Vous êtes sur la page 1sur 6

SPT AND CPT TESTING FOR EVALUATING LATERAL LOADING OF

DEEP FOUNDATIONS
By J. Brian Anderson1 and F. C. Townsend2

ABSTRACT: Current design software (FloridaPier, Com624P) requires p-y curves to estimate the foundation
lateral load resistance. Input parameters used to develop these curves can be obtained from in situ [standard
penetration test (SPT) and cone penetration test (CPT)] correlations. This paper presents an evaluation of pre-
dictions using input soil parameters from SPT and CPT correlations versus field measured values. A lateral load
test database consisting of 24 SPT and 6 CPT data sets was developed. The comparisons showed that four
different SPT correlations for ␾ coupled with three different k-values all produce similar R-values. (R-value =
measured/predicted ⫻ 100%). Therefore, little difference exists between the SPT correlation combinations, albeit
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 07/05/14. Copyright ASCE. For personal use only; all rights reserved.

the estimated k value has a greater effect on predicted deformation. Similar combinations of CPT correlations
also show little effect among the commonly used correlations. SPT predictions are quite conservative at low
load levels (R-values ⬇ 53%) and remain conservative (R-values ⬇ 87%) at high load levels. Also, the scatter
(standard deviation) is high (⬇40%). CPT evaluations gave unconservative predictions (R-values ⬇ 105 to
154%). In addition, the scatter (standard deviation) is high (⬇34 to 74%).

INTRODUCTION predictions derived from standard penetration test (SPT) and


cone penetration test (CPT) correlations for the common soil
Piles and drilled shafts are structural members used to trans- input parameters (␾⬘, su , ␥, and ε50) and a database of 28 SPT
fer loads to deep strata through skin friction and end bearing. and 6 CPT data sets.
These deep foundations are necessary when the upper soil lay-
ers are too weak to prevent excessive settlement of the struc- LITERATURE REVIEW
ture, the structure is subject to large lateral forces, or the foun-
dation is subject to scour. Ship impact, wind, and earth and The most common theoretical approach to analyzing piles
water pressure are all sources of lateral load on deep founda- under lateral loading is the p-y method of McClelland and
tions. Because the lateral loading of a single pile is a soil- Focht (1958). This method is merely an extension of Winkler’s
structure interaction problem, the deflection of the pile de- solution, which models the soil as a set of nonlinear springs
pends on the reaction in the soil, and the reaction in the soil defined by a load transfer or p-y curve. The p-y curve repre-
depends on the deflection of the pile (Reese and Wang 1993). sents the soil resistance at a particular depth and is defined in
(‘‘Pile’’ refers throughout to both piles and drilled shafts.) terms of soil resistance per unit length (FL⫺1) versus deflection
In present practice, laterally loaded piles are analyzed using (L). The soil resistance will typically rise quickly under small
beam theory to represent the pile, and uncoupled, nonlinear deformations to a maximum where it remains constant or de-
load transfer functions called ‘‘p-y curves’’ to represent the creases with further deformation.
soil. Historically, these curves were derived from full-scale FloridaPier (FDOT 1996) is a nonlinear finite-element pro-
lateral load tests made primarily in Texas during the 1970s. gram developed at the University of Florida by Michael
Computer programs have been developed to aid the engi- McVay, Marc Hoit, and Cliff Hays. The program has the abil-
neer in lateral load analysis. Com624P uses the finite-differ- ity to model an entire bridge pier cap to the foundation ele-
ence approach to model a single pile structurally as linear elas- ments under axial, lateral, and/or torsional loads. FloridaPier
tic with constant stiffness. A newer program, FloridaPier models a single pile as 16 two-node finite elements. Piles can
(FLPIER), models an entire bridge pier system, including the be modeled as either linear elastic—for example, h-piles and
bridge structure, the pile cap, and a single pile or pile group pipe piles—or nonlinear—in the case of prestressed concrete
using the finite-element method. In addition, FLPIER allows piles or drilled shafts. FloridaPier uses axial (t-z, q-z), lateral
for the nonlinear analysis of concrete piles under mixed load- ( p-y), and torsional (T-␪) pile-soil interaction. Since this re-
ing conditions. Both programs require the input of pile and search concerns only single free-headed laterally loaded piles,
soil parameters that must be determined by laboratory testing, the discussion of the program features will be limited to those
in situ testing, or empirical correlation. The scenario facing pertinent to analysis of such piles.
many design engineers is the appropriate selection of these soil Due to cost considerations, it is not always economical
parameters. Consequently, the experience of the engineer in to perform large-scale lateral load tests to evaluate specific
determining these properties often dictates the accuracy of curves. Therefore, several normalized p-y curves have been
analysis, or more simply stated, ‘‘garbage in = garbage out.’’ suggested based on common soil parameters such as friction
Thus, the objective of this paper is to provide guidance for angle, cohesion, and unit weight, as shown in Table 1. More
parameter selection via a comparison of full-scale lateral load detailed information on these curves can be found in the man-
tests to computer predictions. This paper reports the results of ual for FloridaPier (FDOT 1996).
Soil information typically is given in the form of SPT and
1
Postdoctoral Res. Assoc., Dept. of Civ. Engrg., Univ. of Florida, Gaines-
sometimes CPT boring logs. Unfortunately, the p-y models
ville, FL 32611-6580. require soil parameters such as friction angle and undrained
2
Prof., Dept. of Civ. Engrg., Univ. of Florida, Gainesville, FL 32611- shear strength that are not readily apparent from boring logs.
6580. Thus, an educated engineering judgment must be made as to
Note. Discussion open until April 1, 2002. To extend the closing date what SPT blowcount corresponds to a friction angle of, say,
one month, a written request must be filed with the ASCE Manager of 30⬚. Fortunately, previous research has been done to correlate
Journals. The manuscript for this paper was submitted for review and
possible publication on February 4, 2000; revised May 30, 2001. This
these parameters with in situ tests such as SPT and CPT. The
paper is part of the Journal of Geotechnical and Geoenvironmental sand p-y curves require ␾⬘, k, and ␥ ⬘. Similarly, the clay
Engineering, Vol. 127, No. 11, November, 2001. 䉷ASCE, ISSN 1090- curves require su , ε50 , ε100 , ␥ ⬘, and k. While ␥ ⬘ is typically
0241/01/0011-0920–0925/$8.00 ⫹ $.50 per page. Paper No. 22272. assumed to be between 6.02 ⫻ 10⫺9 and 9.22 ⫻ 10⫺9
920 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2001

J. Geotech. Geoenviron. Eng. 2001.127:920-925.


TABLE 1. Summary of Input Parameters
Soil type Soil stiffness Soil location Parameters Model (reference)
Sand Loose-dense Above and below GWT ␾⬘, k, ␥ O’Neill and Murchison (1983)
␾⬘, k, ␥ Reese et al. (1974)
Clay Soft/medium stiff Above GWT su , ε50 , ε100 O’Neill and Gazioglu (1984)
Below GWT su , ε50 , ␥ Matlock (1970)
Stiff Above GWT su , ε50 , ␥ Welch and Reese (1972)
Below GWT su , ε50 , ␥, k Reese et al. (1975)
Note: ␾⬘ = friction angle; k = horizontal subgrade modulus; ␥ = unit weight; su = undrained shear strength; ε50 = strain at 50% maximum deviator
stress ⫺ UU test; ε100 = strain at 100% maximum deviator stress ⫺ UU test.

TABLE 4. Subgrade Modulus of Sand from SPT N (Johnson and


Kavanaugh 1968)
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 07/05/14. Copyright ASCE. For personal use only; all rights reserved.

Blows/300 mm
(N) 8 10 15 20 30
k a (kN/mm3) 2.67E-06 4.08E-06 7.38E-06 9.74E-06 1.45E-05
a
At 300 mm below ground surface.

FIG. 1. Friction Angle from Overburden Pressure, Relative Density,


and SPT N (Gibbs and Holtz 1957)

TABLE 2. Sand Parameters from SPT (Terzaghi 1955)


Relative Density of Sand
Parameters Loose Medium Dense
Blows/300 mm (N) 4 to 10 10 to 30 30 to 50
␾ (degrees) 30 34 39
Dry or moist sand k (kN/mm3) 2.20E-06 6.60E-06 1.76E-05
Submerged sand k (kN/mm3) 1.26E-06 4.40E-06 1.07E-05

TABLE 3. Soil Parameters (Teng 1962)


Very Very
Compactness loose Loose Medium Dense dense
Relative density (%) 0 to 15 15 to 35 35 to 65 65 to 85 85 to 100
Blows/30 cm (N) 0 to 4 4 to 10 10 to 30 30 to 50 50⫹
␾ (degrees) 28 28 to 30 30 to 36 36 to 41 41⫹

FIG. 2. Friction Angle from Overburden Pressure and CPT Tip Resis-
kN/mm3, the remaining parameters are not so obvious. Several tance (Robertson and Campanella 1983)
correlations from the literature are shown below.

Sand Friction Angle from CPT


Sand Friction Angle and Subgrade Modulus from
SPT The deBeer equation from bearing capacity theory (San-
glerat 1972) for ␾⬘ from qc is as follows:
The following equation is from Peck et al. (1974), using
uncorrected N-values: qc = 1.3␴⬘V0 Kp exp(␲ tan ␾)
Fig. 2 comes from Robertson and Campanella (1983).
␾⬘ = 53.881 ⫺ 27.6034 ⭈ e⫺0.0147 ⭈N
GENERAL DATABASE INFORMATION
Fig. 1 is from Gibbs and Holtz (1957); Table 2 from Ter-
zaghi (1955); Table 3 from Teng (1962); and Table 4 from The database currently contains 24 SPT records and 6 CPT
Johnson and Kavanaugh (1968). records, each of which consists of the structural and soil data
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2001 / 921

J. Geotech. Geoenviron. Eng. 2001.127:920-925.


TABLE 5. Summary of Database Case Histories ‘‘prediction’’ for deformations at 25, 50, 75, and 100% of the
In situ maximum applied lateral test load are plotted on an x-y graph.
Record Location Description data A perfect agreement between the two methods plots on the 45⬚
line. A mathematical term called R-value (not to be confused
1 Oregon Inlet, N.C. 66⬙ cylinder pile SPT
2 Oregon Inlet, N.C. 66⬙ cylinder pile CPT
with R 2) is defined here as the ratio of the two methods eval-
3 Thorofare Bay, N.C. 24⬙ square prestressed SPT uated: R = method on y-axis/method on x-axis ⫻ 100. Table
4 Onslow Co., N.C. 48⬙ drilled shaft SPT 6 summarizes the R-values.
5 New Bern, N.C. 48⬙ drilled shaft SPT
6 Nevada 96⬙ drilled shaft SPT CPT-Correlated Input Parameters
7 Skyway Bridge Site 1 48⬙ drilled shaft SPT
8 Skyway Bridge Site 2 48⬙ drilled shaft SPT Although only three cases containing CPT data were ob-
9 Roosevelt Bridge, Fla. 30⬙ PS square SPT tained, a limited analysis similar to the SPT cases was con-
10 Roosevelt Bridge, Fla. 30⬙ PS square CPT ducted. Specifically, ␾⬘ from Robertson and Campanella
11 California 96⬙ drilled shaft SPT (1983) and ␾⬘ from deBeer (Sanglerat 1972) along with Ter-
12 None given 36⬙ drilled shaft SPT zaghi’s k were used. The R-values for the three CPT cases are
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 07/05/14. Copyright ASCE. For personal use only; all rights reserved.

13 None given 20⬙ PS pile SPT summarized in Table 7. As with the SPT cases, the CPT results
14 Oregon Inlet, N.C. 54⬙ drilled shaft SPT
15 Oregon Inlet, N.C. 54⬙ drilled shaft CPT from the correlations used seem similar.
16 Fuller Warren LLt1 72⬙ drilled shaft SPT
17 Fuller Warren LLt-2 72⬙ drilled shaft (114.5 ft) SPT
EVALUATION OF ENTIRE DATABASE
18 Century Freeway CDOT 96⬙ drilled shaft (shaft A) SPT All SPT Database Cases
19 Century Freeway CDOT 48⬙ drilled shaft (shaft C) SPT
20 Century Freeway CDOT 50⬙ drilled shaft (shaft D) SPT The R-values and standard deviations for the full database
21 Hartford 64⬙ drilled shaft SPT of 24 SPT cases using PHT’s ␾ and Terzaghi’s k are sum-
22 Houston 20⬙ prestressed square SPT
23 Houston 24⬙ PS pipe pile SPT
marized in Table 8. Typically, the measured values are about
24 Century Freeway CDOT 96⬙ drilled shaft (shaft B) SPT 70% of the predicted values. The standard deviation for the
25 Broadway Bridge—US92 60⬙ drilled shaft SPT SPT cases was very high at around 40. This is attributed to
Halifax R. the conservative nature of the correlations. Thus, the SPT-
26 Daytona 60⬙ drilled shaft CPT based predictions are conservative, but appear to be poor due
27 Salt Lake City 12⬙ pipe pile SPT to the high scatter in the results.
28 Salt Lake City 12⬙ pipe pile CPT
29 Auburn 36⬙ drilled shaft SPT All CPT Database Cases
30 Auburn 36⬙ drilled shaft CPT
Note: 1 in. = 2.54 mm. The R-values and standard deviations for the full database
of six CPT cases using Robertson and Campanella’s ␾⬘ and
Terzaghi’s k are summarized in Table 9. Typically, the mea-
for a single free-headed pile. Six load tests have multiple sured values are over 100% of the predicted values. The stan-
records due to the availability of SPT and CPT data for the dard deviation for the CPT cases was extremely high, at
site. Using the macros in the database, input files were gen-
erated for FloridaPier for each load test. The structural and TABLE 6. Summary of R-Values for SPT Correlations
material pile properties were recorded for each record straight
Method R25 R50 R75 R100
from the plans and given information. The lateral load test data
were also entered verbatim from the test data. One difficulty ␾⬘—PHT 56.2 70.9 86.6 87.5
is that the generic p-y curves are for either ‘‘sand’’ or ‘‘clay,’’ k—Terzaghi
␾⬘—PHT 68.2 83.6 97.8 96.8
and therefore judgment is required. An effort was made to k—Johnson and Kavanaugh
make a best guess at the soil properties based on the in situ ␾⬘—Terzaghi 56.3 71.6 88.7 89.5
tests (SPT and CPT) and the correlations presented in the pre- k—Terzaghi
vious section. Since no axial loads were applied to any of the ␾⬘—Gibbs and Holtz 54.6 73.1 98 94.5
piles, the axial parameters are of little significance in the anal- k—Terzaghi
ysis, and a single correlation was used throughout each of the ␾⬘—Teng 56.2 71.1 86.7 87.8
k—Terzaghi
runs. Table 5 summarizes the 30 database cases.

SENSITIVITY TO INPUT PARAMETERS TABLE 7. Summary of CPT Predicted versus Measured

SPT-Correlated Input Parameters Method R25 R50 R75 R100


␾—Robertson and Campanella 95.3 136 186.8 194.5
Given the premise that Florida sites will always have SPT k—Terzaghi
data, we focused our initial efforts on using SPT correlations ␾—DeBeer 103.9 148.3 260 224.5
as input for the p-y curves. Initially 12 of the database sandy k—Terzaghi
soil cases were used to evaluate various ␾⬘ and k input com-
binations; specifically these were, Peck et al. (1974), Terzaghi
(1955), Gibbs and Holtz (1957), and Teng (1962)’s values for TABLE 8. Summary of SPT Predicted versus Measured
␾⬘, with k values estimated via Terzaghi or Johnson and Method R25 R50 R75 R100
Kavanaugh (1968). From these results it is apparent that the PHT’s ␾; Terzaghi’s k 53.3 66.5 84.5 86.8
various combinations tried are all fairly equal. This similarity Standard deviation 24.9 39.9 46.2 35.9
is attributed to our observation that all of the ␾⬘ to k corre-
lations, particularly k, fall within a narrow range. Therefore;
Peck, Hanson, and Thornburn (PHT)’s ␾⬘ (Peck et al. 1975) TABLE 9. Summary of CPT Predicted versus Measured
and Terzaghi’s k were selected for future predictions. Method R25 R50 R75 R100
No capacity analogous to ‘‘Davisson’’ exists for pile under
R-value 105.2 119.4 145.3 154.2
lateral load. In order to assign a numerical measurement of Standard deviation 34.3 47.1 86.4 73.9
the quality of the predictions, the ‘‘measured-load test’’ versus
922 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2001

J. Geotech. Geoenviron. Eng. 2001.127:920-925.


Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 07/05/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 3. Statistics Sheet for All SPT Cases

around 50% or more. The CPT-based predictions are typically (Terzaghi 1955; Johnson and Kavanaugh 1968), all pro-
unconservative, with a very high degree of scatter. Fig. 3 duce similar R-values (R-value = measured/predicted ⫻
shows the statistics for the SPT cases, while Fig. 4 contain the 100%). Therefore, little differenc exists between the SPT
statistics for the CPT cases. correlation combinations, albeit the estimated k value has
a greater effect on deformation.
CONCLUSIONS 2. Similarly, an evaluation of CPT correlations for ␾⬘ [Rob-
ertson and Campanella 1983; deBeer (Sanglerat 1972)]
1. An evaluation of the database shows that SPT correla- coupled with Terzaghi (1955)’s k all produced similar R-
tions for ␾⬘ (Peck et al. 1974; Terzaghi 1955; Gibbs and values. Therefore, little difference exists between the
Holtz 1957; and Teng (1962), coupled with k-values CPT correlation combinations.
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2001 / 923

J. Geotech. Geoenviron. Eng. 2001.127:920-925.


Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 07/05/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 4. Statistics Sheet for All CPT Cases

3. An evaluation of the 28 SPT database cases using ␾⬘ 154%). In addition, the scatter (standard deviation) is
(Peck et al. 1974, and Terzaghi 1955)’s k reveals that high (⬇34 to 74%).
SPT predictions are quite conservative at low load levels
(R-values ⬇ 53%) and remain conservative (R-values ⬇ ACKNOWLEDGMENTS
87%) at high load levels. Also, the scatter (standard de-
This research was supported by the Florida Department of Transpor-
viation) is high (⬇40%). tation (99700-7691-0101), with Peter Lai as technical coordinator and a
4. An evaluation of the six CPT database cases using Rob- major contributor to the research effort; Braulio Grajales and Marc Novak
ertson and Campanella (1983)’s ␾⬘ and Terzaghi’s k re- assisted with data analysis and reduction, and their help is appreciated.
veals unconservative predictions (R-values ⬇ 105 to The North Carolina Department of Transportation Soils and Foundation

924 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2001

J. Geotech. Geoenviron. Eng. 2001.127:920-925.


Design Section provided case histories; the assistance of Brian Keaney, Peck, R. B., Hanson, W. E., and Thornburn, T. H. (1974). Foundation
Jamey Batts, and Don O’Toole was also greatly appreciated. engineering, Wiley, New York.
Reese, L. C., Cox, W. R., and Koop, F. D. (1974). ‘‘Analysis of laterally
REFERENCES loaded piles in sand.’’ Paper No. OTC 2080, Proc., 5th Annu. Offshore
Technol. Conf., Houston, (GESA Rep. No. D-75-9).
Florida Department of Transportation (FDOT). (1996). FloridaPier Users
Manual. Tallahassee 具http://www.dot.state.fl.us/structures/proglib.htm典. Reese, L. C., Cox, W. R., and Koop, F. D. (1975). ‘‘Field testing and
Gazioglu, S. M., and O’Neill, M. W. (1984). ‘‘Evaluation of p-y relation- analysis of laterally loaded piles in stiff clay.’’ Paper No. OTC 2312,
ships in cohesive soils.’’ Analysis and design of pile foundations, J. R. Proc., 7th Offshore Technol. Conf., Houston.
Meyer, ed., ASCE, New York, 192–213. Reese, L. C., and Wang, S. (1993). Com624P—Laterally loaded pile
Gibbs, H. J., and Holtz, W. G. (1957). ‘‘Research on determining the analysis program for the Microcomputer Version 2.0, Publ. No. FHWA-
density of sands by spoon penetration testing.’’ Proc., 4th Int. Conf. on SA-91-048, U.S. Department of Transportation, Washington, D.C.
Soil Mech. and Found. Engrg., Vol. 1, London, 35–39. Robertson, P. K., and Campanella, R. G. (1983). ‘‘Interpretation of cone
Johnson, S. M., Kavanaugh, T. C. (1968). The design of foundations for penetration tests: Parts 1 and 2.’’ Can. Geotech. J., 20(4), 718–745.
buildings. McGraw-Hill, New York. Sanglerat, G. (1972). The Penetrometer and Soil Exploration. Elsevier,
Matlock, H. (1970). ‘‘Correlations for design of laterally loaded piles in Amsterdam.
soft clay.’’ Paper No. OTC 1204, Proc., 2nd Annu. Offshore Technol. Teng, W. (1962). Foundation Design, Prentice-Hall, Englewood Cliffs,
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 07/05/14. Copyright ASCE. For personal use only; all rights reserved.

Conf., Houston, Vol. 1, 577–594. N.J.


McClelland, B., and Focht, J. A. (1958). ‘‘Soil modulus for laterally Terzaghi, K. (1955). ‘‘Evaluation of coefficients of subgrade reaction.’’
loaded piles.’’ Trans., ASCE, Vol. 123, Paper No. 2954, 1049–1086. Geotechnique, London, 5(4), 297–326.
O’Neill, M. W., and Murchison, J. M. (1983). ‘‘An evaluation of p-y Welch, R. C., and Reese, L. C. (1972). ‘‘Lateral load behavior of drilled
relationships in sands.’’ Res. Rep. No. GT-DF02-83, Dept. of Civ. shafts.’’ Res. Rep. No. 3-5-65-99, Ctr. for Hwy. Res., University of
Engrg., University of Houston. Texas at Austin.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2001 / 925

J. Geotech. Geoenviron. Eng. 2001.127:920-925.

Vous aimerez peut-être aussi