - Pressure at center of marl layer.
M> - Pressure increase due to dead load.
eO - Initial void ratio.
Cc - Compression index.

Equation. 3

Where;
PI - Effective vertical stress at pile tip.
N - Bearing capacity factor.
Ai - Area of pile tip.

Design Approach

The design challenge at hand was to construct a new bridge over a compressible soil
layer that would accommodate HS20-44 loading and not undergo excessive
settlement. The most efficient solution was to penetrate the marl layer using pilings to
transfer loads to the denser sands below. The owner wanted a minimum 0.91 meter
clear height over the water surface to permit small boat passage. This required
approximately 382 m3 offill material to be placed above existing approach grades.
The new fill added a surcharge pressure of approximately 14.4 kPa. Due to saturated
soil conditions, and low blow counts, lateral pressure conditions were significantly
high. The critical load condition was obtained by placing the HS20-44 truck rear
wheels at mid-span. This positioned the front wheels of71.2 kN about 0.3 meters in
back of the bridge abutment. To accommodate lateral earth pressures, it was decided
that soil reinforcement was necessary. This was accomplished by designing a
geotextile wall that would act independently of the bridge. In addition to minimizing
lateral pressures acting on the pile abutments, the geotextile wall prevented lateral
spreading of the embankments. The proposed bridge plan and profile are shown in
Figure 3.

Byrne
1012
...... ~\II.R
-;;?'Wr..
>1//I~ ,d',._

----------- ...............
oU' .......... .. ...
'--,'

....---~.~

--
o '\\ /'/r:1----~-------- ---""\

-- '~'
...... >
-----
(.oJ
/
L _
PLAN

_-==::::_=====~EO~:~LE WALL
II \I
\I \I
\I I'- TIMBER PILES
.JL -.II.-

E-ROFILE
OJ FIGURE 3
"< NOT TO SCALE
I""l
::J
ro
The bridge embankments could theoretically settle a total of 19.3 cm. However, the
bridge was not expected to exceed a total settlement of 7.62 cm. The geotextile
reinforcement was not fastened to the bridge abutments because the marl layer was of
differing thickness on either side of the bridge. Differential settlement of the
embankments may have resulted in undesirable loads transferred to the bridge.

To prevent excessive settlement of the geotextile walls, a lightweight concrete sand


was specified to replace the existing fill soils, The fill sand specified had a dry density
of 1362 kglm 3 . A woven geotextile was used to construct a wall in the embankment.
The woven geotextile specified had a wide width tensile strength of 103 x 93 kn./m, a
vertical flow rate of 0.95 cm/sec'ian apparent size opening of 0.149 mm, a thickness
of 1.91 mm and mass of678 glm ,

Design of Geotextile Wall

The effect of the geotextile reinforcement was to constrain the embankment soils from
spreading laterally if the weak foundation soils experienced collapse. If collapse of the
weak foundation soils occurred, the horizontal thrust would be transmitted to the
reinforcement. In this manner, the lateral pressures are reduced. A 30.5 cm wide
column of loose, noncompacted sand separated the abutments and geotextile walls to
allow movement of the backfill during compaction. This had the effect of prestressing
the fabric. It was believed that this would allow the geotextile reinforcement to
mobilize the tensile strength as deformation of the compacted backfill occurred,
thereby reducing lateral pressures exerted on the abutments. A paper presented by
McGown, Loke and Murray [2], presents experimental results that confirm lateral
pressures are reduced by the placement of a compressible soil column that separates
the wall and the backfill.

The geotextile wall was designed using the methods reported by Koerner [3]. First,
the wall was designed with regard to internal stability to determine the required lift
thickness, embedment length, and overlap. Then, external stability was examined with
regard to overturning, sliding, and foundation bearing capacity.

Internal Stability

Lateral earth pressures were determined using conventional Rankine theory which
assumes no wall friction, Boussinesq elastic theory was used to determine live loads
from the HS20-44 wheels on the soil backfill. The lateral force exerted on the
abutment in the active state was 39.8 kN/m of wall. The vertical spacing of the
geotextile layers was determined by examining the pressure profile along the wall
using Equation 4 for a specific geotextile. The horizontal forces were summed and the
free body diagram was taken at a given depth. This resulted in the determination of a

Byrne
1014
vertical zone of influence that the geotextile reinforcement layer could accommodate
at a given pressure. It assumes that 100% of the allowable tensile strength of the
fabric is available which depends on sufficient deformation to mobilize strain in the
fabric and embedment to mobilize soil friction. The subgrade soils were very poor and
were suitable for sufficient deformation to mobilize the fabric tensile strength.

T
(j (d) x S = allow Equation. 4
h Y FS

Where;
O'h(d) - Lateral pressure as function of depth.
d - Depth below grade.
S - Vertical spacing of geotextile layer.
FS - Factor of safety, typically 3.
Tallow-Allowable stress in the fabric.

The vertical spacing computed at the greatest depth and pressure was 71 cm.
However, 46 cm spacings were specified to provide a more convenient length for
construction. A total of six layers were required using this spacing throughout the
wall profile.

Next the embedment lengths were determined using a similar approach. This method
investigates the embedment length required to accommodate the lateral forces at a
given depth. The frictional resistance along the embedded geotextile is a function of
vertical pressures and the interface friction angle. The embedment lengths were
determined using Equation 5.

L = Sy X (jh X FS
Equation. 5
e 2 x yxz x tan(b)

Where;
L - Required embedment length.
e
"y - The unit weight of backfill soil.
z - Depth from ground surface.
o - Interface friction angle between soil and
fabric.
Sv - The vertical spacing from Eq. 4.
O'h - Total lateral pressure at depth considered.
FS - Factor of safety ( typical 1.5 )

1015 Byrne
The required length of embedment computed was 3 meters at the greatest depth and
pressure. The required length of embedment at the water table interface was 1.5 m.
The embedment lengths are computed beyond the assumed wedge failure plane and,
therefore, additional length is required from the vertical face of the wall to reach the
failure plane. This length varies with depth and is greatest near the top of the wall. A
total length of 3 meters was used for each layer in the geotextile wall. The required
overlap length was computed in a similar manner as the embedment length at about
half the level of maximum horizontal pressure and was 1.2 m. A schematic of the
abutment and g~otextile wall system is shown in Figure 4.

Bridge Abutment

Assumed Failure Plane

I , ',- Geotextile Wall


i :
I :
I I
! ,
, I
, !
! I
I I
I I Lr - length of runout
I
: :
....--T1mber Pile
' ...

I I
Le -length of embedment
I I
I I
Lo - length of overlap
---.Yi-
Sv - vertical spacing

Figure 4. Abutment and Geotextile Wall System.

External Stability

A computer program called PC STABL 5M developed at Purdue University [4] was


used to analyze the non-reinforced embankment fill overlying the very loose sands and
soft compressible marl layer. The results indicated that the embankment would be in
imminent failure with a factor of safety of 0.98. The safety factor was calculated using
the Bishop method of moments. The theoretical failure surface was assumed to be
circular and resulted in a base failure occurring above the loose sand layer.

Byrne
1016
Conventional geotechnical methods were used to determine the external stability of
the geosynthetic reinforced wall. The geotextile wall was considered a rigid body and
the interfacial friction angles were incorporated into the analysis. External stability is
examined by determining the ratio of the resisting forces ( interfacial friction, mass of
wall) to the driving forces ( Rankine active pressures). The ratio determined in the
analysis resulted in the factor of safety. Factors of safety were examined for the case
of overturning, sliding along the base, and bearing capacity. The resulting factors of
safety were 7.1,2.5, and 2.4 respectively.

Construction of the Bridge and Approach Embankments

Removal of the existing bridge presented little difficulty. The bridge abutments and
footings were completely removed. In the vicinity of the bridge, the approach return
walls and footings were also completely removed. However, for the remainder of the
bridge approaches, the approach retaining walls were knocked off approximately 15
cm above the footings. The embankment approaches were stable in those areas and it
was decided they would be of benefit.

Construction of the new bridge began with the driving of 35.6 cm diameter timber
piles. A Vulcan Iron Works, Model No. 1-106, single acting, air driven hammer was
specified to drive the piles. It delivered a maximum rated energy of 15,000 ft-Ibs.
( 20,325 N'm) at 60 blows per minute. Blow counts of3-4 blows per inch ( 1-1.5
blows/cm ) were achieved for the last six inches of driven length. The allowable
bearing value determined from Equation 6 below and was 61.1 kips ( 272 kN). The
effects of negative skin friction ( 8,640 lbs. or 38.4 leN) are accounted for in
Equation 6. The ratio of the allowable bearing value to the working load 10,305 lbs.
( 45.8 leN) is 6 .

WxH 1
Qall= x--Fn Equation. 6
(S+O.I) FS

Where;
Qall - Bearing value of pile load in kN.
W - Weight of the striking parts in kN.
H - The effective height of fall in meters.
S - Average net penetration in cm per blow for the
last 15.24 cm of driven length.
Fn - Negative Skin friction in leN.
FS - Factor of safety taken as 6'*

* Note: Findings from the Michigan State Highway Commission (1965) Test Program revealed that
ultimate test load capacities exceeded design capacities by 2 to 6 times the dynamic pile formulae
depending on the formula.

1017 Byrne
The maximum driven pile length was 14.6 meters. The bearing stratum soils
experienced increased stiffness as more piles were driven and resulted in shorter driven
lengths to achieve blow counts of 1-1.5 blows per centimeter. The shortest driven pile
length was 13.4 meters.

Upon completion of pile driving, the contractor began installing the pile backing which
would serve to retain the soils. The contractor chose not to utilize a steel sheet pile
cofferdam to retain soils and maintain dry conditions. Instead, the contractor opted to
construct an earth dike and to continuously pump the water from the enclosure to
maintain dry conditions. A cofferdam was not specified as part of the contract and
was considered incidental to the construction of the bridge.

During construction of the earth dike, which consisted of lightweight fill sand placed
around the pile group, a classical bearing failure occurred. Approximately 12 meters
from the embankment, soils were observed above the water surface. It was believed
that the driving weight of the newly placed fill soils caused failure to occur along a
circular plane through the embankment, extending to a depth just above the loose sand
layer. This was supported by slope stability findings from the computer program PC
STABL SM. A schematic of the predicted and observed embankment failure are
shown in Figure S.

~-,----,------.. New embankment


Observed failure surface
Water surface

50 30 10 10 30 50
Predicted failure surface
Loose sand layer

Marl Layer

Figure 5. Embankment Failure

It was determined that the best course of action was to construct the lower portions of
the geotextile walls to increase bearing capacity prior to installing the pile backing.
This was very effective and facilitated completion of the pile backing. The remaining
portions of the geotextile walls were then completed. The construction procedure
required that a layer of geotextile fabric be placed on a compacted foundation. Next,

1018 Byrne
the required vertical lift of backfill was placed and compacted to 95% maximum
density as determined by the modified proctor method (ASTM D1557 ).
The required overlap length was then embedded in the sand lift and the next layer of
fabric was placed. The procedure continued until construction of the wall was
complete. A photograph of the completed bridge and approaches are shown in Figure
6.

Post Construction Monitoring

Settlement of the bridge and the approaches was monitored using a survey level. The
first year after construction, the vertical settlement of the bridge was observed to be
50 mm. The approach embankments were monitored at a point 1.7 meters in back of
the abutments. The west approach settled 1.3 cm while the east approach settled only
60 cm. The second year after construction, no additional settlement of the bridge was
observed. However, the west approach had settled a total 5.1 cm while the east
approach settled a total of3.2 cm. The settlement of the bridge corresponds well with
theoretical elastic settlement analysis if negative skin friction is not included in the
analysis. However, the settlement of the embankments was significantly less than that
determined from conventional elastic settlement analysis and consolidation theory.

It is believed that this can be attributed to a number of factors. Excavation of the


heavier fill soils and replacement with lightweight fill sands allowed placement of an
additional three feet of fill without an increase in overburden pressures. Removal of
the heavy concrete structure lessened overburden pressures and allowed the
compressible soils to relax. The new timber bridge loads were transferred through the
compressible marl layer to the lower, dense sands and presented no additional loads to
the compressible marl layer. Since overburden pressures of the marl layer were
reduced, drowndrag or negative skin friction was minimized on the timber piles and
theoretical and observed settlements were in close agreement. Incorporation of the
geotextile wall served primarily to reduce lateral spreading of the embankments.
Because the geotextile walls were independent of the bridge abutments, differential
settlement occurring in the embankments was not experienced by the bridge.

Conclusions

Incorporation of a geotextile wall in the approach embankments of the timber bridge


proved to be a very effective means of stabilizing the embankment over the soft
compressible foundation soils. By virtue of the woven geotextile's tensile strength,
lateral spreading of the embankment was avoided. This had the effect of minimizing
lateral pressures exerted on the abutments and confining the soils. By accommodating
lateral pressures, compressive stresses in the bridge deck were also minimized.

1019 Byrne
The collapse of the non-reinforced embankment during construction confirmed the
highly unstable conditions at the site. Subsequent placement of portions of the
ge6textile wall had the effect of increasing the bearing capacity of the soils and
facilitated completion of the project. Post construction monitoring of the bridge
revealed that settlements were significantly less than predicted settlements and those
observed in the old bridge. Using woven geotextiles allowed construction of a
lightweight "floating" wall that had sufficient strength and did not undergo excessive
settlement. Because the geotextile walls were constructed independent of the
abutments, undesirable differential settlements of the bridge were avoided.

Figure 6. Photograph of Completed Bridge

References

1. Transportation Research Board, National Research Council, 1983. Reinforcement


ofEarth Slopes and Embankments. National Cooperative Highway Research .
Program Report No. 290.. Washington, D.C.

1020 Byrne
2. McGown, A., Loke, K.H., and Murray, R.T., 1992. The Behavior ofReinforced
Soil Walls Constructed by Different Techniques. In Grouting, Soil Improvements
and Geosynthetics, Volume 2, Geotechnical Special Publication No. 30. Roy H.
Borden, Robert D. Holtz and Ian Juran, ( eds. ), American Society of Civil
Engineers in Cooperation with International Society of Soil Mechanics and
Foundation Engineering. New Orleans, pp. 1237-1248.

3. Koerner, R.M., 1990. Designing with Geosynthetics. 2nd edn., Prentice Hall,
Englewood Cliffs, New Jersey, pp. 164-173.

4. Achilleos, E., Purdue University, School of Civil Engineering, 1988. User Guide
for PC STABL 5M. Joint Highway Research Project. International Report. JHRP-
88119. West Lafayette, Indiana.

1021 Byrne
Design of Deep Foundations
for Cut-and-Cover Tunnels

Brian Brenner, PiE.'


Cindy Gagnon 3
C.K. Shah, P.E.

Abstract

This paper discusses analysis and design of deep foundations for


cut-and-cover tunnels. Cut-and-cover tunnels are typically designed to
float on their mat foundations. However, for some design conditions,
deep foundations can be required. Examples are discussed, with
illustrations of design conditions from the Central Artery/Tunnel Project
in Boston.

Introduction

In this paper, we discuss analysis and design of deep foundations


for cut-and-cover tunnels.

Cut-and-cover tunnels are typically designed to float on their mat


foundations. The weight of the tunnel section is less than the weight
of the soil it replaces. However, there are many special design
conditions where deep foundations can be required. Examples include:

Tunnel sections where cofferdam walls are part of the final tunnel
structure. In this case, most of the loads from above are
transmitted to subgrade by bearing on the slurry walls, not on the
tunnel base slab. Complications include design for staged
construction, "cantilever ramps" where only one side is supported
by a slurry wall, and tunnels with multiple boxes of varying
geometries.

Tunnels which pass beneath existing bridges and buildings requiring


deep underpinning. The deep foundations of the underpinning
interact with the tunnel structure.

, Professional Associate, PBQD, Bechtel Parsons Brinckeroff, One South


rtation, Boston MA 02110
Structural Engineer, B/PB
3 Senior Structural Engineer, Mistry Associates, B/PB

1022 Brenner
Design considerations for future structures in the air-rights space
above the tunnels. The design must provide for load transfer to
subgrade of a future conceptual building that may be constructed
years after the tunnel design is complete.

Included is a discussion of conceptual, preliminary, and final designs


from the Central Artery/Tunnel Project in Boston. This massive tunneling
project features miles of highway tunnel design. The project has
provisions for future air rights design, tunnel sections with and without
cofferdam walls, and a particularly complex and difficult underpinning
requirement in which an existing mile long viaduct expressway is to be
underpinned while the tunnel is built below it.

We begin with a discussion of types of excavation support walls and their


effect on the tunnel design.

Cofferdam Walls

Two broad categories of excavation wall support for cut-and-cover


tunnels can be considered: non-rigid cofferdam walls and rigid cofferdam
walls. Examples of non-rigid cofferdam walls are sheet piling, soldier
piling and lagging, and other relatively flexible, temporary wall
systems. Examples of rigid cofferdam walls include concrete slurry
walls, tangent pile walls, and soldier pile tremie concrete (SPTC) walls
(similar to slurry walls, with steel rolled sections used as the main
re inf orc ing) .

Factors affecting the decision for selection of cofferdam wall type


include the following:

Non-rigid cofferdam walls permit more soil deformation outside the


area of excavation.

Non-rigid cofferdam walls tend not to be watertight.

Installation of non-rigid cofferdam walls may impose more vibration


at the site than rigid cofferdam walls, depending on the method of
construction. Sheet piles are typically driven, for example.

Rigid cofferdam walls can be used for underpinning existing


structures.

Site conditions and restrictions may favor selection of one type


over the other. For example, low head room restrictions or
existing utilities can make slurry wall construction or sheet
piling difficult in some areas, whereas soldier piling and lagging,
if acceptable for the other reasons, can be more easily installed.

1023 Brenner
Incorporation of Cofferdam Walls in Tunnel Structure

For those areas where rigid cofferdam walls are to be constructed,


they can be considered with the overall tunnel structure in one of two
ways: integrated with the structure or structurally separate from the
rest of the tunnel (Figure 1). When the rigid walls are not part of the
final structure, separate structural walls are built within the
excavation.

///~ I//~ II/~

.. I . P

.1 K~
. I . : :.' r---- Wat ,srProoling
1 P, I'~
I . I" 1"'---. ......-Lean Concrete
I . I.'
1 I.,'~
.1 1 . '--Slu,rywali

I I
1'-=-=---=---=---=--=--=--=--=-=--=--=--=-=-~=-c=-,,~=-""",~d1 '

-
SUPPORT OF EXCAVATION NOT INCORPORATED
IN THE FINAL STRUCTURE

///~
L/~ - - - - -
I//~
~'J
I//~

. .
0

.
P
~. '--- --.....
0
r-- WaterProofing
Membrane

.
~
~SIUrryWall

. P
. o .

I SUPPORT OF EXCAVATION INCORPORATED


l
IN THE FINAL STRUCTURE
- -

Figure 1: Integrating Slurry Wall in Tunnel Section

Factors influencing the decision whether or not to include rigid


cofferdam walls in the final tunnel structure include the following:

Structurally separate walls require additional room, in plan


dimensions, for construction. This could impose additional costs
and right-of-way acquisitions.
1024 Brenner
Some studies imply that it is more economical to integrate rigid
walls with the tunnel structure.

Tunnel sections with rigid cofferdam walls included in the final


structure cannot have "blind side" waterproofing. Waterproofing
treatment is particularly important for these types of tunnel
sections.

Connections to tunnel roof and base slabs are more difficult when
the cofferdam walls are part of the final structure.

A further distinction can be made in the position of the rigid


cofferdam wall when it is not incorporated in the final structure. The
tunnel inside walls may be cast against the slurry wall, or the slurry
wall can be placed a few feet away and the inside walls formed. This
decision is based in part on considerations of how well the slurry wall
can prevent seepage. If leakage is prevented, tunnel outside
waterproofing can be placed directly against the wall without it being
damaged, and the rest of the tunnel cast.

A shear key can be provided at the base slab for additional mass
against buoyancy. However, in many cases specif ications require the
contractor to keep the water table depressed against the base slab until
backfilling operations are completed. Therefore, the completed tunnel
section would never be negatively buoyant. There would be no advantage
in including a shear key in the section to resist buoyancy.

A typical "floating" cut-and-cover box rests on its mat foundation.


The traditional design assumption was that the total weight down was
uniformly distributed as a reaction on the mat (Figure 2). With
computer-aided analysis, the basic method has recently been modified.
The subgrade soil is modeled by springs (beam on elastic foundation) or
by finite elements. This generally results in a load distribution that
is no longer uniform, but tends to be concentrated near the walls (Figure
2). The exact shape of the upward pressure distribution depends on the
relative stiffnesses of the modeled materials, and the geometry of the
tunnel section.

When cofferdam walls are incorporated in the final tunnel


structure, the overall behavior of the section changes (Figure 3). In
this case, the slurry walls act as hard points in the structural model.
Most of the vertical load from the roof, backfill and walls above are
transmitted to subgrade by the slurry walls. The base slab spans between
hard points. If the water table is high, the slab is loaded mostly by
pore pressure acting upward.

In some cases, the tunnel box section may have a slurry wall on one
side and a cast-in-place wall on the other. This introduces an
unbalanced condition to the support. In a structural analysis, the box
tends to hang as a cantilever section off the slurry wall. The degree
of cantilever action depends on the stiffness of the subgrade material
beneath the tunnel. Soft clays, for example, provide little support
relative to the hard point of the slurry wall on one side. The resulting
cantilever action leads to large moments in the slurry wall. On the
'025 Brenner
other hand, when founded on a hard till, the base slab may have enough
support so that the moments on the slurry wall are not excessive.

Surcharge

rrmml111
Soil Load

mmmm
Deadlood

t
61maginary Supports 0
illlliilllli
Pore water pressure

t11t11t11ill
Delta lood

Delta load = Deadlood + Surcharge +


Soil Water

TRADITIONAL METHOD ACTUAL SOIL REACTION


(Vertical load balancing) (Beam on elastic foundation)

Figure 2: Traditional Versus BOEF Base Slab Design

In cases with a slurry wall on one side and a cast-in-place wall


on the other, large moments in the slurry wall can be reduced by
constructing caissons. The caissons act as hard points, relieving the
cantilever effect on the wall. The caissons must be spaced closely
enough to match the stiffness provided by the continuous slurry wall in
the longitudinal direction. Typically, caissons spaced at 15 to 20 feet
(4.6 to 6.1 m) on center provide the needed stiffness. If the spacing
is much greater, the tunnel box begins to "sag" under loading between the
caissons, leading to moments building up in the slurry wall.

Another variation of this design condition is the case of multiple


tunnel boxes with slurry walls on either side (Figure 4). For this
condition, it is possible to design the tunnel box section as a
Vierendeel truss between hard point supports. The design decision is
based on the depth of the tunnel and backfill loading, the width of the
tunnel, and the foundation support conditions. It may be more effective
to introduce a row of caissons beneath the interior walls to reduce the
overall span of the tunnel section.

1026 Brenner
LOAD PATH

..

!
".

Figure 3: Load Path for Section with Slurry Walls

Analysis in the Longitudinal Direction

Cut-and-cover tunnel sections are typically analyzed in section.


Longitudinal effects are of lesser concern, because the stiffness of the
box is so much greater than the stiffness of the individual slab elements
that are loaded in section analysis. However, there are special
conditions when longitudinal analysis must be considered. The conditions
are largely related to changes in deep foundation support, and include:

Change in subgrade material strength. For example, in the


longitudinal direction, the subgrade may vary abruptly from a hard
till to a soft clay.

Change in wall types, from a tunnel section with slurry walls to


a section with cast-in-place walls. A tunnel section designed to
"float" on its mat may be connected to a stretch of tunnel with
slurry walls or caissons. Longitudinal analysis may indicate that
the "floating" portion of tunnel tends to hang off the hard points
in the adjacent section, resulting in a large moment and shear to
be transferred in the longitudinal direction. This can
particularly be a problem if the floating section is on soft
material, and there are concerns about the position of the water
table after construction is complete.

Localized change in vertical loading conditions, requiring deep


foundations and introduction of hard points along the box. For

1027 Brenner
\\\\\ \ \ \'II///~/I/;::::'1II~IIk:::'1II~

Figure 4: Box Section with Vierendeel Framing

example, the design may need to include provision for viaduct piers
or buildings on top. The localized heavy load requires deep slurry
walls or caissons to transmit the overhead loading to competent
subgrade. Unfortunately, the rest of the tunnel is connected to
the hard point of the foundation elements, leading to potential
overloads in the longitudinal direction .

Connections between cut-and-cover tunnels and immersed tubes.


Because of the change in type of structure and design loadings,
longitudinal effects are a concern.

Design for longitudinal effects can be handled in several ways:

A movable joint can be introduced between sections. For example,


a joint can be placed between a floating tunnel segment and the
tunnel segment founded on hard points. The floating segment is
free to move independent of the neighboring tunnel box. This
solves the problem of longitudinal shear and moment build up, but
it introduces additional maintenance concerns and adds cost.
Movable joints in tunnels usually should be avoided because they
Brenner
1028
promote leakage.

A gradual transition can be designed into the tunnel box in the


longitudinal section from "floating" to deep foundation section.
For example, at the transition from hard point to "floating"
segment, additional rebars for moment and shear can be included in
the longitudinal direction. The slurry walls or caissons at the
transition region can be designed to handle the extra loads from
the hanging segment on soft foundation. This approach is more
effective when the transition between floating and hard point
foundation is not as extreme in the stiffness analysis. A segment
floating on a mat founded on hard till will be more amenable to
this type of transition design.

For the case of a "floating" to deep foundation longitudinal


connection, deep foundations can be introduced in the section that
was on a floating mat. This is the "brute force" design method
which gets the job done but is probably the most expensive of the
three options.

Some designs have considered construction staging effects for the


longitudinal analysis. For example, if two adjacent segments of tunnel
are scheduled to be constructed years apart, settlements on the earlier
segment may be "locked in" on caissons or other foundation elements.
When the later segment is constructed, settlements from backfill and
other vertical loads can tend to drag down the previously constructed
section. Large longitudinal moments and shears may need to be
transferred at the interface. One solution to this problem is to specify
a last pour delayed shear joint. The last section of tunnel cast, maybe
a 50 foot (15.2 m) longitudinal section, is the connecting section
between the two segments. It is built after the second segment is cast
and backfilled.

Underpinning

The tunnel analysis and design will often need to consider existing
structures along the right-of-way. Those structures directly in the path
that can not be removed or relocated must be underpinned during
construction. Examples of existing structures that may require
underpinning include buildings, highway viaducts, and utilities,
particularly sewer pipes, steam lines, and gas lines, which are the most
difficult to deal with.

The tunnel design may be complicated by the introduction of


underpinning, because heavy underpinning loads can require deep
foundations. A tunnel section that could have been designed to float on
its base slab may now interact with slurry walls or caissons that extend
to a firm stratum. The design can be detailed in one of two ways:

Structurally isolate the underpinning elements from the tunnel


structure. This could mean, for example, that a detail would be
developed to prevent shear transfer from an underpinning caisson
to a tunnel slab. Or, the deep underpinning foundations could be
placed to avoid the tunnel structure altogether.
1029 Brenner
Design for the temporary or permanent underpinning loads in the
tunnel structural analysis. For this strategy, it is desirable to
use the most direct load path from the underpinned structure to the
hard bearing points. For example, loads should be independently
transferred to tunnel slurry walls, avoiding the roof and therefore
minimizing the impact on the tunnel structure.

For very wide tunnels, it can be difficult to avoid affecting the


tunnel structure below. The underpinning requirements may make it
advantageous in this case to use rigid cofferdam walls. In addition to
excavation support and potential use as part of the final tunnel section,
these walls can be used as underpinning foundations.

When using slurry walls for underpinning support, bearing becomes


an important consideration in the design. Usually, designing slurry
walls for bearing is a lesser concern when considering foundation support
and overall excavation stability. When the slurry walls are to be used
to support heavy loads during construction, vertical deflections must be
limited. Bearing capacity is increased by designing deeper walls which
take advantage of undisturbed areas of subgrade for skin friction, and
which have higher overall allowable end bearing values with increased
depth.

Design for Future Air Rights

Design for air rights structures above the tunnel introduces


similar concerns as for underpinning, but with an important difference.
The underpinning analysis involves existing structures with quantifiable
geometries and loadings. For future air rights structures, however, the
buildings and parks that must be accounted for are usually not designed
until after completion of the tunnel. The planning process during the
tunnel design must include enough flexibility to allow possible changes
to the air rights plans after the tunnel is built. Too much flexibility,
such as structurally providing for support of a sixty-story building
above the tunnel everywhere, would be prohibitively expensive.

Structurally, design for air rights is a trade-off. Without an air


rights building, the tunnel is subjected to the loadings from the soil
backfill and from the vehicle live loads on the surface streets. It can
be assumed that backfill would be removed from the roof of the tunnel
prior to construction of an air rights building. Thus, the capacity of
the tunnel designed to support soil backfill is available to support the
future building.

As with underpinning considerations, the best strategy for


supporting a future air-rights structure is to avoid affecting the tunnel
altogether. If this can not be done, tunnel sections with deep
foundations already included are better able to support air-rights
structures than floating mat sections. The deep foundations provide a
readily available load path to competent subgrade.

For future buildings, column location determines in large part the


impact on the tunnel below. Columns located at midspan of the roof could
overstress the slabs. On the other hand, columns located on slurry walls
1030 Brenner
would have less impact on the tunnel since there would be a direct load
path to the subsurface bearing strata. Columns located at interior walls
would have less impact than midspan locations but more impact than those
at slurry walls, since they would impart a kind of Vierendeel bending on
the structure, because of the compressibility of the soil underlying the
tunnel.

Impacts to the tunnel can be lessened through selection of an


appropriate structural type and shape for the air rights building.
Columns can be located strategically to minimize internal loads in the
tunnel structure. Transfer girders or grade beams can be used, for
example to apply loads at columns located at slurry walls instead of
directly on the tunnel roof.

The choice of air rights building framing type also has an impact.
A heavy cast-in-place concrete frame will have a greater impact on the
tunnel than a lighter structural steel frame. Not so obvious, however,
is the difference in lateral earthquake loads resulting from different
types of structures. For example, the Massachusetts State Building Code
prescribes lower seismic loads from moment resisting space frames,
because of their greater ductility, than for braced frames or box
structures. Also, braced frames impart more concentrated horizontal
shear loads and vertical couples resulting from the seismic loadings than
moment resisting space frames. The way future air rights loads are
distributed in the longitudinal direction of the tunnel has a significant
impact on the design of the tunnel and deep foundations below.

Examples from Central Artery/Tunnel Project

Three examples from the Central Artery/Tunnel Project illustrate


some of the design concerns discussed above.

Figure 5 shows a cut-and-cover tunnel section from the proposed


depressed Central Artery in the downtown Boston segment of the Project.
In this section of tunnel, soldier pile tremie concrete (SPTC) walls are
designed for the support of excavation. The walls also form part of the
final tunnel structure. The walls are constructed by excavating a slurry
trench, placing steel rolled sections in the trench, and then filling
with tremie concrete. The steel sections form the primary structural
reinforcement of the wall.

In this tunnel section, Ramp C was determined to develop large


cantilever forces in the SPTC wall. The foundation below the base slab
was relatively soft Boston Blue Clay. To reduce the moments in the SPTC
wall, caissons were designed at 20 feet (6.1 m) on center.

Figure 6 shows a plan for a cut-and-cover tunnel ramp in the South


Bay segment of the Project. The northern section of tunnel was founded
on caissons and slurry walls ("Contract B"). The southern section of
tunnel was originally designed to float on its mat foundation (Ramp DN
in "Contract A"). Further analysis determined that, because of the very
soft clay in the South Bay, the southern tunnel segment could be expected
to settle several inches. This would lead to a large build up of shear
forces at the boundary area, more than the tunnel section could handle.
1031 Brenner
I'zj
f-'.

~
11
(I)

..
U1
J
5-)58'

n s
SEE NOTE S
OJ (TYP I I I
:J I
rt- 1

f-'. 1 I
f-'
(1) I I
< I I
(1)
11 L s CANS
E 8ARS I I
1 I
~ G 8ARS
RAMP C "111)6"
1
I
~
Ef
1

"C I
CIl
C 8ARS
SoIIlTYPJ I
.... (1)
n 1 I
0 rt- SlITYP>! 1
(,J f-'.
I\) 0
R7tflZ"
I
:J 1'0 '1 ITYPI
1_ _ SPTC WA.lL
I I BY OTHERS I

A BAAS
15EE NOTE 9 I
I
SPTC WALl:
I
I
I
(A.I SSON
I
I I
~
to REINFORCEMENT DETAILS FOR TUNNEL SECTION 5
'1
ro STAIION 80+'i1,~1 TO STAIION 92+08"
;::l NIS
;::l
ro
'1
Slurry Wall

I - 9 3

RAMP cl
Caisson

CONTRACT A CONTRACT B

Figure 6: Tunnel Plan

1033 Brenner
The designer conservatively chose to introduce caissons in the southern
segment of tunnel to match the deep foundation conditions of the two
segments. Furthermore, a "delayed shear" joint was introduced at the
boundary between the two segments. The joint was specified as a final
pour, to be completed after both sides had been constructed and
backfilled.

Figure 7 shows another section of tunnel from the downtown


depressed Artery segment. In this segment, the existing Central Artery
viaduct expressway is to be underpinned while the new tunnel is
constructed below. The existing expressway is carried by a six-lane,
steel bent viaduct constructed in the 1950's. The bents are supported
on groups of piles extending to glacial till. The underpinning and
tunnel construction sequence can be described as follows:

Construct segments of slurry wall adj acent to each bent.


slurry walls will also be used as the tunnel walls.
The

Construct grade beams and needle beams on either side of each pile
bent cap, extending from slurry wall to slurry wall.

Jack against the needle beams and transfer loads to grade beams.

Sequentially excavate and construct the tunnel box.

Considering the analysis and design of the tunnel, this sequence


requires that the underpinning loads be considered in the design of walls
which are also to be used for the tunnel box. The deep slurry walls must
be analyzed for a variety of conditions. They must perform in bearing
to support underpinning loads. They must behave as excavation support
walls, with moments sequentially locked into the wall as excavation
proceeds to subgrade. They must act as a water cut-off, to prevent a
drop in the water table outside the cut as the excavation is dewatered.
They need to handle final design loading conditions from the finished
tunnel box. Finally, it may be required for the walls to support future
air rights.

1034 Brenner
Figure 7: Tunnel Section with Underpinning of Viaduct

1035 Brenner
Acknowledgements

The authors express their gratitude to the Massachusetts Highway


Department, with Central Artery/Tunnel Project Manager Peter Zuk, Manager
of Engineering Curtis Davis, and Manager of Structures Anthony Ricci, for
their support in publication of this paper.

References

Bonanno, P., Goldberg, D. Mehta, A. (1987). "Slurry Wall Construction for


a Cut-and-Cover Tunnel,"Civil Engineering Practice, the Journal of the
Boston Society of Civil Engineers, Spring.

Brenner, B., Druss, D. and Nessen B. (1993). "Managing the Impacts of


Ground Movement in Urban Tunneling", Transportation Facilities Through
Difficult Terrain. Brookfield, A.A. Balkema.

Iffland, J. (1980). "Design of Slurry Walls as Part of Permanent


Structures" Proceedings, Symposium on Slurry Walls For Underground
Transportation Facilit~es. Federal Highway Administration, March, 1980.

Liao, S. (1993), "Using Finite Element Analysis to Derive The Coefficient


of Subgrade Reaction," Proceedings, Fifth International Conference on
Computing in Civil and Building Engineering.

Moyer, P. and Brenner, B. (1992). "Central Artery Air Rights Design"


Proceedings, The Future of the Underground Industry in North America.

Slurry Walls as an Integral Part of Underground Transportation Structures


(1981). U.S. Department of Transportation, Federal Highway
Administration, November.

1036 Brenner
The Computational Method of Settlement and Loads of Pile Group

Li Jie, Qin Si Yin, Zhou Guo Ming

SYNOPSIS

This paper mainly describes computational method of


settlement and loads of pile group by computer program. The
method is based on Geddes's solution. The total settlement
consists of compression deformation of pile and soil deformation
below the pile bottom. In computing,it is considered that
compatible working among the pile cap,pile,soil below the pile
bottom and interaction among the pile group. For the case of a
prefectly flexible pile cap, valus of the maximum settlement and
maximum differential settlement are obtained. For the case of a
prefectly rigid pile cap, the sett1ement in center of pile group
and individual loads on the pile are given.
The influence caused by depth of foundation pit is
important for settlement. The distributed coe.fficients of the
total loads between the bottom of pile and the pile-side are
obtained by expriment. It is reasonable agreement that the results
of theoretical computing are compared with values obtained by
observation.

INTRODUCTION

The reliable prediction of pile foundation settlement at


working load remains a major civil engineering problem in soft
soil.

The Ninth Design & Research Institute, 303 Wuning Rd. Shanghai
200063, P. R. C h in a.

1037 Li Jie
In this Paper the analysis of a single pile is extended to
the consideration of pile group. For a general pile group two
limiting cases are assumed: that of a perfectly rigid pile cap
where all piles settlement are equal but loads carried by pile
are not equal and that of a perfectly flexible pile cap where
all piles carry equal load but all piles settlement is not
equal. For normal working loads assumption of elastic conditions
of pile and soil appears tobejustified. In the following,all
analysis is based on the assumption:
(a). No slip is assumed to occur between the pile and adjacent
soil under the working loads.
(b). The pile cap is assumed to be perfectly rigid or flexible.
(c).The effects of interaction of stress among piles are
assumed to add with liner manner.
(d).The resistance around pile-side is assumed to be distri-
buted in ladder-shaped along axis of pile~
(e) . The load ratio of bottom of pile to pile-side doesn't vary
with identical piles.
In fact, the ratio of the load for the bottom of piles and
pile-sides varies with identical piles, but the influence of
this varying is too small to neglect for settlement. If the load
ratios among piles are not the same, the- analysis of the problem
become more complex.

ANALYSIS OF A SINGLE PILE

The first step in the analysis of the settlement of pile group


IS considered the a single pile condition. Refering to Fig.!,

!P
I I I I I I I I
1 I --
I I Ps 1 ~ 1 ~
+ -- + +
1 ~ 1 ~ 1 I J ,

1 1 l ~ 1 I 1 .~
" v v v Pu 'y Pv
lit

Figure 1.

1038 Li Jie
dividing load P on the pile to two loads Pp and Ps as following:
P=Pp+Ps
where P is load on the pile top.
Pp is resistance at pile bottom and Pp= Ci. P.
Ci. obtained by expriment is the ratio of load of bottom
of pile to total load P.
Ps is resistance of pile-side along the axis of pile
and Ps=Pu+Pv.
Puis resistance of uniform d istribu tion an d Pu = {3 P.
f3 obtained by the expriment is the ratio of load of
resistance distributed in uniform for pile-side to total
load P.
Pv is resistance of triangle distrbution and Pv= y P.
Y obtained by the expriment is the ratio of load of
resistance distributed in triangle for pile-side to total
load P.
Whereas
Ci.+{3+y=l
The vertical stress in anywhere in soil below the pile bottom
may be expressed as following:
O"z=O"zp+O"zs-O"zf (1)
in formula
2
0" zp=Ip Pp/L
2 2
0" zs= 0" zu+ 0" zv=Iu Pu/L +Iv Pv/L
2
0" zf=Ip Pf/L
where
0" zp is vertical stresss produced by resistance of pile
bottom.
0" zs is vertical stress produced by resistance distributed
in uniform and triangle around pile-side.
0" zf IS vertical stress produced by weight of original
clay above the pile top.
Ip, Iu, Iv are Geddes's stress factors.
The settlement of pile top is made of two parts that are
pile compression deformation Sp and soil deformation Ss below the
pile bottom. It is may be expressed as
S=Sp+Ss (2)
Since compression deformation of pile is more smaller than soil
deformation. So Sp is approximately writen as
Sp=PL/2EpAp (3)
where
Ep is Young's modulus of the pile material.
Ap is the cross-sectional area of the pile.
The influencing depth of addition stress is calulated

1039 Li Jie
until addition stress is equal to 20% stress caused by the
weight. of soil. The soil settlement below pile bottom can
be given by relation of stress-strain.
S s =./ (j z/ E s d z (4)
where
Es is Young's modulus of soil below the pile bottom.
substituting equation (1) and equations related into(4),
equation(4) is changed as
S s = C( P - P f) / L 2J ./ K/ E s d z ( 5)
in formula
K= Ci. (Ip-Iv)+ fJ (Iu-Iv)+Iv
K is function of both Z and R.
So far,the settlement of a single pile IS obtained by
substituting equations (3) and (5) into equation (2). The
equation (2) may be writen as
S=PL/2EpAp+ CCP-Pf)/L 2 J '/K/Es dz (6)

ANALYSIS OF PILE GROUP

For a group consisting of n piles the settlement of any pile i


in the group is
Si=Spi+Ssi
where
Spi=PiLi/2EpAp
. Ssi =./ =./ c
2: (j z/ E s d z 2: (P j - P f) / L ~ K j. J J d z/ E s
Pj is the load on the pile j.
Lj is the lenght of pile j.
Kij is the effect factor produced by pile j to pile i.
Pf is constant load produced by soil above pile top.
So the settlement on the pile top may be expressed as
S i =P i L i/ 2 EpA p +./ C2: (P j P f) / L ~ K j. J J d z/ E s (7)
For a general pile group, two limiting cases may be considered:
(i)equal load Pj in all piles-- this case corresponds to
a perfectly flexible pile cap.
(ii)equal settlement of all piles - - t h i s case corresponds
to a perfectly rigid pile cap.
Perfectly Flexible Cap:
If the total load on the pile cap is Pg, then Pj=Pg/n.Hence
equation (7) may be used directly to caculate the settlement in
anywhere in the pile group and the maximum differential
settlement among piles.
Perfectly Rigid Cap:
Now the deformation of pile top is still consist of both pile

1040 Li Jie
compression and soil deformation below the pile bottom.So
deformation compatible equations may be writen as
. Sc+ () xYi+ () yXi=Spi+Ssi (8)
where
Bc is the settlement of cap in z direction.
() x an d () yare rotating angle of cap revolving x or y axis.
Xi and Yi are cordinator of pile i.
Substituting equation (7) into (8), the equation (8) can be
writen as following:
Sc+ f) xYi+ f) yXi=PiLi/2EpAp+.! (r(Pj-pf)/L~ K 1J J dz/Es (9)
For n piles, the n equations similar to equation (9) can be
obtained as
S c+ f) xY 1 + f) yX 1= P IL l/2EpAp+.! (r (Pj-Pf)/L~ K 1.JJ d z/E s

.........................
Sc+ f) xYi+ f) yXi=PiLi/2EpAp+.! (r(Pj-pf)/L~ K J dz/Es (10)
. . . . . . . . . . . . . . . . -. . . . . . . . .1J

.........................
Sc+ () xYn+ () yXn=PnLn/2EpAp+.! (r(Pj-pf)/L~ K~jJ dz/Es
For rigid cap, balance equations can be established as
G=rPi (11)
Mx=rPiYi (12)
My=rPiXi (13)
where
G is vertical total load on the cap.
Mx is total moment revolving x axis on the cap.
My is total moment revolving y axis on the cap.
By means of equations (10), (ll), (12), (13), the PI, Pn, Sc,
f) x, f) y can be solved.

Application of Engineering
Shanghai district is known as soft clay. The maximum settlement
and differential settlement of building are important for
structural design. There are five practical engineering examples
in Shanghai. The settlement values of observation have been
obtained. The theoretical results and observed settlement
are shown as table 1.

1041 Li J ie
table 1.

project length number theoretical observed


of pile of pile settlement settlemen t

Center of
material 60m 249 3.6cm 2.5cm

Building of
ChaoYang 27m 312 6.7cm 5.9cm

Building
of Nantai 35.2m 202 16cm 12.4cm

Building
of Ship 24m .256 4.6cm 4.0cm

Building
of Diediao 52m 163 3. Scm 2. Scm

The other example is an independent foundation. The length of


piles is 20 meter. Space of piles IS 3 meter. Total load on the
pile cap is 9000KN. The piles is located in Fig. 2

-f
4# +?# -f~

E
c<""l

+t +2#- -+1#

-~ 3m
+ 3m
~

Figure 2.

1042 Li J ie
The theoretical result is shown as table 2.

table 2.

No. 1# 2# 3# 4# 5# 6# 7# 8# 9#

loa d
(KN) 1026 988 1026 988 941 988 1026 988 1026

settle-
ment
(cm) 1. 10 1. 10 1. 10 1. 10 1. 10 1. 10 1. 10 1. 10 1. 10

By means of table 1, the theoretical result is more than


data of observation. The principal reason for this is that the
rigid of cap and building is ignored for perfectly flexible cap
and interaction range among piles is less than practical
interaction range.
From table 2, the distribution of load is shown that the loads
on the piles located in out-side are larger than that in middle,
the load on the pile located in angle is maximum and the load on
the pile located in center is minimum.These principle are
compatible with results measured.

DISCUSSION OF RESULTS

In this Paper it does not take into account the deformation of


soil within range of pile length and recognizes only
deformation of pile compression within this range.
In solving deformation of soil, the soil Young's modulus Es
varies with differential layers soil.
The ratio factor of IX generally changes fr"om 0.0 to 0.3. It
varies with type of pile and length of pile, but the effect of IX
on maximun settlement in group is not greater than effect of yin
soft soil. Generally, this effect on maximum settlement is about 5%
, but for single pile the effect is about 30%. The effect of the
length of pile on the settlement is obvious. If the total load
gi ven and other can d i tion don at ch ange, the settlem en t for
short pile arranged in denseness is greater than that for long
pile arranged in thin. So if long pile is used in tall building
short pileis convenient for aid building to decrease
differential settlement. Beacuse the cap is assumed to be
perfectly rigid, the theoretical ratio of load of pile for
aeverage to load of pile distribu ted in angle is smaller than

1043 Li Jie
the measured ratio of that. For the building with complete shear
wall this ratio is almost agreement.

CONCLUSION

The method of analysis presented in this Paper enables to


determind the settlement and the load distributed in group
with perfectly flexible cap and perfectly rigid cap. The
solution have been obtained for settlement and load
distribu tion in group. Comparisons are made between reported
observations on the behaviour of pile group and the behaviour
predicted by the theory. The theoretical values for settlement
are reasonable agreement with the observed values. The
theoretical ratio ofloadofpile for average to load of pile
distributed in angle is smaller than the measured ratio of that.

REFERENCE

Poulos, H. G., 1968. Geotechnique, 18, 449-471. Analysis of the


settlement of pile groups.
Butterfield, R. & Banerjer, P.K., 1971. Geotechnique 21,.No. 1,
43-60. The elastic analysis of compressible piles and pile groups.
Geddes, J.D. Geotechnique Vol.l6, pp231-235. Stresses in
foundation soil due to vertical subsurface.
Cheng Zhong Ni, 1987 , Foundation engineering.
Chao Xi Hong, The design theory for foundation of pile and box
in Shanghai.
The design code of foundation in Shanghai, DBJ08-11-89.

1044 Li Jie
ABSTRACf

GROUP LOAD TEST: 9-TIMBER PILES


by
Robert Alperstein 1 and Carlos Dobryn 2

Renovation of an 80 year old building required increasing the


foundation loads by about 20 percent. The fooodations were
known to consist of timber piles in 9-pile groups loaded to about
8 tons/pile (71 kN/pile). The nearest subsurface data was several
hundred feet away and variable in plan and profile. A load test
was deemed appropriate to justify the proposed load increase and
to ascertain the condition of the pile group. The structural
geometry did not allow for the testing of a single pile so a group
test was designed and implemented.

The group test involved the design and fabrication of a special


structural steel load application framework to allow load transfer
from adjacent columns. Load measurements were made with
calibrated jacks and movement measurements were made by optical
survey accurate to 0.001 ft. (.3 mm).

Acceptability criteria were based on a range of anticipated soil


conditions, estimated timber pile properties, conventional criteria
for single piles, and anticipated group action. The load test results
satisfied the acceptable criteria.

1 Principal, RA Consultants, Glen Rock, NJ

2 Partner, DeSimone, Chaplin & Dobryn, PC, New York, NY

1045 R. Alperstein et ale


Group Load Test: 9 Timber Piles

by
Robert Alperstein and Carlos Dobryn

Introduction

Renovation of existing structures often causes changes in loads or load distribution


resulting from the function of the "new" structure. This can create "new"
foundation loads that exceed the original design or code criteria. Strengthening the
original foundations in response to the new loads can be prohibitively expensive or
extremely difficult to implement. In such cases full scale load tests may not only
be appropriate, but may be the least costly approach. The paper describes a case
where the new foundation loads were somewhat larger than the original loads.

The project, an existing concrete and steel structure dating back to 1917, was
formerly known as the Bullard Engineering Works and was originally used for the
manufacturing of heavy weapons such as cannons and tanks. The building was to
be renovated and rehabilitated for its new use, which included multiplex movie
theaters.

An analysis of the existing structure indicated that the loads imposed by the new
use were generally equal or smaller than the original ones, with the exception of
several locations where the new loads were as much as 20% larger than the
original ones.

The original building was founded on piles in the locations under consideration.
Other than that, data on the type and number of piles was lacking. It was possible
to infer, however, that the piles were of the wood type and that their number
correlated well with the shape and dimensions of the pile caps and the loads
imposed by the existing structure.

The pile groups were under a "crawl space" which made a "probe" type of approach
extremely time consuming and expensive. It was decided to test a pile group
directly to a load well above the expected new loads. A reaction frame was devised
and the load test successfully completed, thus providing sufficient additional
evidence of the adequacy of the pile groups to support the new loads.

1046
Available Geotechnical Data

Figure 1 shows a foundation plan of the existing structure. The locations of two
borings are shown with respect to the pile supported columns. These borings were
drilled to obtain data where new construction was to be added to the existing
building footprint. These borings are located approximately 100-200 feet from the
pile group test and are within a glaciated area, indicating a potential for variable
conditions.

The boring logs are summarized as follows:

o - 1 ft. (0 - .3m) Asphalt/crushed stone

1 - 7 ft. (.3 - 2.1m) Fill (granular) (N=6-20)

7 - 12 ft. (2.1 - 3.7m) Sandy Silt (N=12)

12 - 14.5 ft. (3.7 - 4.4m) Weathered rock (N=100+)

8.5 ft. (2.6m) Water level

o - 1ft. (0 - .3 m) Asphalt/crushed stone

1 - 10ft. (.3 - 3m) Fill (granular) (N)45)

10 - 30 ft. (3m - 9.1m) Sandy silt or Silty sand (N=0-4)

30 - 35 ft. (9.1m - 10.7m) Sand & gravel (N)20)

35 - 36 ft. (l0.7m - 11m) Weathered rock (N=100+)

12 ft. (3.7m) Water level

Although the basic stratigraphy of the two borings is similar, the N-Values are
radically different and the depth to weathered rock varies by about 20 ft. (6.1m).

1047
No borings were available inside the building near the column renovation.
Therefore the two available boring logs were used to bracket the expected pile
group performance and establish criteria for acceptability of the load test
performance.

Development of the Load Test

The existing timber piles were believed to be over 80 years old. They were
suspected to be lO-inch (25.4cm) butt diameter with 8-inch (20.3cm) tip diameter.
No driving records, construction history or settlement records were available.
However, inspection of the building revealed no obvious or deleterious signs of
settlement. The variable soil conditions made the use of a static pile analysis (in
addition to its usual uncertainty) so uncertain so that it would not serve as a
verification method. We concluded that a load test of an existing individual pile
or pile group would be necessary to verify the pile capacity under the new loads.

The results of two new borings indicated that the piles were likely bearing on
weathered rock. Analysis of the existing structure and the new loads indicated that
an additional load of 150 tons (1334 kN) should be imposed on the existing pile
group to prove the adequacy of the pile group to a load well in excess of that
expected. A proper location was selected and the adjacent columns used as reaction
points (Plan, Fig. 2A). A steel frame consisting of W36 wide flange beams (Fig.
2) was connected to the adjacent columns by means of epoxy glued steel jackets
and threaded tension bars to develop additional dead weight from the lower "crawl
space" slab. The existing "dead load" on the pile group was estimated at 70 tons
(622 kN). Two jacks, at 75 tons (667 kN) each, provided the additional load for
a total of 220 (1957 kN) tons for the pile group.

Acceptance Criteria

Acceptance criteria were based on a "reasonable" estimate of the settlement of the


pile group under the test load. The first assumption in this process is that the piles
are end bearing on weathered rock, Secondly no group action will occur as the
settlement of each pile will equal the group settlement and the load in each pile will
be the same as the other piles. The settlement of each pile can be estimated as :

Sl == settlement of pile tip

se == "elastic" pile shortening (PLlAE)

L, A, and E had to be estimated because of the age of the foundations and the lack
of "as built" data. L was assumed as 25 ft. (7.6 m), "A" was assumed as the area
of a 9 inch (22.9 cm) shaft and E was assumed as 1.2 x 10 6 lb./sq. in. 8.27 x 10 6
kla. The tip movement was estimated from the relationship that the net pile

1048
movement may be 0.01 x pile load as given in most Building Codes or 0.17 inches
(0.43 cm).

Summing the components and allowing a 15% contingency for uncertainties and
assumptions produced an allowable butt movement at full test load of 0.38 inches
(9.7 mm). Allowable residual settlements after removal of the test load was set at
0.12 inches (3 cm) by assuming some rebound at the tip.

Load Test Results

A load test was conducted in February 1993. Loads were applied to the group in
37.5 tons (333.6 kN) increments with a recently calibrated hydraulic jack. The full
test load of 150 tons (1334 kN) was held for 24 hours. All other increasing loads
were held for 2 hours. Unload decrements were held for 30 minutes. Group
settlements at all loads were measured by optical survey accurate to 0.001 ft. (.3
mm.). The survey instrument was set up approximately 100 ft. from the test group
thereby eliminating interference between the measurements and the instrument. The
measured settlements were proportionately increased to account for the bending
stiffness of the floor slabs.

Figure 3 shows the load-deflection curve for the group. The maximum settlement
was 0.24 inches (6 mm) compared to the acceptance criteria of 0.38 inches (9.7
mm). The residual settlement was 0.08 inches compared to acceptance criteria of
0.12 inches (3.0 mm).

Conclusions

We conclude that the load frame behaved as intended and was able to transfer the
loads to the pile group. Second, the pile group's ability to sustain the changes in
loads was verified by the load test. A'n important part of the load test was the
selection of reasonable and realistic criteria.

1049
FIG. 1 BORING LOCATION PLAN

1050
~.",
1-"'0 ~DtI :z.",. ~

.
Q

(t
~~lJ
~LlJl'1'" ;[W~W ~ ~
~T\DtoJ
{~L-'JW\N
- "Il I
1..
or..
'\

Q
, ,.,. ~fGrLlJ
t'\\~""" ~
.,
~
L. J""'" ~E.r

FIG. 2A LOAD TEST SET UP PLAN

FIG. 28 LOAD TEST SET UP ELEVATION

1051
FIG. 2C LOAD TEST SET UP SECTION A-A

1052
Pile Group Load Test
Column G.3-31

160 r---~---.--__.--__r_-___,

140 f-~-f----+---+ _ _-+T--j'-j

120 +----+--+---10''"---+--+---1

100
CD
C
0
to-
'0
as
80
....0

60

40 +-----i~---+--+_1--_+_-___.,

20 t---1~-+---+-i'----l
__ _+_-___.,

o I I
0.00 0.05
0.15 020 025

Settlement (inches)

FIG. 3

1053
JJ'OOTTNGS WITH SE1'TLEMENT ~RJ1~:OtTCJNG PILF~S
IN NON-COHESIVE SO'L
Phung Due Long ,,)

Alth(lllgli tne design con<:~pt based on the idea of liruiting the s~ttlM1ent: Of' footings
by S~JtlM'El\'1.t"'t'e;ducing piles is gaining more and mOfe sUjJPqtt, U'el'e have beerl
very few e,cperiruental sl:udies of ti'e behaviOl' of lJiled k'01ings ill (li}(l-"c;onltsive soiL
'fhe ll1nllenf:e g( the C(llltact between the pile cap :md the soil 011 trJe c..l.p<l.city and
t:he loa(H:\et:tl.erl1~nt
behavil)!, of a piled fOiJting 3.1'e ~ijnsidef':j.ble I,~ut this h:J,s not been
well llnrlet~tQOtl By ~tfo!'mjng the field m('K1el tests, the Author h<\s tried to ae<lte
~. b~t,ter uniJerst:rnding of the load-transfer Inec:oaJi\sn i and uf the )o:jd"settleI1H~tit
b~haviof of a piled footLng in non-cohesive soiL

I; Introdudion

The GUrJ.'erll: design pmGtke fot' piled footings is based on the aSSlli11ptir)11 that the
piles atE} fl'ee-st~lidif1g, and tMt {lJI the ~xternaJ lO<l.d is cal"ried by the piles, with
~.ny Gontribution of the rooting being igftOfjYi, this approach is oveH:onservative,
since th~ footin~ itself is :;letH~J.1y in <fiteGt eQotact with th~ soil. and thus tarries a
sjgnH1G~.t1.t f~ction of the load, The p!1ll tJSQphy of <l.esi.gn IS !'ec;ently lHldergoing on
~. gi~f;hl:'lJ d'~ng~, The i<i~::t~ of using piJes ::t.S ~el:tl\i'ef1t t'e(\\J(~~f~ W8.S sf.::uied in the
s~vel'lti.es (allrlMd et al" 197'''7: H::tfl~ho et al" 1973), There 8Je number Of reasons
why tht;l ide::l. of design of s{Jf~d footiflgS wHh s~ttlement"'f~(\uGing piles has not
oecC:H'ne widely used. Orle of the reasons i.s the la.cIt: rJf feHable calculation methods
(lj!' pr~ditir!g the settlement Mel the behavior of such foundations, Bes.i(i.~s, thel'e
have be~n very few ex~ri!tlental stud.ies of the behavior of the piled fiYjUngs with

"')Ph,tJ. CnQ,lmer~; University oj Tech/IQlogy, Gorhenburg, and Swedish Geotethnlcal Imtitute,


.t.inldJping, Sweden

1054 Phung Duc Long


the cap being in contact with the soil surface. The influences of the footing (cap) in
contact with the soil on the bearing capacity of piles and on the load-settlement
behavior of a piled footing are considerable, but this has not been well understood,
especially for footings in non-cohesive soil. The mechanism of load transfer in a
piled footing involves a highly complex overall interaction between piles, pile cap
and surrounding soil, which is considerably changed due to pile installation and to
the contact pressure at the cap-soil interface.

In order to clarify the overall cap-soil-pile interaction and the load-settlement


behavior of a piled footing in non-cohesive soil, three extensive series of large-scale
field model tests were performed. Through the study, the Author has tried to create
a better understanding of the load-transfer mechanism and of the load-settlement
behavior of a piled footing in non-cohesive soil, especially the settlement-reducing
effect of the piles (Phung, 1993). In this paper, some of the main results obtained
from this study are presented.

2. Large-scale Field Model Tests

The overall pile-cap-soil interaction of a piled footing in sand includes interaction


between the piles, named as pile-soil-pile interaction, as well as between the piles
and the pile cap (footing), which is in contact with the soil surface, named as pile-
cap interaction. Comparisons of the test results on free-standing pile groups with
those on single piles under the same conditions show the pile-soil-pile interaction,
while comparisons of the test results on piled footings with those on free-standing
pile groups and on shallow footings show the pile-cap interaction. Three different
series of large-scale model tests (denoted as T1 1'2 and TI) were carried out, each
of which consists of four separate tests comprising a shallow footing (denoted as
C), a single pile (S), a free-standing pile group (G) and a piled footing (F) under
~ual soil conditions and with equal geometry, see Table 1. In this table, the
relative density of sand I D is defined as: I D = (elfUU- e)/(elfUU - enti,J, where e, emax
and emin are the initial, maximum and minimum void ratio of sand, respectively.

The tests were performed at Grabo, 35 km north-east of Gothenburg. Soil


investigations, both field tests (including CPT, pressuremeter, and dilatometer tests)
and laboratory tests, were carried out before and after each test series. The results
of the soil investigation can be found elsewhere (phung, 1993). In Test series T1,
sand was quite loose, while in Test series 1'2 and T3, sand was dense and medium
dense. The footings (caps) were made of prefabricated reinforced concrete. The
model piles were hollow steel piles with a square cross-section, 60 mm by 60 mm.
The length of the model pile was about 2.3 m and the depth of penetration of the
piles lp in each test is shown in Table 1. The model piles were covered by sand
which was glued to the pile surface. All the pile groups were square, and consisted
of five piles: one central and four corner piles. As the main purpose of the research

1055 Phung Due Long


was to study the settlement-reducing effect of the piles, the center-to-center pile
spacing was chosen to be relatively large: 4b, 6b, and 8b, where b is the pile width.
The piles were driven mechanically by a hollow ram.

Table 1. Summary ofthe larf!e-scale field model tests


Test Pile Group and Sand Separate tests in one Pile length
Series Cap (Footing) ID % test series 10 (m)
square group of five piles TIC, shallow footing -
TI pile spacing S =4b I D = 38% TIS, single pile 2.0
cap: 46cmx46cmx3Ocm TIG, pile group 2.1
TIF, piled footing 2.3
square group of five piles TIC, shallow footing -
T2 pile spacing S=6b ID = 67% TIS, single pile 2.0
cap: 63cmx63cmx35cm TIG, pile group 2.1
TIF, piled footing 2.3
square group of five piles T3C, shallow footing -
T3 pile spacing S =8b I D = 62% T3S, single pile 2.0
cap: 80cmx8Ocmx6Ocm T3G, pile group 2.1
T3F, piled footing 2.3

All the tests were performed using the same standard procedure, the quick
maintained load test (ML test). The point of failure was determined according to the
method suggested by Vesic (1969), which can be used both for piles and footings.
For the tests on piled footings, two different test procedures were used: I) starting
the test when the cap was already in contact with soil, and 2) at the start of the test,
the cap was about 20 mm above the ground surface. The first procedure was
applied in series TI, while the second in series TI and T3. Axial pile loads were
measured by means of load cells, at the base and the head of every piles. The load
was also measured in the middle' of one corner pile in order to investigate the
distribution of the axial load along the pile. The total applied load was monitored
by an independent load cell. The load carried by the cap in a piled footing was then
obtained by subtracting the load taken by the piles from the total load.
Displacements of test footings (or piles) were measured by means of electric
resistance transducers. The measurements were governed by a PC-based data
acquisition system. The lateral earth pressure against the pile shaft was measured
along the central pile by means of Glatzl total stress cells.

3. Basic Results of Field Model Tests

The load-settlement curves obtained from separate tests are compared with each
other for one and the same test series, Fig. 1 through 3. In the second and the third
test series, tests on piled footing were performed using the second test procedure.

1056 Phung Duc Long


This test procedure shows very clearly the effect of the cap in contact with soil on
the pile behavior. However, the results can not be directly compared with those
from the other tests in the same series. For the sake of comparison, the original
load-settlement curves for the piled footings TIF and T3F were modified, basing on
the remark from test series T 1 that the initial stiffness of the piles in the piled
footing (TlF) is almost the same as that of the free-standing pile group (TlG). Both
the original curves and the modified curves for the pile footings TIF and T3F were
then compared with other test results in the same series, Figs 2 and 3.

80 ..,.---....,.......-----,---,---, TEST SERIES T1


<,~ ~~...............
7-
60 +---+-~~=----+__--+__--1 Sand: ID= 38 %

/V --- --
Fooling: 46cm x 46cm
Group: 5 piles, S = 4b

se 40 / __ - --
o r- 11f' - c!:,,:._
~ I
I ::::~~ ~- ~::;~;:-=.-
_.-.-.-. .-.-.- _.-.-.
...
20 4J-.L----+--,.,.-~;...-~~-__i---___1 T1C: Shallow fooling
~.6..,."'- -~~clc~P)
.,' TlS: Single pile
,,~
" TlG: Pile group
o r,...__--
. T6
._._._.- - ' - ' - ' - ' _._._._.- TlF: Piled fooling
o m w ~ ~
SETTLEMENT mm I

Fig. 1 Test series Tl - Comparison ofseparate tests.


400 I (a) (b)
I...J
I~

300
l:r
I~
I.....
~ .-

z
~

0200
I~
I~
Iu
I~
Y "\:v'

iY
~';,:i~:"C';P__~
o....J IU ~'i'\-
\1.c).c~~~.~_ .
1
I . .... -~.........
;:'::,..--
-- '"",if:- PiLt.s
Ii /~
100
1["- I ,..,..--
_.~~

T2G(PILES)
-
~;::.-

- - - - - - - - - - - - - - riG ~LES
I~;;'~~~
J
I!
o (,-.-._.-. .+-_.-
1.1
-!7~. __ f- ..
~.-'-'-' ._._._. __
T~
.__ ._._._._._.

o 20 40 60 80 0 10 20 30 40
SETTLEMENT mm I SETTLEMENT mm I

TEST SERIES T2 Sand: 10 = 67 7- T2C: Shallow fooling T2G: Pile group


Foaling: 6.3cm x 6.3cm T2S: Single pile T2F: Piled fooling
Group: 5 piles, S = 6b

Fig. 2 Test series 12 - Comparison of separate tests, using:


a) original 12F curves; b) modified 12F curves.

Phung Duc Long


1057
400

300
=II
81
i:1
~I
I

1:;1
~I
-
l?"
, -
/'
(a) ( b )

al
UI
~I / .-'

-~'
.~.~."".

100
514I
I
II
L
~~>
...."f',.".
,
..,:~_
. ,
'
..... ;>~
'./~~ _ C.
1\ . ''\

..... ~
'i3f-!\~~
--
~-~

I' J
I '17'-~--" .-,'lGJP!hE~ -----
/~--''''''I,
~J
-q~
o "
__~t~__~ "

o 20 40 60 80 0 40
SETTLEMENT mrn I

TEST SERIES T3 Sand: 10 _= 62 % nc,: Sholiow footing TJC: Pile g(OIJP


Footing: /jDcm ~ aDem
T35: Single pile OF: Piled footing
G,' uup: 5 !liles, S = 80
Fig, 3 Tesl series T3 - Comparison a/separate lests, using "
a) origilUlI T3F curves,' b) modified T3F curves.

From the results shown in Figs 1 through 3, two importailt remarks can be drawn
for all the thtee test series: a) the load~seltlement curve of the cap in a piled footing
is very similar to that of a corresponding shallow footings; b) the load eanied by
the piles in a piled footings is much larger than that the load eamed by a
corresponding free-standing pile group,

4. Load Efficiency and Bearing capacity

The concept of group efficiency is conventionally used to compare the ultimate


bearing capacity of a free-standing pile group with that of a single pile under equal
soil conditions, This efficiency shows the pile-soil-pile interaction and is suit.able
only for free-standing pile group. Many researchers, however, used t.he same
concept also for piled footings, This is not logical as the contribution of the cap is
quite independent of the number of piles, pile spacing; pile length, and mainly
depends on its size. In order to compare be<lring capacities of a single pile, a free-
standing pile group, a piled footing and a shallow footing under the same
conditiol1S, seven different load efficiency factors, instead of group efficiency, were
suggested by the Author (Phung, 1993). The most pra.cricaI load efficiency factors
are defined in Table 2. In this table, Ps ' Pgr and Pc and are loads applied on single
pile, free-stlnding pile group, and shallow footing; PIP Pjp, and Pic are the total
load. load taken by the piles and load carried by the cap in a piled footing; and n is
the number of piles in the group. From Table 2, it can be seen that efficiency TIl
shows the pile-soil-pile interaction; efficiencies 1]4 and 1']6 show the influence of the
cap-pile interaction on the pile behavior and on the cap behavior, respectively. The

Phung Duc Long


1058
load efficiency factors are evaluated by comparing loads at the same settlement in
the two tests involved.

Table 2. De
S

Comparison of free-standing pile groups and single piles

The conventional group efficiency TJ is defined by the ratio between the failure
loads of a free-standing pile group and that of a single pile. The use of different
failure criteria will therefore lead to different values of group efficiency. The
conventional group efficiency TJ has almost the same meaning as the load efficiency
TJr. defined in Table 2. The group efficiency TJ, estimated using Vesic's criteria, is
shown in Table 3 for the pile head, shaft and base loads. The indices "s" and "b"
indicate pile shaft and base. From this table, it can be seen that the base efficiency
TJb is close to unity in medium dense to dense sand, Test series 1'2 and T3. This is
in good agreement with the results obtained by Vesic (1969). In Test series 1'2, it is
slightly less than unity. This can be explained by the fact that in the test on free-
standing pile group 1'2G, the pile base penetrated into a softer soil layer, which can
be clearly seen from CPT test results. However, in Test series T1 (loose sand), the
base efficiency TJb is much higher than unity, possibly because the soil below the
pile base was considerably compacted by pile driving. The shaft efficiency TJs is
always larger than unity, showing a compaction effect due to pile driving on the
pile shaft capacity, both in loose and dense sand. In Test series T3, the pile spacing
is as large as eight times the pile width, but TJ. is still much higher than unity.

Table 3. Load en ciencv n ( old definition), usinJ! Vesic's failure criteria


Test Sand Ratio Single pile load, kN Group efficiency TJ
series ID, % Sib Head Shaft Point Head, n Shaft, n. Base, TJb
TI 38 4 2.2 1.4 0.8 2.36 2.56 2.00
1'2 67 6 12.9 2.1 10.8 1.10 3.19 0.85
T3 62 8 8.4 1.5 6.9 1.20 2.00 1.03

The load efficiency TJJ, estimated according to the new definition in Table 2, is
plotted versus settlement in Fig. 4 for the pile head, shaft and base loads. It can be
seen that at settlements larger than 5 mm, the load efficiency TJJ becomes almost
constant and close to the old group efficiency TJ in Table 3. However, the new
definition (TJJ) has an advantage over the old one (TJ) that it is free from the choice

1059 Phung Duc Long


of failure criteria, and shows very clearly the change of the efficiency in the whole
loading process.

4 4......---....,.....-----,----r------,
(a) (a)

-3
I:'"
.3 +----+----+---+-----j
I:'"
>.. >..
u
cC1) y- TEST TI u
c:
~2 :g 2 +------t----+----+----
UJ
i
i
-
..-_._-
i UJ
~
g \.>....._- t-----........ i---......-...
TEST 13 ~

.....J '7' 8
.....J
TEST T2-I

o 0-1----+-----;.---+-------'1
o 10 20 30 40 0 10 20 30 40

8 ......---,-----,---""1"----, 4...,......-----,.----,-----...,.--_
(b) (b)

I:'" 6 4------+----+----f------i
.. 3 -I------tI----+---+-------1
'"
I:'"
>.. >..
g g
C1)
'u 13 ....... ."
~ 4 +-...-==--1---__+----+-------1 ~ 2 -I------t----+---==-.::.; ..-.+-::=--.....:;".----j
~
ol
f-'--_ i') .' ---
Wool .......

15 i - - - - _T[ST T2 15 it.S~ . ,' '1fsi i'2


.3 ! TESTTi- ----- .3 ._......_..... ;;.;:~ ~-
=a 2 -+"I:'....:.':.:..-....:.~.::.-.:.=.-.'"'""-I'--=d:=====___i =a 1~"::.:.,.~;.=--=-:.....----+----+---__+---__j
6i -._.- -'-TE'sFrT' .-.-.-.._.-. .c: U'l

o+----+----+----+------i 0 +---+----4-----+-----i
o 10 20 30 40 0 10 20 30 40

4 ......---....,.-----.----r---..., 4
(c) (c)

I:'" 3 +-------,1----+----+-------1
..Q

1:'"3
u
>.. >..
u
c:
C1)
c
C1.l
'u TEST T1 'u
:= 2 -Ue.----=-+--====~~==:j::====1
UJ i ==2
UJ
~ i
.3
o i
i "8
.....J

~
\,..-.-.-...._._._._._._.. _.I~?l.I~_.._._..__._._. i: - lEjI,..Tl_

~ ['- ...---- ----- -TEs'TT2- ... ----- & TEST T}...!

o+---+----+-----+-----j 0
o 10 20 30 40 0 10 20 30 40
Set t1ement . mm Settlement mmI

Fig. 4 Load efficiency 7]/. Fig. 5 Load efficiency 7]4.

Phung Duc Long


1060
Comparison of piled footings and free-standing pile groups

Comparison of a piled footing with a free-standing pile group, using the load
efficiency T'/4, shows the effect of cap-soil contact pressure on the pile capacity. If a
test on piled footing is carried out according to the fIrst test procedure, no matter
how carefully the test is performed, it is almost impossible to avoid errors caused
by several effects, such as recompression effect, a deeper pile penetration, or time
effects. However, if the test is performed according to the second test procedure,
the effect of cap-soil contact pressure on the pile behavior can be discerned by
taking the test results before cap-soil contact as signifIcant for the free-standing pile
group, thereby eliminating time and penetration effects. The results for the
efficiency T'/4, evaluated in this way is plotted in Fig. 5. Test series Tl is not
included because the test on piled footing TIF was performed using the fIrst test
procedure. From this fIgure, it can be seen clearly that the base efficiency T'/4b is
almost equal to unity, which means that the cap-soil contact pressure has no or very
little effect on the pile base capacity (if the pile length is large enough). The shaft
efficiency T'/4s is quite constant and slightly higher than unity at settlements less than
about 6 to 7 mm; thereafter it increase very quickly to approximately 2.0 to 2.5 at a
settlement of 40 mm. The total (head) efficiency T'/4 has the same tendency as the
shaft efficiency T'/4s, but with a smaller magnitude.

Comparison of piled footings and shallow footings

The bearing capacity of a piled footing and that of a shallow footing (cap) with an
equal size in plan can be compared through the load efficiency T'/7. The load carried
by the cap in the piled footing is compared with that carried by the shallow footing
by using the load efficiency T'/6.
2 ALL SERIES
(n, T2 and 0)

'"
l=""

---
>..
u TEST T1
c:
J,
:Q --- ---
C1)

-
w
"'0
-~~ _":":.:=.
'< "_._.- .".,=-

TEST T2
f--
o f-- TEST TJ
o
~

o
o 10 20 30 40
Set tlement mm
I

Fig. 6. Load efficiency T'/6'

1061 Phung Duc Long


5 ALL SERIES

... 4 f\ (T1, T2 and T3)

>.
u
~- TEST Tl
c: 3 .\
v \
.\
~ \ \
I::j \'-
~
0
2 " --
............. ~----- --_ ............... -1------
._ .. -.-._._. _JESILL _..................
0
...J TEST TJ

o
o 10 20 30 40
Settlement, mm
Fig. 7. Load efficiency 777'

In all the three test series, the efficiency 776. plotted versus settlement in Fig. 6, is
very close to unity. It seems, however, that 776 is higher than unity for loose sand
(Test series T1), and lower than unity for medium dense to dense sand but
approaching unity at a large settlement (Test series T2 and T3). From Fig, 7, it is
found that the 777-settlement curve has the same shape for all the three test series: at
small settlements, less than 2 to 3 mm, 777 has a quite high magnitude; afterwards,
it drops rather quickly and approaches a constant value at a settlement larger than
10 mm. The constant value depends mainly on the contribution of the cap to the
capacity of the piled footing: the higher the capacity of the cap (larger cap size,
denser sand), the lower the 777 value.

Bearing capacity of a piJed footing

As a summary, the bearing capacity of a piled footing Pft can be estimated as


follows:

(1)

where, n is the number of piles in the group; Pss and Psb are the shaft and base
capacities of a reference single pile; other symbols, see Table 2.

The efficiencies 171s and .""" which show the influence of the pile-soil-pile
interaction on the pile shaft and base capacities, can be estimated by comparing the
load per pile in a free-standing pile group with that of a single pile at a certain
settlement, e.g., s = 10 mm. The efficiency 77lb can be taken as unity for medium

1062 Phung Due Long


dense to dense sand, and higher than unity for loose sand. The efficiencies 7]45 and
7]40' which show the influence of the pile-cap interaction on the pile shaft and base
capacities, can be determined by tests on piled footings performed according to the
second test procedure. However, for piles long enough (lp > 2.5B c , in which Ip is
the pile length, and Be is the cap width), we can take 7]4b as unity. The efficiency 7]6
can be taken as 1.0 for loose sand and 0.9 for medium dense to dense sand.

50 Settlement Ratio

The conventional concept of settlement ratio is used to compare the settlement of a


free-standing pile group with that of a reference single pile. For comparison of the
settlement of a single pile, a free-standing pile group, a piled footing, and a shallow
footing under equal conditions, different new settlement ratios were suggested in
Table 4. As discussed by the Author (phung, 1992 and 1993), the conventional
settlement ratio ~ depends very much on the choice of failure criterion and safety
factor. The use of different definitions also leads to different ~-values even for one
and the same test. To avoid this, in this study, the settlement ratio of a free-
standing pile group or a piled footing as compared with a single pile will be based
on settlement comparison at the same load per pile, and the settlement ratio of a
piled footing as compared with a free-standing pile group or with a shallow footing
will be based on comparisons at the same applied load.

Table 4. De actors
s

In Table 4, Ss is the settlement of a single pile, and SgT' Sc, and Sf are the average
settlement of a free-standing pile group, a shallow footing and a piled footing under
equal conditions. The ratios ~1 and ~J' estimated by comparing the settlement of a
pile group or a piled footing with that of a single pile, are similar to the
conventional settlement ratio ~. These ratios have little practical meaning in
estimating settlement of piled footings, and will not be discussed here.

Comparison of settlement of a piled footing with that of a free-standing pile group


can be made using the ratio ~5' The ratio is estimated at the same applied load,
which is given as percentage of the failure_load of the corresponding free-standings
pile group Pgf, see Fig. 8. The figure indicates that the increase in stiffness of the
piles footing, as compared with the corresponding free-standing pile groups, is

Phung Duc Long


1063
considerable. This conclusion is contrary to that drawn in most of the theoretical
studies, based on the theory of elasticity (Butterfield & Banerjee, 1971; Poulos &
Davis, 1980; and Randolph, 1983). The increase seems to depend on the cap size:
the large the cap, the larger the increase.
1.0 ALL SERIES
C (Tl, T2 and 0)
0.8
l(l C
....... C C
C
Pg( fOll... e lood of

C
'ree-slcnding pie grOll'
.Q-
"0 0.6
Test T~ Pgt = 26 kN

a:: A C Test T2, Pgt = 71 kN

c ~ Test n, Pgf= 50.5kN

E 0.4 A

C
~ ~
VA
~ VA ~ C
U'l V 4 .0. 0.0.

0.2 VV,
~~ ~
v C A
CCC

AAA
Test series Tl
Test series T2
VV~
~""""'"
V
0.0 vvv Test series TJ
o 20 40 60 80 100 120
Load level PIPgf, %
Fig. 8. Settlement ratio ~5'

The ratio ~7 defined by comparing the settlement of a piled footing and that of a
corresponding shallow footing at the same applied load, seems to be the most useful
settlement ratio. This ratio means the reduction in settlement of a piled footing as
compared with that of a shallow footing under equal conditions. The ratio ~7 is
plotted versus applied load level, given as percentage of the failure load of the
corresponding shallow footings pc!, see Fig. 9. As expected, the ~7 value is always
lower than unity.
0.5 ALL SERIES
(Tl, T2 and TJ)
0.4 C
roo Pet = foil...e lcod of Ctl'
....... C v
v Vv v
v Tesl n, PeI= 20 kN
C
.Q-
"0 0.3
Test T2, Pet= 1.30 kN
a:: Vv A
V
Tesl n, Pct= 200 kN
CC1J ~I<..~
vvv
~
E 0.2 ,~
'I

~
A
....
C1J ~ A
U'l C A A
0.1 C
CDC Test series T1
C
C [
C C C AAO Test series T2
Pee C
Test series TJ
0.0 vvv

o 20 40 60 80 100 120
Load level PiPet, %
Fig. 9. Settlement ratio ;7.
1064 Phung Duc Long
Let us define the so-called relative cap capacity a as the ratio of the load applied on
the shallow footing (cap) to that applied on the corresponding piled footing at a
certain settlement close to failure. The relative cap capacity shows the relative
contribution of a cap to the total bearing capacity of a piled footing. With a chosen
settlement of 5 mm, the a value is 0.27, 0.48 and 0.55 for Tests T1, 1'2 and T3,
respectively. It is noted that in the case of small footings on a large number of high-
capacity piles (the contribution of the cap to the total capacity is very small as
compared with that of the piles), a and ~7 are both very small, and can be
considered equal to zero. In the case of shallow (unpiled) footings, both a and ~7
are equal to unity. The ratio ~7 can then be plotted versus the relative cap capacity
a, for different load levels from 60% to 120% of Pcf, see Fig. 10.

1.0 ALL SERIES


I (T1, T2 and 13)

0.8 I
~
VI
I
Pd = failure lood of
Test T\ Pd= 20 kN
cop

~
~
Test T2, Pel= 1.30 kN
.Q n. Pd = 200 kN
'0 0.6
Test
a:::
V'l
u.J 1/ V'l
~
u.J
/1 If'.
C<U >- 'I 1
~ ~
E 0.4 :>; '/ I

-~ {~ 1
I

J
<U
V1
.-' I
.i /
0.2 - - - - P/Pd 60% =
fJ'
/' .
. --~
_._.-_.- P/Pcf = 80%

0.0
_'"':":-~_-::: ?
~~~.-.
- - P/Pcf = 100%
_._.- P/Pcf = 120%

0.0 0.2 0.4 0.6 0.8 1.0


Relative Cap Capacity ex
Fig. 10. Settlement ratio ~7 versus relative cap capacity a.

This figure shows a clear tendency that when a is smaller than about 0.5, the
settlement ratio ~7 decreases slowly with a decreasing a value. In other words,
with a less than 0.5, a considerable increase in pile capacity (induced by increasing
the number of piles or the pile length) will not lead to a significant further reduction
in the settlement of the footing. However, with a higher than 0.5, i.e. when the
cap contributes a major part to the capacity of a piled footing, the presence of piles
has a clear effect in reducing the settlement of piled footings. It can be illustrated
by a simple example: a) the first case, a = 0.5 (i.e. the capacity of the piles is
equal to that of the cap), the ~7 value is about 0.15, which means that the
settlement of the piled footing is only 15 % of that of the unpiled shallow footing; b)
the second case a = 0.25 (i.e. the capacity of the piles is three times that of the
cap), the c;7 value is about 0.05 to 0.07. Supposing the piles have the same length,

1065 Phung Duc Long


and therefore the same capacity, the number of piles used in the latter case is three
time larger than that used in the former case. This, however, leads to a further
reduction in settlement of only 8 to 10% in comparison with the first case.

6. Prediction of Settlement and Design of Piled Footings

Based on the test results, simplified methods of estimating settlement of piled


footings in non-cohesive soil are suggested below. Settlement of a piled footing can
be calculated using the test results of a shallow footing and a single pile under equal
conditions. In both loose and dense sand, it can be considered that the load-
settlement curve of the cap in a piled footing (the load carried by the cap plotted
versus settlement) is approximately identical with that of the shallow footing, Figs
1 to 3. The settlement of the piled footing can therefore be approximately estimated
as the settlement of the shallow footing under a load equal to that carried by the cap
in the piled footing. In a piled footing, the load carried by the cap Pfc is equal to the
total applied load Pft minus the load taken by the piles Pfp :

(2)

The key factor is now to estimate Pfp. Based on the results from the load test on
single piles, depending on if only 771 or both 71J and 774 are known, the load taken by
the piles in a piled footing can be estimated according to:

(3)

or (4)

The factors 771 and 774 can be obtained from load tests or by experience, Section 4.
If we do not know about these factors, both the factors can be taken as unity, and
the calculated results will be on the safe side.

The proposed method of settlement analysis is exemplified for all the three test
series, using the results from the tests on single piles and on shallow footings, see
Section 3, and the efficiencies 77 J and 774 obtained from Figures 4 and 5. The
estimated settlements are in good agreement with the measured results. Comparison
between estimated and measured settlement for Test T3F is shown in Fig. 11.

The relationship between the settlement ratio ~7 and the relative cap capacity a,
shown in Fig. 10, can be used as the basic for a quick estimation of the settlement-
reducing effect. This method is also exemplified for all the three test series, and a
good agreement is also obtained between the predicted and the measured settlement,
(Phung, 1993).

1066 Phung Duc Long


400 TEST T.3F
PILED FOOTING

.300 n ~ Sand: 10 = 62 %
~ Footing: 80cm x 80cm

~
Group: 5 piles, S = 8b

r
~200
<t:
o..J '"
'"
100
/ - - measured
"''''''' calculated, Eq. (3)
o 000 calculated, Eq. (4)

o 10 20 .30 40
SETTLEMENT mm
I

Fig. 11. Comparison of calculated and measured settlements (Test series T3)

A practical procedure of design of piled footings can be proposed as follows: a) the


cap can take a certain load Pjc, which is equal to that can be carried by the unpiled
footing Pc without causing unacceptable settlement, and b) the piles will carry the
remaining load Pfp = Pft - Pjc. As the allowable settlement of the piled footing is
large enough for pile shaft resistance to be fully mobilised, the pile load can be
estimated close to the ultimate load of a single pile Ps ' This design procedure is still
conservative because under the cap-soil contact pressure the pile shaft resistance
increases considerably. This is not taken into account in the design procedure yet.

7. Conclusions

The results of this study strongly support the new idea of using piles as settlement
reducers for footings on non-cohesive soil. The reduction in settlement of a piled
footing in non-cohesive soil, in relation to a shallow footing under equal conditions,
is significant even with a large pile spacing. This reduction depends clearly on the
relative cap capacity, which indicates the contribution of the cap to the total bearing
capacity of a piled footing. The use of piles as settlement reducers is most effective
when the relative cap capacity is large (a> 0.5), i.e. the cap has a bearing capacity
not less than the total capacity of the piles. The number of piles required as
settlement reducers is therefore reduced considerably as compared with
conventional design.

1067 Phung Duc Long


8. References

Burland, J.B., Broms, B.B., De Mello, V.F.B. (1977). Behaviour of foundations


and structures. Proc. 9th ICSMFE, Tokyo, Vol. 2, 495-546.

Butterfield, R., & BaneIjee, P.K. (1971). The problem of pile group - pile cap
interaction. Geotechnique, Vol. 21, No.2, 135-142.

Hansbo, S., Hofmannn, E., Mosesson, J. (1973). Ostra Nordstaden, Gothenburg.


Experience concerning a difficult foundation problem and its unorthodox
solution. Proc. 8th ICS:MFE, Moscow, Vol. 2, 105-110.

Hansbo, S. (1993). Interaction problems related to the installation of pile groups.


Proc. 2nd Int. Geotech. Seminar on Deep Foundations on Bored and Auger
Piles, Ghent, Belgium, 59-66.

Phung, D. Long (1992). Tests on piled footings and pile groups in non-cohesive
soil - A literature survey. Swedish Geotechnical Institute, Varia No. 369,
Link6ping, Sweden.

Phung, D. Long (1993). Footings with settlement reducing piles in non-cohesive


soil. Chalmers University of Technology, Department of Geotechnical
Engineering, Ph.D. Thesis, Gothenburg.

Poulos, H.G., & Davis, E.H. (1980). Pile foundation analysis and design. John
Wiley and Sons, New York, etc.

Randolph, M.F. (1983). Design of piled raft foundations. Cambrige University,


Engineering Department, Research Report, Soils TR143.

Vesic, A.S. (1969). Experiments with instrumented pile groups in sand.


Perfonnance of Deep Foundation, ASTM STP 444, 177-222.

1068 Phung Duc Long


THE MECHANICS OF PILED EMBANKMENT

2 3
Bujang B. K. Huat 1 , William H. Craig & Faisal Ali

Abstract

The use of timber or concrete piles offers one


solution to the problem of stability and settlement
posed by construction of road or highway embankments on
soft ground. The idea is to transfer most of the
applied load on to a grid of piles, and for the reason
of economy the piles should be spaced as widely as
possible. and capped individually rather than by a
continuos slab. However the mechanism of load transfer
from the ground .to the top of the pile cap appears not
to be fully understood. Uncertainties exist regarding
the design of pile cap; its size and spacing, and types
of fill. Based on the results of centrifugal and
conventional laboratory model tests, this paper examines
the influence of embankment height, pile geometry and
soil properties on the efficiency of the pile support.
Comparisons with data obtained from field studies and
the results of a closed form solution are also made.

1
Lecturer, Faculty of Engineering,
Universiti Malaysia Sarawak, 94300 Kota Samarahan
Sarawak, Malaysia.
2
Lecturer, Department of Civil Engineering,
University of Manchester, Manchester
M13 9PL, United Kingdom
3
Lecturer, Department of Civil Engineering,
Universiti Malaya, 59100 Kuala Lumpur
Malaysia.

1069 Huat, Craig & Ali


Introduction

Highway embankment construction on soft clay


deposits poses problems of instability during
construction and long term and persistent settlement
subsequently. This is particularly so-in cases where
the embankment is high, as for example at bridge
approaches. Movement of a structure such as the bridge
abutment will be limited by its piled foundations,
whereas the adjacent embankment may settle significantly
causing differential settlement and lateral
deformations. One method of solving these problems is to
use a grid of piles to support the embankment. In
theory piles driven into soft ground have two effects -
they stiffen the soft subsoil and reduce stresses on the
upper subsoil by transferring load down' to lower
elevations, reducing settlement and lateral movement,
and allowing the possibility of rapid, single stage
construction without the danger of undrained failure
while minimizing land take.

Where an embankment leads to a structure such as


bridge abutment differential settlement between the
structure and the embankment can also be avoided. Recent
examples of piled embankment construction are given by
Holmberg (1978), Reid & Bu~hanan (1984), Chin (1985),
Ooi et al. (1987) and Combarieu & Pioline (1991) .
However, the complexity of the soil-piled structure
interaction problem is such that no fully developed
theoretical relationship between the characteristics of
the soil in the field and of the pile versus behavipr of
the embankment as a ,whole, appears to have been well
established. Design methods are empirical, based on
past experiences and on the results of low stress level
tests on sand rather than clay. Broms & Hansbo (1981)
and Chin (1985) referred specifically to Figure 1, a
relationship which was published by the Swedish Road
Board (1974), as a general guidance for selection of
pile cap size, a, and spacing, s. In addition to Figure
1, reference is also made to a summary of current
practice in Sweden, published by Broms and Hansbo
(1981), summarized as follows:

(a) Each pile is provided with separate concrete cap


typically 0.8 m x 0.8 m to 1.5 m x 1.5 m in size.

(b) For embankments of greater than 2 m to 3 m,precast

1070 Huat, Craig & Ali


2 2
concrete piles are used with area ratio, a /s of
the order 0.25.

(c) The piles are designed to carry the total weight of


the embankment which is assumed to be uniformly
distributed over the pile caps, i.e. r H S2 (where
r is the unit weight of fill and H is the height of
fill above the pile head) is assumed to act on
each pile, and the allowable load on each pile is
1.5 times that of the same pile in building
construction.

(d) All piles below the embankment slope are raked,


with inclination of the order 1 (horizontal) to 4
(vertical) .

(e) Rockfill is generally used as embankment fill, and


the method of construction is only used when the
soil below the embankment has low sensitivity (Sr <
20), and undrained strength which exceeds 10 kPa
so as to provide sufficient lateral resistance to
the piles.

Combarieu (1989) proposed a design chart for


selecting pile spacing for a given pile cap size and
hLight of embankment (Figure 2), similar to that of the
Swedish Road Board. The advantage is that the
properties of the fill are accounted for - expressed as
K tan ~ where K is an empirical constant, and ~ is the
fill angle of friction. Efficiency E (pile load/total
available load) is taken as equal to 0.8 rather than
unity.

Based on the results of centrifuge and laboratory


model tests, this paper examines the influence of
embankment height, pile geometry and soil properties on
the efficiency of the pile support. Comparisons with
data obtained from field studies, and the results of a
closed form solution are also made.

Methods of Study

Centrifuge model

A series of centrifuge model test has been


performed, all nominally at 1 100 scale. Figure 3

Huat, Craig & Ali


1071
Figure 1. Empirical design chart (Swedish Road Board,
(1994)

o ..l...-I..l...-~l---'_I'---...........l...-....L...-'
l....-.........

o 2 4 6 8 10
HIe

Figure 2. Empirical design chart (Combarieu, 1989)

shows typical cross section of the model embankment and


its instrumentation. Troll clay consolidated from
slurry using the hydraulic gradient method (Zelikson,

1072 Huat, Craig & Ali


1969) to a graded strength profile was used to simulate
a soft clay foundation. Clay and a sand/clay mix were
used to simulate cohesive embankment fill. The model
piles were' made of aluminum rod, 4.6 mm in diameter,
rigidly connected to 18 x 18 x 5 rran aluminum caps.
Instrumentation installed consisted of displacement
indicators, pile load cells, pore pressure transducers
and LVDTS. The models were spun at 100 g for 1 - 10
hours in the centrifuge using gravity turn-on technique.
Details of the model preparation and test procedures
have been described by Bujang et al. (1991).

C rain simulator

SI 53 54

I ./
Fill i I v-sa nd layer
1'~3 ; I ~5
I
PI .R2 , P8

Ifj6
I
Soft I .R4 I 'P9
Clay
II R7
I I I I
Section "---Sand layer
950mm J

DOpe 0 q p~ GP
PI P2 !fB~~ 0 t1 P~
51
54
5pa~hettl t! t1
i
~_ _-=-'---=--.L.. ---J. J

S - LVDT P - Druck transducer


LC - Load Cell CP - Cone Penetrometer

Figure 3. Typical cross section of centrifuge model


embankment

Laboratory model

In addition to the centrifuge testing, tests were also


carried out using simple laboratory model (Figure 4).
The influence of pile area ratio, fill thickness and
ground settlement on the failure mechanism and load

1073 Huat, Craig & Ali


carrying characteristics of the pile support was
studied. The model consists of perspex-walled box,
which is square in plan with dimension 910 x 910 x 910
rom. piles are represented in the tests by blocks of
wood (38 mrn in diameter) and pile caps are made of
square pieces of plywood. The base-board represents the
ground which is settling and transferring load to the
piles. First, the piles with caps were inserted through
the holes in the base-board which could be lowered. The
walls were greased to minimize friction. Sand was then
placed and compacted in layers inside the box. The
settlement of the soft ground was simulated by lowering
the base-board. This was performed slowly and steadily
in order to simulate the actual settlement as closely as
possible. As the base-board was lowered, loads taken by
the board and piles were monitored.

lC lC lC lC

PILE
lC lC lC lC
LUBRIC:ATED
GROUP
SURFAC:E
x

:3
x X
h

x lC lC

Figure 4. Cross section of laboratory model

Results and Discussion

Mechanism of load transfer

Figure 5 shows a typical pile load - time relation


obtained from the centrifuge model. During loading
pause period, the total load carried by the pile was
observed to increase with time.

A similar observation has also been reported at


field sites by Reid and Buchanan (1984) and Ooi et al.
(1987) indicating that as the ground in between the pile

1074 Huat, Craig & Ali


caps settles, greater stresses are being transferred
from the ground to the top of the pile cap.

The mechanism of this may be best explained by


referring to Terzaghi's (1943) classic descriptions of
a trap door which are stated as follow: "If one part of
the soil mass yields while the remainder stays in place,
the soil adj oining the yielding part moves out of its
original position between the adjacent stationary
masses. This relative movement within the soil is
opposed by shearing resistance within the zone of
contact, between the yielding and stationary masses.
Since the shearing resistance tends to keep the yielding
mass in its original position, it reduces pressure on
the yielding parts and increases pressure on the
adj oining stationary part". In a piled embankment
problem, the fill in between the pile caps, as in the
trap door, can be visualized as moving downwards due to
yield and consolidation of the ground underneath, while
the pile caps provide stationary support. The transfer
of fill stresses from the ground to the top of the pile
cap is expected to involve an abrupt increase of the
intensity of vertical pressure along the edges of the
pile cap, the discontinuity of which requires the
existence of a zone of shear springing from the edges of
the pile cap, associated with lateral expansion of the
soil located within the high pressure zone above the
pile cap towards the low pressure zone above the soft
ground in between the pile cap.

Figure 6 shows a typical result obtained from the


laboratory model for case of His > 1 and pile cap
occupying 25 % of the base-board area. The figure_shows
changes in efficiency E (defined as ratio of pile load
over total available load, y H S2) as the base-board is
lowered. As in the above, the proportion of load or E
of the pile increases with settlement of the base-board
until peak. The figure also shows drop in E to constant
residual value after the peak. Both peak and residual
values of E are dependent of the fill heights with
higher values for higher fill.

During the test the formation of the arches is


clearly seen in between the pile heads. The weight of
fill above the arches is transferred to the pile caps
which remain stationary during test.

1075 Huat, Craig & Ali


Figure 7 shows the behavior obtained from the
laboratory model for pile caps occupying 12.5 % of base-
board area. It can be seen that the proportion of fill
weight taken by the pile drops continuously after the
peak. No stable arches seem to form during the test.
When the test was stopped the pile almost punched
through to the surface. At this stage, as in the above,
humps with star formed cracking were observed directly
above the pile caps.

Failure by punching has also been observed by


Sutherland (1988) for the case of a shallow (push-out)
plate anchor in cohesive soil where failure occurred by
a combination of yielding above the anchor due to
excess shear and tensile stresses, and cracking in the
soil material, resulting in low pull-out load.
Sutherland also observed that for the case of a deep
anchor, failure was restricted to the vicinity of the
anchor plate, resulting in high pull-out load. Pushing
out of a plate anchor is analogous to settlement of the
ground and fill relative to a stationary pile cap.

It is apparent that in a piled embankment design


in order to prevent failure by punching and to encourage
arching or high efficiency of the pile support, the
fill, thickness, depending on the pile area ratio, needs
to be greater than the pile spacing.

220.-------------------,

Z 160 134 Q, r
dacelorotion
't:l
c 120
0
-I
lI.I
80
c::
40

6 12 14 16 x10 3

Figure 5. Typical pile load results of centrifuge model

1076 Huat, Craig & Ali


and throws more load on to the subsoil. This is shown
in the values of E of the centrifuge models with
different area ratios in Figure 8. Plotted also on
Figure 8 are the values of E from field site and that of
a closed form solution based on limiting equilibrium of
forces required to sustain an arch suggested by Hewlett
and Randolph (1988). The agreement between the
experiment, field data and calculation appears to be
reasonable.

It is of interest to note that as shown in Figure


8, for a given type of fill there is no further
significant increase in E once the fill thickness
exceeds 1 - 2 times the pile spacing, depending on the
pile area ratio. Although arching that occurred between
adj acent piles necessarily meant that all overburden
beyond certain elevation was transmitted to the piles,
the limiting condition in this case is the bearing
capacity of the pile cap . The efficiency of the pile
support for a given type of fill can however be
increased by increasing the area ratio of the pile
support, either by enlarging the pile cap or reducing
the pile spacing.

Effect of fill properties

Since the transfer of fill load from ground to the


top of the pile cap relies on the shearing resistance of
the fill, type or quality of the fill can be expected to
affect the efficiency of the pile support. Better
quality fill of high strength and stiffness would
facilitate more efficient transfer of fill load to the
pile. This is shown in the value of E of the centrifuge
models with equal area ratio but of different fill types
summarized in Table 1. It may be suggested that rock
fill which has higher strength and stiffness - hence
high bearing capacity and low strain required to
mobilize large shearing resistance would allow more
efficient transfer of fill load to the pile than the
materials (compacted cohesive fill and sand) dealt with
in this study. Layers of stiff geomembrane placed
immediately on top of the pile head will also help to
improve local bearing capacity of the pile support.

It is also of interest to note that as mentioned


earlier, there is no significant increase in rate of

1078 Huat, craig & Ali


10 (I )

r:::= (II)
(1111

07 5-
~
Curve ( I I : fill heIght = 2.0 s
>. " (II) : " = 1.~ I
u
l:l 0.00 l- II (III) :
" =1.0 I
G.I
'r!
U
'r!
~
~
~
02 5-
Area of pile cops =0'2~
MaterioJ of fill = sand
o I I
o 50 100 150
Settlement (mm I

Figure 6. Efficiency vs settlement, laboratory model

1,0..-----------------------...,

(I)
~

>.
u
l:l O'~O
G.I
'r!
U
.r!
~ Curve ( I): fill hel~ht a 2' Os
~
(II)
~ O'2~
II (11):.. -1'05
Area of pile cap =0125
Material of fill = sand
Ol- ...L..... --J- --'

o 50 100 150
Settlement (mm)

Figure 7. Efficiency vs settlement, laboratory model

Effect of pile area ratio

Reducing the area ratio may lead to higher load on


indi vidual piles but naturally reduces the value of E

1077 Huat, Craig & Ali


09
, --\.."O I 5 0'51
08

o ~----r--'-----"2----""'3----""'4----i
HIs

Calculation
Field Studies
P-Field Trial (ave. E), als = 0.51 Free stand-
O-Ooi et al. (1987), als = 0.52 ) ing Vertical
RB-Reid & Buchanan (1984), als = 0.37 ) piles

Centrifuge Models (ave. E.@ 100g)


o PE3, a/s = 0.51 ]
o PE4, als = 0.33 ]
o PE8, als = 0.28 ] Free standing vertical piles
PES, als = 0.33 ]
PE6. als = 0.51 ]
PE7, als = 0.33 - Vertical piles with tie beams

Figure 8. Comparison of pile support efficacy, E

increase in E once the fill height exceeds the pile


spacing. It may therefore be advocated that a high
quality fill (such as rock fill) need not extend higher
than the pile spacing above the pile heads. A lower
grade fill may be placed above this elevation, the
weight of which will be transferred to the piles via
arching of the lower layer.

1079 Huat, Craig & Ali


Table 1 Efficiency of Centrifuge Model PE 1, PE 2 and
PE

Model Fill Efficiency, E


at 100 g after ------ hrs at 100 g

a.7 hr (0. 8 yr) 1 hr (1.1 yr)

PEl Mixed 0.52 0.57 0.68

PE2 Clay 0.40 0.44

PE3 Mixed 0.51 0.59 0.65

Confinement to the arches

Lateral spreading of the embankment must be


prevented as confinement to the arches is needed. One
method to prevent lateral spreading of the piles
involves installation of raked piles with inclination
of 1 : 4 beneath the embankment toe (Broms & Hansbo,
1981) . A similar scheme has also been used at other
field sites (Holmsberg, 1978; Ooi et al. 1987). Chin
(1985) however pointed out that utilization of a raked
pile may give rise to difficulty associated with large
bending moments resulting from the structural system of
a horizontal pile cap cast on top of the pile. The
alternative is to use vertical piles with the piles
heads connected to one another by tie beams. - The
effectiveness of this arrangement, i . e . in increasing
the efficiency of the pile support, has been
demonstrated in centrifuge model PE 7~ in Figure 8.

Transition between piled to unpiled section of the


embankment

Design of an embankment for maximum available load


as dealt with above is suited for situation of say close
to a bridge where the settlement need to be minimized.
But there will also be a transition between the
untreated and treated sections of the embankment. Out
of necessity to avoid sudden change in the embankment

1080 Huat, Craig & Ali


support stiffness, the weight of the embankment must be
gradually transferred from being carried by the piles to
the ground. Current procedures involved staggering of
piles both in length (shortened or floating) and spacing
(increased) examples are presented by Holmsberg
(1978), Reid and Buchanan (1984), and Combarieu and
Pioline (1991) with piled length determine from loading
test, or increasing the pile spacing only with all piles
end bearing (Chin, 1985). The second option however has
led to punching failures of several recently constructed
piled embankments in Malaysia, the reason of which can
be found with reference to the above.

Conclusions

The mechanism of load transfer or efficiency of


the pile support is a complex relationship between the
strength and stiffness and height of fill above the
pile head, geometry of the pile support and
consolidation characteristics of the foundation clay.

Settlement of the foundation increased the


efficiency initially due to transfer of stresses from
the ground to the top of the pile cap, until the shear
strength of the fill was exceeded local to the pile caps
when load was again transferred to the clay, foundation.
Ultimately however, stable 'arches' formed above the
pile head.

Reducing the area ratio increases the proportion


of load carried by the ground in between the pile caps,
reducing the efficiency with the possibility of punching
failure.

High quality fill of high strength and stiffness


resulted in more efficient transfer of load .from ground
to the piles, giving high efficiency.

Installation of tie beams offers a means not only


to increase the area ratio of the pile support but also
to help to contain the arches.

Any abrupt change in the embankment support


stiffness cannot be tolerated and must be avoided. In
the transition zone, the weight of the embankment must
be gradually transferred from being carried by the piles
to the ground in between.

1081 Huat, Craig & Ali


REFERENCES:

Broms, B.B. & S. Hansbo (1981): Foundations on Soft


Clay. Chapter 6 in Soft Clay
Engineering, (editors) Brand & Brenner. Elsevier
Publishing Company, pp. 417 - 480.
Bujang, B. K.H., W.H. Craig & C.M. Merrifield (1991):
Simulation of a Trial Embankment Structure in
Malaysia. Centrifuge 91. (Editors) Ko & McLean.
Balkema, Rotterdam, pp. 51 - 58.
Chin, F.K. (1985): The Design and Construction of High
Embankment on Soft Clay. Proceedings 8th South
East Asian Geotechnical Conference, Kuala Lumpur,
Vol. 2, pp. 42 - 60.
Combarieu, O. (1989) : Embankments on Soft Soils
Improved by Vertical Rigid Inclusions.
Proceedings 12th International Conference on
Soil Mechanics and Foundation Engineering, Rio de
Janeiro, pp. 1723 - 1724.
Combarieu, O. & M. Pioline (1990) : Reinforcement
des remblais d'acces du futur echanger de Carrere
(Martinique) . Etudes et reccherches 1989
Laboratoire Central des Ponts et Chaussees, Paris,
pp. 32 - 33.
Hewlett, W.J. & M.F. Randolph (1988): Analysis of
Piled Embankment. Ground Engineering, Vol. 21,
No.3, pp. 12 - 18.
Holmsberg, S. (1978): Bridge Approaches on Soft
Clay Supported on Embankment Piles.
Geotechnical Engineering, Vol. 10, pp. 79 - 90.
Ooi, T.A., S.F. Chan & S.F. Wong (1987): Design,
Construction and Performance of Pile Supported
Embankments. Proceedings 9th South East -Asian
Geotechnical Conference, Bangkok, pp. 2.1 - 2.12.
Reid, W.M. & N.W. Buchanan (1984): Bridge Approach
Support. Piling. Piling and Ground Treatment,
Thomas Telford Limited, London, pp. 267 - 274.
Sutherland, H.B. (1988): Uplift Resistance of
Soils, Geotechnique, Vol. 38, No.4, pp. 493 -
516.
Swedish Road Board (1974) : Embankment Piles.
Report No. TV 121.
Terzaghi, K. (1943): Arching in Ideal Soil. Chapter 5
in Theoretical Soil Mechanics, John Wiley & Sons,
New York.
Zelikson, A. (1969): Geotechnical Models using
the Hydraulic Consolidation Similarity Method.
Geotechnique, Vol. 19, pp. 495 - 508.

1082 Huat, Craig & Ali


DYNAMIC TESTING OF AN INSTRUMENTED DRILLED SHAFf
by
F.C. Townsend!
J.F. Theos!
D. Bloomquist!
M. Hussein2

ABSTRACT

The purpose of this paper is to report a case history used to assess the effec-
tiveness of dynamic testing as a viable alternative to static load testing of drilled
shafts. Wave equation (CAPWAPC) analysis was used to predict the axial capacity
and settlement a priori of a 24-in. (61cm) diam. instrumented drilled shaft dynam-
ically tested using a 10 ton (89kN) drop weight. Subsequently, a static load test was
performed to verify these predictions.

TheCAPWAPC predicted capacity of 319 tons (2.84MN) was within 6 per-


cent of the static capacity of 300 tons (2.67MN). Along the shaft, the load distribu-
tions of the two methods differed, but the total skin friction magnitudes agreed
within 10 percent. The CAPWAPC predicted shaft settlement was approximately
60 % of the static settlement at a working load of 200 tons (1. 78MN). The dynamic
testing was less .time consuming than static load testing.

INTRODUCTION

Drilled shafts are frequently used for building and bridge foundations and .
although their design and construction has become highly skilled, there is still some
uncertainty in design estimates of their capacity and settlement. Therefore, static .
load tests are sometimes performed on the shafts. However, such load tests are
expensive and time consuming, and often only a small proportion of the drilled
shafts on a site are tested.

1 Professors, and Graduate Student, University of Florida, Gainesville, Florida.


2 GRL, Orlando, FL
1083 TOWNSEND
Alternatively, dynamic high and low strain testing have been used success-
fully on driven piles to verify capacity and integrity. Hence dynamic testing to
determine both integrity and capacity could be useful for drilled shafts.

Although dynamic testing is commonplace on driven piles, since the pile


driver/hammer is used to provide the dynamic force, drilled shafts obviously are not
driven and thus dynamic loading is not common. In addition, the high capacity of
drilled shafts and the dynamic load testing "rule of thumb" of using a loading force
greater than 1 % of the pile (shaft) axial capacity, necessitates rather large weights
and equipment for dynamic testing.

The project consisted of; (1) designing and constructing a fully instrumented
load test and two reaction drilled shafts, (2) subsequently performing low and high
strain dynamic tests using Pile Integrity Test (P.1. T.) and CAPWAPC methods, and
a 89kN drop weight, respectively, and (3) comparing results with the static load
test.

SITE CHARACTERIZATION

The soil profile at the Gainesville, Florida test site consists of: 9.4m of sand
(SPT-Nvalues 6-17, CPT qe < lOMN/m2 ), a 1. 8m clay layer (SPT-Nvalues """ 10,
CPT qe """ 1.3MN/m2), and very irregular soft limestone (unconfined compression,
qu = 2.55MN/m2 , and split tensile, qt = 0.53MN/m 2). A "pullout" test 6-in-diam.
by 6.5 ft long (15.2cm by 2m) at elevation 49.0 to 43.0 ft. (14.8 to 13m) was con-
ducted in the soft limestone, as illustrated in Figure 1. This test resulted in an esti-
mated skin friction of 7.84tsf (0.75Mpa).

SHAFT DESIGN

The skin friction of the sand and clay overburden layers was designed using
the Federal Highway Administration's (FHWA) method of Reese and O'Neill
(1987). For the sand layer, a unit weight of 1.762 g/cc with no water table was
used. The 6.1m casing (See figure 2) was assumed to provide zero friction. For the
clay layer, a = 0.55, and Cu estimated from CPT was 0.074MN/m 2 . For the lime-
stone socket, McVay's (1992) correlation (1/2.Jqu.JqJ resulting in f su = 0.58MN/m 2
was used. The end bearing was estimated as 0.5qu' Consequently, the shaft
capacity was estimated as 756.2KN + 145.0KN + 2,407KN + 379.0KN =
3,687.2KN (415 tons).

1084 TOWNSEND
,....j4:~ _ _ NUT (4' ABOVE GROUND SURfACE )

STEEL PLATES

~~ 1 1/4' DY~IDAG BAR

CASING
1----(5' DIAHETER)

TOP OF GROUT LIHESTONE SURFACE

:::::':::::':I__NG

6.5' GROUT PLUG


~ I~'~i~ : : :T~E
-----..-1,11
::P:H

1a:..:.1J:.i.LJ._ _-----
BOTTOI1 OF HOLE
DEPTH (49.0')

Figure 1 Schematic of Pullout Test

1085 TOWNSEND
SHAFT CONSTRUCTION

One test shaft and two reaction shafts were installed at the site. The test
shaft construction details are shown in Figure 2. A 30 inch (76cm) diameter auger
was used to a depth of 16 feet (4.8m), after which a 20 foot (6. 1m) long, 28 inch
(71cm) outside diameter (OD) steel casing was set at 16 feet (4.8m) below ground
level. The casing was used to reduce sufficiently the skin friction so the 500 T
(4.45MN) loading jack would be adequate. Subsequently, wet hole construction
techniques with Florigel--an attapulgite drilling mud was initiated with a 24 inch
(61cm) diameter auger. At approximately 43 feet (l3m) below ground a 24 inch
(61cm) bailer was used to start cleaning out the excavation. The final depth of the
shaft was 44 feet (13.3m) below ground level, and the casing was pushed down so
that only 1 foot (0.3m) remained above ground level. Prior to tremie placing the
concrete, the mud at the bottom of the shaft excavation was sampled to confirm
adequate clean-out. In addition, the bottom of the drilled shaft was sounded to
check the depth for settled solid debris.

NORTH REACTION TEST SHAFT SOUTH REACTION


SHAFT SHAFT
30'" ~ Ground level

13 Ties
13 ne,_, 18'" Spacing
'.-T,--"," 18" Spatlng
t:"<C '0
:z:
...
CIl
CIl

...o~20
...iii
10 '11 Ban Vertical
ReinlOleel'Mnl
~30
9w
III
:z:
...
::;40
o

U'

Figure 2 Construction Details of Drilled Shafts

1086 TOWNSEND
LOW STRAIN INTEGRITY TESTS

Low strain integrity tests (P.I.T.) were performed on all three shafts prior
to the high strain test by GRL, Inc. (Townsend, et al., 1991) A 2.7 kg hammer was
used to strike the shaft top to induce the required stress waves. an accelerometer
was attached to the top of the shaft, and the signal amplified/digitized for PIT soft-
ware analyses.

The P.1. T. velocity-depth results presented in Figure 3 were obtained


utilizing a 13,000 ft/sec wave speed. For the Load Test Shaft (LTS), the positive
velocity reflection at 22 ft. (6.6m) represents an impedance/area reduction, which
matched the end of the casing. The next major velocity increase occurs at 46 ft.
(13.9m) or the end of the shaft. For the North Reaction Shaft (NRS), the velocity
reflection at 44 ft. (13.3m) shows the shaft tip. Secondary reflections at 25 ft.
(7.6m) suggest the cold joint and small cavities encountered during shaft construc-
tion. For the South Reaction Shaft (SRS) the velocity reflection at 44 ft. (13.3m)
marks the shaft tip, while the reflections at 25 ft. (7. 6m) indicate an increase in
shaft impedance/area due to filling a cavity during construction. The presence of
these NRS and SRS cavities were clearly noted during construction, yet unknown
to the test engineer.

Co 1<1
Joint (

r.avit;e(

Figure 3 Results of P.I.T. and Soil Profile

1087 TOWNSEND
Figure 4 presents velocity-depth traces for the P.I.T. results "before" and
"after" the high strain dynamic testing of the LTS. These results indicate the same
shaft structural characteristics, which suggests the LTS was not damaged by the high
strain tests. The greater negative amplitudes near the shaft tip may be indicative of
a higher soil stiffness after dynamic loading.

45 Ft

--Before - --- After

Figure 4 "Before" and "After" P.I.T. Traces of High Strain Test

HIGH STRAIN DYNAMIC LOAD TESTS

High strain dynamic tests conforming to ASTM D4945 were performed prior
to the static load test using a 10 ton (89KN) drop hammer impacting a 3-in. (7.6cm)
thick steel plate over a 8-in. (20.3cm) thick plywood cushion. Drop heights of 3,
7,8, and 8 ft. (0.9,2.1, & 2.4m) produced measured transferred energies of 7.1,
12.9,20.4, and 22.0 kip-ft, or 9 to 14% of the potential energies. This low energy
transfer is probably due to the thick soft plywood cushioning.

Both PDA (Rausche, et al, 1985) and CAPWAPC (Likins, et al,1988)


analyses were performed as summarized in Table 1. The agreements between PDA
and CAPWAPC for the 7+ ft drops are quite good. CAPWAPC also determined
average quakes of 0.06 and 0.18 inches for skin and tip, respectively, and Smith
damping constants of 0.18 and 0.14 sec/ft. Dynamic stresses during impact loading
were less than 1. 6ksi in compression and less than O.lksi in tension. Thus indi-
cating that the high strain test did not cause structural distress.

STATIC LOAD TEST RESULTS

Subsequent to the dynamic tests, a static load test was performed following
ASTM D1143. Figure 5 presents the static load-settlement curve vs the CAPWAPe
(blow 4) comparison.

1088 TOWNSEND
Table 1 High Strain Dynamic Tests

Drop Pile Capacity


Height Set PDA CAPWAPC
(ft) (in) (tons)
(tons)

3.0 0.02 282 305


7.0 0.31 316 323
8.0 0.22 340 333
8.0 0.12 337 328
Mean 319 322

CAPWAPC vs Static Load Test Data


-'i>pUod lood/Sholl Top Mo.... m.nl
800, -.---------.:...::.:.:.:.:..:......:..:~----:~:....----------__,

600

til
.....0.
~ 400 -
-c
III
0
....l

200

O .....---.------r--.,--..---tf---..--.,--e-~---r-_J
o 0.2 0.4 0.6 0.8 1
Shaft Top Movement. Inches

Figure 5 Comparison of CAPWAPC (blow 4 ) vs Static Load Test

1089 TOWNSEND
A procedure developed by Fellenius (1989) was used to evaluate the accuracy
and reliability of the vibrating wire strain gage sister bar measurements. The
method is to plot the increment of load over the increment of strain versus the
strain, i.e., the tangent modulus of the stress-strain curve, from which the modulus
may be obtained. Utilizing this approach, the "good" sister bar data was selected
and the load shedding curves are shown in Figure 6. As may be seen there is no
load shed for the first 20 feet (6.1m) depth because of the casing. The major load
shedding occurs at approximately 23 feet (7.0m) depth. This is most likely due to
the bearing capacity of the casing at the area change. Consequently, the unit skin
friction for this region is in error. Continuing downward hardly any load is trans-
ferred until approximately the 37 feet (11.2m) depth at the top of the limestone
socket.

Load Distribution
0
~ Q 101.246 tons ~ 200.7300 ions [J 300.8591 tons
QI
m
....
.....
-10 )II 350.191 I 1 tal5

~ ~ j
....
r.::l
OJ
-20
'- ~
a
0...
a
t-
):
-30
a
-m
CD
.J:. -40
....
0-
m
0

-50
0 100 200 Xi) 400
Load in Oril led Shaft. tone

Figure 6 Load Distribution Profile

1090 TOWNSEND
COMPARISONS OF STATIC AND HIGH STRAIN DYNAMIC TESTS

Bearing Capacity - Table 2 summarizes the load distribution from the static and high
strain dynamic (CAPWAPC) tests. The average total static bearing capacity
of the shaft as determined by the Case Method (PDA) was 319 tons, while
the CAPWAPC method predicted 322 tons. The static design estimated the
capacity at 415 tons. The Davisson failure load from the static test gave a
value of 300 tons, (see Figure 5). Thus, the averages of the Case Method,
CAPWAPC and the static design over-predicted the axial capacity by 6.2,
7.3, and 38 percent respectively, when compared to the measured static load
test. The end bearing percentages of the measured static applied load pre-
dicted by CAPWAPC and the static design were 43, and 10 percent, respec-
tively.

Table 2 Summary of Load Distributions

Method Ultimate Side End


Load Load Bearing
(tons) (tons) (tons)

Static 300 204 968


CAPWAPC 322 185 137
Design 415 372 43

a) From Fig. 5 @ 300.9 Tons

Load Distribution - Table 3 compares the load distribution in kips along the shaft
as predicted by CAPWAPC (blow 4) and the static design, versus the mea-
sured values. The depths selected correspond to the vibrating wire sister bar
elevations, and interpolation was used to estimate predicted values at these
elevations. Casing the upper 19 ft., which provided zero friction in this
region, and the area reduction from 28-in-diam. to 24-in-diam. at the casing
tip certainly confuses the analysis. As shown, CAPWAPC predicts shedding
of 282 kips along the casing (producing a calculated f su = 2.23 ksf) plus an
additional 8.3 kips (291 kips total) at a depth of 23.3 ft. The calculated
design load at this depth assuming zero friction for the casing and a unit
casing tip bearing of 27.3" results in 85.3" total. By comparison, the mea-
sured shed load was 438 kips at this depth. Accordingly, the predicted loads

1091 TOWNSEND
and skin friction for this cased section are not in agreement with the mea-
sured values.

In the sand layer (23.3 to 30.6 ft., 7.1-9.3m), CAPWAPC predicted a load
shedding of 38.6\ while the FHWA's method for sands estimates shed loads of
108.5k By comparison, the measured shed load was only 6.5k These correspond
to interpolated friction stresses of: CAPWAPC = 0.84 ksf, FHWA = 2.37 ksf, and
measured = 0.14 ksf.

In the clay layer (30.6 to 36.2 ft., 9.3-11m) CAPWAPC predicts a shed load
of 13. 8k , while FHWA's method estimates 30. ()< as being shed. By comparison, the
measured load shed for this region was 7.7k These values correspond to interpo-
lated friction stresses of: CAPWAPC = 0.39ksf, FHWA = 0.87 ksf, and measured
= 0.22 ksf.
These comparisons reveal that for the upper sand and clay layer, the mea-
sured values are considerably less than those predicted by CAPWAPC or the FHWA
design, with the former being the better estimator. It appears that the casing area
reduction causes a "shadowing" effect on the underlying skin friction values.

In the weathered limestone layer (36.2 to 42.4 ft., 11-12.9m), CAPWAPC


predicted a load shed of 19 kips, while the design estimate was 475 kips. The
measured load shed was 73 kips. These values correspond to interpolated friction
stresses of: CAPWAPC = 0.49ksf, FHWA = 12.2ksf, and measured = 1. 87ksf.

Table 3 Comparison of Load Distribution

Sister Bar Mat'l Static Load Test CAPWAPC FHWA Design


Depth
(ft) cap. fsu cap. fsu cap. fsu
(kips) (ksf) (kips) (ksf) (kips) (ksf)
23.3 Casing 438.5 290.6 2.23 85.3 b 2.15
23.3-30.6 Sand 6.5 0.14 38.6 0.84 108.5 2.37
30.6-36.2 Clay 7.7 0.22 13.8 0.39 30.6 0.87
36.2-42.4 LIS 72.7 1.87 19.0 0.49 475 12.2
(L. Stone)
Tip 176 56.0 274 87.2 85.1 27.1

Blow #4
b Includes casing end bearing qt = 0.6N = 12tsf

1092 TOWNSEND
The end bearing estimated by CAPWAPC was 274 kips, while our design
estimate was 85 kips. By comparison, the measured end bearing for the 350 ton
applied load was 176 kips.

It is very obvious that the rock cores and pull-out test did not represent the
actual socket material of the shaft, and the design friction is poorly estimated. This
error possibly could be attributed to the poor quality limestone at the site, and diffi-
culty in sampling. CAPWAPC's prediction of the measured friction for this layer
was better the our design estimate. Our design estimate of the end bearing as qb =
0.5qu is conservative, while CAPWAPC overestimated the measured value.

CONCLUSIONS

The low strain dynamic tests using a hand-held hammer successfully detected
the irregularities (area changes) of each reaction shaft and load test shaft and cor-
related well with the construction records.

Excellent agreement was obtained between PDA and CAPWAPC high strain
dynamic capacities, 319 tons (2.84MN), and 322 tons (2.87MN), respectively, and
the measured static load of 300 tons (2.76MN). The high strain testing over pre-
dicted the static bearing capacity only by approximately 7 percent, but under pre-
dicted the settlements by 2mm at working loads.

The estimated design capacity (415 tons, 3.69MN) over predicted the mea-
sured capacity by 38 percent. This error is principally due to severely
overestimating the unit skin friction in the limestone socket.

Although the measured static capacity and high strain dynamic capacities
agreed well, the load distribution and end bearing values were not in agreement.

Likewise, the estimated design load distribution and end bearing were in poor
agreement with the measured values. These disagreements are possibly due to the
added bearing resistance at the casing tip (19 ft.).

The dynamic load tests were performed in only one day; whereas construc-
tion of the reaction shafts, assembly and disassembly of the .static test load frames,
and performance of the static test required five days.

ACKNOWLEDGEMENTS

This research was funded in part by the Association of Drilled Shaft Contrac-
tors, Dallas, TX, and Florida Department of Transportation. The research effort
would not have been possible without; Coastal Cassion Drill Co.; Clearwater, FL,
and GRL, Inc., Orlando, FL. Messrs. Tom Brown, and "Bud" Stebbins, both of
CCDC and M. Hussein of GRL are acknowledged for their contribution.. Dr. I.A.

1093 TOWNSEND
Caliendo, FDOT, was the Technical Co-ordinator and assisted in the dynamic and
static tests.

REFERENCES

Fellenius, RH. (1989), "Tangent Modulus of Piles determined from Strain Data"
ASCE Foundation Engineering: Current Principles and Practice, Vol. 1, pp 500-
510.

Hussein, M., Townsend, F., Rausche, F., and Likins,G. (1992) "Dynamic Testing
of Drilled Shafts" TRB Record 1336.

Likins, G., Hussein, M., and Rausche, F. (1988) "Design and Testing of Pile Foun-
dations", Third International Conference on Stress Wave Theory to Piles, Ottawa,
Canada.

McVay, M., Townsend, F. and Williams, R. (1992) "Design of Socketed Drilled


Shafts in Limestone" ASCE, Jrn. of Geotechnical Engineering, Vol. 118 No. 10,
Oct.

Rausche, Frank M., Garland E. Likins Jr. and Mohammed Hussein, (1988) "Pile
Integrity by Low and High Strain Impacts," Third International Conference on
Stress Wave Theory to Piles, Ottawa, Canada.

Rausche, Frank M., George G. Gobel and Garland E. Likins Jr., (1985) "Dynamic
Determination of Pile Capacity, " Journal of Geotechnical Engineering, ASCE, Vol.
111, No.3, March.

Reese, Lymon C. and Michael W. O'Neill, (1987) Drilled Shafts: Construction Pro-
cedures and Design Methods (DRAFT), Office ofImplementation, Federal Highway
Administration, McLean, Virginia.

Townsend, F., Theos, J.F., Shields, M.D., Hussein, M. (1991) "Dynamic Load
Testing of Drilled Shaft", Report to ADSC, Dallas, TX, and FDOT, Tallahassee,
FL. .

APPENDIX I
Conversion to SI Units

To Convert to SI Multiply by
Kips Kn 4.45
Tons MN 0.009
tsf MN/m2 0.096
in cm 2.54
ft m 0.305

1094 TOWNSEND
validity of predicting pile capacity
by pile Driving Analyzer

Peter Lai 1 and ching L. Kuo 2

Abstract

The Florida Department of Transportation (FDOT) has


routinely used the pile Driving Analyzers (PDA) on pile
load test programs since 1982, beginning with the
construction of the Sunshine Skyway Bridge. Recently the
FDOT has used PDA extensively to monitor test pile driving
to establish production pile lengths and driving criteria.
Comparison between the capacities predicted by the PDA and
the results of static load tests reveals that the PDA
prediction pile capacities range between overestimation of
15% and underestimation of 150% the pile capacity.

In reviewing the theoretical background of the PDA and


pile driving operations, three factors significantly affect
the prediction accuracy of the PDA; (1) the selection of
analytical solutions of the Wave Equation used in the PDA,
(2) the selection of CASE damping value to account for the
dynamic effects of pile driving, and (3) energy transferred
into the pile-soil system to mobilize the pile capacity.
The evaluations of these factors and the recommendations
for using the PDA to control of pile driving operations are
presented.

1 Assistant State Geotechnical Engineer, Florida Department


of Transportation, Tallahassee, Florida

2 Senior Geotechnical Consultant, PSI/Jammal and Associates


Division, Tampa, Florida

1095 Lai, Kuo


Introduction

The conventional practice of determining pile capacity


in the field is to count the number of hammer blows and
compare it with the driving criteria set by the wave
equation analysis. However, the actual pile capacity is
unknown by just counting blows during pile installation
unless a static load test is done. Therefore, pile capaci~y
evaluation in the field by dynamic measurements is very
attractive because the "real time" capacity assessment can
save the construction time by eliminating many
uncertainties, such as pile stresses and hammer efficiency
etc., and it can also reduce the amount of static load
tests.
Extensive studies of dynamic measurements were
conducted by the Michigan state Highway Commission in 1961.
Further studies were sponsored by the Ohio Department of
Transportation and the Federal Highway Administration in
1965. This led to the development of a commercial systems
which enables engineers to acquire and analyze dynamic pile
responses during driving fairly easily. The most popular
system used in the United state is the Pile Driving
Analyzer (PDA) manufactured by PDI, Inc.
The Florida Department of Transportation (FOOT) has
routinely used pile Driving Analyzer (PDA) on pile load
test programs since 1982; beginning with the construction
of the Sunshine Skyway Bridge. Recently the FOOT has used
the PDA extensively to monitor test pile driving to
establish production pile lengths and driving cr iter ia.
However, before adopting the PDA monitoring as a standard
procedure, the limitations of the PDA need to be carefully
evaluated.

Basic Concepts of PDA

The techniques used for PDA today for both field


measurement and analysis were developed in the 1960s at
Case Institute of Technology(now Case Western Reserve
University) under the direction of Professor G. G. Goble.
It provides on-site real-time prediction of pile capacity.
The PDA records the acceleration and strain reading at the
pile top under each hammer blow and utilizes the CASE
method to predict the pile capacity.
The CASE method (PDA Manual, 1990) was derived based
on the assumption of a uniform linear elastic pile and
simplified one-dimensional wave propagation. Using the
measured pile top force and velocity, the CASE method
1096 Lai, Kuo
calculates the ultimate pile capacity, RTL, by the
following equation,

where F(td, F(t 1+2Ljc) = pile top forces at


time t 1 and t 1+2Ljc
V(t 1 ) , V(t 1+2Ljc) = pile top
velocities at time
t 1 and t 1+2Lj c
L, M = length and mass of
pile
c = speed of wave
propagation in the
pile

RTL is the sum of static, RS, and dynamic, D resistances.


The dynamic resistance, or damping force is considered as
viscous in nature. It is assumed to be a function of pile
toe velocity (V~) and CASE damping factor (J c ) and can be
expressed as follows,

D = JC * Me
L * V toe

J c is a non-dimensional constant and is related to the soil


type at the pile toe. It can be expressed in the following
equation;

J
where
K

where J is the Smith damping factor, K is the pile


impedance, E is the modulus of elasticity of the pile
material and A is the pile cross sectional area. This
constant is val id only for pile with uniform impedance.
Thus, if the pile is not uniform, in either cross section
or material properties, the PDA capacity result may not
reliable(Likins, 1994).

1097 Lai, Kuo


Limitations of PDA prediction

There are three factors that significantly affect the


validity of using the PDA to estimate pile capacity: method
used to estimate pile capacity, damping factor, and
mobilization of soil resistance.

1. The CASE method is a simplified analytical solution


of wave propagation along an end bearing pile. To account
for different soil types and pile responses under hammer
impact force, it uses different methods to estimate pile
capaci ty according to the characteristics of force and
velocity curves. These methods include total resistance
method (RTL), damping factor methods (RSP or RSl), maximum
resistance method (RMX), minimum resistance method (RAN),
unloading method (RSU), and automatic method (RAU, RA2).
The variations of pile capacity estimated using these
methods are very high. Table 1 shows an example of a set of
PDA data and the pile capacities calculated using different
methods for various J c values. The actual failure capacity
(Davisson criteria) was 1060kN(238 kips) and its
corresponding damping factors were 0.06, 1.10, and 0.04 for
RS1, RMX,and RAU, respectively.

2. The CASE damping constant is an empirical


correlation of dynamic effects. It is used to estimate the
static resistance during pile driving. The J c values
recommended by Goble et ale (1975) are shown in Table 2.
These values were based primarily on data for small
diameter closed end pipe piles where a major portion of the
total soil resistance occurred at or near the pile toe.
These values were revised in the PDA manual for Model GCPC
(1990) as a result of continued research. Table 2 also
shows the revised J c values. Instead of giving a single J c
value, the revised recommendations provided a range of J c
values for each soil type.
In accordance with the database, compiled by Paikowsky
et ale (1993), the J c values can vary from 0.0 to more than
1.0 for any type of soils. These damping values were back
calculated from the failure loads as determined using the
Davisson's failure criteria from the 208 cases of static
load tests in the database. To select a proper damping
value for estimating pile capacity in such a wide range of
values is not an easy task. The predicted pile capacity is
highly dependent on the value selected.

Furthermore, soils subjected to dynamic loadings


(during pile driving) behave differently than that of
static loadings. Therefore, the static capacity estimated

1098 Lai, Kuo


during pile driving may not be compatible with the failure
load from the static load test. The dynamic resistance, and
consequently the J c value, is not only dependent on the soil
type but also on the pile size and the effective overburden
pressures along the pile. Currently, there is no simple
method to determine the J c value except the empirical
recommendations. The best method for selecting a J c value is
to calibrate the PDA results with the CAPWAP analyses.

3. The mobilization of soil resistance is also a key


factor affecting the accuracy of the pile capacity
prediction by PDA and CAPWAP analysis. Figure 1 shows the
relationship between the Davisson failure loads from static
load tests and the CAPWAP predictions. The predictions
range between overestimation of 15% and underestimation of
150% of the static load test pile capacities. The
underestimation may be due to insufficient hammer impact
force to mobilize the soil resistance. This is usually
indicated by very small set. As a rule of thumb, a set of
at least 2.5mm(0.1 in.)/blow is considered sufficient to
mobilize the true failure load. Figure 2 shows the
relationship between the CAPWAP prediction and the hammer
impact force. It shows that the upper limit of the CAPWAP
predictions is about 1.3 times the impact force.

Another cause of underestimation of pile capacity is


due to the loss of effective soil strength from continuous
build-up of excessive porewater pressures by repeated
impact loading. The pile capacity will gradually increase
after dissipation of the excess porewater pressures and is
referred to as "pile freeze or set-up". This problem can be
solved by redriving the pile. However, the waiting period
required to allow the excessive porewater pressure to fully
dissipate is unknown. Therefore, several redrives may be
necessary to accurately predict the true static pile
capacity.

Conclusions and Recommendations

The validity of PDA results mainly depends on how


accurately the three factors discussed above are assessed.
However, to improve and/or completely master these factors
is not a simple task. It requires continued research
through data collection before the most accurate method is
established .. To overcome these obstacles, the following
guidelines are recommended and may provide a more
reasonable estimate of the pile capacity during driving
when the PDA is used.

1099 Lai, Kuo


1. Examining set and maximum pile displacement closely
when doing PDA analysis of pile capacity. This is done
to ensure that the soil resistance is fully mobilized.

2. Use a CAPWAP analysis to calibrate the CASE damping


factor.

3. Make sure the appropriate CASE method is selected to


estimate the pile capacity.

4. Redrive to evaluate the pile freeze or set-up effect.


It is very important that the soil resistance is fully
mobilized during redriving in order to obtain the true
static pile capacity.

5. A static load test is necessary for a large scale


project. The benefits are not only to verify the
design but also to calibrate the PDA and CAPWAP
analysis to validate the input parameters for
controlling the remainder of the project. This will
substantially reduce the number of static load tests
required for a large site.

Reference

Goble, G.G., Likins, G. and Rausche, F. (1975) " Bearing


Capacity of Piles from Dynamic Measurements", Final Report,
Ohio Department of Transportation, Ohio DOT-05-75

Likins, G. (1994) "Helpful Hints for Field Testing and Data


Interpretation Using the pile Driving Analyzer ll , Users Day,
February, 1994 Orlando, Florida

Paikowsky, S. G., Regan, J. E. and McDonnell, J. J. (1993).


A Simpilified Field Method for capacity Evaluation of
Driven Piles (1993) FHWA, Report No. FHWA-RD-94-042

Pile Dynamics Inc., 1990. Model GCPC Pile Driving Analyzer


Manual

1100 Lai, Kuo


TABLE 1

PORT ORANGE RELIEF BRIDGE, BENT 2, PILE #6


CASE :METHOD CAPACITY RESULTS
(in kilonewton)

I J c I 0.0
I 0.1
I 0.2
I 0.3
I 0.4
I 0.5
I 0.6
I 0.7
I 0.8
I 0.9
I
RSl 1260 920 570 230 o. o. o. o. o. o.
RMX 1670 1580 1490 1420 1720 1300 1260 1230 1200 1170

RSU o. o. o. o. o. o. O. o. O. O.

RAU 950 1250

TABLE 2

RECOl\1MENDED CASE DAMPING FACTOR J e

Soil type Goble et ale (1975) PDA Manual (1990)


Clean sands 0.05 0.10 - 0.15
Silty sands 0.15 0.15 - 0.25
silts 0.30 0.25 - 0.40
Silty clays 0.55 0.40 - 0.70
Clays 1.10 0.70 - 1. 00

1101 Lai, Kuo


I CAPACITY VS. IMPACT FORCEI
14 .-------r----""T""7--~.._____,----_,_---_____,

121----------j

1-------+--------;~-7f----___t----_____i_--=II'------______j
~
10
U "0(/)
8 I-------+-~----;;;;E_-+---___ _ _ t - - - -__- - - - I

:ii(/)
a..
u 5 6 I------~_+_ _____==__-+----___t----____r_---_____l
a.. ~
~
a.. 4 1------:1:.

u
2

5 10 15 20 25
Thousands
MAXIMUM IMPACT FORCE, kN

FIGURE 1 Maximum im~act 10rce and CAPWAP predicted capacti~

20
I PILE CAPACITY COMPARISON I
,----==;:====::;:::====:;:::==::::;;::::===========-
~ 1-1150% II~//
(3
<C
a..
15 1 - - - - - + _ - :
I
i //
f/
/0
:
L _
,.-- -- -

<c I / ! II
U i!l ! //
LU c , //
0: ~ 10 I-----+_------:;f----r-----+---~
:3 ] ..
-<c ~
z
L1.
/ - - l - - - - -' -
I
I 15%
@ 5 I-----~= II i
C/)
:>
<c
o
2 4 6 8 10 12 14
Thousands
CAPWAP CAPACITY. kN

FIGURE 2 Dynamic measured capacity vs static load test Davisson capacity

1102 Lai, Kuo


DYNAMIC AND STATIC TESTS ON DRIVEN AND CAST-IN-PLACE PILES

Mohamad H. Hussein 1 and William M. Camp2

Abstract

A pile installation and testing program was undertaken at the site of


a proposed bridge replacement project in South Carolina, USA.
Subsurface conditions at the test site consisted of 14 m of sand and clay,
0.6 m thick cap rock layer over a deep underlaying stratum of calcareous
sand. Three prestressed concrete piles, one steel H-pile and one drilled
shaft were studied. All driven piles were dynamically monitored during
installation with a Pile Driving Analyzer according to the Case Method.
Some piles were also dynamically tested during restrike to evaluate time
dependent pile capacity changes. All concrete piles, including the drilled
shaft, were additionally tested using the P.I.T. dynamic low strain method
for structural integrity assessment. Two of the driven piles were statically
load tested to failure. The drilled shaft was tested up to the capacity of the
loading system (8900 kN). One of the driven concrete piles was not
subjected to a static loading test because of pile damage during
installation. The steel pile was not statically tested due to misalignment
between pile head and reaction system configuration.

This paper presents discussions on the low strain and high strain
dynamic testing and static loading test methods along with comparative
evaluation of results. Ultimate pile static resistances determined from
dynamic tests were within 8% of those measured by full scale static loading
tests. Additionally, predicted and measured pile head load-movement and
pile shaft forces at ultimate resistance relationships also agreed well.
Structural damage in the shaft of one of the driven concrete piles was
evident in records of both high and low strain dynamic tests.

1partner, Goble Rausche Likins (GRL) and Associates, Inc., 8008 South
Orange Ave., Orlando, FL 32809; Tel: (407) 826-9539, Fax: 859-8121
2S en ior Project Engineer, S&ME, Inc., 840 Low Country Blvd., Charleston,
South Carolina 29464; Tel: (803) 884-0005, Fax: 881-6149

1103 M. Hussien et al.


Introduction

Foundation settlements of more than 300 mm and corresponding


structural distress required replacement of the swing span bridge carrying
traffic along Highway 802 over Battery Creek in the eastern coastal region
of South Carolina, USA. Settlements occurred when the supporting deep
foundations (octagonal prestressed concrete piles with steel H-section
extensions) punched through the bearing stratum which consisted of a
relatively thin limestone caprock. The new replacement structure will be a
segmental concrete bridge with a total length of approximately 915 m
(including a 43 m long center span) and a width of 26.2 m. Navigational
clearance dimensions will be 13.7 m vertical and 18.3 m horizontal near the
middle of the bridge.

A pile installation and testing program was undertaken at the project


location to determine relevant foundation design parameters and evaluate
the performance of various pile types and sizes. The test site location was
chosen for its accessibility and convenience in order to minimize the cost
of the testing program. A total of five piles of various types and sizes were
evaluated. Additionally, four shafts were installed as reaction piles. All test
piles were dynamically tested and three were load tested statically. The
specific objectives of this preconstruction testing program included:
determination of pile installation characteristics, verifying minimum factors
of safety against failure, determining pile load-movement behavior,
analyzing pile-soil load interaction behavior, and to compare several
different pile types to determine which would be the most suitable to satisfy
project requirements. This paper presents discussions on pile installations
and testing procedures and results.

The original work for this project was done in the English units, soft
conversions were used to convert values to the SI units for this paper.

Subsurface Conditions

Twenty six soil test borings ranging in depth between 10 and 30 m


were drilled to explore the subsurface conditions along the proposed
bridge alignment. The pile driving and testing site was located on a small
peninsula near an approach to the existing bridge. This location was
chosen for its convenience allowing for economic accommodation of the
test program and its required equipment.

One soil test boring was drilled to a depth of 27.4 m within the
confines of the test site. The boring initially encountered a 1.8 m thick
layer of firm sandy clay which was underlain by a 3.3 m thick layer of loose
silty sand followed by a 1.8 m thick layer of soft plastic clay. At a depth of
7.0 m, a 2.4 m thick layer of loose sand was encountered. The Hawthorn

1104
Formation (very loose to loose very silty sand) was found at a depth of 9.5
m. A thin (approximately 0.6 m) layer of hard, very sandy limestone
("caprock") was encountered at approximately 12.5 m depth. From a
continuous coring run of the caprock only one 50 mm long sample was
recovered. Due to the insufficient size of this sample, the compressive
strength of the caprock could not be determined. A fine to coarse,
calcareous sand displaying some cementation was encountered below the
caprock and extended to the boring termination depth of 27.4 m, A
simplified log of the boring showing soil strata including Standard
Penetration Test N-values is presented in Figure 1.

Test Piles

Three prestressed concrete piles, one steel H-pile and one drilled
shaft were studied. Additionally, four drilled shafts (890 mm in diameter)
ranging in lengths between 27 and 32 m were installed as reaction piles for
the static loading tests. The piles tested are referred to as Test Piles A, B,
C, D and E. Test pile A was a steel HP 14x73 section (area = 138 cm 2 )
with a length of 24.4 m. Test Pile B was a 457 mm square (area = 2090
cm 2 ) prestressed concrete pile with a length of 19.5 m. Test Piles C and
D were 610 mm octagonal (area = 3077 cm 2 ) prestressed concrete
sections with lengths of 19.5 and 24.1 m, respectively. The 890 mm
diameter, 17.7 m long drilled shaft was Test Pile E. Each of the precast
concrete piles was cast with a steel "stinger" which extended 0.76 m
beyond the tip of the pile. An HP 1Ox57 section was embedded 1.8 minto
the center of the square concrete pile and HP 12x74 sections were
embedded 2.3 m into the octagonal piles.

Vibrating wire strain gages were installed at various points in each of


the test piles for measurement of pile strains at different sections during
static loading tests. Sixteen Geokon model VSM 4000 weldable gages
were welded to the web of the H-pile. Gages and leads were protected by
steel angles welded on both sides of the web. Geokon Model VCE 4200
gages were tied in the center of the concrete precast piles before casting,
and tied to the reinforcing cage of the drilled shaft. Geokon VSM 4000
gages were used to instrument the H-pile extensions. Gages were located
at five to seven locations along pile lengths corresponding to the different
soil layers. Two gages were used at each location.

Installing the driven piles was accomplished with a Vulcan 520 single
acting air hammer. This particular hammer had a ram weight of 89 kN and
was fitted with a slide bar allowing for a 0.91 and 1.52 m strokes
(corresponding rated energies of 81.6 and 136.0 kJ, respectively). Sheets
of plywood with thicknesses ranging between 150 and 300 mm were used
as pile top cushions when driving the concrete piles.

1105
Depth (m) SPT Description
N Value
0
1.8 I 15 Sondy Clay (Cl/SC)

~ Fine Sand (SP/SP-SM)


4
5.2
I 2/18" Highly Plastic Clay
70
::>
W Fine Sand (SP-SM/SM):
~~
4
Slightly Silty to Silty
4
13.4
~ t%80~ 6_
ore Limestone
:zy--
140 ~~/ 24
20
. 53
27
Calcareous Sand (SP-SM)
Fine to Coarse
.
26
21
With Shell Hash and Cementation

26
24
274

FIGURE 1: SOIL BORING

91 m

__ -t_4~_,
TP-A~ I
'~: If;'
/

TP-( m

''1 TP-9 ,. \' lTP-C


~il_-l--~
~~6m~
@ Reaction Shaft

FIGURE 2: TEST AND REACTlor~ P!L.E LOCATIONS

1106
A 610 mm diameter auger was used to form a 9 m deep hole before
inserting and driving Test Pile A (i.e., H-pile). The pile toe advanced to a
depth of 13.4 m under the weight of the pile and hammer. The pile was
driven with a 0.91 m hammer stroke for 30 cm before reaching refusal blow
counts on top of the cap rock. The pile was extracted and damage (slightly
bent flanges) was evident both at the pile head and toe. Additionally,
during extraction the "choaker" collapsed the flanges over a one meter
length of the pile within the upper seven meters. The top 0.5 m of the pile
was removed and the collapsed flanges reinforced with steel plates. A spud
weighing 107 kN was dropped thirty times from a height of 12 m to break
through the caprock. Test Pile A was again inserted in the hole and driving
commenced until pile top flanges were again bent. Another 0.75 m of the
pile top was removed and pile driving continued (still with the short stroke)
to a final depth of 21.6 m and driving resistance of 8 blows/ 0.25 m. During
the installation process, the pile head rotated approximately 15 degrees
and moved approximately 30 cm horizontally.

Prior to driving each of the concrete piles, a 610 mm diameter auger


was used to form a hole 12 m deep and the spud was used to break
through the caprock; a process similar to that of driving the steel pile. Test
Pile B was driven to a depth of 19.2 m. End of driving resistance suddenly
dropped from 33 blows/150 mm to 5 blows /250 mm. Dynamic tests
showed that this pile was indeed broken.

Test Piles C and D were installed with no problems. Pile C was driven
to a final penetration of 18.9 m and a resistance of 6 blows/25 mm using
a 0.91 m hammer stroke. Towards the end of driving of Test Pile D, the
hammer was operated with the short stroke, and the final pile penetration
was 23.5 m with a resistance of 11 blows/102 mm.

The installation process for all drilled shafts (i.e., Test Pile E and
reaction shafts) was similar. To retain the unstable overburden soils, a 914
rnm outside diameter steel casing with a 12.5 mm wall thickness was first
installed with a vibratory hammer to a depth of approximately 13.7 m which
corresponds to the location of the caprock layer. An 890 mm diameter
auger was then used to excavate each shaft to the desired depth. The
excavation proceeded in the dry until the caprock was reached at which
point the casing was filled with water. Two 50-lb bags of Flourigel were
typically added shortly thereafter to maintain side-wall stability within the
calcareous sand stratum. Following excavation, a clean-out bucket was
used to remove the cuttings before the reinforcing cage was lowered into
place and concrete placed by tremie method. As soon as the hole was
filled witb concrete, the casing was removed with the vibratory hammer;
except for the test shaft where the casing was left in place.

"07
Dynamic Pile Testing

All driven Riles were dynamically tested during installation with a Pile
Driving Analyzer (PDA) according to the Case Method. Some piles were
also dynamically monitored during restrike. Subsequent dynamic data
analysis was performed according to the CAPWAp Method. All concrete
piles, includin3- the drilled shaft, were additionally tested using the Pile
Integrity Tester M (P.I.T.) dynamic method for structural integrity evaluation.

Since dynamic pile testing during installation is performed under the


pile driving hammer blows, this type of test is commonly referred to as a
"high strain" test. Testing is often performed for the purposes of evaluating
hammer and driving system performance (Likins and Rausche 1988),
assessment of pile driving stresses and structural integrity (Hussein and
Rausche 1991), and pile driving resistance and static bearing capacity
(Rausche et a!. 1985). Dynamic measurements of strain and acceleration
under hammer impacts are the basis for modern dynamic pile testing. The
equipment consists of two each reusable strain transducers and
accelerometers, bolted at opposite sides of the pile approximately one
meter below its head, and a Pile driving Analyzer. The PDA is a state-of-
the-art, user friendly, field digital computer. Basically, it applies Case
Method equations to pile force and velocity data in real time between
hammer blows after providing signal conditioning, amplification, filtering,
calibration to measured signals and data quality assessment. High strain
dynamic pile testing has been incorporated into many standards and
specifications (ASTM 1989) and is now routine procedure in modern deep
foundation practice worldwide (Goble and Hussein 1994).

The PDA computes some 40 different dynamic variables according to


the case method after each hammer blow. The most interesting values are:
maximum energy transferred to the pile, maximum dynamic pile
compressive and tensile stresses, a structural integrity assessment factor,
pile driving resistance and static bearing capacity. dynamic testing is also
performed during pile restrikes to evaluate time dependent soil resistance
changes and their effect on pile load carrying capacity.

Dynamic data obtained in the field can further be analyzed according


to the CAse Pile Wave Analysis Program (CAPWAP) for a more
comprehensive understanding of the soil and pile behavior during pile
driving and under static loading conditions (Rausche et al. 1994). The
analysis is done in an interactive environment using measured pile data
and wave equation type analysis as a system identification process
employing signal matching techniques. Results from a CAPWAP analysis
include static pile capacity, soil resistance distribution along pile shaft and
under toe, soil damping and quake (maximum elastic deformation) values,
forces along pile length at ultimate resistance and a simulated static

1108
loading test relating pile top and toe load-movement relationships.

Low strain dynamic tests are performed on concrete piles (driven or


cast-in-place) for the main purpose of assessing shaft structural integrity.
This type of test requires minimal pile preparation, and is simply done by
affixing an accelerometer to the pile top and impacting the pile head with
a small hand held hammer. The measured data of pile top motion is
processed, analyzed and digitally stored by a dedicated system called the
Pile Integrity Tester (P.I.T.). The premise of this method is that changes in
pile impedance (elastic modulus times area divided by stress wave speed)
and soil resistance produce predictable wave reflections that can be
measured at the pile head. Since the pile impedance includes both
material strength and geometric parameters, a reduction in impedance then
represents a weakening in the pile shaft. Data evaluation may be done by
visual inspection of the records in either time or frequency domain, or by
more rigorous dynamic analysis (Rausche et al. 1994). Testing is usually
performed shortly after pile installation so that deficient piles may be
identified and corrective measures taken before construction of the
superstructure. Application of this method was recently expanded to test
piles under existing structures for determination of unknown pile lengths
(Hussein et al. 1992).

Static Loading Tests

Two of the driven concrete piles (Test Piles C and D) were statically
load tested to failure; the drilled shaft (Test Pile E) was tested up to the
capacity of the loading system. Test Pile A was not load tested due to
misalignment between the pile head and the reaction system and Test pile
B also was not statically tested due to damage in the pile shaft. Figure 2
shows test and reaction piles relative locations.

All static loading tests were performed in general accordance with


ASTM 0-1193 quick load procedures. Load was applied in increments of
approximately 15% of the anticipated ultimate value for each case. The
load was held (typically for three minutes) to take two readings of four dial
gages measuring vertical movement and two dial gages measuring lateral
movement, one reading of all strain gages and one level/ruler reading.
Loads were applied until the pile plunged or the capacity of the loading ram
was reached. Loads were applied via an 8900 kN hydraulic ram and were
measured with a 7120 kN electronic load cell. When the applied load
exceeded 7120 kN, which only happened during the drilled shaft load test,
the hydraulic pressure and a previous jack calibration were used to
estimate the actual load. In each case, reaction for test loads was
provided by means of a steel wide beam and two reaction drilled shafts.

1109
Immediately before each test, all strain gages were read to obtain a
reference "zero" value. All subsequent readings were subtracted from the
initial readings to obtain pile strain values. Since there were usually two
strain gages at each location in each pile, data from both gages were
averaged. Readings from the top gages near the pile heads and load cell .
were used to compute pile elastic modulus. This computed value was then
used at all strain gage locations. It was assumed that the gages at all
locations were under plane strain conditions. Consequently, for concrete
piles with steel stingers, the strain measured in the concrete was assumed
equal to that in the steel (and the opposite when the weldable strain gages
data was used). Therefore, at the pile toe, the total load was the sum of
the values in the concrete and steel.

For the drilled shaft, loads were calculated for the concrete and
reinforcing steel, neglecting the steel casing. If the steel casing was
included, then the load calculated from the measured strain would be
higher than the applied load. This finding indicates that the plane strain
assumption is not valid for the drilled shaft/casing combination.

Discussion of Testing Results

During the first attempt to drive the H-pile refusal driving resistance
was met at the top of the limestone at which point the PDA computed a pile
capacity of 2670 kN. Maximum pile compressive stresses at the
transducers locations was 193 MPa. As evident by the yielding of the
flanges, stresses at both pile ends were higher probably due to non-
uniform hammer impacts. Compression stresses were similar during the
second driving attempt. At a penetration of 19 m, the pile encountered its
maximum resistance and capacity (1600 kN according to CAPWAP). The
pile capacity decreased with increasing pile penetration until an end of
driving capacity of 445 kN was computed at a final pile penetration of 21 .6
m. These capacities values were higher than anticipated for this type of
pile. The pile head moved laterally during driving making it impossible to
perform a useful static loading test under safe conditions.

Test Pile B (457 mm square concrete pile) was broken during


installation. Dynamic data obtained during pile driving indicated pile
damage 17.7 m below the pile head when the pile penetration was
approximately 16.5 m. This damage location corresponds to the location
of the embedded end of the steel "stinger' in the bottom of the pile. It is
worth mentioning that it was observed prior to driving this pile that the
stinger was not straight or parallel with the pile's longitudinal axis. During
driving, maximum pile compressive and tensile stresses reached 14 and 2
MPa, respectively. Due to site constraints, the PDA gages had to removed
from the pile and the pile was driven an additional one meter without
dynamic monitoring. Apparently, as evident by the P.I.T. test results

1110
performed after pile installation the pile was broken again at a location 11.6
m below its head during its last meter of penetration. P.I.T. test results are
presented in Figure 3 in the form of pile top velocity record plotted as a
function of length. Time to length conversion was done by using a material
stress wave speed of 4000 m/s; an exponential amplification factor was
applied to the measured data in order to compensate for pile and soil
damping effects. This pile was not statically load tested.

During the installation of Test Pile C (short octagonal concrete pile),


maximum pile compression and tensile driving stresses were 17.9 and 2.6
MPa, respectively. This pile was driven to a final penetration of 18.6 m and
an end of driving resistance of 240 bl/m and a PDA computed capacity of
4940 kN. Two days after installation, the pile was subjected to a static
loading test. Testing results in the form of a pile head load-movement
curve are included in Figure 4. The pile "failed" under a static load of 4895
kN (Davisson's criteria), but supported an ultimate load of 5340 kN.
CAPWAP analysis performed with dynamic data obtained during restrike
one day after the static loading test indicated a static pile capacity of 5010
kN, which is just 2.5% higher than "Davisson's" failure load. The low strain
dynamic test result performed the same day of restrike is included in Figure
3. Pile-soil load transfer curves were computed from measured pile strains
at various pile cross sections. These results are presented in Figure 6 and
show that the pile developed very little friction above the caprock and that
approximately 60% of total pile capacity was in end bearing. The total pile
head movement at half the failure load (i.e., 2447 kN) was approximately
8 mm.

Test Pile D (long octagonal concrete pile) was driven to a penetration


of approximately 23.5 m and was statically load tested two days after end
of installation. During driving, maximum pile compressive stresses reached
20 MPa and tension stresses were generally below 6 MPa. The pile "failed"
under a static load of 2270 kN (Davisson's criteria), but supported an
ultimate of 2447 kN. The CAPWAP capacity computed from a restrike blow
one day after the static test indicated a pile capacity of 2447 kN, only 8%
higher than the Davisson's failure value. Figure 4 includes the pile top load
movement curve from the static loading test. CAPWAP analysis plotted
results are presented in Figure 5. Load transfer curves developed from pile
strain measurements during the static loading test are presented in Figure
6. Testing indicates that the soils above the caprock provided virtually no
resistance and that 65% of the pile capacity was due to end bearing. The
"gross" pile head movement at half the failure load (i.e., 1135 kN) was
approximately 5 mm. Results of the P.I.T. test performed after pile
installation are included in Figure 3.

During the installation of Test Piles C and D, maximum transferred energies


to the piles (at the transducers locations) averaged 45 kJ while the hammer

1111
---I.SO ~

12 VII

I
iJ\
18
0

/ ~
\-- ~

- - - 2.00 II.

0 6 12 18 m
24

V\ ('
I~~
""'\ i.--'

f\/
l/
Test Pil e C

E===~;;;;~d:9.5 m

- - - 2.00 II.
m
12 24

If\
6 8

I 1\
II V

\ "l\./ ~ l.---

Test Pile 0

~========~;;;;~:=J~ :4.1 m

_ - - 1.50 M.
r-...,.._ _"""T""_ _-r-_ _.,.....-_ _r--_---,r---~--_,--...,...-...,m
6 12 18 24

Test Pi Ie E 25

c================:J=== 17.7 m

Figure 3: Low strain dynamic testing (P.I.T.) results.

1112
Applied Load (kN)
o 1000 2000 3000 4000 5000 6000 7000 8000 9000

- 10
-.w
~
Q) Legend

20
~ Pile C
Pile D
--*-
~ Pile E

30
Figure 4: Pile head load-rroverrent relationships from
static loading tests

_ _ For- Msa 2000 _ _ For- Mea


2000
_____ Fa,. KIDS _____ 'Iel Mea
K1DS CDt

1000 1000

me 0
SUc

-1000 -1000
LOBel In KIDS
o 250 500 750 1000 _ _ Pile Too ' -_ _---') Pile

.25 \'''r\ _____ Botto,"

~ut. 550.0 K1DS


200
KiDS
Sk1n ~es1Stance

\
\
\ 01str-1Dut10n
\
\ ~Sk 350.1 KIDe
.50
I

~ ~to 199.9 K1DS Pile For-ces st Rut


.75 1\ Oy - .72 Incn
-.......:
~ 2000
OOlX .65 Incn
1.00 (1 inch = 25.4 mm, 1 kip = 4.45 kN)
Tag ~ovemene in 1ncn

Figure 5: CAPWAP analysis results - Test Pile D.

1113
operated at a 0.91 m stroke, which translates to a transfer ratio (transferred
divided by potential) of 58 percent. Dynamic data from both PDA and P.I.T.
did not indicate any damage in the shafts of these piles.

The drilled shaft (Test Pile E) was statically load tested 19 days after
its installation. Prior to the load test, the piles was tested with the P.I.T. to
verify its structural integrity. Low strain test results are included in Figure
3 and static load test result (load-movement curve) in Figure 4. The pile
was loaded to the maximum capacity of the load cell (7120 kN) and then
unloaded. The load cell was then removed and the shaft was loaded to
8900 kN (as determined by the hydraulic pressure gage on the jack). The
pile head movement under this load was only 8 mm, which is approximately
equal to the theoretical pile elastic shortening. The pile did not appear to
be close to failure, although it is not known if it was on the verge of
plunging. As mentioned earlier, analyzing data from the embedded strain
gages was a complex process. Three sets of calculated load transfer
curves are possible depending on the assumptions employed. The gages
located in the pile within the length of the steel casing usually gave erratic
results while those located below the casing and caprock produced more
reasonable and stable results. Regardless of how the load was being
carried above and within the cap rock, less than 15% of the applied load
was transferred to below the caprock. Pile-soil load transfer curves
obtained from computed forces in the concrete, and ignoring the steel
casing, are included in Figure 6.

Summary

A pile driving and testing program was undertaken at a proposed


bridge replacement project site. Three prestressed concrete piles, one
steel H-pile and one drilled shaft were studied. Evaluations included
observations during pile installation, dynamic high- and low-strain testing
and static loading tests. One of the concrete piles was damaged during
driving and was not load tested. The steel pile also was not statically
tested due to pile head lateral movement during driving. The other two
driven concrete piles were load tested to failure. The drilled shaft was
loaded to the capacity of the loading system and did not fail. Data analysis
showed that dynamic predictions of static pile capacities, pile head load-
movement and pile-soil load transfer relationships agree very well with
those obtained from static loading tests. Figure 7 presents plots of actual
(i.e., static loading test) and predicted (i.e., CAPWAP analysis) pile head
load- movement relationship for Test Pile D. Static loading test determined
and CAPWAP analysis simulated pile-soil load transfer curves for Test Pile
C are presented in Figure 8. Results of the program were instrumental in
designing a proper and economical foundation system.

1114
load (kN) wad (kN) wad (kN)
o 2000 4000 6000 o 12S0 2S00 o 3000 6000 9000
o i i i I I I I o i I I I I I i I I I I o i I I I I I i

5
5 5

--.
--.
'8 10 6
6
---.c 10 --- (
---
.c 10
l.
.c
.w .w
.w
.....
.....
->.
I~
0..
~ IS ' '
~
(Jl
I

1St- of t t I 20
l- f f I I 15

20 I I
25 I I
20 I I

Test Pile C Test Pile D Test Pile E


Figure 6: Load transfer CUIVes
3000
Thst Pile D

2500 ~
I
Static Iarl 'lest

-B-- ~

-~ 2000

- 1500
~
S 1000
500

a ~_ ...J--..........- - ' -...........'--'-...J--..........- - ' -...........................L...-.............J--~-'-............................L-.L.....J.-'-..J.....J

a 5 10 15 20 25 30
Pile Head Movement (mm)
Figure 7: Measured and simulated pile head load - rrovement curves

Load (kN)
a 1100 2200 3300 4400 5500
a
Test Pile C

-E
-..c: 10
.w
~
15

20 L- -----J

Figure 8: load transfer curves including ~ simulated cw:ve

1 '116
Acknowledgments

The writers wish to express their sincere gratitude to their own


organizations for allowing them the time and resources to prepare this
manuscript and to the South Carolina Department of Transportation for their
guidance and support during all phases of this project.

References

ASTM 04945-89. "Standard test method for high strain dynamic testing of
piles," Annual Book of American Society for Testing and Materials, Volume
4.08,1018-1024.

Goble, G.G. and Hussein, M. (1994). "Dynamic pile testing in practice,"


Thirteenth International Conference on Soil Mechanics and Foundation
Engineering (XIII ICSMFE), ISSMFE, New Delhi, India.

Hussein, M. H. and Rausche, F. (1991). "Determination of driving induced


pile damage," Proceedings of the International Conference on Deep
Foundations, Presses L'Ecole Nationale des Pants et Chaussees, Paris,
France, 455-462.

Hussein, M., Likins, G. and Goble G.G. (1992). "Determination of pile


lengths under existing structures," Proceedings of the 17th Annual
Members Conference, Deep Foundations Institute, 195-208.

Likins, G. and Rausche, F. (1988). "Hammer inspection tools," Proceedings


of the Third International Conference on The Application of Stress Wave
Theory to Piles, Bengt H. Fellenius ed., BiTech Publishers, 659-667.

Rausche, F., Goble, G. and Likins, L. (1985). "Dynamic determination of


pile capacity," J. of Geotechnical Engrg., ASCE, Vol. 111, No.3, 367-383.

Rausche, F., Likins, G. and Hussein, M. (1994). "A formalized procedure


for quality assessment of cast-in-place shafts using sonic pulse echo
methods," 73rd Annual Meeting of the Transportation Research Board,
Washington, D.C.

Rausche, F., Hussein, M., Likins, G. and Thendean, G. (1994). "Static pile
load-movement from dynamic measurements," ASCE Settlement'94
Conference, Austin, Texas; Geotechnical Special Publication No. 40.

1117
A Rational and Usable Wave Equation Soil Model
Based on Field Test Correlation

Frank Rausche,1 Garland Likins,2 and George Goble 3 Members ASCE

Abstract

Dynamic soil modeling of pile driving is presented. To improve the


commonly used model, both static and dynamic data have been measured
with the Standard Penetration Test (SPT) and a data base with correlations
to full scale pile tests was generated. The literature was investigated for
so-called rational soil models whose parameters can be derived from
standard geotechnical soil properties and a correlation study was made
to relate standard soil constants with dynamic model parameters.

Several ideas for model improvements were found in the literature.


However, these suggestions were complex and results not proven by
measurements. A simple, improved dynamic soil model for pile driving
which has been compared with dynamic testing of both the full scale pile
data base and SPT results and how it can be implemented into standard
wave equation practice are discussed.

Introduction

The main pile driving question is how to quickly, safely and


economically drive a pile to sufficient capacity with acceptable
settlements. This paper attempts to explain the basics of past modeling
efforts and why changes should be made to existing technology. Smith
(1960) devised the current pile driving analysis model which is successfully

1 President, Goble Rausche Likins and Associates Inc., 4535 Emery


Industrial Parkway, Cleveland, OH 44128.
2president, Pile Dynamics, Inc., 4535 Emery Industrial Parkway,
Cleveland, OH 44128.
3principal, Goble Rausche Likins and Associates, Inc., 5398 Manhattan
Circle, Boulder, CO 80303.

Rausche, Likins, Goble


1118
used in many countries and commonly referred to as the "wave equation".
His goal was to replace pile driving formula (based on energy concepts)
relating bearing capacity to blow count with a more accurate numerical
algorithm. Wave equation modeling uses a one dimensional mathematical
representation of hammer, driving system, pile and soil which allows an
accurate calculation of (a) the progress of pile penetration into the ground,
(b) the relationship between pile bearing capacity and pile penetration, (c)
the stresses in the pile during driving and (d) the mechanics and/or
thermodynamics in a hammer. Smith did not consider the driveability
problem which includes blow count and stresses as a function of pile
penetration.

Many subsequent efforts at improvement were directed at a soil


model which (a) can be physically explained and (b) whose component
parameters can be derived from standard geotechnical engineering soil
properties. Several theoretical studies are primarily based on Novak et al.
(1978) who derived soil stiffness and soil damping from the soil shear
modulus. Randolph and Simons (1986), Chow, et al. (1988) and Lee, et
al. (1988). included this concept in their proposed soil resistance models.
Related models were described by Corte and Lepert (1986), Holeyman
(1985), Middendorp et al. (1984) and others.

Smith's model relates elevated resistances due to high loading


rates with a velocity dependent resistance in addition to the displacement
dependent, static resistance. Coyle and Gibson (1970) used the same
concept, but with a dynamic resistance varying exponentially with velocity
as determined by laboratory testing. Briaud and Garland (1984) used a
time to failure or an average loading velocity raised to some power to
define a ratio of total dynamic capacity to static capacity. The maximum
applied load is also displacement dependent. None of these models
proposed by the academia have been subjected to extensive correlation
with a database of field results and thus remain unproven. However, there
is unanimous agreement that the current practice may lead to errors
particularly for situations which are beyond the traditional data bases
established with hammers of relatively low impact velocities.

Finally, significant errors in dynamic pile predictions are made


because of an inaccurate assessment of the losses or gains of soil
strength caused by pile driving (e.g., Heerema, 1979, Svinkin et al. 1994).
Skov and Denver (1988) proposed the direct measurement of these
effects; however, the prediction of soil strength changes with time is still
very difficult. A proposal on implementing these effects in driveability
analyses will be made.

1 1 19 Rausche, Likins, Goble


Background of Problem Statement

Several phenomena contribute to the behavior of the soil during pile


driving and each must be clearly understood and accurately modeled if the
analysis is to properly predict the pile driving process. The major effects
are velocity or rate dependent effects, soil movement, soil degradation or
set-up, and creep. These effects will be discussed individually.

Rate effects

The static resistance is a function of the relative pile-soil


displacement. Unfortunately, the soil resistance does not behave
identically during static and dynamic load applications. When loads are
applied rapidly as in pile driving, additional velocity and acceleration
dependent resistance components are generated. These dynamic
resistance components increase the apparent resistance of a quickly
penetrating pile compared to a slowly advancing one. In this paper, the
total resistance or elevated resistance is the sum of static and dynamic
resistance.

For long piles with resistance distributed along the shaft and for any pile
with high resistance, the maximum velocities along the pile shaft may be
highly variable and generally much lower than at the pile top. Under these
conditions, the non-linearity of dynamic resistance vs. velocity becomes
very important and would require very high damping factors with traditional
linear damping models. Conversely, new model hammers with higher
strokes or greater efficiencies produce much higher velocities than
contained in the original data base used to develop parameters for the
original wave equation model. An improved method of accounting for rate
effects appears to be desirable.

Soil movement

Smith made a simplifying assumption that the soil is fixed in space.


Soil motions can be included in the calculations with a so-called radiation
damping model (the soil motion radiates energy away from the pile soil
interface). Unfortunately, radiation damping and viscous damping are
terms which are often used interchangeably. CAPWAP (Rausche, et al.
1985, GRL 1993 and Rausche et al. 1994) contains a radiation damping
model and extensive parameter studies indicate certain narrowly bounded
model parameters for good correlation with static load test capacities.
However, attempts to incorporate this new model into the wave equation
analysis GRLWEAP (GRL 1991) have not succeeded because of sensitivity
of the calculated blow count to the radiation damping parameters,
Therefore, the soil in the model proposed in this paper is considered fixed.

1120 Rausche, Likins, Goble


This model has limited accuracy for refusal cases such as driving into hard
rock or dynamic loading of drilled shafts, etc. where soil motions may be
as large as pile motions.

Static soil resistance degradation (set-up)

When a hammer strikes a pile, soil particles around the pile are
suddenly displaced as the pile penetrates under a hammer blow.
Moreover, the pile also rebounds a considerable distance. In fact, during
hard driving, the upward and downward pile movements are much greater
than the net permanent penetration into the ground due to the elasticity of
pile and soil. This relentless down and up pile motion affects the ground
pore water pressures and/or destroys the natural fabric of the soil, at least
temporarily. The resulting loss of soil strength leads to a static soil
resistance which is less than the long term value under sustained loads.
The soil resistance generally increases after driving and therefore the term
set-up describes what is really only a return to a long-term strength as
might be calculated from a static analysis. Usually, a set-up factor is used
to calculate the long term capacity from the temporarily reduced capacity
at the end of pile installation using an exponential approach such as
proposed by Skov and Denver (1988).

(1 )

In this expression, A is a non-dimensional quantity defining the


capacity increase between time to (when the capacity is Ro) and time t.

Creep

Compared to short term static or dynamic loads, a pile will


experience greater settlements due either to soil creep or to soil
consolidation under loads maintained for a considerable time. Under short
duration dynamic loads and quick static tests, these so-called secondary
settlements are hardly noticeable. However, under long term loads they
may affect a pile's serviceability. Moreover, in evaluating a static load test,
creep deformations may make the apparent capacity lower for a
maintained load test than for a quick test. Therefore, for a correct
prediction of a pile's load-set behavior, an estimate of the creep
deformation as a function of loading rate should be added 'to the
dynamically predicted values perhaps using an exponential expression.
Unfortunately, little has been done to solve this problem.

1121 Rausche, Likins, Goble


Description of the Improved Model

Basically an extended Smith model is proposed as in Figures 1a and


1b for pile shaft and toe soil resistances respectively. The models
represent the forces in the pile-soil failure zone. Thus, radiation damping
is ignored. The components of the models are described below,

+----- pile bo-t-torrl


segrlent 1""10.$5

elClsto-plClstiC spring

__________ COr"1preSSion
li' _... only MCl5S

""'---- res;sta.nce
gap

"'---hyperbOliC-PlastiC
~~7=:- spring
2 "-- pile segMent MCISS

~ " - - pile spring

Figure 1a. Shaft Resistance Model Figure 1b. Toe Resistance Model

Consideration of rate effects and damping

Compared to a slow (static) penetration into the ground, the


dynamic resistance may be higher than the (possibly temporarily reduced)
static resistance due to the inertia of the displaced soil particles, and due
to the high relative pile-soil velocities. Smith considered these velocity
effects using a simple approach based on the pile velocity, v.

(2)

where Rd is the dynamic resistance, J s is the Smith damping factor and Rs


is the static pile soil resistance which is a function of time. The total
resistance, RIo that the pile has to overcome is then

(3)

Many correlation studies, summarized by GRL (1992), have shown


considerable scatter for the damping parameter, J s ' using Smith's
approach. No direct relationship between J s and soil type has been clearly
observed. In fact, the damping term is often adjusted to absorb all of the
uncertainties in a correlation study. For example, where inertia effects, soil

1122 Rausche, Likins, Goble


set-up and relaxation (including different times of pile installation, static
load test and/or restrike test) and incomplete capacity activation (high
blow counts) have not been properly considered, improper corrections are
often made to the shaft damping parameter rather than to the computed
capacity value.

Coyle and Gibson (1970) suggested that the maximum dynamic


resistance contribution varies not linearly but rather exponentially to the
pile velocity. Thus,

(4)

where the exponent, N, typically is less than 1. While this seems inherently
good in that it matches laboratory measurements of maximum damping
resistance, actual application of this equation for the wave equation
creates large unrealistic damping oscillations when the velocity during the
unloading portion of the blow. Note also that the damping constant J G in
this approach has dimension [s/m]1/N. A conversion of a damping factor
from Smith's linear system to a nonlinear system is therefore not a simple
operation.

A somewhat different approach (Briaud and Garland 1984) relates


the total dynamic capacity to the static one using a "static" velocity, Vs
(e.g., the load test velocity).

(5)

There is a very basic difference between Briaud's approach and


those proposed by Smith and Coyle. Both Smith and Coyle approaches
include a separate damping component, usually explained as related to
the viscous behavior of soil; for Smith and Coyle, the elevated resistance
is simply the sum of the static and the damping components. Briaud's
approach determines an overall elevated resistance, instead of defining
individual static and damping components. If Briaud's elevated resistance
is determined by a dynamic test, then it must be reduced to a long term
static value using the ratio of testing velocities which are themselves time
variable. To be useable, Briaud's method requires an average velocity,
i.e., the failure set divided by the time expended to reach failure. An
additional damping component must still be added as per Randolph to
produce overdampened behavior seen in dynamic pile test records.

In order to satisfy the need for (a) an elevated resistance (b) I

viscous damping and (c) an exponential relationship between loading


velocity and capacity Rausche et al. (1992) proposed the following
equation.

1123 Rausche, Likins, Goble


(6)

This proposed approach is the improved model and contains only


one variable in the damping term, namely the pile velocity, v. The second
term in Eq. 6 represents linearly viscous damping. The resistance R a is the
maximum static resistance component activated during the blow prior to
the time under consideration (starts at zero for every blow, increases until
the failure load is reached, and then remains constant). Similarly, Vx is the
maximum velocity achieved up to a particular time during the blow. Both
Ra and Vx usually reach their maxima during the blow very quickly and then
remain constant. This approach addresses completely an elevated
resistance and the exponential nature of the maximum viscous damping
as determined in laboratory tests, but avoids the numerical shortcomings
of the Coyle approach since the exponential term, v x ' does not return to
near zero during the blow.

In any event, the damping factor, J R , and the exponent, N, must be


determined from special laboratory or in-situ tests or from values given in
the literature. J R has dimensions (s/m) 1/N and determines the magnitude
of the viscous damping force. The velocity exponent, N, defines the rate
at which the damping increases, given a velocity maximum, v x ' which is
related to the measured velocity, v.

This improved exponential model yields results comparable to


Smith's approach when the maximum pile velocity is within certain narrow
ranges. An approximate recalculation of the required damping factor J R
for the exponential approach from the corresponding Smith damping factor
J s commonly used is easily and automatically possible. For example, after
choosing a reference velocity Vx (say 3 m/s) and a ratio of average
temporary to activated static resistance (say 0.9), J R could be calculated
from

(7)

Details of Resistance-Displacement Relationships

Shaft

The elasto-plastic Smith's shaft resistance model is satisfactory.


Although it has been proposed by Novak and other researchers that the
quake (the elastic dynamic relative pile-soil displacement) be determined
from the shear modulus of the soil, numerous dynamic signal matching

1124 Rausche, Likins, Goble


analyses have shown that Smith's proposed shaft quake of 2.5 mm (0.1
inches) is generally reasonable. Thus, Smith's original shaft quake
appears to be better than one that relies on the soil's shear modulus which
is strongly dependent on the magnitude of the shear deformations.

The authors also investigated a modified, bi-linear shaft unloading


quake to introduce extra hysteresis into the static shaft resistance law.
However, the complexity of the numerical treatment and an additional
unknown did not justify the small gain in realism.

Toe

Correlation studies found in the literature as well as signal matching


by CAPWAP have not shown a relationship between the soil stiffness
(flexibility or quake) and soil type. In fact, the only conclusions supported
to date are that high dynamic (not necessarily static) quakes occur
sometimes in saturated soils (Likins 1983) and quakes larger than the
GRLWEAP recommended 0/120 are often observed. A hyperbolic toe
resistance vs. toe displacement relationship is perhaps more realistic and
can be numerically achieved by introducing a factor, c q , which multiplied
with the quake yields the point where ultimate is reached (Figure 2). Thus,
unlike a pure hyperbola, the improved model has an ultimate resistance
that still can be reached at a finite toe displacement. This load-movement
relationship had been proposed by Eriksson (1990).
R

Au -------,f--------------=-~- /-----,
fl //
I I "
I I /
I I /
/ I ///
I I /
I I /
I I /
I /
I ,,/
I "
'/
/, Y

// :
I I I u
Gap Quake
I Quake (Factor)

Figure 2. Toe Static Resistance Penetration Behavior

Toe Soil Mass Resistance

Force and motion measurements at the pile toe have indicated a


pronounced inertia effect in cohesive soils (Grasshoff 1953 and Rausche
1970). Figure 3 shows a similar effect observed in a modified SPT;
dynamic strain and acceleration records were measured at the SPT top
and the toe resistance force and toe displacement were calculated. The

1125 Rausche, Likins, Goble


observed inertia effect is the result of a soil mass adhering to or moving
with the SPT special tip; the GRLWEAP toe mass model is only active
during the first positive acceleration. Using this model, the viscous
damping was reduced in a number of modified SPT tests and their
subsequent data analysis. These special SPT tests were performed to
confirm suspected soil behavior and are probably not necessary for
standard SPT measurements.

Tip Resistance
. - - SPT

_ C Soil mass effect


r=:==; FlQ t Tip

II \
2
' - - - - - ' . - - 3167 CM

Second impact
~
UI ii TiP ResistQoce

Displacement, U

Figure 3. Resistance Force vs. Displacement at Special SPT Tip from


Dynamic Measurements

Consideration of Soil Strength Changes During Driving

A hammer blow changes the soil strength due to pore water


pressure and other effects mentioned earlier. Regaining the original soil
strength occurs in an logarithmic manner. Skov and Denver (1988)
suggested two restrike tests, one early and one later, to allow for an
extrapolation to the pile's long term bearing capacity; they also
recommend the first restrike be no earlier than 12 hours after driving for a
reliable prediction of long term capacity. Their formula (Eq. 1) therefore
cannot be reversed to precisely predict the capacity during driving, but
does indicate the strength change trend during driving that can be
modeled.

1126 Rausche, Likins, Goble


Static resistance at the time of driving from presumably known long
term resistance should take into account the energy expended on the soil,
Such an approach would be beneficial both for impact and vibratory driven
piles, It also must take into account the "rest periods" between energy
dissipation in the soil (Figure 4), Thus, if at one point along the pile the
ultimate shear resistance is T u ' then the unit energy dissipated in the soil
by pushing the pile relative to the soil a displacement u is

(8)

Static soil strength

Time
between blows

Time
Figure 4, Soil Strength Changes During Driving

There is a certain limit energy, e l , which once reached or exceeded,


causes the soil to reach its residual ultimate strength T ru ' Conceptually,
each time energy is dissipated in the soil, the temporary ultimate shear
strength of the soil reduces by an increment which, as a first
approximation can be considered linearly related to the current shear
strength (although some exponential law may also be appropriate).

Thus,

ee
soil

I
T
old,u
(9)

T new,u = T old,u - f:j,T (10)

with
T ru :s; T new,u :s; Tu (11 )

1127 Rausche, Likins, Goble


During any pause in energy application, such as between hammer
blows or interruption of driving, the soil regains strength. As a time period
of rest, 6t, passes, the shear strength increases to

(12)

where trag is an appropriately chosen lag time from the beginning of the
most recent hammer blow and A is as discussed for Eq.1. If successful,
this approach would make driveability analyses much more accurate,

Static soil resistance increase (relaxation)

The static (displacement dependent) soil resistance component may


increase temporarily during dynamic loading. A good example is the
occurrence of negative pore water pressures at the pile toe in very dense
saturated fine sands and silts. After driving, pore water pressures and
effective stresses return to their natural levels and long term resistance
then is lower; the term relaxation has been used. Thus, by the time a
static test is performed on a dynamically installed pile, relaxation effects
usually have occurred and the engineer is left to wonder why the driving
resistance was so high. There is no current mathematical model
describing this relaxation effect, since it happens very quickly. However,
an approach similar to the proposed approach for the soil resistance
degradation should be applicable since pore pressure dissipation is
generally logarithmic. Relaxation has also been observed for piles driven
into weathered shale, although there the time required is longer and the
logarithmic equation seems applicable.

Recommended Parameters and Procedures

For the shaft, if no experience data exists to suggest otherwise, the


toe quake should be fixed at 0.1 inches. As long as no other experience
data exists, the proposed exponential shaft damping approach (Eq. 6)
should be used as long as no other experience data exists. The shaft
damping factor J R can be computed from Eq. 7 based on the Smith
parameter J s chosen from soil type according to Smith's original
recommendations.

The toe quake recommendation of 0/120, independent of soil type,


still seems to be reasonable. It is recognized, however, that this value is
usually a lower bound. In fact, quakes on small diameter toes like the
modified SPT of Figure 3 suggest substantially greater values than 0/120.
More realism is therefore introduced with the hyperbolic loading behavior,
The displacement at which the hyperbola ends and where pure plasticity
starts should be 2.5 times the toe quake unless more specific data is

1128 Rausche, Likins, Goble


available for a site, The soil mass may be calculated for open profiles
considering a volume that has a diameter equivalent to that of the pile and
a height of 2 m unless there is reason to believe that the soil column is
actually shorter, For large displacement piles a soil mass attached to the
pile bottom (practically extending the pile) may also be calculated based
on a volume that equals a cube with its dimensions equal to the effective
pile diameter. Toe damping would be replaced in part by the inertia of the
soil mass for cohesive soils. The velocity damping force therefore can be
smaller and toe damping parameters can be fixed, independent of soil
type an.d would use the exponential approach as detailed above for the
shaft damping. This approach is in general agreement with results from
signal matching analyses of numerous field tests.

Distinction has to be made between the bearing graph analysis and


a driveability study. Bearing graphs still may be calculated based on
assumed capacity values split into shaft resistance and end bearing
components. For driveability analysis, the data preparation process
requires the ultimate unit static soil resistance, perhaps obtained by
measurement on an SPT for improved predictions. While the Smith
approach was based on resultant force values, more realism and accuracy
can be expected (particularly for non-uniform piles) when the capacity
calculations are based on unit shear stresses and an equivalent
circumference for both shaft and toe. With this additional input
information, the calculation of resultant resistance forces and' of a soil
mass size may be automated.

For the driveability analyses, the soil degradation/set-up/relaxation


effects can now be automatically considered. If proven adequate, this
concept would revolutionize the currently available analysis process.

Summary

A modification of Smith's soil resistance model has been proposed.


For the practicing engineer, these changes will not require any additional
knowledge of soil behavior than currently required, although for driveability
analyses, some measurement on a SPT will be useful. Naturally, the more
accurate the soil exploration, the more accurate the prediction of the static
and dynamic soil behavior.

The proposed model considers the exponential nature of the total


soil resistance increase with loading rate. It also considers a static and
a dynamic resistance component rather than one increased displacement
dependent total soil resistance. Furthermore, the model does not include
radiation damping as an additional refinement. The model does consider
the hyperbolic nature of the resistance vs. toe penetration behavior, and

1129 Rausche, Likins, Goble


toe soil mass effect. This model does not aid in the calculation of plug-
pile interaction forces.

This improved model for the analysis of pile driving has been
selected such that the engineer is not burdened with a totally new
approach or complex additional calculations for input preparation. The
additional model parameters, exponent, N, toe quake factor, c q , and soil
mass, m l , can be easily estimated or, for many standard analyses, ignored.
The new damping factors can be calculated based on current practice. An
effective circumference, providing the pile-soil contact area for both shaft
and toe, is a known quantity. With unit shaft resistance and end bearing
pressures (perhaps measured) specified driveability analysis become a
simple and realistic task.

Recommendations

The model presented here must still be extensively tested. Although


it requires some additional input (exponent, toe quake factor, toe soil
mass, effective pile circumference), comparison analysis would be very
quick and easy. At the same time, however, existing measurement
capabilities and the information available during acquisition of SPT data
should be used.

After the completion of the dynamic analysis, a rational and realistic


static reanalysis should be performed in the future. This analysis would
yield the pile top load-set curve and therefore would allow for an easy
check of the accuracy of the dynamic simulation. For the greatest
accuracy, estimated effects of soil set-up/relaxation (and possibly even
creep) should be included and checked against real measurements.

Acknowledgements

This work is part of a study conducted for the Federal Highway


Administration. The advice, suggestions and support of Messrs. Carl Ealy
and Jerry DiMaggio both of FHWA and Professor Jean Louis Briaud of
Texas A&M University are gratefully acknowledged.

References

Briaud, J. L. and Garland, E. (1984). "Pile response to dynamic loads."


Texas A&M University, College Station, Texas.
Chow, Y. K., Wong, K. Y., Karunaratne, G. P. and Lee, S. L. (1988). "Wave
equation analysis of piles-a rational theoretical approach." 3rd Int'l
Application of Stress-Wave Theory on Piles, Ottawa, Canada.

1130 Rausche, Likins, Goble


Corte, J. F. and Lepert, P. (1986). "Lateral resistance during driving and
dynamic pile testing." 3rd Int'I Conference on Numerical Methods
in Offshore Piling, Nantes, France, 10-33.
Coyle, H. M. and Gibson, G. C. (1970). "Empirical damping constants for
sands and clays." ASCE Journal of Soil Mechanics and Foundation
Division,
Eriksson, H. (1990), "Static behavior of driven piles estimated from stress
wave measurements on dynamic probes." Presented at the 1990
PDA Users Days, Gothenburg, Sweden.
Grasshoff, H. (1953). "Investigation of values of the dynamic penetration
resistance to model piles in sand and clay, obtained from tests." 3rd
Int'I Conference on Soil Mechanics and Foundations Eng., Vol. II.
GRL and Associates, Inc., (1991). GRLWEAP Manual, Cleveland, Ohio.
GRL and Associates, Inc., (1992). "Determination of pile driveability and
capacity from penetration tests." Cleveland, Ohio. Interim report
submitted to the FHWA, Washington, DC.
GRL and Associates, Inc., (1993). CAPWAP Manual, Cleveland, Ohio.
Heerema, E. P. (1979). "Relationships between wall friction displacement
velocity and horizontal stress in clay and in sand for pile driveability
analysis." Ground Engineering.
Holeyman, A. (1985). "Dynamic non-linear skin friction of piles."
Comptes-rendus Symposium Penetrability, Driveabilityof Piles, San
Francisco, CA, 173-176.
Lee, S. L., Chow, Y. K., Karunaratne, G. P. and Wong, K. Y. (1988).
"Rational wave equation model for pile driving' analysis." J. of
Geotechnical Engineering, ASCE, 114(3).
Likins, G. E. (1983). "Pile installation difficulties in soils with large quakes."
ASCE Conference, Philadelphia, Pennsylvania.
Middendorp, B. and van Brederode, P. J. (1984). "Skin friction models for
sand from static and dynamic laboratory load test." 2nd Int'I
Application of Stress-Wave Theory on Piles, Stockholm, Sweden,
210-220.
Novak, M., Nagami, T. and Aboul-Ella, F. (1978). "Dynamic soil reaction for
plane strain case." J. of Engineering Mechanics Division, ASCE,
104(4), 953-959.
Randolph, M. F, and Simons, H. A. (1986). "An improved soil model for one
-dimensional pile driving analysis." 3rd Int'I Conference on
Numerical Methods in Offshore Piling, Nantes, France.
Rausche, F. (1970). "Soil response from dynamic. analysis and
measurements on piles." Ph.D. thesis, Case Western Reserve Univ.,
Cleveland, Ohio.
Rausche, F., Goble, G. G. and Likins, G. E. (1985). "Dynamic determination
of pile capacity," J. of Geotechnical Engineering, American Society
of Civil Engineers, 111 (3).

1131 Rausche, Likins, Goble


Rausche, F., Goble, G. G. and Likins, G. E. (1992). "Investigation of
dynamic soil resistance on piles using GRLWEAP." 4th Int'I
Application of Stress-Wave Theory Model on Piles, Rotterdam.
Rausche, F., Hussein, M., Likins, G. and Thendean, G. (1994). "Static pile-
load movement from dynamic measurements." Conference on
Vertical and Horizontal Deformations of Foundations and
Embankments, ASCE, Special Publication No. 40.
Skov, R. and Denver, H. (1988). "Time-dependence of bearing capacity of
piles." 3rd Int'l Application of Stress-Wave Theory on Piles, Ottawa,
Canada.
Smith, E. A. L. (1960). "Pile-driving analysis by the wave equation." J. for
Soil Mechanics and Foundations Division, ASCE, 86(4), 35-61.
Svinkin, M. R., Morgana, C. M. and Morvant, M. (1994). "Pile capacity as
a function of time in clayey and sandy soils." 5th Int'I Conf. and
Exhib. on Piling and Deep Foundations, Bruges, Belgium.

1132 Rausche, Likins, Goble


Examination of the Energy Approach for
Capacity Evaluation of Driven Piles

Samuel G. Paikowskyl and Valerie A. LaBelle l

Abstract

A simplified field method known as the Energy Approach is presented. The


method enables evaluation of the capacity of driven piles based on dynamic
measurements during driving. An in depth study of this method was carried out as part
of an FHWA supported research study.

The method was validated through an examination of two large scale data sets.
One, PDILT, contained 208 pile-cases related to 120 piles that were monitored during
driving followed by a load test to failure. The other, PD, contained 403 pile-cases of
piles monitored during driving only. As a result of this study, the accuracy of the
method was evaluated and factors of safety were reconunended.

An independent control data set of driven piles monitored during driving and
load tested to failure was gathered. The case studies in the control data set do not
duplicate the pile-cases of the data sets that were used for the analysis of the Energy
Approach method. The control data set was used to evaluate the validity of the
reconunended factors, independently examining the perfonnance of the method and its
accuracy.

It was found that for all the examined new cases, even though a large scatter in
predictions was evident, the use of previous reconunendations would result in
economical pile design and safe perfonnance.

1Associate Professor and Graduate Research Assistant, respectively, Department of


Civil Engineering, University of Massachusetts Lowell, Lowell, MA 01854.

1133 s. paikowsky et al.


Background

Dynamic analyses of piles are methods that predict pile capacity based on
behavior during driving. Evaluation of static capacity from pile driving is based upon
the concept that the driving operation induces failure in the pile-soil system~ in other
words, a very fast load test is carried out under each blow. There are basically two
methods of estimating the ultimate capacity of piles on the basis of dynamic driving
resistance: pile driving formulas (i.e., dynamic equations) and Wave Equation (W.E.)
analysis.

Dynamic equations can be categorized into three groups: theoretical equations,


empirical equations and those that consist of a combination of the two. The theoretical
equations are formulated around analyses that evaluate the total resistance of the pile,
based on the work done by the pile during penetration. These formulations assume
elasto-plastic force-displacement relations, as shown in figure 1. The total work is,
therefore:
Q
W. R (S.-)
II 2 (1)

where R u is the yield resistance, Q is the quake denoting the combined elastic
deformation of the pile and the soil, and S is the set, denoting the plastic deformation
(permanent displacement) under each blow.

Realstance

Ru -- --_._-----j
1/

/ 1/
/ WOI1c 1/
/ 1/
V 1/
1--------1 Set Quak8
I D1s~
Figure 1. Resistance vs. displacement at the top of the pile.

When the work ofthe resisting forces, W, is equated to the energy delivered to
the pile, i. e., W = Ed' we can extract the basic familiar elements of the dynamic
equations:

1134
Ed
R1/
Q
(S. - )
(2)
2

The low reliability ofthe dynamic equations (see, for example, Housel, 1965) is
due to several reasons: (a) their parameters, such as the efficiency of energy transfer and
the quake, are assumed, and, therefore, may not reflect the high variability of the field
conditions (b) the theoretical analysis of the "rational" pile formula (see, for example,
Bowles, 1988) relates the energy transfer mechanism to a Newtonian analysis of ram-
pile impact. This formulation is theoretically invalid for representing the "elastic" energy
transfer mechanism which actually takes place and (c) no differentiation is made between
static and dynamic soil resistances.

A clear distinction is, therefore, required between the underlying valid energy
analysis and additional estimations ofthe different parameters, many of which are either
invalid theoretically or practically limited in their accuracy.

The Wave Equation (W.E.)

Stress propagation in a pile during driving can be described by the following


equation of motion:

(3)

where:
u(x,t) = longitudinal displacement of infinitesimal segment
~,Sp = pile area and circumference, respectively
Ep,p p = modulus of elasticity and unit density of the pile material.

The fiierion stresses (fJ are activated by the pile movement and under free wave
motion (f9 = 0), equation 3 becomes the familiar I-D W.E. (One-Dimensional Wave
Equation). The fiiction stresses are traditionally represented by a soil model suggested
by Smith (1960). The static soil resistance-displacement relationship is assumed to be
elasto-plastic and is represented by a spring and a slider. The dynamic resistance is
assumed to be viscous (soil type related) and, therefore, velocity dependent, represented
by a dashpot parallel to the spring. The resisting soil stress (fJ is a combination of the
two.

The wave equation formulation is used in two ways:


(a) Pre-driving analysis, where the entire system is modeled, including the pile, hammer
and driving system, as was first suggested by Smith (1960).
(b) Post driving analysis, utilizing dynamic measurements obtained near the pile head
1135
during driving. The response ofthe modeled pile-soil system (e.g., force at the pile top)
under a given boundary condition (e.g., measured velocity at the pile top) is compared
to the measured response (force measured). The modeled pile-soil system or, more
accurately, the modeled soil that brings about the best match between the calculated and
measured responses, is assumed to represent the actual soil resistance. This procedure
of wave matching technique was first suggested by Goble et al. (1970), utilizing the
computer program CAPWAP. Similar analyses were developed by others, e.g.,
Paikowsky (1982), (see also Paikowsky and Whitman, 1990) utilizing the computer
code TEPWAP.

Capacity Evaluation in the Field

The procedure of monitoring pile driving by dynamic measurements is well


established. Early large scale studies (e.g., Michigan State Highway Commission, 1965,
Texas Highway Department, 1973, and Ohio Department of Transportation, 1975, see
also Highway Research Record, 1967, and Goble et aI., 1970) led to the development
of conunercial systems which enable complete and relatively easy acquisition of dynamic
measurements and analysis during driving. The PDA (pile Driving Analyzer, see Pile
Dynamics Inc., 1992), which is the most conunonly used device in the USA, utilizes a
simplified pile capacity evaluation method, known as the Case method (see Goble et aI.,
1970). The formulation of the method is based on a simplification of the wave equation
and employs the force and velocity measurements taken at the pile top in order to obtain
the total resistance. The static resistance is then evaluated based on a dimensionless
damping coefficient, Ie' (Case damping) that was correlated to the soil type at the pile
tip (Goble et al., 1967, 1975). The method encountered two fundamental difficulties:
(a) the total resistance is time dependent and different variations of the method produce
different results, (b) the dimensionless damping coefficient was found to have
questionable correlation to soil type (e.g., Paikowsky, 1982, Thompson and Goble,
1988, Paikowsky and Chemauskas, 1992, and Paikowsky et aI., 1994).

The Energy Approach Principle

An energy balance must exist between the total energy transferred to the pile and
the work done by the pile/soil system. This is true for the dynamic equations as well as
for the W.E. type of analyses. Both analyses assume static resistance to follow elasto-
plastic force-displacement relations. While the W.E. formulation distributes the
resistances along the pile as it simulates the propagation of the stresses, the energy
balance equation lumps it for the entire system. Although the losses due to the dynamic
phenomena are considered by viscous damping, they are not represented in the dynamic
equations, as the particle velocity along the pile cannot by evaluated. Practically,
however, the dynamic component of the resisting forces, even though represented by
viscous damping, accounts for other energy losses such as radiation, soil inertia, true
damping, and more. These energy losses are determined by the combination of pile
shape and surrounding soil type in addition to the penetration velocity. The W.E. type
of solutions (including CAPWAP and TEPWAP) consider the damping at each depth,
1136
and, therefore, indirectly account for the energy losses in the different pile types and
surrounding soils. As a result: (a) very little correlation can be found between the soil
type and the damping parameters. This is true for Smith damping and even more so for
the Case method damping (paikowsky et al., 1994), (b) while the total capacity of the
pile can be found accurately by analyses like CAPWAP (as it matches energy delivered
to work produced), the distribution of the resistances is not necessarily accurate (e.g.,
Paikowsky, 1982, Thompson and Devata, 1980, Paikowsky and LaBelle, 1995), and (c)
the parameters used to obtain matches of energy delivered to work produced will be a
function of pile type, especially large vs. small displacement (for difference in parameters
between small to large piles, see, for example, Liang, 1991). A detailed substantiation
of the above is presented by Paikowsky et al., 1994.

The Energy Approach Equation

Equation 2 is used as the basic energy balance equation. The parameters of this
equation are replaced by measured values obtained during driving.
(a) the energy delivered is taken as Ed = E max, the maximum energy obtained by the
maximum value of:

!V(t)F(t)dt (4)
o

where Vet) and F(t) are the measured force and velocity signals at the pile top.
(b) the Quake is evaluated from the difference between the maximum pile top
displacement and the permanent set.
Q D_-S (5)

where the maximum displacement is obtained by the maximum value of:

!V(t)dt (6)
o

(c) the set can be obtained by the final displacement of the integrated velocity signal for
the full measured time:

D /ill !v(t)dt (7)


o

Practically, however, the displacement is already the second integration of a measured


value (acceleration) and the accuracy of the final displacement is questionable. The use
of the field blow count is, therefore, recommended for the set.

The maximum resistance under the above assumptions is obtained by the


following Energy Approach Equation:
1137
EIIllIIl:
R II ---=:....--
(DIIllIIl:- Set) (8)
Set.-~--
2

This equation considers only the elasto-plastic energy losses of the pile-soil system and
can be regarded as the maximum possible resistance. Correlation to the predicted static
capacity is achieved via a single correlation factor which represents all dynamic related
energy losses in the soil:
R,. K",oR Il (9)

where: Ita = static pile capacity


Kap = static pile capacity coefficient.

Energy Losses and Soil Inertia

Soil inertia is an unaccounted for major factor contributing to the energy loss
during driving. As such, a substantial portion of the dynamic resistance should be a
function oftwo parameters: (a) mass via volume of the displaced soil that is a function
of the pile geometry (b) acceleration of the displaced soil, especially at the tip that
conveniently can be examined as a function of the driving resistance.

The volume of the displaced soil is identical to the volume of the penetrating
pile, excluding the cases in which pile plugging takes place (paikowsky and Whitman,
1990). The piles, therefore, can be classified as small (e.g., H and open pipe) and large
(e.g., closed pipe and square concrete) piles. Additional classification of open-pipe piles
can take place according to a tip-area ratio similar to that used for soil samplers
(paikowsky et aI., 1989).

As most of the soil displacement takes place at the tip area, an appropriate
classification of piles can be achieved by looking at the ratio between the piles'
embedded surface area and the area of the pile tip:
As*in
AR - (10)
A ttp

where:
AR = pile area ratio

"kin
~p
=
=
pile's surface area in contact with soil
area of the pile tip.

According to this ratio, a pile that is traditionally referred to as a "large


displacement" pile can behave like a small displacement pile if it is driven deep enough.
Because the fiictional resistance of a pile increases as the pile skin area in contact with
1138
soil increases, the effect of the soil mobilized at the tip decreases. As the pile's
embedded surface area and the skin friction increases, the energy losses resulting from
the mobilization ofthe soil mass at the pile tip will decrease relative to the energy losses
along the side of the pile. For example, the area ratio for cylindrical (closed-end) piles
IS:

A 2D
R R (11)

where:
D = penetration depth
R = pile radius.

For one diameter, this area ratio increases linearly with depth, e.g., a 14-in (356-
mm) diameter pile will have an area ratio of 69 at the depth of20 ft (6.1 m) and an area
ratio of360 at the depth of 105 ft (32 m). It is clear that the effect of soil inertia at the
tip in the second case will be substantially smaller than that in the first case and the pile
may be classified as a "small displacement pile."

The energy loss through the displacement of the soil mass at the tip, is directly
related to the acceleration of this mass. The evaluation of this acceleration is beyond
the scope of the present paper. Indirect evaluation, however, can be performed through
the driving resistance, which is the measure of the pile's permanent displacement under
each hammer blow. With low driving resistance high acceleration will take place at the
pile tip, resulting in high inertia of the tip soil mass. In the case of high driving
resistance (hard driving), there is little mobilization of the tip soil mass, the acceleration
at the tip is very low, and as a result, the corresponding energy loss is small.

Considering the preceding criteria, the Energy Approach should theoretically


produce two distinct trends:

In the case of high "unknown" energy losses, i.e., in easy driving of piles
with small area ratios, the Energy Approach predictions should have a tendency for
over-prediction. Hence, R.u is expected to be higher than the actual resistance as the
large energy losses were not considered. As a result, ~p is expected to be smaller than
unity (Ksp < 1.0).

In the case of small "unknown" energy losses, i.e., hard driving of piles
with large area ratios, the Energy Approach predictions should have a tendency for
under-prediction. Hence, Ru is expected to be smaller than the actual resistance as there
are only small energy losses and the full capacity may not have been developed. As a
result, K sp is expected to be higher than unity (Ksp > 1.0).

These expected patterns may not be applicable for restrikes in which the pile is

1139
driven a very short distance (say less than one diameter) and, hence, the remobilization
of the soil mass at the tip is not complete.

Large Scale Study

General

A large scope study examining the energy approach and the W.E. based methods
was presented by Paikowsky et aI. (1994). The study was based on two extensive data
sets gathered at the University of Massachusetts at Lowell. One, PDILT, contains 208
dynamic measurement cases on 120 piles monitored during driving, foIlowed by a static
load test to failure. The data were obtained from various sources and reflect variable
combinations of soil-pile-driving systems. The other, PO, contains data on 403 piles
monitored during driving and was provided by Pile Dynamics Inc. of Cleveland, Ohio.
All cases were examined and analyzed with the detailed results presented by Paikowsky
et at. (1994). The following sections summarize the major conclusions and
recommendations obtained in this study. Some of the following conclusions will not be
evaluated through the control data set and are included herein for the completeness of
the presentation only.

Conclusions

Four major correlations were examined in the study:


(1) Damping parameters vs. soil type.
(2) Load test results vs. wave matching techniques (CAPWAPrrEPWAP) capacity
predictions using the parameter ~ = load test capacity/wave matching
prediction.
(3) Load test results vs. Energy Approach capacity predictions using the parameter
Ksp = load test capacitylEnergy Approach prediction.
(4) The predicted capacity using CAPWAPITEPWAP vs. that of the Energy
Approach using the parameter Kew = wave matching prediction/Energy
Approach prediction equivalent to K ew = Ks/Ksw.

The following conclusions are based on the graphical and statistical analyses of
the above correlations as presented in the study:

1. Higher accuracy and substantially lower scatter was obtained by the Energy
Approach for all cases compared to the predictions of the wave matching techniques as
summarized in table 1.

2. The End ofDriving (EOD) condition is of special interest as it represents the


ability ofthe methods to predict the capacity during driving. The data presented in table
1 clearly indicate very good EOD predictions and correlations of the Energy Approach
for all area ratios with better performance for small displacement piles. The large mean
and standard deviation ratios for the K sw coefficient suggest limitations of the wave
1140
matching methods for all piles at the end of driving, particularly for large displacement
piles.

3. Viscous damping does not truly represent the physical phenomena through
which energy is lost and, hence, cannot be viewed as intrinsic to soil type. No
correlation was observed between either Case damping parameter (JJ or Smith damping
parameters (for both tip and side) and soil type.

4. The capacity predictions for small displacement piles resulted in higher


accuracy and substantially lower scatter for both dynamic methods when compared to
the predictions and the scatter obtained for large displacement piles.

Based on the obtained results, small and large displacement piles were defined
based on area ratio. Small displacement piles relate to AR>350 and large displacement
piles relate to AR<350.

5. The above conclusion was reinforced by excellent correlations that were


obtained between the prediction methods (CAPWAPITEPWAP vs. the energy
approach) for the small displacement pile cases.

6. Correlations between driving resistance and dynamic predictions did not lead
to definitive conclusions. Analysis of the correction coefficients (Ksw and ~p) for all
times on the basis of blow counts between 0 to 10 BPI (0.39 blows per mm) and over
10 BPI (0.39 blows per mm) are presented in table 1 suggesting the following trends:

Table 1. Summary of the large scale data analysis (paikowsky et aI., 1994)
Area Driving Driving K = Load TestJCAPWAP K = Load TestJEnem- Appr.
Ratio Time Resistance
A., (BPI) no. mean std. dev. no. mean std. dev.

all cases all times all 206 1.367 0.533 208 0.925 0.293

all cases EOn all 95 1.441 0.619 96 1.001 0.316

>350 all times 0-10 25 1.360 0.558 25 0.939 0.279

>350 all times >\0 32 1.159 0.442 32 0.929 0.219

<350 all times 0-10 101 1.353 0.488 101 0.906 0.326

<350 all times >10 43 1.60\ 0.628 43 0.95\ 0.296

1 BPI = 0.039 blows per mm

Small displacement piles with high driving resistance will result in a small loss
ofenergy due to soil inertia and, therefore, more accurate predictions, as the actual pile
resistance is similar to the maximum resistance during driving. As a result, both
methods of analysis performed well for that category.

Large displacement piles with low driving resistance will result in a large loss

1141
of energy due to soil inertia and less accurate predictions, as the actual pile resistance
is the difference between the maximum pile resistance during driving and the large
energy loss. (For this category, the Energy Approach predicts well for EOD and over-
predicts for Beginning OfRestrike (BOR) while the wave-matching techniques seem to
under-predict for EOD and improve with time.)

Factor of Safety Recommendations

The simplicity of the Energy Approach fonnulation together with its high
accuracy at the end of driving makes it an ideal method of analysis to be used in the field
and as a check for the wave matching analyses.

Based on data set PDILT, Paikowsky et aI. (1994) recommended the following
factors of safety to be used with the Energy Approach predictions:

F.S. = 2.50 for all piles in all cases (AAA, mean Ksp = 0.93).
F.S. = 2.00 for all end-of-driving cases (AEA, mean K sp = 1.00).
F.S. = 2.00 for all small displacement piles (AR > 350) in all cases (SAA,
mean ~p = 0.94).
where: - first letter denotes pile type: A = all piles, L = large displacement, and S = small
displacement.
- second letter denotes time of measurement: A = anytime, E = end of driving,
and B = beginning of restrike.
- third letter denotes soil type: A = all soils, S = sand and silt, C = clay and till,
and R= rock.

The Control Data Set

Objective

The above findings were based on the largest available data set comprised of
both dynamic measurements and static load tests to failure. Although supported by a
comprehensive risk analysis, the recommended factors of safety would probably be
considered as "unconservative" under the existing standards of practice. It is of great
importance, therefore, to conduct studies, allowing for the examination of the
recommended factors of safety for independent control data sets.

The Available Pile-Cases

The control data set contains 26 pile-cases related to 11 piles which were
dynamically monitored during driving and load-tested to failure. The piles were driven
in three project sites, two in Florida and one in Wisconsin. For all pile-cases, piles
ranged in diameter from 9.625 inch to 24 inch (24.46 cm to 60.96 cm) and in length
from 90 ft to 175 ft (27.4 m to 53.3 m).

1142
T abl e 2 Summaryo f t he contro I dat a set .
Pile no.! Pile Area Time Blow Static CAPWAPI K.,. Energy ~
Case no. Type Ratio Count LT. TEPWAP Appr.
(BPI) (kips) (kips) (kips)

I-I CEP 12.75" 463 EOD 17.oJ 647 390 1.66 472 1.37

CEP 12.75" 3
2-1 441 EOD 3.8 504 271 1.86 336 1.50

3-1 HP 12x63 7224 EOD 0.8 3 315 110 2.86 202 1.56

3-2 7224 BORI 2.5 270 1.17 366 0.86

3-3 7236 BORJ 6.3 340 0.93 536 0.59

3-4 7240 BOR4 4.0 450 0.70 574 0.55

4-1 HP 12x63 6612 EOD 1.03 214 105 2.04 203 1.05

4-2 6612 BORI 2.0 235 0.91 282 0.76

4-3 6631 BOR2 2.2 220 0.97 361 0.59

4-4 6643 BORJ 2.4 335 0.64 466 0.46

5-1 CEP 14" 532 EOD 1.8 3 237 110 2.15 418 0.57

5-2 533 BOR2 1.5 390 0.61 500 0.47

6-1 CEP 9.63" 711 EOD 0.8 364 150 2.43 193 1.89

6-2 711 BOR 3.3 340 1.07 367 0.99

7-1 CEP9.63" 693 EOD 1.23 554 221 1 2.51 338 1.64

7-2 693 BOR 9.8 530 1.05 524 1.06

8-1 PSC 18"SQ 50 BORI 9.0 284 304 0.93 441 0.64

8-2 53 BOR2 7.3 264 1.08 423 0.67

8-3 55 BORL 8.0 281 1.01 389 0.73

9-1 PSC 18"SQ 41 BOR 16.0 265 300 0.88 547 0.48

9-2 46 EOD 9.2 3 245 1.08 359 0.74

9-3 47 EORL 5.3 3 235 1.13 316 0.84

10-1 PSC 24" 131 EOD 5.5 3 830 658 1.26 1019 0.81

10-2 132 BORL 10.0 587 1.41 994 0.84

11-1 PSC 24" 81 EOD 10.3 14771 756 1.95 1302 1.13

11-2 82 BORL 33.3 1008 1.47 1789 0.83

Average 1.38 0.91

Std Dev. 0.62 0.39


I Determined from TEPWAP. I BPI = 0.039 blows per nun
1 Extrapolated from a load test with a maximum load of 1400 kips. 1 kip = 4.448 kN.
3 Blow count (BPI) based on blows per foot. EOD = End of Driving.
CEP = Closed-Ended Pipe pile, PSC = PreStressed Concrete pile BOR = Beginning Of Restrike
HP = H-pile BORL = Beginning OfRA:strike after static load test.

1143
Data Analysis

A summary of the control data set analyzed cases is presented in table 2. The
data was analyzed in the same fashion as previously described by Paikowsky and
Chemauskas (1992) or Paikowsky et aI. (1994). The static pile capacity is based on five
different methods ofload test interpretation: (1) Davisson (1972), (2) shape of curve as
judged by the tangents to the curve when the load-settlement relations are drawn in a
scale in which the elastic defonnation line is approximately 20 to the horizontal load
Cl

axis (see Vesic, 1977) (3,4) Terzaghi (1942) where the capacity corresponds to limiting
top displacement of! inch (25.4 nun) and 10% of pile diameter and (5) DeBeer (1970)
using the load-settlement curve drawn on a log-log scale.

CAPWAP results were used whenever they were available. In two of the cases,
TEPWAP analysis was perfonned on digitized force and velocity signals. Case no. 7-1
was problematic and, hence, the obtained results are questionable. The Energy
Approach analysis was carried out on the data pertaining to the same blow analyzed by
the wave matching techniques.

Presentation of Results

Figures 2 and 3 present the static load test capacity versus the predicted capacity
by the wave matching technique and the Energy Approach, respectively. A relatively
large scatter exists for both methods.

Figures 4 and 5 present the ratios of static load test capacity over the dynamic
prediction versus driving resistance for the wave matching techniques and the Energy
Approach, respectively. A substantial underprediction of the wave matching technique
is observed for the End Of Driving (BOD) conditions with low driving resistance.

Figures 6 and 7 present the distribution of the prediction ratios for both methods
in the form of relative frequency. Both distributions of the large scale study and the
current control data set are presented on top of each other. The mean prediction values
presented in tables 1 and 2 together with the distributions of figures 6 and 7, clearly
show that although the average predictions of the control data set (Ksw = 1.38 and Ksp
= 0.91) matches very well that of the large data set (Ksw = 1.37 and K sp = 0.93), the
scatter ofthe control data set is much higher. This can be observed, for example, in the
large number of overpredictions by the Energy Approach and underpredictions by the
wave matching techniques.

The larger scatter of the control data set when compared to the original study
may have resulted from the unique piles used in the three sites comprising the control
data set. The H piles of cases 3 and 4 were very long and most likely plugged. As a
result, their area ratio may not reflect their actual state and with very easy driving, this
would result in overpredictions. Four out of nine EOD cases were driven with a blow
count of less than 2 BPI.
1144
Figure 2. Predicted capacity (CAPWAP or TEPWAP) vs. static load test results

o 200 400 600 800 1000 1200 1400 1600 1800


Energy Approach Predictions (kips)

Figure 3. Energy Approach predictions vs. static load test results

1145
3.0
03-1

-~

05-1
2.0 04-1
Oll-1
02-1
01-1
010-2 011-2
03-2 010-1
1.0 06-~t:&.oJ-2D7-z
f)1:{ 03-rti-l 09-1
0f?1~
1 inch = 25.4 nun
1 BPI = 0.039 blows per mm
0.0 -1----~--__.__--____r---.__--~--__.__--__r'----__1

0.0 10.0 20.0 30.0 40.0


Blow Count (BPI)

Figure 4. Ksw ratio vs. blow count

3.0

goJ
c

.-'i-
.g
-=
.-
goJ

~ ~
Col

'"
2.0
06-1
=:.- .c
Col 7-\
fIl
~
E-c
=
c:>
J- 02-1
01-1
"'0 '"
Q,
=
c:>
Q,
< 1.0 04:.106_2 - B-I-l
J- - - - - - - - - - - - - - -
~
~ &;' i&9i:B?~o-, Oll-2
'"=
~
5-a~t~J-40 - 8-1
09-1
~ 5-2 4-
II
Co
0.0
~
0.0 10.0 20.0 30.0 40.0
Blow Count (BPI)
Figure 5. Ks p ratio vs. blow count

1146
0.20 - . , . . . . . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

~ Original Study - 206 cases


mean = 1.37
I I Control Study - 26 cases
mean = 1.38

0.0 0.5 1.0 1.5 2.0 2.5 3.0


K sw = Load Test I CAPW AP or TEPWAP

Figure 6. Frequency distributions ofKsw for the original and control data sets.

0.20 - . , . . . . . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

~ Original Study - 208 cases


mean =0.93
C 0.15
I I Control Study - 26 cases
c~ mean =0.91
0=
"

~
e 0.10

.-.>...
~

~ 0.05

0.0 0.5 1.0 1.5 2.0 2.5 3.0


Ks p = Load Test I Energy Approach

Figure 7. Frequency dishibutions of Ksp for the original and control data sets.
1147
Conclusions

The analyzed control data set resulted in mean prediction ratios of both dynamic
analyses identical to the original large scale study but with a substantially higher scatter.

The application of the recommended factor of safety for all EOD cases (for
which the Energy Approach is primarily aimed) would result in a safe performance. The
application of a factor of safety of 2. 0 to all cases of small displacement piles proves to
be marginal in several Beginning Of Restrike (BaR) cases (see cases 4-4, 5-2 and 9-1)
and should be used cautiously.

Acknowledgements

The original study referred to in this paper was supported under FHWA contract
no. FHWA-DTFH61-92-C-0003 8 and Pile Dynamics, Inc. of Cleveland, Ohio. The
presented control data set was obtained from Florida DOT and CH2M-Hill of Oregon.
The authors thank Mssrs. William Knight, Peter Lai, and Paul Pessy of the Florida DOT,
Mr. Paul Bullock ofSchmertmann and Crapps and Mr. Richard Riker ofCH2M-Hill for
their help in obtaining the information for the control data set. Further developments
are currently being studied under Massachusetts Highway Department contract no.
6780. The assistance and support of Mr. Nabil Hourani, head of the MHD Geotechnical
section, are highly appreciated.

Appendix - References

Bowles, lE. 1988. Foundation Analysis and Design. McGraw-Hili, 4 th ed.


Davisson, M.T. 1972. "High Capacity Piles", Proceedings, Soil Mechanics Lecture
Series on Innovations in Foundation Construction, American Society of Civil
Engineers, Illinois Section, Chicago, March 22, 1972, pp. 387-411.
DeBeer, E.E. 1970. "Proefondervindellijke bijdrage tot de studie van het
grandsdraagvermogen van zand onder funderinger of staal". English version,
Geotechnique Vol. 20, NO.4. pp. 387-411.
Goble, G.G., Scanlan, and Tomko, J.J. 1967. Dynamic Studies on the Bearing Capacity
ofPiles, Phase II, Volume I and II, Case Western Reserve University.
Goble, G.G., Likins, G., and Rausche, F. 1970. Dynamic Studies on the Bearing
Capacity of Piles Phase III, Report No. 48, Division of Solid Mechanics,
Structures and Mechanical Design. Case Western Reserve University.
Goble, G.G., Likins, G., and Rausche, F. 1975. Bearing Capacity of Piles from
Dynamic Measurements, Final Report, Ohio Dept. of Trans., OHIO DOT-05-
75.
Highway Research Record No. 167. 1967. Bridges and Structure, Highway Research
Board, Washington, D.C.
Housel, W.S. 1965. "Michigan Study of Pile Driving Hammers", Proceeding of the
ASCE Journal of Soil Mechanics and Foundations, Volume 91, No. SM5,
September 1965, pp. 37-64.
1148
Liang, R. Y. 1991. "In Situ Determination of Smith Soil Model Parameters for Wave
Equation Analysis", Geotechnical Engineering Congress, Colorado, June 1991,
pp.64-75.
Michigan State Highway Commission. 1965. A Performance Investigation of Pile
Driving Hammers and Piles.
Ohio Department of Transportation. 1975. Bearing Capacity ofPiles from Dynamic
Measurements, Research Report OmO-DOT-05-75, Final Report.
Paikowsky, S.G. 1982. Use ofDynamic Measurements to Predict Pile Capacity Under
Local Conditions, M.Sc. Thesis, Dept. of Civil Engineering, Technion-Israel
Institute of Technology, July 1982.
Paikowsky, S.G., Whitman, R.V., and Baligh, M.M. 1989. "A New Look at the
Phenomenon of Offshore Pile Plugging", Marine Geotech., 8(3): 213-230.
Paikowsky, S.G. and Whitman, R.Y. 1990. "The Effects of Plugging on Pile
Performance and Design", Canadian Geotechnical Journal, Vol. 27, No.3,
August 1990.
Paikowsky, S.G. and Chernauskas, L.R. 1992. "Energy Approach for Capacity
Evaluation ofDriven Piles", 4th International Conference on the Application of
Stress-Wave Theory to Piles, The Hague, The Netherlands, pp. 595-601.
Paikowsky, S.G., Regan I.E. and McDonnell J.1. 1994. "A Simplified Field Method for
Capacity Evaluation of Driven Piles", FHWA, report no. FHWA-RD-94-042.
Paikowsky, S.G. and LaBelle, Y.A. 1995. "Static Pile Settlement from Dynamic
Measurements During Driving", under review for possible publication in the
ASCE Geotechnical Journal.
Pile Dynamics Inc. 1992. PDI Pile Driving Analyzer, Model P AK, Manual.
Smith, E.A.L. 1960. "Pile-Driving Analysis by the Wave Equation", Journal of Soil
Mechanics and Foundations American Society of Civil Engineers, August
1960, pp. 35-61.
Terzaghi, K. 1942. "Discussion ofthe Progress Report ofthe Committee on the Bearing
Value of Pile Foundations." Proceedings ASCE, Vol. 68:311-323.
Texas Highway Department. 1973. Bearing Capacity for Axially Loaded Piles,
Research Report 125-8-F Sept. 1967 - Aug. 1973, pp. 134.
Thompson, C.D. and Devata, M. 1980. "Evaluation of Ultimate Bearing Capacity of
Different Piles by Wave Equation Analysis", Proceedings ofthe 2"d Conference
on the Application ofStress-Wave Theory on Piles, Stockholm, Sweden, June
1980, pp. 1-33.
Thompson, C.D. and Goble, G.G. 1988. "High Case Damping Constants in Sand", 3rd
International Conference ofStress-Wave Theory in Piles, Ottawa, Canada, pp.
464-555.
Vesic, AS. 1977. Design ofPile Foundations, National Cooperative Highway Research
Program. Synthesis of Highway Practice, Publication No. 42.

1149
Influence of Pile Parameters on Pile Driveability

Mark R. Svinkin, l Member, ASCE

Abstract

This paper presents some new observations of the role that pile properties
play in the pile driveability process. A relationship is derived between maximum
force at the pile head and pile stiffness. As pile stiffness increases so does the
force generated in the pile, but the effect of that force on the pile penetration
resistance is limited. It is shown that in consequence of hammer-pile
misalignment coupling of horizontal and rocking pile motions are involved in
total resistance to pile driveability.

Introduction

For proper pile installation, it is important to overcome soil resistance


without pile damage. Pile driveability depends on energy and dynamic force
transferred to a pile, pile geometry and material, and soil resistance. Various
problems of pile driveability have been studied and published, for example
Fellenius and Samson (1976), Heerema (1978), Holloway et al. (1978), Goble
and Rausche (1980), Hannigan and Webster (1987), Brucy et al. (1988), Li et al.
(1988), Tang et al. (1988), Agrawal and Chameau (1992), Holeyman (1992) and
others. Software, such as GRLWEAP (1993), is available for wave equation
analysis applied to driveability study, equipment selection and pile design before
actually going in the field.

It is common to consider pile driveability to be a function of pile stiffness


and pile strength, Vanikar (1985). Proper choice of both quantities for given
pile length and soil conditions should insure successful driveability. Sufficient

lConsulting Engineer, Goble Rausche Likins and Associates, Inc., 4535 Emery Industrial
Parkway, Cleveland, OH 44128

1150 M. R. Svinkin
pile stiffness assures that driving forces in the pile will be able to overcome soil
resistance. Sufficient pile strength allows the pile to withstand the driving forces
without damage. However, the role of pile properties in driveability is not
restricted to aforementioned reasons. The effect of a pile as a solid body is
more complicated and substantial. This paper presents an approximate method
for the assessment of the expected maximum forces generated at the pile head
during driving and shows influence of pile mass, mass moment of inertia and
cross-sectional dimension on pile driveability in cases of hammer-pile
misalignment.

Relationship between Impact Force and Pile Dimensions

The force generated at the pile head affects significantly on pile


penetration resistance during blow measured usually as blow count per 0.3 m of
pile penetration. In order to overcome soil resistance, this force imposes stresses
in a pile and is transmitted to the surrounding soil. Evaluation of the effect of
pile parameters on a maximum value of the impact force is performed below.
Axial impact on a rod (or a pile) creates a stress wave which travels away from
the point of impact. As long as there are no reflection arriving at the pile head,
where measurements are being taken, the compressive stress, a, is proportional
to the velocity, v, of particle motion as follows, Timoshenko and Goodier, (1951):

a = E~ (1)
c

where E is the modulus of elasticity of pile material, c is the longitudinal stress


wave speed. Measured records demonstrate that equation (1) is fair almost
always for maximum values of force, F, and velocity. Equation (1) can be
rewritten as

F EA
= __ v (2)
c

where A is pile cross-sectional area.

For determination of the maximum velocity at the pile head, it was


assumed that kinetic energy of pile obtained from hammer impact is equal the
maximum energy transferred to the pile, WI' and soil does not affect the
maximum values of the measured pile velocity and transferred energy. So,
Mv 2 (3)
= WI
2

where M is the pile mass. This approach is quite acceptable for finding the
relationship between the maximum measured velocity at the pile head and the

1151
maximum transferred energy. At first, a particle velocity usually reaches its
maximum values during a duration of impact except cases when soil conditions
create a free pile end for the pile toe during easy driving. Secondly, this
approach is similar in certain degree to the elementary way, Timoshenko and
Goodier (1951), for determination of strain energy of a rod struck by moving
body when strain energy of the rod is equated to the kinetic energy of a striking
body. Thking into account that M= pAL (p is mass density of a pile, L is a pile
length) and c2 =E/p we derive the expression for measured pile velocity at the
pile head as a function of pile parameters and transferred energy

(4)

Substitute equation (4) in equation (2) and write down the expression for force
measured at the pile head

EA (5)
F = J2 L WI

This expression revealed that maximum force at the pile head is a function of
pile cross-section area and length, pile modulus of material, and transferred
energy. An increase of pile stiffness results an increase of force. Measured
energy transferred to the pile is typically only 20-60 % of rated hammer energy
and mostly in the range of 30-50 %. In equation (5), transferred energy can be
replaced with rated energy, Wo multiplied by the efficiency, ~, of the entire
driving system.

Computed values of the force according to equation (5) were verified with
measured force at the pile head obtained on fifteen piles tested at three sites.
Actual values of driving system efficiency were used for calculations. The
description of sites is the following.

Site 1. Five prestressed concrete piles, after DiMaggio (1991), were driven on
site with predominate silty sands. The water table was at a depth of 0.6 m from
ground surface. Pile length, embedment, cross-section area, wave propagation
velocity, elastic modulus, impedance, mass and mass moment of inertia are
presented in Thble 1. Driving data for end of initial driving (EOID) including
blow count per 0.3 m, hammer model and type, rated and transferred energy,
efficiency of the entire hammer assembly, measured and computed maximum
force are shown in Thble 2.

1152
Site 2. Two prestressed concrete piles, one H-pile, and one closed end steel pipe
pile, after DiMaggio (1989), were installed on site with soil conditions consisted
of 16 to 19 m of silty clay overlying a clayey glacial till deposit. A groundwater
was found at depth of 11 m. Pile and driving data for EOID are presented in
Table 1 and 2 respectively.

Site 3. Seven prestressed concrete piles, after Svinkin et al. (1994), were driven
in soil consisted of about 25.6 m of mainly gray clays followed by a bearing layer
of silty sand. Water table was at the ground surface. Pile and driving
information for EOID are shown in Thble 1 and 2 respectively.

6000

5000 x
x
1/
Z
~ 4000
;/
CD
...0
(J

u.
3000
;7
- I~
"C
CD
::J
a.
E 2000

1/
0
()

1000

o 1/
o 1000 2000 3000 4000 5000 6000
Measured Force (kN)

FIG. 1. Comparison Computed and Measured Force

From comparison of computed and measured forces for each pile, the
average quantity of adjustment factor was found to be 0.85. Equation (5) can
be rewritten with the adjustment factor and transferred energy replaced by rated
one as

~
F = 0.85 2-W ~
A (6)
L r

Maximum measured forces at the pile head and computed ones according to
equation (6) are shown in Thble 2 and Figure 1. It can be seen that accuracy of
computed forces is pretty fair for piles tested at all three sites and mostly within

1153
TABLE 1. Pile Data
Mass
Moment
No. Description Length Embdt Area Wave Elastic Impedance Mass of
Velocity Modulus Inertia
L 1;, A
(cm 2)
c E Z M r..
(m) (m) (m/s) (MPa) (kN/m/s) (t) (t-m 2)

1.1 Prestressed Concrete 20.4 19.7 2090 4084 43063 2126 10.26 357
457 x457 mm

1.2 Prestressed Concrete 23.5 22.9 2090 4083 43063 2126 11.79 541
457 x457 mm

1.3 Prestressed Concrete 20.4 19.5 3155 4191 43689 3233 15.73 547
610 x610 mm
(267 mm D. void,
solid ends)
-..
CJ"I 1.4 Prestressed Concrete 23.5 22.9 3155 4266 43689 3233 18.04 828
~ 610 x 610 mm
(267 mm D. void,
solid ends)

1.5 Prestressed Concrete 22.6 22.3 5794 4236 43063 5894 33.71 1431
915 x 915 mm
(572 mmD. void,
solid ends)

2.1 HP 254x63 22.9 22.3 80 5124 210000 323 1.43 62


depth = 246 mm
ftg. width = 254 mm

2.2 Prestressed Concrete 19.8 19.8 929 3956 37564 883 4.42 145
305 x 305 mm

2.3 Prestressed Concrete 19.8 17.1 1264 3780 40885 1254 6.02 196
356 x 356 mm
TABLE 1 continued. Pile Data
Mass
Moment
No. Description Length Embdt Area Wave Elastic Impedance Mass of
Velocity Modulus Inertia
L ~ A c E Z M 1m
(m) (m) (cm 2) (m/s) (MPa) (kN/m/s) (t) (t-m 2)

2.4 324 mm O.D. by 13 mm 21.3 20.1 124 5124 210000 500 2.14 81
thick CEP

3.1 1372 X 127 mm 25.6 24.1 4966 4219 42718 5034 30.55 1670
Cylinder

3.2 1372 x 127 mm 25.6 24.7 4966 4462 47796 5325 30.55 1670
Cylinder

3.3 610 x 610 mm 25.6 24.8 2986 4080 42801 3031 18.37 1004
........
01
(305 mm D. void)

01 3.4 762 x 762 mm 25.6 24.8 4035 4080 39962 3957 24.82 1357
(475 mm D. void)

3.5 762 x 762 mm 25.6 25.0 4035 4080 39962 3957 24.82 1357
(475 mm D. void,
spliced)

3.6 914 x 127 mm 25.6 24.7 3142 4267 41340 3133 19.33 1056
Cylinder

3.7 914 x 127 mm 25.6 24.7 3142 3962 39962 3081 19.33 1056
Cylinder
(spliced)
TABLE 2. Driving Data for EOID
Hammer
Measured Computed Coefficient
Blow Ratio Ratio
Pile Model Rated 'llansferred Efficiency Force Force
Count F/M F/kI~
No, & Energy Energy F F k
(blows/O.3 m) (kNIt) (kN/t-m')
lYpe E, E, (kN) (kN)
(kJ) (kJ)

1.1 18 KOBE 76.67 23.77 0.310 3369 3823 1.50 328 6.29
K-45

1.2 42 KOBE 84.74 28.77 0.339 3884 3918 1.50 329 4.79
K-45

1.3 34 KOBE 96.85 30.90 0.319 4585 5446 2.00 291 4.19
K-45

.......
....... 1.4 77 KOBE 95.50 25.84 0.271 4605 4681 2.00 255 2.78
(J"I
(J) K-45

1.5 92 DELMAG 201.90 50.25 0.249 9370 8956 3.00 278 2.18
062-22

2.1 34 DELMAG 51.91 24.39 0.470 1495 1595 0.83 1045 29.05
030

2.2 60 DELMAG 46.54 17.27 0.371 2119 2096 1.00 479 14.61
030

2.3 110 DELMAG 46.54 13.37 0.287 2482 2151 1.17 412 10.82
030
TABLE 2 continued. Driving Data for EOID
Hammer
Measured Computed Coefficient
Blow Ratio Ratio
Pile Model Rated Transferred Efficiency Force Force
Count F/M Flkl m
No, & Energy Energy F F k
(blows/O.3 m) (kN/t) (kN/t-m 2)
1)'pe E, E, (kN) (kN)
(kJ) (kJ)

2.4 30 DELMAG 51.01 22,67 0.444 2285 1985 1.06 1066 26,61
D 30

3.1 38 DELMAG 102,97 17.38 0.169 4256 4564 4.50 139 0.566
D 46-13

3.2 48 DELMAG 102.97 14.56 0,141 4011 4419 4.50 131 0.534
D 46-13

--"

U1
3.3 10 DELMAG 89.24 34.13 0.382 3626 4881 2.00 197 1.805
-.J D 46-13

3.4 14 DELMAG 89.24 31.39 0.351 6154 5348 2.50 248 1.816
D 46-13

3.5 23 DELMAG 82.38 15.24 0,185 3517 3726 2.50 142 1.036
D 46-13

3,6 15 DELMAG 89.24 20.69 0,232 4002 3951 3.00 207 1.263
D 46-13

3.7 32 DELMAG - 9.17 - 2894 2513 3.00 150 0,913


D 46-13
15 %. This is acceptable for practical applications. In the office and field,
equation (6) can be used for quick assessment of the maximum force
corresponding to rated energy and efficiency of the driving system in time of
dynamic testing for specified piles.

In equation (6), the force is proportional to square root of pile cross-


section area and inversely proportional to square root of pile length. Actually,
the force equals the square root of pile stiffness multiplied by value of
transferred energy and factor of two. Equation (6) confirmed the known fact
that an increase of pile stiffness increases the pile force which is transmitted to
the surrounding soil to overcome soil resistance. As a consequence, the blow
count decreases. What stiffness changes do affect pile penetration? To
evaluate this question, it was reasonable to consider a pipe pile with variable wall
thickness. A 1220 mm outer diameter steel pipe pile with length of 43 m was
selected for study. A wave equation analysis was performed for the piles with
real and imaginary wall thickness of 6.5, 13, 19, 25, 32, 38, 51, 57, and 63 mm.
All analysis were computed for pile capacity of 4448 kN with energy adjustment
to value of 77.6 kJ. Percentage of skin friction was taken as 20 % with triangle
distribution over the lowest 20 percent of each pile.

1220 mm 0.0. Steel CEP Pile


1200

-
E
C')
1000

\
a-... 800
CII
~
0
e- \
-
0
c:
:::::l
600

() 400
~
0
ffi
1\
200
L I I
o
o 10 20 30 40 50 60 70 80
Pipe wall thickness (mm)

FIG. 2. The Effect of Pile Stiffness on Pile Penetration Resistance

Result of wave equation analysis revealed that for all piles maximum
stress was in allowable limits, decreasing with an increase of the wall thickness.

1158
For pile with the smallest wall thickness of 6.5 mm and termination criterion of
driving, no actual pile penetration, pile capacity reached 4448 kN and pile
penetration resistance was conditionally taken as 1000 blows per 0.3 m. For pile
capacity 4448 kN, the blow count per 0.3 m as a function of the pipe wall
thickness is shown in Figure 2.

It can be seen that an increase of wall thickness from 6.5 to 19 mm


strongly decreased blow count down to 78 blows per 0.3 m. However, further
wall thickness increase did not change amount of blows per 0.3 m. Thus, in this
example reasonable stiffness increase should not be greater than three. The
limited effect of the increased force on the pile penetration resistance probably
can be explained by augmentation of pile mass simultaneously with augmentation
of pile stiffness. Consequently, the pile mass inertia force increment becomes
in balance with the increment of acting force.

The Effect of Horizontal and Rocking Pile Motions on Pile Driveability

Proper hammer-pile alignment should be maintained for driven piles.


However, sometimes deviations from the required axial pile alignment,
Committee of Deep Foundations of ASCE (1984), could be caused by
misalignment of equipment or pile, the use of inadequate equipment or
extremely flexible piles, subsurface obstructions or ground compaction. As a
result, the pile could either be driven off the required axial alignment but with
its axis battered straight or driven with some portion of its axis lying off the
straight line from butt to tip. Actually, vertically driven piles are often in extent
degree out-of-plumb during driving and dynamic forces act with certain small
eccentricities.

The primary consequence of the hammer-pile misalignment is the


development of some lateral component of the total dynamic force perpendicular
to the pile shaft. Always acting in combination with axial component of the
dynamic force it exerts bending stress in a pile and imposes coupling of
horizontal and rocking pile motions as a solid body. These pile motions involve
pile mass moment of inertia and also horizontal and rocking soil stiffness in the
total resistance to pile driving.

The pile penetration resistance, or blow count per 0.3 m (BC), is the
important characteristic of pile driveability. BC flexibly reflects the changes of
the pile and soil resistance to the acting force for certain energy level.
Apparently, BC depends on the force applied to a pile. Lateral component of
the force generated in a pile during blow decreases the share of axial force
component which causes pile penetration and also induces the additional
resistance to pile driveability. The effect of horizontal and rocking pile motions
on pile driveability was evaluated using pile mass moment of inertia as the
measure of an increase of the pile penetration resistance. Assessment was

1159
Site 1 Site 2 Site 3
'OIl r
. 120

50

E
~
Q.
80
. E
C')

~
'OIl E
~
40

~ ~
~ 80
80
~
e . e 1IO
e
30

40'r- 20
o
o 8 40
<3

~
iii
20I
i 20 ~
iii
1

o
o 1234587
- 8
o
o 4 8 12 18 20 24 28 32 Q 0.3 0.8 0.9 1.2 1.5 1.8 2.1 2.4
Normalized Force = F(k 1m) (kN(1m2) Normalized Force ... F(1< 1m) (1<N(t-m2) Normalized Force = F(k 1m) (1<N(t-m2)

Site 1 Site 2 Site 3


..... ,
..... lOCh I I i i I I I 120 50
I
en I
0 E
C"l
80 . E 100 E 40I

.
C') C"l
s2III ci ~
~ 80 UI
~ 60 ~ ~
0 30I
e .
0
a 80
0
:0
~

c:
::J 40 c:::J C 20I
0 8
.
0 40
0 0
.
0
~ ~ ~
0 20 0 0 10
iii iii 20 iii
0
a sao 0 o
100 200 300 400
600 700 800
0 150 300 450 800 750 900 1050 '200
o 35 70 lOS 140 175 210 245 280
Normalized Force = F/M (kNIt) Normalized Force = F/M (kNIt)
Normalized Force = F/M (kNIt)

FIG. 3 Pile Penetration Resistance at EOID as a Function of Measured Normalized Force


Site 1
100

'E eo \
M
I BC .. (150 k !mlfF l
ci
iii
~
60
\
e \
c ~
<3
~
40

~ "--.... '---.
iii 20

o
o 2 3 456 7 8
Normalized Force". F/(k 1m) (kN/I-m2)

Site 2
120

100
1\
M
Q. 60
\ I BC .. (900 k ImlfF 1
'~"
e
c:l
o
()
60

40
"'" ~ ........
r----, ~
~
1IJ
20

o
o 4 8 12 18 20 24 28 32
Normalized Force ". F/(k 1m) (kN/I-m2)

Site 3
so
\
4O
'E
M
g
en
~ 30
\ I BC .. (25 k ImlfF I

e
c:l 20
~
8 ~
"----.,
~
III 10
r--- t-

o
o 0.3 0.8 0.9 1.2 1.5 1.8 2.1 2.4
Normalized Force". F/(k 1m) (kNII-m2)

FIG. 4 Computed Blow Count per 0.3 m as a Function of Normalized Force


11 G1
performed on the basis of dynamic data at EOID obtained for described above
piles on three sites. Every site was studied separately. For each pile, BC was
considered as a function of the measured force normalized with mass moment
of inertia multiplied by the coefficient, k, taking into account the cross-sectional
pile dimension. The normalized force determines how much strength is
necessary for penetration a unit of pile mass moment of inertia for the certain
pile penetration resistance. Also for the purpose of comparison, the measured
force was normalized with pile mass. Values of the normalized force are shown
in Thble 2.

For determination of pile mass moment of inertia, the pile can be


considered as a slender rod. Pile cross-section dimensions are on 1-2 orders
smaller than pile length. Therefore, only one pile dimension, namely pile length,
L, affects pile moment of inertia, 1m , taken relatively of the center of gravity of
a pile

(7)

where M is pile mass. Pile and soil resistance to horizontal and rocking pile
motions depends on the pile cross-sectional dimension. The coefficient, k, was
assumed equal to one for pile with cross-sectional dimension of 0.3 m. For the
rest of the piles on the basis of existing experience, Tsytovich et al. (1970), the
coefficient, k, was determined as the ratio of the pile cross-sectional dimension
to 0.3 m.

Derived graphs of BC as a function of the normalized measured force are


depicted in relatively similar scale in Figure 3. These graphs reveal the soil
penetration resistance for each pile for the same, rather very close, soil
conditions on every site. It can be seen that the force normalized with pile mass
moment of inertia and the coefficient, k, reflects better the changes of the pile
penetration resistance than the force normalized with pile mass. This difference
is especially pronounced for site 1. Acquired experimental data shows the
tendency to decrease blow count per 0.3 m with an increase of normalized force
outlay. The blow count per 0.3 m and the normalized force are inversely
proportional quantities and their relationship can be expressed as a hyperbolic
function
a k I
BC = m_ (8)
F

where a is coefficient distinctive for each site. Coefficient, a, is equal 150, 900,
and 25 for site 1, 2, and 3 respectively. These functions presented in Figure 4.
Described procedure is quite fair for short piles. However, obtained results have

1162
demonstrated the acceptance of suggested procedure for long piles in order to
assess expecting penetration resistance of different piles at close soil conditions.

Conclusions

The effect of the pile role on driveability process is more complicated and
substantial than just to withstand driving stresses and transfer force to
surrounding soil.

The quantity of force generated at the pile head during impact is essential
for pile driveability. The relationship between the dynamic force and pile
stiffness has been derived. Maximum value of the dynamic force is equal square
root of pile stiffness multiplied by value of transferred energy and factor of two.
Wave equation analysis of the pipe pile with different wall thickness has showed
that a force increase at the pile head, stipulated by a stiffness increase, strongly
decreases the pile penetration resistance only in some range and further
augmentation of pile stiffness and force does not change the pile penetration
resistance. This fact can probably be explained by augmentation of pile mass
simultaneously with an augmentation of pile stiffness. In consequence of that
the pile inertia force increment become in balance with the increment of the
acting force.

An attempt is made to assess the effect of hammer-pile misalignment on


the pile penetration resistance. Hammer-pile misalignment develops some lateral
component of the total dynamic force. Besides bending stresses in a pile, the
lateral force imposes coupling of horizontal and rocking pile motions in a vertical
plane. These motions involve pile mass moment of inertia and shear resistance
mobilized in soil surrounding the pile in the total resistance to pile driving and,
as the result, increase in the pile penetration resistance. Blow count per 0.3 m
at EOID is expressed as hyperbolic function of the force normalized with pile
mass moment of inertia and coefficient characterized the pile cross-sectional
dimension.

Presented results can be used in addition to wave equation analysis which


is the major method for determining of pile driveability.

Appendix 1. References

Agrawal, G. and Chameau, J.-L.A. (1992). "Adjustment factors for field


control of pile driving." Proc. Forth Int. Con! on the Application of SWT to Piles,
The Haque, The Netherlands, 389-394.
Brucy, F., Meunier, J., and Nauroy, J.-F., (1988). "Analysis of pile driving tests
in calcareous sand sites." Proc. Forth Int. Con! on the Application of SWT to
Piles, Ottawa, Canada, 706-716.

1163
Committee of Deep Foundations of ASCE (1984). Practical guidelines for the
selection, design and installation of piles. ASCE, New York, USA.
DiMaggio, I. (1989). Dynamic pile monitoring and pile load test report.
Demonstration project No. 66, 1-80/480 Interchange-Omaha, Nebraska. FHWA,
Washington.
DiMaggio, J. (1991). Dynamic pile monitoring and pile load test report.
Demonstration project No. 66, 1-165 1(2) Mobile County, Alabama. FHWA,
Washington.
Fellenius, B.H. and Samson, L. (1976). "Testing of driveability of concrete piles
and disturbance to sensitive clay." Canadian Geotech. Journal, 13, No.2, 139-
160.
Hannigan, PJ. and Webster, S.D. (1987). "Comparison of static load test and
dynamic pile testing results." Second International Symposium, DFI,
Luxembourg, 85-108.
Heerema, E.P (1978). "Predicting pile driveability: Heather as an illustration of
the "friction fatigue" theory." European Offshore Petroleum Conference and
Exhibition, London, UK, 413-422.
Holeyman, A.E. (1992). "Keynote lecture: Technology of pile driving testing."
Proc. Forth Int. Con! on the Application of STtT to Piles, The Hague, The
Netherlands, 195-215.
Holloway, D.M., Audibert, I.M.E., and Dover, A.R. (1978). "Recent advances in
predicting pile driveability." The 10th Annual Offshore Technology Conference,
Houston, 1915-1924.
Goble, G.G. and Rausche, F. (1980). "Pile drivability predictions by CAPWAP"
Numerical methods in offshore piling, ICE, London, UK,29-36.
Goble Rausche Likins and Associates, Inc. (1993). GRLWEAP - Wave Equation
Analysis of Pile Driving. Manual, Cleveland, Ohio, USA.
Li, J.e., Yao H.-L.,and Ong B. (1988). "Hammer selection and stress-wave
equation behavior of piles." Proc. Third Int. Con! on the Application of STtT to
Piles, Ottawa, Canada, 601-612.
Svinkin, M.R., Morgana, e.M., and Morvant, M. (1994). "Pile capacity as a
function of time in clayey and sandy soils." Proc. Fifth Int. Con! and Exhib. on
Piling and Deep Foundations, Bruges, Belgium, 1.11.1-1.11.8.
Thng, N.-e., Yuan, J.-H., Wang, Y, and Lu, T-S. (1988). "Driveability analysis
of long steel pipe piles - Case history studies." Proc. Third Int. Con! on the
Application of STtT to Piles, Ottawa, Canada, 513-524.
Timoshenko, S. and Goodier, J.N. (1951). Theory of elasticity. McGraw-Hill Book
Co., 2nd edition, New York, USA.
Tsytovich, N.A., Berezantsev, VG., Dalmatov, B.I., and Abelev, M.Y. (1970).
Bases and Foundations, Concise Course (in Russian). Higher Schools Publishing
House, Moscow, USSR.
Vanikar, S.N. (1985). Manual on Design and Construction of Driven Pile
Foundations. U.S Department of Transportation, Federal Highway
Administration.

1164
Soil Modeling for Pile Vibratory Driving

Alain E. Holeyman 1 and Christian Legrand2

Abstract

A rational procedure to model the dynamic nonlinear behavior of the skin friction of
piles and sheet piles during vibratory driving is presented. The model is based on the
fundamental analysis of the dynamic behavior of a cylinder embedded in a semi-
infinite medium. Cylindrical shear waves propagating away from the vibrated pile are
evaluated using a one-dimensional radial discretization of the soil surrounding the
pile. Elements of earthquake engineering normally used to assess liquefaction
potential are applied to evaluate skin friction degradation upon cyclic shear stress.
Degradation and excess pore-pressure buildup charts are presented based on
correlations derived from the friction ratio as measured in a CPT test.

Introduction

This paper presents the results of the preliminary development of a detailed model
aimed at representing the dynamic response of soils under cyclic loading induced by
vibratory driving. The source of the cyclic loading acting upon the soil is a pile or
sheet pile being activated by a vibrator. The vibrations are essentially vertical and as
a first approximation the vibration pattern of the surrounding soil can be considered
to possess cylindrical symmetry.

The sheet pile will be represented in a first step by a rigid mass acted upon by the
inertial effects of the vibrator eccentric masses, and deriving restrain from the
dynamic reactions of the surrounding soil. The model used to represent the soil
reactions is described below.

1Principal, EarthSpectives, 6262 Sierra Palos Rd., Irvine, CA 92715-3945


2Head of Geotechnical Division, BBRI, Brussels, Belgium

1165 Holeyman
Loading Conditions and Model Geometry

Although stresses are often used as the primary boundary condition in laboratory
experiments, it is our opinion that in the case of a vibrating sheet pile, the governing
boundary condition should be kinematic rather than dynamic: calculations performed
with a single degree of freedom (SDOF) program (Holeyman, 1993a) show that the
vibratory behavior (i.e. the amplitude of the movement) of the sheet pile itself is not
strongly influenced by the soil resistance. In soft soils the shear stress resisting the
sheet pile movement is small, while it is higher in stiffer soils. It is therefore of
interest to base the present soil model on strain-controlled cyclic shear tests rather
than stress-controlled shear tests.

For the purpose of the present analysis, the soil reactions will be separated into skin
friction and toe resistance. Toe resistance will be represented by a SDOF, commonly
utilized in wave equation calculations (Holeyman, 1988). Because of its
preponderance in the study of vibration and penetration of sheet piles, the skin
friction will be addressed by a more complex model that aims at encompassing the
fundamental aspects of the vibratory behavior of the soil around the sheet pile.

The geometric shape of the proposed soil model surrounding the sheet pile or pile has
cylindrical symmetry, as shown in Fig. 1. It is a disk with a thickness that slightly
increases linearly with the radius. Normalized to the penetration depth of the sheet
pile, it has a thickness provided by the following equation:

(1)

The increase of the disk thickness with radial distance r tends to simulate the
geometrical damping provided by the half space of soil located below the toe of the
sheet pile. The equivalent radius ro of the sheet pile is obtained from perimeter
considerations. The outer boundary of the model is set at a radial distance Rm based

ltm - . --;

~ PILE: SI-IM"]'

Fig. I - Model Geometry

1166 Holeyman
on a trade-off between calculation time and zone where the evaluation of the
vibrations is of interest. An energy absorbing boundary condition in accordance with
plane-strain elasticity theory (Novak et aI, 1978) limits the lateral extent of the model
at a distance large enough to ensure that deformations stay within the elastic range
and to avoid artificial energy reflections.

The system of cylindrical waves propagating within the geometric model will be
calculated by discretizing the medium into concentric rings that possess individual
masses and that transmit forces to their neighbors. The shear force-displacement
relationship between successive rings will be established based on the stress-strain
relationship. Inter-ring reactions Ti are obtained based on ring displacements ui using
the following relationships:

(2)

with G' representing the generalized secant shear modulus as discussed below.
Movement of the rings is evaluated from the time integration of the laws of motion,
and in particular from the acceleration resulting from the net unbalanced loads acting
on each ring.

Constitutive Relationship

The constitutive relationship proposed for the representation of the large-strain,


dynamic and cyclic shear stress-strain behavior of the medium surrounding the
vibrating sheet pile will be described by several laws addressing the following
elements:

Static stress-strain law expressing nonlinear behavior under monotonic loading


and hysteresis upon strain reversal,
Shear modulus at small strains and ultimate shear strength based on soil
characterization: nature, void ratio, overconsolidation ratio or based on
correlations with qc and FR obtained from CPT tests,
Softening and increase of hysteretic damping with increasing strain based on soil
characterization,
Effect of strain rate on initial shear modulus and ultimate strength,
Degradation of properties resulting from the application of numerous cycles,
Generation of excess pore pressure leading to liquefaction and substantial loss of
resistance, and
Accommodation of variable strain amplitude history.

The following paragraphs address these components of the constitutive relationship.

1167 Holeyman
~-
""Ls- L- -z:-s
~--:...-........~
- --= _JIME
-- - - - <;;.;? - --

-'{

1.- -J T1ME
---..

-u

' \ -L.--==-----
"\ DECRADED BACKBDNE
"-- CURVE AT CYCLE N

INITIAL LOADING
BACKBONE CURVE AT N.I

Fig. 2 - Soil Behavior under Constant Cyclic Shear Strain Amplitude Loading
(Sketch from Vucetic, 1993; 1994)

1168 Holeyman
Static Stress-strain behavior

A typical soil response to uniform cyclic strains is represented in Fig. 2, which


highlights or allows one to derive the following fundamental parameters:

Gmax: initial (or tangent) shear modulus


Smax: ultimate shear strength, revealed at large strains
Gs: secant (or equivalent) shear modulus
A.: hysteretic (or intrinsic) damping ratio = fj, W /21tY c't c,,
with fj, W = Energy lost in a given cycle

Both Gs and A. are strain-dependent parameters that need to be described by specific


laws within a given cycle. Gmax and Smax are shown to decrease with the number
of cycles (cyclic degradation).

Initial Shear modulus and ultimate shear strength

Numerous studies have dealt with the initial shear modulus to be used in earthquake
engineering. However, because most of them are supported by parameters determined
in the laboratory, we recommend an empirical approach based on correlations with
CPT data (cone resistance qc, local skin friction fs, and friction ratio FR), as follows:

Gmax=K. qc with K=15 (3)

Smax = Beta. fs with Beta = 0.65 + 0.35 . Tanh 1.5 (FR-2 %) (4)

Secant Shear Modulus and Hysteretic Damping

As can be observed in Fig. 2, Gs decreases with the shear strain during the initial
monotonic loading. The curve that represents the initial monotonic loading is referred
to as the initial "backbone" curve, because it also serves as the basis to generate the
family of curves corresponding to unloading and reloading. Kondner's mathematical
formulation (1963) is frequently employed to describe the initial backbone curve in
earthquake engineering:

11 = 't / 't max = 0/ (0+ 1) with 0= Y/ Yr = y.Gmax / 't max (5)

It is of interest to show the hyperbolic law using reduced variables 11, the
mobilization ratio and 0, the relative shear, as shown in Fig. 3. Yr is called the
reference strain. Two of the parameters Gmax, Yr, and 't max are generally adjusted
from laboratory experiments. In the case of CPT data, we propose to use 't max =
Smax per equation (4) and

Yr = Smax/Gmax = Beta. FR / K (6)

1169 Holeyman
~III 0.9
a:
>. 0.8
E'
w
! 0.7

~ 0.6 - - - - Mobilization Ratio


III

g 0.5 --0-- Rei. En. loss

~ 0.4 ------.- GsiGmax


:c
~ 0.3
gj' 0.2
"5
-g 0.1

:liE 0 6==t'l:;;:;:;:;~i:::::::+=--------+-----+-~-~===~~
0.01 0.1 1 10 100
Strain Ratio (Delta = gam/gamr) [oJ

Fig. 3 - Hyperbolic law to evaluate G/Gmax and Relative Energy Loss

From the point of maximum straining, the unloading curve is described by the
following equation, in accordance with Masing's rules 1 and 2 (Masing, 1926):

't - 'to = (y - Yo) I (lIGmax + (y - Yo)/2't max ) (7)

The energy contained in a loop depends for a given soil on the amplitude of the
cyclic strain. Empirical data collected in laboratory tests indicates that the damping
ratio increases with Yc as the soil undergoes higher plastic deformations.

We propose to utilize the unifying approach recently developed by Dobry and


Vucetic (see Dobry and Vucetic (1987), Vucetic and Dobry (1991), and Vucetic
(1993 and 1994) to accommodate the influence of the nature of the material
characterized by the plasticity index, as indicated in Fig. 4. The relative energy loss
(or 1t x damping ratio) as obtained by directly integrating the area defined by a loop
in the stress-strain diagram has also been represented in Fig. 3 to demonstrate the
ability of the hyperbolic law to reproduce experimental observations. The PI
influencing the value of the reference strain Yr available from other studies was
correlated to the friction ratio using:

PI = 50 .(1 + Tanh (FR - 3.5%)) (8)

Strain Rate Effects

Although it is well known that undrained modulus and shear strength increase with
increasing strain rate (y), experimental data generated under different apparatuses and
loading conditions lead to different conclusions. Based on our review of the
literature, we believe that a viscosity mechanism would provide a satisfactory

1170 Holeyman
framework for understanding the strain rate effect observed when comparing fast and
slow undrained monotonic stress-strain curves, as well as for explaining the
roundness of the loop tips during a sinusoidal strain-controlled cyclic test. Evidence
would point to the fact that sands and non plastic silts have very small viscosity in
that their stress-strain loops exhibit sharp rather than rounded tips (Dobry and
Vucetic, 1987).

1.0 r---.....:;:::::::l"'-=:::::---=:::::::".,..----~---__:_---~

0.8
SAND and SILT
....
i 0.6
~
~
~ 0.4

IOCR:11
0.2

O.O':":':':---:"':":-:-----.J..--_..l....-.....;".;..~....;E:... _ ____J
0.0001 0.001 0.01 0.1 1 10
CYCLIC SHEAR STRAIN AMPLITUDE, )',(%}

0.10------..,..-........- - - - - - -
ci
W
I-
W
::::E
c(
a:
c(
a..
Z 0.05
o
~
Q
c(
a:
C)
w
o OCR=1
o.o.u:.....-----="=------~
0.5 1
CYCLIC SHEAR STRAIN AMPLITUDE, "Yc(%)

Fig. 4 - Effect of the Plasticity Index on (a) Relative Secant Modulus


(b) soil degradation (Vucetic, 1993)

1171 Holeyman
The mathematical functions proposed in the literature to represent the nonlinear
viscosity also depend on the type of experimental observations. We propose to adopt
a power law:

'tdyn = 'tstat . (1 + J . yn ) (9)

The advantage of that mathematical form is that resistance does not become zero at
zero strain rate. The power law also requires the strain rate to vary by orders of
magnitude to provide tangible increases in both the modulus and the ultimate
strength. The J coefficient and n exponent depend on the nature of the soil. Based on
pile driving data, we propose to use n=0.2 and J=0.3 s-0.2 for plastic soils. J should
therefore essentially depend on the plasticity of the soil, and thus on the FR obtained
from CPT tests, as proposed in first approximation below:

J=0.10FR (10)

Degradation Law

When subjected to undrained cyclic loading involving a number N of large strain


cycles, the soil structure continuously deteriorates, the pore pressure increases, and
the secant shear modulus decreases with N. This process known as cyclic stiffness
degradation can be best characterized on the basis of strain controlled tests for the
type of loading involved with the vibratory penetration of sheet piles. Typical results
of strain-controlled tests are sketched in Fig. 2, where the degradation is clearly
expressed by the decrease of the amplitude of the peak stress mobilized at successive
cycles.

The quantification of the degradation process calls for the introduction of the
degradation index~, defined by:

'tn = ~ . 1'1 (11)

Laboratory results conducted at constant cyclic strain show that in many soils, the
degradation index can be approximated by the following relationship as suggested by
Idriss et al (1978):

(12)

The exponent t, called degradation parameter, depends mainly on the amplitude of


the cyclic strain and the nature of the material (PI), as suggested by Dobry and
Vucetic (1988) and as indicated in Fig. 4 (Vucetic, 1993). It is noteworthy that the
degradation parameter assumes a zero value at strains smaller than a cyclic
"threshold" shear strain, Y1'u' The threshold strain increases with the plasticity of the
soil, as suggested in Fig. 4.

1172 Holeyman
Laws to represent the degradation based on CPT results are proposed as follows:

Yru = Beta. FR 130 (13)

t= (yIYTu - 1)Y2 1 (PII2 + 25) (14)

Soil liquefaction

Vibration induced compaction of saturated sands has received attention not only from
the earthquake engineering community, but also from vibro-compaction specialists.
Recent advances tend to indicate that build up of pore pressures (eventually leading
to liquefaction) and volume reduction of cyclically loaded materials are the
expression of the same phenomenon, i.e. the irreversible tendency for a particulate
arrangement to achieve a denser packing when sheared back and forth.

Under drained conditions, the volume reduction is immediate. Under undrained


conditions, the tendency for volume reduction is expressed by an increase in the pore
water pressure (see Fig. 2), such that the effective stress is reduced to a value that
may be close to zero. It is then necessary to wait for the sample to consolidate in
order to see the volume reduction take place.

The strain driven evaluation of the build up of pore pressure as suggested by Dobry
et al. (1979) has been adopted as it allows a direct transposition to the problem of the
vibrations induced by a vertically vibrating sheet pile. It also allows one to evaluate
potential changes of the void ratio based on a cyclic strain rather than stress history,
as evidenced in the early tests conducted on drained sands by Youd (1972). Finally,
this framework of analysis enables the threshold cyclic strain to encompass in a
single concept the intrinsic relationship between degradation and pore pressure build-
up, with the advantage that it can be applied to general categories of soils (sands to
clays).

The excess pore pressure generated during cyclic loading has been shown (see Fig. 5)
to increase with the shear strain and the number of cycles for a given soil type. We
have adopted the damage parameter K approach (Finn, 1981) to evaluate the excess
pore pressure OU resulting from a particular strain history, as characterized by the
following equations:

ouJcr' = 14 (Rei. En. Loss) . In (1 + Y2 K ) (14)

with Relative Energy Loss given by Fig. 3, and K = damage parameter given by:

with ~ = length of strain path (15)


= 4 Nyc, for constant amplitude cycles
A=5

1173 Holeyman
o
I~ 1.0 Dr =60(%)
U<J
::J II -
U* 0'0 =96 kN/m 2
>- :: 0.8
u w
-leI: N SYMBOL
:5O(j)55 0.6 1 t:J.

-w 5 0
f:3 eI: 0.4
cr:e. 10 0
ocr: 30
WW 100
N ~ 0.2
;i~
~ w 0
cr:a:
00
ze.
10- 3 3 5 10- 2 3 5 10- 1 3 5 1
CYCLIC SHEAR STRAIN AMPLITUDE, Yc(%)

Fig. 5 - Build up of residual pore pressure in different sands in undrained cyclic


triaxial strain-controlled tests (Dobry et aI, 1982)

Generalized strain history

Because the constitutive relationship parameters are generally established on the


basis of constant strain, laboratory controlled tests, it is necessary to formulate a
means to follow the dynamic behavior of the medium under the non regular types of
loading present during the vibratory penetration of a sheet pile: start-up and tum-off,
or progressive modification of soil properties resulting from degradation. Masing
rules 3 and 4 have thus been applied to accommodate non-repetitive straining paths.

Degradation can also be represented under irregular loading, provided that a


degraded backbone curve is used instead of the initial backbone curve. The degraded
backbone curve is completely defined by the degradation index ~, the equivalent
value of which must be ascertained based on the amplitude and number of previous
straining cycles. This will require the updating of the current degradation index using
the following equations:

N eq = ~(-l/t) (13)

1174 Holeyman
Modeling Results

The CPT has been chosen as the basic sounding upon which the evaluation of the
vibratory penetration of sheet piles is to be conducted. It provides simple, yet ad hoc
parameters that can be related directly or indirectly to the parameters necessary to
model the soil behavior. Because the proposed approach is based on desk generated
correlations, full scale tests will be used to refine the correlations by matching the
penetration speeds and vibration levels calculated by the model with those measured
in the field.

At this stage of reporting of the development of the model, the reasonableness of the
proposed constitutive law and correlation can be ascertained by graphically
representing numerical results derived from our comprehensive set of assumptions.
Samples of those graphical results are presented in Figs 6, 7, and 8. Comparison of
Fig. 6 with Fig. 4c, Fig. 7 with Fig. 4a, and Fig. 8 with Fig. 5, respectively, provides
a measure of the preliminary satisfactory agreement between the general trends
depicted by the constitutive law proposed herein and available laboratory results.

0.1 ~ ~~/~
~QOO, f /
! ~/
::
QJ
0.08 ; I ./
/
~ 0.07, ~ / /
co 0.00 I I /
~0.05:/
00.04 i.
/. ~ ~v

~O.Q3llil / ~
"002'1/. /~v
~ 0:01 i jf I/------ Plasticity Indices Correlated to Friction Ratios of 1, 2, 3, and 4%, respectively
o JJJ, o-?----j I I

o 0,2 0.4 0.6 0.8 1.2 1.4


Shear Strain riD]

Fig. 6 - Degradation Parameter versus Shear Strain for different Friction Ratios

Conclusions

The constitutive relationship governing the nonlinear cyclic behavior of soil under
cylindrical shear as produced by sheet pile vibratory driving has been developed and
presented. The developed relationship highlights degradation of the soil resistance
under cyclic loading as a key parameter in modeling vibratory penetration
Confirmation of the proposed correlations between constitutive parameters and
results of CPT tests requires incorporating the constitutive relationship into a radial
discrete model and the comparison of calculated physical features, such as
penetration speed and ground vibration levels, with those measured during full scale
penetration tests.
1175 Holeyman
09

0'
1ri~~2
it -D--- 5

-0-10

~ 0.7 -----0---- 20

~
"C
0.6 ~50

III
C, 0.5 -----fr-- 1 00
Q>

~ 04 --200
III

inE 0.3 --0-- 500

0.2 -x-1000
Number of Cycles N
0.1 - x - 2000

o --f-- 5000
0.001 0.01 0.1 10
- - 10000
Cyclic Shear Strain Amplitude rio]

Fig. 7 - Degradation versus Strain (Friction Ratio = 2%)

III
E
.~ 09 Number of Cycles N
"C
~ 0,8
lP
~
c..~
0.7

0.6 - D - - - 10
/

/I/}
ell
(5::::::: 0.5 ---100
c..
1!l 0.4 --0-- 1000
~
an 0.3 ~10000

~
iii
o
E
0.2

0.1 /' /
Z O .........--w---=. .=~=~==~~==I:F===
0.01 0.1 1 10 100
Strain Ratio (delta =gam/gamr) [.]

Fig. 8 - Pore Pressure versus Relative Strain for different Numbers of Cycles

Acknowledgments

The study reported herein was completed within the framework of BRITE/EURAM
research contract CT91-0561, "High Performance Vibratory Pile Drivers Based on
Novel Electromagnetic Actuation Systems and Improved Understanding of Soil'
Dynamics", subsidized by the European Union. The joint-venture research team
included the following prime contractors: the University of Sheffield, UK., Jan de
Nul, Belgium, Belgian Building Research Institute (BBRI), Belgium, and Procedes
Techniques de Construction (PTC), France. The soil modeling portion of the research
was subcontracted by BBRI to the first author.
1176 Holeyman
References

Dobry, R., Ladd, R.S., Yokel, F.Y., Chung, R.M., and Powell, D. (1982). "Prediction
of Pore Water Pressure Buildup and Liquefaction of Sands During Earthquakes
by the Cyclic Strain Method." National Bureau of Standards Building Science
Series 138, July 1982, 150 pp.

Dobry, R. and Swiger, W.F. (1979). "Threshold Strain and Cyclic Behavior of
Cohesionless Soils. " Proc. 3rd ASCE/EMDE Specialty Conference. Austin,
Texas, pp. 521-525.

Dobry, R. and Vucetic, M. (1987). "State-of-the-Art Report: Dynamic Properties and


Response of Soft Clay Deposits. " Proceedings ofthe IntI. Symposium on
Geotechnical Eng. ofSoft Soils, Mexico City, Vol. 2, pp. 51-87.

Drnevich, V.P., Hall, lR., Jr., and Richart, F.E., Jr. (1967). "Effects of Amplitude of
Vibration on the Shear Modulus of Sand." Proceedings ofthe International
Symposium on Wave Propagation and Dynamic Properties ofEarth Materials,
Albuquerque, N.M., pp. 189-199.

Finn, W. D. L. (1981) "Liquefaction Potential: Developments Since 1976",


Proceedings, IntI. Con! on Recent Advances in Geotechnical Earthquake
Engineering and soil Dynamics, Sf. Louis, Missouri, Vol. II, pp. 655-681.

Hardin, B.a. and Black, W.L. (1968). "Vibration Modulus ofNonnally Consolidated
Clay." Journal ofthe Soil Mechanics and Foundations Division, ASCE, Vol. 94,
No. SM2, Proc. Paper 5833, pp. 353-369.

Holeyman, A. (1985) "Dynamic non-linear skin friction of piles," Proceedings ofthe


International Symposium on Penetrability and Drivability of Piles, San
Francisco, 10 August 1985, Vol. 1, pp. 173-176.

Holeyman, A. (1988) "Modeling of Pile Dynamic Behavior at the Pile Base during
Driving," Proceedings of the 3rd International Conference on the Application of
Stress-Wave Theory to Piles, Ottawa, May 1988, pp. 174-185.

Holeyman, A. (1993a) "HYPERVIB1, An analytical model-based computer program


to evaluate the penetration speed of vibratory driven sheet Piles", Research report
prepared for BBRI, June, 23p.

Holeyman, A. (1993b) "HYPERVIBIIa, An detailed numerical model proposed for


Future Computer Implementation to evaluate the penetration speed of vibratory
driven sheet Piles", Research report prepared for BBRI, September, 54p.

1177 Holeyman
Idriss, I.M., Dobry, R, and Singh. RD. (1978). "Nonlinear Behavior of Soft Clays
during Cyclic Loading." J Geotechnical Engineering Div., ASCE, 104(GTI2),
pp. 1427- 1447.

Kondner, R. L., (1963). "Hyperbolic Stress-Strain Response: Cohesive Soils."


Journal o/the Soil Mechanics and Foundations Division, ASCE, Vol. 89, No.
SM1, pp. 115-143, Jan.

Masing, G. (1926), "Eigenspannungen und Verfeistigung beim Messing",


Proceedings o/Second International Congress 0/Applied Mechanics, pp. 332-
335.

~ovak:,M., Nogami, T., and Aboul-Ella, F. (1978). "Dynamic Soil Reactions for
Plane Strain Case", J. Engrg. Mech. Div., ASCE, 104(4),953-959.

NRC (1985). "Liquefaction of Soils During Earthquakes." National Research Council


Committee on Earthquake Engineering, Report No. CETS-EE-001, Washington,
D.C.

Seed, H.B. and Idriss, I.M. (1970). "Soil Moduli and Damping Factors for Dynamic
Response Analyses." Earthquake Engineering Research Center, College of
Engineering, University of Califomia, Berkeley, Report No. EERC 70-10.

Seed, H.B. and De Alba, P. (1986) "Use ofSPT and CPT Tests for Evaluating the
Liquefaction Resistance of Sands" Proc. INSITU 86, VA, 22p.

Vucetic, M. and Dobry, R (1988). "Degradation of Marine Clays Under Cyclic


Loading. ASCE Journal o/Geotechnical Engineering,Vol. 114, No.2,pp.133-l49.

Vucetic, M. and Dobry, R (1991). "Effect of Soil Plasticity on Cyclic Response."


ASCE Journal o/Geotechnical Engineering, Vol. 117, No.1, pp. 89- 107.

Youd, L.T. (1972). "Compaction of Sands by Repeated Shear Straining." Journal of


the Soil Mechanics and Foundations Division, Proc. ASCE, Vol. 98, No. SM7,
pp. 709-725.

Vucetic, M. (1993). "Cyclic Threshold Shear Strains of Sands and Clays", Research
Report, UCLA Dept. of Civil Engineering, May 1993.

Vucetic, M. (1994). "Cyclic TIueshold Shear Strains of Sands and Clays", Paper in
print, ASCE Journal ofGeotechnical Engineering.

1178 Holeyman

Vous aimerez peut-être aussi