Shear Strength of Rock

THE SHEAR STRENGTH OF ROCK MASSES
by
Kurt John Douglas
BE(Hons) USyd

A thesis submitted in partial fulfilment of
the requirements for the degree of
Doctor of Philosophy

School of Civil and Environmental Engineering
The University of New South Wales
Sydney, Australia

December 2002

Thesis/Project Report Sheet

Surname or Family Name: DOUGLAS
First Name: Kurt Other name/s: John
Abbreviation for degree as given in the University calendar: PhD
School: Civil and Environmental Engineering Faculty: Engineering
Title: The Shear Strength of Rock Masses
Declaration relating to disposition of project report/thesis
I am fully aware of the policy of the University relating to the retention and use of higher degree project reports and theses, namely that the University
retains the copies submitted for examination and is free to allow them to be consulted or borrowed. Subject to the provisions of the Copyright Act 1968,
the university may issue a project report or thesis in whole or in part, in photostate or microfilm or other copying medium.
I also authorise the publication by the University Microfilms of a 350 word abstract in Dissertations Abstracts International (applicable to doctorates
only)
...................................................................... ...................................................................... ......................................................................
Signature Witness Date
The university recognise that there may be exceptional circumstances requiring restrictions on copying or conditions on use, Requests for restriction for
a period of up to 2 years must made in writing to the Registrar. Requests for a longer period of restriction may be considered in exceptional circumstances
if accompanied by a letter of support from the Supervisor or Head of School. Such requests must be submitted with the thesis/project report.

FOR OFFICE USE ONLY Date of completion of requirement for Award:
Registrar and Deputy Principal
THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS
Abstract 350 words maximum:

The first section of this thesis (Chapter 2) describes the creation and analysis of a database on concrete and
masonry dam incidents known as CONGDATA. The aim was to carry out as complete a study of concrete and
masonry dam incidents as was practicable, with a greater emphasis than in other studies on the geology, mode of
failure, and the warning signs that were observed. This analysis was used to develop a method of very
approximately assessing probabilities of failure. This can be used in initial risk assessments of large concrete and
masonry dams along with analysis of stability for various annual exceedance probability floods.

The second and main section of this thesis (Chapters 3-6) had its origins in the results of Chapter 2. It was found
that failure through the foundation was common in the list of dams analysed and that information on how to
assess the strength of the foundations of dams on rock masses was limited. This section applies to all applications
of rock mass strength.

Methods used for assessing the shear strength of jointed rock masses are based on empirical criteria. As a general
rule such criteria are based on laboratory scale specimens with very little, and often no, field validation.

The Hoek-Brown empirical rock mass failure criterion was developed in 1980 for hard rock masses. Since its
development it has become almost universally accepted and is now used for all types of rock masses and in all
stress regimes. This thesis uses case studies and databases of intact rock and rockfill triaxial tests collated by the
author to review the current Hoek-Brown criterion. The results highlight the inability of the criterion to fit all
types of intact rock and poor quality rock masses. This arose predominately due to the exponent, a, being
restrained to approximately 0.5 to 0.6 and using rock type as a predictor of m
i
. Modifications to the equations for
determining the Hoek-Brown parameters are provided that overcome these problems.

In the course of reviewing the Hoek-Brown criterion new equations were derived for estimating the shear
strength of intact rock and rockfill. Empirical slope design curves have also been developed.

Abstract Page i
ABSTRACT
The first section of this thesis (Chapter 2) describes the creation and analysis of a
database on concrete and masonry dam incidents known as CONGDATA. The aim was to
carry out as complete a study of concrete and masonry dam incidents as was practicable,
with a greater emphasis than in other studies on the geology, mode of failure, and the
warning signs that were observed. This analysis was used to develop a method of very
approximately assessing probabilities of failure. This can be used in initial risk
assessments of large concrete and masonry dams along with analysis of stability for
various annual exceedance probability floods.
The second and main section of this thesis (Chapters 3-6) had its origins in the results of
Chapter 2 and the general interests of the author. It was found that failure through the
foundation was common in the list of dams analysed and that information on how to
assess the strength of the foundations of dams on rock masses was limited. This section
applies to all applications of rock mass strength such as the stability of rock slopes.
Methods used for assessing the shear strength of jointed rock masses are based on
empirical criteria. As a general rule such criteria are based on laboratory scale
specimens with very little, and often no, field validation.
The Hoek-Brown empirical rock mass failure criterion was developed in 1980 for hard
rock masses. Since its development it has become virtually universally accepted and is
now used for all types of rock masses and in all stress regimes. This thesis uses case
studies and databases of intact rock and rockfill triaxial tests collated by the author to
review the current Hoek-Brown criterion. The results highlight the inability of the
criterion to fit all types of intact rock and poor quality rock masses. This arose
predominately due to the exponent a being restrained to approximately 0.5 to 0.62 and
using rock type as a predictor of m
i
. Modifications to the equations for determining the
Hoek-Brown parameters are provided that overcome these problems.
In the course of reviewing the Hoek-Brown criterion new equations were derived for
estimating the shear strength of intact rock and rockfill. Empirical slope design curves
have also been developed for use as a preliminary tool for slope design.
Acknowledgements Page iii
ACKNOWLEDGEMENTS
The support of the sponsors of the research project - Dams Risk Project - Estimation of
the Probability of Failure, the Australian Research Council, and the Faculty of
Engineering at the University of New South Wales is acknowledged. The sponsors of
the Dams Risk Project were:
ACT Electricity and Water;
Australian Water Technologies, Sydney Water Corporation;
Dams Safety Committee of NSW;
Department of Land and Water Conservation;
Department of Land and Water Conservation - Dams Safety;
Electric Corporation of New Zealand (ECNZ);
Goulburn Murray Water;
Gutteridge Haskins and Davey (GHD);
Hydro-Electric Commission, Tasmania;
Melbourne Water;
NSW Department of Public Works and Services;
Pacific Power;
Queensland Department of Natural Resources;
Snowy Mountains Engineering Corporation (SMEC);
Snowy Mountains Hydro-Electricity Authority.
South Australia Water Corporation;
Water Authority of Western Australia;
The access to, and assistance with collection of data provided by the United States
Bureau of Reclamation and BC Hydro is also acknowledged.
Thanks are made to Pells Sullivan Meynink Pty Ltd who provided the data for most of
the case studies for the rock mass component of this thesis. Thanks also to their staff
who were always available to provide assistance and encouragement. Particular thanks
to Alex Duran who worked with the author on the creation of the slope design curves.
Acknowledgements Page iv
A general acknowledgement is made to the organisations that provided data on various
mine slopes and test results that have been used in the development of this thesis, yet for
the purposes of confidentiality their contributions cannot be properly referenced.
Marcus Helgstedt and Anna Tarua are acknowledged for their assistance with computer
analysis of some of the rock mass strength case studies.
To my fellow PhD comrades: Mark Foster, James Glastonbury and Gavan Hunter, a
major thankyou for your encouragement through example and your friendship.
To my friends, who have not only provided support but also nagging questions e.g. Are
you finished yet? Yes.
To my parents, thanks for your values, your encouragement and your genes.
To my supervisor and co-supervisor Garry Mostyn and Robin Fell respectively,
thankyou not only for your invaluable assistance but also for the invaluable practical
experience you gave me in the areas of rock and dam engineering.
Finally, to my fiance, Rebecca: thanks for hanging in all these years waiting for (and
putting up with) me.

Table of contents Page v
TABLE OF CONTENTS
CHAPTER 1: INTRODUCTION
1.1 THESIS OBJECTIVES.......................................................................................... 1.1
1.2 THE BACKGROUND TO THIS THESIS ............................................................ 1.2
1.3 THE CHAPTERS IN THIS THESIS ..................................................................... 1.3
1.3.1 The analysis of concrete and masonry dam incidents.......................................... 1.3
1.3.2 The shear strength of rock masses ....................................................................... 1.4
1.4 PUBLISHED PAPERS/REPORTS ....................................................................... 1.6
CHAPTER 2: THE ANALYSIS OF CONCRETE AND MASONRY DAM
INCIDENTS
2.1 OUTLINE OF THIS CHAPTER........................................................................... 2.1
2.2 STRUCTURE AND ASSEMBLY OF CONGDATA DATABASE..................... 2.4
2.2.1 Sources of Data .................................................................................................... 2.4
2.2.2 CONGDATA Layout ............................................................................................ 2.9
2.2.3 Data Entered into CONGDATA......................................................................... 2.12
2.2.3.1 Definitions of Failures/Accidents............................................................... 2.12
2.2.3.2 Types of Dams............................................................................................ 2.14
2.2.3.3 Failure Types............................................................................................. 2.14
2.2.3.4 Incident Time.............................................................................................. 2.15
2.2.3.5 Type of Foundation.................................................................................... 2.15
2.2.3.6 Dam Height ................................................................................................ 2.15
2.2.3.7 Detection Methods..................................................................................... 2.16
2.2.3.8 Classification of Causes of Incidents of Dams And Reservoirs................. 2.17
2.2.3.9 Classification of Remedial Measures......................................................... 2.22
2.2.4 Selection of Additional Variables...................................................................... 2.24
2.2.4.1 Time of Incidents........................................................................................ 2.24
2.2.4.2 Foundation Incident Mode......................................................................... 2.25
2.2.4.3 Dam Incident Mode.................................................................................... 2.26
2.2.4.4 Comments on Incidents.............................................................................. 2.26
Table of contents Page vi
2.2.4.5 Description of the Failure or Accident ...................................................... 2.26
2.2.4.6 Additional Geological Information............................................................ 2.26
2.2.4.7 Dam Dimensions........................................................................................ 2.28
2.2.4.8 Valley Shape............................................................................................... 2.29
2.2.4.9 Radius of Curvature................................................................................... 2.29
2.2.4.10 Monitoring and Surveillance Data ............................................................ 2.31
2.2.4.11 Warning Rating.......................................................................................... 2.32
2.2.4.12 Warning Time............................................................................................. 2.32
2.2.4.13 Other Design Factors................................................................................. 2.32
2.2.5 Assumptions Made in Assembling the Database............................................... 2.33
2.2.6 Data on the Population of Dams ........................................................................ 2.36
2.3 RESULTS OF ANALYSIS OF THE DATABASE ............................................ 2.40
2.3.1 Summary of Incidents ........................................................................................ 2.40
2.3.2 Year Commissioned of Dams Experiencing Incidents ...................................... 2.45
2.3.3 Height................................................................................................................. 2.54
2.3.4 Age at Failure..................................................................................................... 2.60
2.3.5 Incident Causes.................................................................................................. 2.78
2.3.6 Monitoring and Surveillance Data ..................................................................... 2.85
2.3.6.1 Using ICOLD Terms .................................................................................. 2.85
2.3.6.2 Details of Warnings ................................................................................... 2.89
2.3.7 Remedial Measures............................................................................................ 2.98
2.3.8 Geology............................................................................................................ 2.101
2.3.8.1 Geology of Dam Foundations Experiencing Incidents............................ 2.101
2.3.8.2 Geology of the Population of Dams......................................................... 2.106
2.3.9.3 Geology - Comparison Between Incidents and Population..................... 2.112
2.3.9 Other Design Factors in Failed Dams .............................................................. 2.123
2.4 METHOD OF FIRST ORDER PROBABILITY ASSESSMENT .................... 2.132
2.4.1 Probability of Failure....................................................................................... 2.132
2.4.1.1 Introduction.............................................................................................. 2.132
2.4.1.2 Population of Dams ................................................................................. 2.133
2.4.1.3 Dam Year ................................................................................................. 2.135
2.4.1.4 Probabilities of Failure............................................................................ 2.135
2.4.1.5 Gravity Dams - Separation of Concrete and Masonry Dams.................. 2.146
Table of contents Page vii
2.4.2 General Approach for Estimating the Probability of Failure for Individual
Gravity Dams ................................................................................................... 2.153
2.4.3 Details of the Method for Estimating the Probability of Failure for
Individual Gravity Dams ................................................................................. 2.154
2.4.4 Gravity Dam Probability Multiplication Factors............................................. 2.158
2.4.4.1 Soil/Rock Foundation Factor, f
SF
and f
PF
................................................ 2.158
2.4.4.2 Geology Types - Sliding on Rock, f
SG
....................................................... 2.161
2.4.4.3 Geology Type - Piping on Rock, f
GE
......................................................... 2.164
2.4.4.4 Height on Width Ratio, f
H/W
...................................................................... 2.164
2.4.4.5 Other Observations, f
O
............................................................................. 2.169
2.4.4.6 Surveillance, f
S
......................................................................................... 2.169
2.4.5 Results.............................................................................................................. 2.170
2.5 DISCUSSION AND CONCLUSIONS.............................................................. 2.172
CHAPTER 3: THE SHEAR STRENGTH OF INTACT ROCK
3.1 INTRODUCTION.................................................................................................. 3.1
3.2 FAILURE CRITERIA FOR INTACT ROCK....................................................... 3.2
3.3 LABORATORY TEST DATABASE FOR INTACT ROCK............................. 3.12
3.4 AN ANALYSIS OF THE ANALYSIS OF DATA ............................................. 3.14
3.5 HOEK-BROWN CRITERION FOR INTACT ROCK........................................ 3.24
3.6 GENERALISED CRITERION FOR INTACT ROCK ....................................... 3.32
3.7 GLOBAL PREDICTION..................................................................................... 3.49
3.8 COMPARISON OF CRITERIA.......................................................................... 3.60
3.9 SYSTEMATIC ERROR IN HOEK-BROWN CRITERION............................... 3.64
3.10 APPLICATION TO SLOPE ENGINEERING.................................................... 3.73
3.11 CONCLUSION.................................................................................................... 3.77
CHAPTER 4: THE SHEAR STRENGTH OF ROCKFILL
4.1 OUTLINE OF THIS CHAPTER........................................................................... 4.1
4.2 FACTORS AFFECTING THE SHEAR STRENGTH OF ROCKFILL ............... 4.2
Table of contents Page viii
4.2.1 Confining Pressure............................................................................................... 4.2
4.2.2 Particle Strength................................................................................................... 4.6
4.2.3 Uniformi ty Coefficient ........................................................................................ 4.8
4.2.4 Density............................................................................................................... 4.10
4.2.5 Maximum Particle Size...................................................................................... 4.12
4.2.5.1 Increasing d
max
with Constant D.................................................................. 4.13
max
with Constant d
max
/D.......................................................... 4.13
4.2.6 Silt and Sand Fines versus Gravel and Larger Particle Content ........................ 4.14
4.2.7 Particle Angularity............................................................................................. 4.16
4.2.8 Other Factors...................................................................................................... 4.17
4.2.9 Summary of Factors Affecting the Secant Friction Angle ................................ 4.18
4.3 SHEAR STRENGTH CRITERIA ....................................................................... 4.19
4.4 DATABASE OF TRIAXIAL SHEAR TESTS.................................................... 4.26
4.5 DATABASE ANALYSIS.................................................................................... 4.29
4.5.1 Analysis Methodology....................................................................................... 4.29
4.5.2 Secant Friction Angle,
sec
, Versus Normal Stress,
n
...................................... 4.30
4.5.2.1 General Assessment of Database............................................................... 4.30
4.5.2.2 Statistical Analysis of Database................................................................. 4.40
4.5.3 Maximum Principal Stress,
1
, versus Minimum Principal Stress,
3
............ 4.41
4.5.3.1 Secant Friction Angle Versus Normal Stress............................................. 4.58
4.5.4 Hoek-Brown Criterion....................................................................................... 4.61
4.6 CONCLUSION.................................................................................................... 4.64
CHAPTER 5: EMPIRICAL ROCK SLOPE DESIGN
5.1 INTRODUCTION.................................................................................................. 5.1
5.2 REVIEW OF THE ROCK MASS RATING SYSTEMS ...................................... 5.2
5.2.1 Methods for Estimating the Basic Rock Mass Rating......................................... 5.3
5.2.1.1 The Rock Mass Rating, RMR, and Geological Strength Index, GSI............ 5.3
5.2.1.2 Mining Rock Mass Rating, MRMR.............................................................. 5.6
5.2.1.3 Rock Mass Strength, RMS............................................................................ 5.9
5.2.1.4 Slope Rock Mass Rating, SRMR................................................................ 5.11
5.2.1.5 Modified Rock Mass Classification, M-RMR............................................. 5.14
Table of contents Page ix
5.2.1.6 Basic Quality, BQ...................................................................................... 5.17
5.2.2 Adjustment Factors to basic rock mass ratings.................................................. 5.19
5.3 A REVIEW OF SLOPE DESIGN METHODS WHICH ARE BASED ON
ROCK MASS RATINGS .................................................................................... 5.25
5.3.1 Correlations with Shear Strength Parameters and Slope Angles....................... 5.25
5.3.2 Available Slope Performance Curves................................................................ 5.30
5.3.3 Pells Sullivan Meynink Slope Performance Curves.......................................... 5.36
5.4 ANALYSIS OF CASE STUDY DATA.............................................................. 5.39
5.4.1 Case Studies Used.............................................................................................. 5.39
5.4.2 Correlations of MRMR, SRMR and RMS with GSI......................................... 5.40
5.4.3 General Assessment of the Parameters in GSI .................................................. 5.51
5.4.4 Development of Generalised Slope Design Curves........................................... 5.54
5.4.4.1 Use of MRMR in Haines and Terbrugge Method...................................... 5.54
5.4.4.2 Revised Method Using MRMR................................................................... 5.54
5.4.4.3 Method Based on the Use of the Geological Strength Index, GSI ............. 5.55
5.5 CONCLUSION.................................................................................................... 5.62
CHAPTER 6: THE SHEAR STRENGTH OF ROCK MASSES
6.1 INTRODUCTION.................................................................................................. 6.1
6.2 ESTIMATING THE SHEAR STRENGTH OF A ROCK MASS......................... 6.3
6.2.1 Predicting Rock Mass Strength from Discontinuities.......................................... 6.3
6.2.2 Predicting Rock Mass Strength using Empirical Formulae................................. 6.8
6.2.3 Predicting Rock Mass Strength using the Hoek-Brown Criterion....................... 6.8
6.3 A DISCUSSION OF THE HOEK-BROWN CRITERION WITH
PARTICULAR REFERENCE TO SLOPES....................................................... 6.16
6.3.1 Calculation of GSI ............................................................................................. 6.16
6.3.1.1 Intact Strength............................................................................................ 6.16
6.3.1.2 RQD ........................................................................................................... 6.17
6.3.1.3 Defect spacing............................................................................................ 6.18
6.3.1.4 Joint condition............................................................................................ 6.18
6.3.1.7 GSI from published figures........................................................................ 6.20
6.3.1.8 A Note on Schistose Rocks......................................................................... 6.20
Table of contents Page x
6.3.2 Estimation of Parameters from GSI................................................................... 6.26
6.3.2.1 The Rock Mass Disturbance Factor, D...................................................... 6.26
6.3.2.2 The Variation of the Hoek-Brown Parameters with GSI........................... 6.27
6.4 VALIDATION OF THE HOEK-BROWN CRITERION ................................... 6.30
6.4.1 Chichester Dam.................................................................................................. 6.30
6.4.2 Nattai North Escarpment Failure ....................................................................... 6.30
6.4.3 Katoomba Escarpment Failure........................................................................... 6.35
6.4.4 Aviemore Dam Insitu Shear Tests ..................................................................... 6.39
6.4.5 Discussion of the Results of the Analysis.......................................................... 6.43
6.5 A NEW ESTIMATION OF ROCK MASS STRENGTH ................................... 6.46
6.5.1 Development of a Modified Criterion ............................................................... 6.46
6.5.1.1 Exponent .............................................................................................. 6.47
6.5.1.2 Parameter m............................................................................................ 6.47
6.5.1.3 Parameter s ............................................................................................. 6.48
6.5.2 Development of the Equations to Estimate the Parameters in the Hoek-
Brown Criterion................................................................................................. 6.49
6.5.2.1 A New Equation for m
b
............................................................................ 6.50
6.5.2.2 A New Equation for s
b
............................................................................. 6.54
6.5.2.3 A New Equation for
b
............................................................................ 6.56
6.5.2.4 The Overall Equation................................................................................. 6.61
6.5.3 Summary of Method.......................................................................................... 6.69
CHAPTER 7: CONCLUSIONS AND RECOMMENDATIONS
7.1 CONCLUSIONS....................................................................................................7.1
7.1.1 The Analysis of Concrete and Masonry Dams .................................................... 7.1
7.1.2 The Shear Strength of Intact Rock....................................................................... 7.3
7.1.3 The Shear Strength of Rockfill ............................................................................ 7.4
7.1.4 Empirical Slope Design ....................................................................................... 7.4
7.1.5 The Shear Strength of Rock Masses.................................................................... 7.5
7.2 RECOMMENDATIONS FOR FURTHER RESEARCH.....................................7.7
7.2.1 The Analysis of Concrete and Masonry Dams .................................................... 7.7
7.2.2 The Shear Strength of Rock Masses.................................................................... 7.7

Table of contents Page xi
REFERENCES
APPENDICES
APPENDIX A CONGDATA DATABASE ................................................... CD-ROM
APPENDIX B DAM LIST - FAILURES .................................................................. A.1
APPENDIX C POPULATION OF DAMS................................................................ A.3
APPENDIX D CAUSES OF INCIDENTS.............................................................. A.11
APPENDIX E SHEAR STRENGTH OF INTACT ROCK DATABASE ...... CD-ROM
APPENDIX F SHEAR STRENGTH OF ROCKFILL DATABASE ............. CD-ROM
Table of contents Page xii
TABLE OF FIGURES
Figure 1.1. Thesis structure.......................................................................................... 1.2
Figure 2.1. Definition of dimensions in CONGDATA............................................... 2.30
Figure 2.2. Definition of dimensions in CONGDATA - section across river............. 2.31
Figure 2.3. The distribution of reported dam incidents vs country............................ 2.43
Figure 2.4. Reported incidents as percentage of countrys dam population from
ICOLD (1984) ......................................................................................... 2.44
Figure 2.5. Year commissioned vs concrete gravity dam incidents .......................... 2.47
Figure 2.6. Year commissioned vs masonry gravity dam incidents .......................... 2.48
Figure 2.7. Year commissioned vs all dam incidents ................................................ 2.49
Figure 2.8. Year commissioned for world population data obtained from ICOLD
(1979) ...................................................................................................... 2.50
Figure 2.9. Year commissioned vs percentage of gravity dams constructed in the
USA......................................................................................................... 2.51
Figure 2.10. Year commissioned - failures/population per period .............................. 2.52
Figure 2.11. CONGDATA - height ranges for all dam significant incidents ............... 2.55
Figure 2.12. CONGDATA - Height ranges for concrete gravity dam significant
incidents................................................................................................... 2.56
Figure 2.13. CONGDATA - height ranges for masonry gravity dam significant
incidents................................................................................................... 2.57
Figure 2.14. Height of failed dams - failures/population (%)...................................... 2.59
Figure 2.15. World dams - height ranges for all concrete & masonry dams ............... 2.60
Figure 2.16. Age at incident - all dams ........................................................................ 2.64
Figure 2.17. Age at incident - concrete gravity dams .................................................. 2.65
Figure 2.18. Age at incident - masonry gravity dams .................................................. 2.66
Figure 2.19. Time to significant incident - gravity dam incidents/population (%)...... 2.67
Figure 2.20. Time to significant incident - all dam incidents/population (%)............. 2.67
Figure 2.21. Failure mode: age at failure versus year commissioned (all dams)......... 2.73
Figure 2.22. Over topping: age at failure versus year commissioned (all dams)......... 2.74
Figure 2.23. Dam type: age at failure versus year commissioned ............................... 2.75
Figure 2.24. Age at significant incident versus year commissioned............................ 2.76
Figure 2.25. Age at significant incident versus year commissioned............................ 2.77
Figure 2.26. Causes of significant incidents (rock & unknown foundations) ............. 2.84
Table of contents Page xiii
Figure 2.27. Causes of significant incidents (soil foundations)................................... 2.85
Figure 2.28. Warning types - gravity dams ................................................................. 2.86
Figure 2.29. Warning Types - All Dams ..................................................................... 2.87
Figure 2.30. Most common remedial measures - all dam incidents ............................ 2.99
Figure 2.31. Foundation incidents age, type and year commissioned - all dams ...... 2.104
Figure 2.32. Foundation incidents geology - all incidents......................................... 2.105
Figure 2.33. Geology for incidents in the foundation and dam population all
dams....................................................................................................... 2.115
Figure 2.34. Geology for incidents in the foundation and dam population
concrete gravity dams............................................................................ 2.116
Figure 2.35. Geology for incidents in the foundation and dam population
masonry gravity dams ........................................................................... 2.117
Figure 2.36. Foundation geology type as a percentage of the geology population
all dams.................................................................................................. 2.118
Figure 2.37. Foundation geology type as a percentage of the geology population
gravity dams .......................................................................................... 2.119
Figure 2.38. Foundation geology type as a percentage of geology population
arch dams ............................................................................................... 2.120
Figure 2.39. Foundation geology type as a percentage of geology population
buttress dams ......................................................................................... 2.121
Figure 2.40. Foundation incident geology and population mode of
failure/accident ...................................................................................... 2.122
Figure 2.41. Average failure stresses for dams with failure through the foundation. 2.131
Figure 2.42. Average failure stresses for Bhandardara Dam..................................... 2.131
Figure 2.43. h
d
/W versus year commissioned............................................................ 2.166
Figure 2.44. h
d
/W versus h
d
....................................................................................... 2.167
Figure 2.45. h
d
/W factors........................................................................................... 2.168
Figure 2.46. Range of annual probability of failure for concrete gravity dams ......... 2.170
Figure 2.47. Range of annual probability of failure for masonry gravity dams ........ 2.171
Figure 3.1. Generalised failure criterion...................................................................... 3.2
Figure 3.2. Comparison of test results with theoretically based failure criteria
(Johnston & Chiu, 1984) ........................................................................... 3.3
Figure 3.3. Comparison of Hoek-Brown criterion (solid) and Johnston criterion
(dashed) for Melbourne mudstone (Johnston, 1985)................................. 3.4
Table of contents Page xiv
Figure 3.4. Fits to artificial data (a) full range (b) low stress range .......................... 3.20
Figure 3.5. m
i
from literature against m
i
from test results and Hoek-Brown
Equation................................................................................................... 3.26
Figure 3.6. Rock type against m
i
from test results and Hoek-Brown equation.......... 3.28
Figure 3.7. Unconfined compressive strength against that predicted by the Hoek-
Brown equation ....................................................................................... 3.30
Figure 3.8. Uniaxial tensile strength against that predicted by the Hoek-Brown
equation................................................................................................... 3.31
Figure 3.9. m
i
i
from test results and generalised
equation................................................................................................... 3.34
Figure 3.10. m
i
i
from test results and generalised
equation................................................................................................... 3.36
i
from test results and generalised equation......... 3.37
Figure 3.12. Unconfined compressive strength against that predicted by
generalised equation................................................................................ 3.39
Figure 3.13. Uniaxial tensile strength against that predicted by generalised
equation................................................................................................... 3.41
Figure 3.14. against m
i
.............................................................................................. 3.43
i
categorised by
c
................................................................. 3.45
Figure 3.16. Family of failure envelopes..................................................................... 3.46
Figure 3.17. Results showing failure envelopes crossing............................................ 3.47
Figure 3.18. Residuals for global fit with constant against
3
/
c
categorised by
-
c
/
t
........................................................................................................ 3.50
Figure 3.19. Residuals for global fit with variable against
3
/
c
categorised by -
c
/
t
......................................................................................................... 3.52
Figure 3.20. Three dimensional plot of global fit ........................................................ 3.54
Figure 3.21.
1
/
c
with fits for variable against
3
/
c
categorised by -
c
/
t
for
high stress................................................................................................ 3.56
Figure 3.22.
1
/
c
3
/
c
categorised by -
c
/
t
for
low stress ................................................................................................. 3.57
i
showing cases with measured or reported
3
and
t
......... .3.59
Figure 3.24.
1
/
c
with fits for published m
I
against
3
/
c
categorised by m
I
for
high stress................................................................................................ 3.62
Table of contents Page xv
Figure 3.25.
1
/
c
I
against
3
/
c
categorised by m
I
for
low stress ................................................................................................. 3.63
Figure 3.26. Pattern of residuals for Hoek-Brown fits ................................................ 3.64
Figure 3.28. Hoek-Brown fits to artificial data............................................................ 3.67
Figure 3.29. Hoek-Brown fits to actual data................................................................ 3.69
Figure 3.30. Residuals for Hoek-Brown fits for weak rock against
3
/
3max

categorised by ...................................................................................... 3.71
Figure 3.31. Residuals for generalised fits for weak rock against
3
/
3max

categorised by ...................................................................................... 3.72
Figure 3.32. Residuals against
3
for various fits ....................................................... 3.75
Figure 4.1. Methods for representing the shear strength envelope.............................. 4.4
Figure 4.2. Variation of secant friction angle,
sec
, with respect to cell confining
stress,
3
, for (a) dense and (b) medium dense crushed basalt from
triaxial tests (Al-Hussaini, 1983)............................................................... 4.5
Figure 4.3. Average strength of rockfills from large-scale direct shear tests
(Anagnosti & Popovic, 1982).................................................................... 4.5
sec
, with normal stress
n

(Indraratna et al., 1993) ............................................................................. 4.6
Figure 4.5. Shear strength and grain size curves for crushed (a) limestone and (b)
marble (Anagnosti and Popovic, 1982)..................................................... 4.7
Figure 4.6. Shear strength and grain size curves for crushed flysch sandstone-
marl rockfill (Anagnosti and Popovic, 1982)............................................ 4.7
Figure 4.7. Shear strength and grain size curves for different gradings of
limestone gravel (Anagnosti and Popovic, 1982) ..................................... 4.9
Figure 4.8. Shear strength and grain size curves for (a) crystalline schist and (b)
sandstone gravels (Anagnosti and Popovic, 1982).................................... 4.9
Figure 4.9. Scalped rockfill gradings (Marachi et al, 1969)...................................... 4.11
Figure 4.10. Strength porosity relationships with
3
= 88kPa (Marachi et al, 1969) .. 4.11
Figure 4.12. Void ratio vs angle of friction (modified from Nakayama et al., 1982).. 4.12
Figure 4.13. Friction angle vs ma ximum particle size (Thiers & Donovan, 1981) ..... 4.13
Figure 4.14. Effect of gravel content on for silty gravel based on USBR (1966) .... 4.15
Figure 4.15. Effect of gravel content on for clayey gravel based on USBR
(1961) ...................................................................................................... 4.15
Table of contents Page xvi
Figure 4.16. 4.76mm content vs secant angle of friction (Nakayama et al., 1982) ... 4.16
Figure 4.17. Gonzalez (1985) shear strength of rockfill equations.............................. 4.22
Figure 4.18. Friction angle, , vs normal stress,
n
. .................................................... 4.23
Figure 4.19. Frequency distributions for coefficients B versus porosity (Sarac &
Popovic, 1985)......................................................................................... 4.23
Figure 4.20. Dependence between coefficient A and gravel unit weight (Sarac &
Popovic, 1985)......................................................................................... 4.24
Figure 4.21. Dependence between coefficient A and mean grain diameter, d
50

(Sarac & Popovic, 1985) ......................................................................... 4.24
Figure 4.22. Secant friction angle,
sec
vs normal stress,
n
........................................ 4.33
sec
vs normal stress,
n
, sorted on angularity
rating........................................................................................................ 4.34
sec
vs normal stress,
n
, sorted on rock type =
basalt........................................................................................................ 4.35
sec
vs normal stress,
n
, sorted on coefficient
of uniformity, c
u
...................................................................................... 4.36
sec
vs normal stress,
n
, sorted on maximum
particle size, d
max
..................................................................................... 4.37
sec
vs normal stress,
n
, sorted on percent fines
(passing 0.075mm) content ..................................................................... 4.38
sec
vs normal stress,
n
, sorted on unconfined
compressive strength of the rock substance, UCS (MPa) ....................... 4.39
Figure 4.29. RFI
e
versus void ratio.............................................................................. 4.43
Figure 4.30. RFI
UCS
versus unconfined compressive strength..................................... 4.43
Figure 4.31. RFI
FINES
versus percent fines................................................................... 4.44
Figure 4.32 Residuals versus unconfined compressive strength of intact rock ........... 4.46
Figure 4.33. Residuals versus d
max
............................................................................... 4.47
Figure 4.34. Residuals versus void ratio...................................................................... 4.48
Figure 4.35. Residuals versus angularity rating........................................................... 4.49
Figure 4.36. Residuals versus fines content................................................................. 4.50
Figure 4.37. Residuals versus sample diameter........................................................... 4.51
Figure 4.38. Effect of unconfined compressive strength on
1
.................................. 4.52
Figure 4.39. Effect of angularity on
1
(7 = sub-angular to angular; 8 = angular)..... 4.52
Table of contents Page xvii
Figure 4.40. Effect of fines content on
1
................................................................... 4.53
Figure 4.41. Effect of maximum particle size on
1
................................................... 4.53
Figure 4.42. Effect of initial void ratio on
1
............................................................ .4.54
Figure 4.43.
1
vs
3
showing data used in analysis and RFI relationship................ 4.55
Figure 4.44.
1
vs
3
showing all data and RFI relationship..................................... 4.56
Figure 4.45.
1
vs
3
showing all data and RFI relationship (
3
up to 1.5MPa)...... 4.57
sec
.................................. 4.59
sec
(7 = sub-angular to angular; 8 = angular).... 4.59
sec
.................................................................. 4.60
sec
.................................................. 4.60
sec
............................................................. 4.61
Figure 4.51. Statistical analysis results using Hoek-Brown formula and for a = 0.6
and a = 0.95............................................................................................. 4.63
Figure 5.1. Estimate of GSI based on geological descriptions. (Hoek, 2000)............. 5.7
Figure 5.2. SRMR strength correlation (a) Island Copper Mine (b) Getchell Mine
(Robertson, 1988).................................................................................... 5.14
Figure 5.3. Example of Planar Failure Case with High SMR.................................... 5.21
Figure 5.4. Slope height, H, vs slope height factor, (after Chen, 1995) ................. 5.22
Figure 5.5. RMR versus slope angle (Orr, 1996)....................................................... 5.27
Figure 5.6. Observed cases (ESMR) vs (a) SMR, (b) CSMR (Chen, 1995) ............. 5.29
Figure 5.7. Upper bound slope height versus slope angle curve for rock masses
(Hoek & Bray, 1981)............................................................................... 5.31
Figure 5.8. Slope angle versus slope height with regression curves (modified
after McMahon, 1976)............................................................................. 5.33
Figure 5.9. Slope height vs slope angle for MRMR (Haines & Terbrugge, 1991).... 5.34
Figure 5.10. Haines & Terbrugge (1991) slope design replotted on basis of slope
angle versus slope height showing Haines & Terbrugge (1991) slope
data. ......................................................................................................... 5.35
Figure 5.11. Slope performance curves for case studies (Duran & Douglas, 1999).... 5.37
Figure 5.12. Correlations of GSI with MRMR, SRMR, RMS rating (mod. Duran
and Douglas, 2002).................................................................................. 5.42
Figure 5.13. GSI versus slope height for failed and stable slopes............................... 5.51
Table of contents Page xviii
Figure 5.14. GSI defect spacing rating versus slope height for failed and stable
slopes....................................................................................................... 5.52
Figure 5.15. GSI defect condition rating versus slope height for failed and stable
slopes....................................................................................................... 5.52
Figure 5.16. GSI RQD rating versus slope height for failed and stable slopes ........... 5.53
Figure 5.17. GSI UCS rating versus slope height for failed and stable slopes............ 5.53
Figure 5.18. Haines & Terbrugge (1991) slope design curves & slope data (Figure
5.10) with additional case studies (Duran & Douglas, 1999)................. 5.56
Figure 5.19. Suggested slope design curves for MRMR (Duran & Douglas, 1999)... 5.57
Figure 5.20. Slope height vs slope angle case study data and the authors proposed
design curves for a dry slope................................................................... 5.58
Figure 5.21. Slope height vs slope angle case study data and a comparison of
design curves for a dry slope................................................................... 5.59
design curves for moderate pressures...................................................... 5.60
Figure 5.23. Slope height vs slope angle case study data and a comparison of
design curves for moderate pressures...................................................... 5.61
Figure 6.1. Heavily jointed rock mass ......................................................................... 6.1
Figure 6.2. Example of shear failure through rock mass at the toe of a slope -
Nattai Escarpment Failure ......................................................................... 6.2
Figure 6.3. Assessment of Barton and Bandis (1982) JRC
0
vs JRC
n
.......................... 6.7
Figure 6.4. Values of the parameter m
i
for intact rock (Hoek, 1999) ........................ 6.11
Figure 6.5. Estimation of GSI (Hoek, 1997).............................................................. 6.12
Figure 6.6. History of the Hoek-Brown criterion (Hoek, 2002) ................................ 6.13
Figure 6.7. Scale effect of Intact Rock (Hoek and Brown, 1980) ............................. 6.17
Figure 6.8. Slope failure block size ........................................................................... 6.19
Figure 6.9. Effect of scale on defect properties ......................................................... 6.19
Figure 6.10. GSI Table (Hoek et al, 1998) .................................................................. 6.22
Figure 6.11. E
d
versus GSI case study data and Hoek et al (1995) equation for
c

100MPa and
c
= 10MPa ..................................................................... 6.24
Figure 6.12. E
d
test from case studies versus E
d
pred from Hoek et al (1995)
equation................................................................................................... 6.25
Figure 6.13. Variation of a, s and m
b
/m
i
with GSI....................................................... 6.28
Figure 6.14. Typical section of the Nattai North failure (Pells et al, 1987)................. 6.31
Table of contents Page xix
Figure 6.15. Illustration of the failure mechanism at Nattai North (Helgstedt,
1997)........................................................................................................ 6.32
Figure 6.16. Katoomba Escarpment Failure ................................................................ 6.36
Figure 6.17. Katoomba Escarpment Failure, column prior to collapse....................... 6.37
Figure 6.18. Direct Shear Test Set up (Foster & Fairless, 1994)................................. 6.41
Figure 6.19. Example of mesh used (Helgstedt, 1997)................................................ 6.44
Figure 6.20. Close up of the simulated jack (Helgstedt, 1997) ................................... 6.44
Figure 6.21. Back analysis results using Figure 6.5 for GSI ....................................... 6.45
Figure 6.22. Test results for tectonised quartzitic sandstone (Habimana et al,
2002)........................................................................................................ 6.52
Figure 6.23. The authors statistical fits to Habimana et al (2002) data...................... 6.53
Figure 6.24. m
b
versus GSI .......................................................................................... 6.54
Figure 6.25. GSI for an intact or massive rock structure (Hoek, 1999)....................... 6.55
Figure 6.26. s
b
versus GSI ........................................................................................... 6.56
Figure 6.27.
b
versus GSI .......................................................................................... 6.57
Figure 6.28. Graphical representation of the equations for and m........................... 6.58
Figure 6.29. Transition of and m from intact rock to rock mass .............................. 6.59
Figure 6.30. Original and modified relationships for
i
and m
i
................................... 6.60
Figure 6.31. Shear strength curves for tectonised quartzitic sandstone (Habimana
et al, 2002)............................................................................................... 6.61
Figure 6.32. Results of analysis of Habimana et al (2002) data using the new
equation and parameters from equations ................................................. 6.63
Figure 6.33. Results of global analysis of Habimana et al (2002) data using new
equations.................................................................................................. 6.64
Figure 6.34. Non-dimensionalised plot of new shear strength curves for m
i
= 40
and varying GSI....................................................................................... 6.65
i
= 4
and varying GSI....................................................................................... 6.66
Figure 6.1. Comparison of the authors criterion and the Hoek-Brown criterion
(Hoek, 2002) for m
i
= 40 ......................................................................... 6.67
Figure 6.2. Comparison of the authors criterion and the Hoek-Brown criterion
(Hoek, 2002) for m
i
= 4 ........................................................................... 6.68

Table of contents Page xx
TABLE OF TABLES
Table 2.1. Bibliography for failed dams in CONGDATA .......................................... 2.6
Table 2.2. CONGDATA parameters........................................................................ 2.10
Table 2.3. Causes of incidents of concrete dams ...................................................... 2.18
Table 2.4. Causes of incidents of masonry dams ..................................................... 2.19
Table 2.5. Causes of incidents to appurtenant works ............................................... 2.20
Table 2.6. Causes of incidents of reservoirs............................................................. 2.21
Table 2.7. Causes of incidents downstream of dam................................................. 2.21
Table 2.8. Classification of remedial measures........................................................ 2.23
Table 2.9. Number of dam incidents in database by type ........................................ 2.40
Table 2.10. Number of significant dam incidents in database by type....................... 2.40
Table 2.11. Number of dam incidents reported in each country................................ 2.42
Table 2.12. Year commissioned - failures vs population per period.......................... 2.53
Table 2.13. Year commissioned - accidents vs population per period....................... 2.54
Table 2.14. Percent of concrete & masonry dam fails vs population for height ........ 2.58
Table 2.15. Percent of concrete & masonry accidents vs population for height ........ 2.58
Table 2.16. No. of dam foundation sliding & piping failures vs age at failure.......... 2.62
Table 2.17. No. of structural (shear or tensile) failures vs age at failure ................... 2.63
Table 2.18. Time to significant incident - incident/population of dams surviving
period (%)................................................................................................ 2.68
Table 2.19. Time to significant incident..................................................................... 2.69
Table 2.20. Details of dam failure water levels.......................................................... 2.70
Table 2.21. Failure types ............................................................................................ 2.78
Table 2.22. Main causes of incidents in all dams ....................................................... 2.79
Table 2.23. Main causes of incidents in concrete gravity dams ................................. 2.80
Table 2.24. Main causes of incidents in masonry gravity dams ................................. 2.80
Table 2.25. Main failure causes for dams with soil foundations ................................ 2.81
Table 2.26. Main significant incident causes for dams with rock or unknown
foundations .............................................................................................. 2.82
Table 2.27. Warning types vs dam type - failures...................................................... 2.88
Table 2.28. Warning types vs dam type - accidents ................................................... 2.88
Table 2.29. Warning types vs dam type - major repairs............................................. 2.89
Table 2.30. Warning ratings for failed dams .............................................................. 2.90
Table of contents Page xxi
Table 2.31. Details of dam failures and descriptions of warnings ............................. 2.91
Table 2.32. Details of dam significant accidents and descriptions of warnings......... 2.95
Table 2.33. Remedial measures - all dam incidents ................................................. 2.100
Table 2.34. Geology for dams with failure in the foundation.................................. 2.102
Table 2.35. Geology for dams with accidents in the foundation.............................. 2.103
Table 2.36. Foundation geology for Australia, New Zealand, Portugal and USBR
(percent and number for each group) .................................................... 2.109
Table 2.37. Foundation geology for Australia, New Zealand, Portugal & USBR
dams - totalled Figures .......................................................................... 2.111
Table 2.38. Failed dams with grouted foundation.................................................... 2.123
Table 2.39. Crest length/height for failed dams and population .............................. 2.125
Table 2.40. Upstream and downstream slopes for failed dams ................................ 2.126
Table 2.41. H
d
/W for failed dams ............................................................................. 2.128
Table 2.42. Back analysed shear strengths for failed dams (mod. from Rich,
1995)...................................................................................................... 2.129
Table 2.43. Calculated normal stresses along the failure plane of back analysed
gravity dams .......................................................................................... 2.130
Table 2.44. Number of dams as at 1992................................................................... 2.133
Table 2.45. Population of dams by dam type and year commissioned .................... 2.134
Table 2.46. Number of dams (excluding China) in the population.......................... 2.134
Table 2.47. Annual probability of failure (1992, exc. China) - all failure types...... 2.136
Table 2.48. Probability of failure (as at 1992, exc. China, non-annualised) - all
failure types ........................................................................................... 2.137
Table 2.49. Annual probability of failure (as at 1992, excluding China) - sliding
failures................................................................................................... 2.138
Table 2.50. Probability of failure (as at 1992, excluding China, non-annualised) -
sliding failures ....................................................................................... 2.139
Table 2.51. Annual probability of failure (as at 1992, excluding China) - piping
failures................................................................................................... 2.140
piping failures........................................................................................ 2.141
Table 2.53. Annual probability of failure (as at 1992, excluding China) -
tension/shear failures through dam body............................................... 2.142
Table of contents Page xxii
tension/shear failures through dam body............................................... 2.143
Table 2.55. Number of failures during overtopping where the failure mode was
unknown................................................................................................ 2.144
Table 2.56. No. of failures where the failure mode was unknown (no
overtopping) .......................................................................................... 2.145
Table 2.57. Distribution of concrete and masonry gravity dams in the USA........... 2.146
Table 2.58. Distribution of concrete and masonry gravity dams chosen for
analysis .................................................................................................. 2.148
Table 2.59. Summary of annualised probabilities of failure for gravity dams (exc.
China) .................................................................................................... 2.148
Table 2.60. Suggested values for annualised probabilities of failure for gravity
dams (excluding China)......................................................................... 2.149
Table 2.61. Annualised probabilities of failure for gravity dams - all failures ........ 2.150
Table 2.62. Annualised probabilities of failure for gravity dams - sliding failures . 2.150
Table 2.63. Annualised probabilities of failure for gravity dams - piping failures .. 2.151
Table 2.64. Annualised probabilities of failure for gravity dams - dam body
tension/shear failures............................................................................. 2.151
Table 2.65. Number of failures during overtopping where failure mode was
unknown................................................................................................ 2.152
Table 2.66. Number of failures where failure mode was unknown......................... 2.152
Table 2.67. Foundation types - USBR...................................................................... 2.158
Table 2.68. Foundation types - Australia/New Zealand........................................... 2.158
Table 2.69. Foundation types - Portugal .................................................................. 2.159
Table 2.70. Gravity dam foundation types - combined............................................ 2.159
Table 2.71. Foundations for gravity dam failures by sliding or piping.................... 2.159
Table 2.72. Gravity dam factors for piping and sliding failure on soil and rock,
f
SF
and f
PF
............................................................................................... 2.160
Table 2.73. Foundation types - accidents ................................................................. 2.160
Table 2.74. Weighting factors used for weighted average (ICOLD (1984) dam
population)............................................................................................. 2.161
Table 2.75. Adopted gravity dam factors for sliding on a rock foundation, f
SG
....... 2.162
Table 2.76. Gravity dam factors for sliding on a rock foundation........................... 2.163
Table of contents Page xxiii
Table 2.77. Multiplication factors for structural height/width ratio of gravity
dams, f
H/W
............................................................................................... 2.165
Table 2.78. Monitoring and surveillance multiplication factors, f
S
.......................... 2.169
Table 3.1. Various intact rock failure criteria............................................................. 3.6
Table 3.2. Empirical estimates of exponents for the equations in Table 3.1............ 3.10
Table 3.3. Suggested values of constant k (Yudhbir et al, 1983 and Bieniawsi,
1974)........................................................................................................ 3.10
Table 3.4. Values of M and B for a range of materials (Johnston, 1991) ................ 3.10
Table 3.5. Intact rock database descriptors .............................................................. 3.13
Table 3.6. Results of different regression methods on artificial data....................... 3.19
Table 3.7. Error in approximating
ut
as -
c
/(m
i
+1)................................................. 3.42
Table 3.8. Comparison of predictions ...................................................................... 3.60
Table 3.9. Errors in fitting Hoek-Brown criterion to materials with 0.5 ........... 3.65
Table 3.10. Variation of
c
and m
i
with
3max
for exact simulated results ................. 3.66
c
and m
i
with
3max
for data set 434.................................. 3.68
Table 3.12. Triaxial component of strength ............................................................... 3.73
Table 3.13. Comparison of predictions for weak rocks at low stress......................... 3.74
Table 4.1. Increase in with d
max
/D from Marsal (1973) data for different
n
........ 4.13
Table 4.2. Reduction in with particle size, d
max
, from Marsal (1973) data ........... 4.13
Table 4.3. Summary of factors affecting the secant friction angle........................... 4.18
Table 4.4. Parameters obtained using De Mello (1977) (Charles, 1991)................. 4.19
Table 4.5. Various shear strength criteria for rockfill .............................................. 4.20
Table 4.6. List of parameters in triaxial shear strength database ............................. 4.27
Table 4.7. Summary of basic statistics from the rockfill database........................... 4.28
Table 4.8. Changes in
sec
on Figure 4.46 to Figure 4.50 for
n
=1MPa and
n
=
0.5MPa .................................................................................................... 4.58
Table 4.9. Results from the statistical analysis of the rockfill database using the
Hoek-Brown equation ............................................................................. 4.62
Table 5.1. Comparison of weightings for various rock mass rating methods ............ 5.2
Table 5.2. Rock mass rating (Bieniawski, 1989)........................................................ 5.5
Table 5.3. Geological strength index, GSI (Hoek et al, 1995) ................................... 5.6
Table 5.4. MRMR (Laubscher, 1977) ........................................................................ 5.8
Table 5.5. Defect condition rating for MRMR (Laubscher, 1977)............................. 5.8
Table of contents Page xxiv
Table 5.6. RMS Classification and Ratings (mod. Selby, 1980).............................. 5.10
Table 5.7. SRK Geomechanics Classification or Slope Rock Mass Rating
(SRMR) ................................................................................................... 5.13
Table 5.8. SRMR strength correlation (Robertson, 1988)........................................ 5.14
Table 5.9. Joint condition index I
JC
(nal, 1996)..................................................... 5.16
Table 5.10. Ratings for joint condition parameters (nal, 1996)............................... 5.17
Table 5.11. The basic quality, BQ, rock mass classes (Lin, 1998) ............................ 5.18
Table 5.12. Adjustment rating for joints (after Romana, 1985) ................................. 5.20
Table 5.13. Adjustment Rating for methods of excavation of slopes (after
Romana, 1985) ........................................................................................ 5.20
Table 5.14. Tentative description of SMR classes (after Romana, 1985).................. 5.21
Table 5.15. Discontinuity condition factor (Chen, 1995)........................................ 5.22
Table 5.16. Blasting adjustment, A
b
(nal, 1996) ...................................................... 5.23
Table 5.17. Major plane of weakness adjustment, A
w
(nal, 1996)........................... 5.23
Table 5.18. Rock mass properties for RMR
76
(Bieniawski, 1976)............................. 5.25
Table 5.19. Stable slope angle versus MRMR (Laubscher, 1977) ............................. 5.26
Table 5.20. Case records for SMR (after Romana, 1985) .......................................... 5.28
Table 5.21. Parameters for McMahons (1976) slope relationship............................ 5.32
Table 5.22. Summary of slope data from case studies ............................................... 5.39
Table 5.23. Correlation between rating methods authors case studies................... 5.41
Table 5.24. Summary of best estimate GSI data for mine cases ................................ 5.43
Table 5.25. Summary of defect condition for GSI ..................................................... 5.44
Table 5.26. Summary of best estimate of Laubschers MRMR data for mine
cases......................................................................................................... 5.45
Table 5.27. Summary of best estimate of SRMR data for mine cases ....................... 5.47
Table 5.28. Summary of best estimate of RMS data for mine cases.......................... 5.49
Table 6.1. Estimation of Hoek-Brown co-efficients .................................................. 6.9
Table 6.2. Rock mass deformability case studies..................................................... 6.23
Table 6.3. Guidelines for estimating disturbance factor D (Hoek et al, 2002)......... 6.26
Table 6.4. Maximum and minimum values of Hoek-Brown parameters using
Figure 6.10............................................................................................... 6.29
Table 6.5. Joint orientation, spacing and persistence for Nattai North .................... 6.34
Table 6.6. Summary of parameters used for the Nattai North Escarpment
Failure...................................................................................................... 6.34
Table of contents Page xxv
Table 6.7. Summary of Hoek-Brown parameters for Nattai using RMR and the
Hoek-Brown chart ................................................................................... 6.35
Table 6.8. UDEC output: average shear and normal stresses along the predicted
failure plane ............................................................................................. 6.35
Table 6.9. Summary of parameters used for the Katoomba Escarpment Failure..... 6.38
Table 6.10. Summary of Hoek-Brown parameters for the Claystone in the
Katoomba Escarpment Failure using RMR and the Hoek-Brown
chart ......................................................................................................... 6.38
Table 6.11. Summary of the Joint Properties from the Joint Survey carried out by
Read et al (1996) ..................................................................................... 6.40
Table 6.12. Hoek Parameters for Aviemore Shear Tests using RMR, and the
Hoek-Brown Chart .................................................................................. 6.40
Table 6.13. Intact material parameters ....................................................................... 6.43
Table 6.14. Defect and Interface Material Parameters ............................................... 6.43
Table 6.15. Results of statistical analysis of Habimana et al (2002) test data............ 6.50
Introduction Page 1.1
1 INTRODUCTION
1.1 THESIS OBJECTIVES
The overall objective of this thesis was to improve design procedures for large-scale
structures constructed on or in rock.
The first objective of this thesis was the development of a comprehensive database and
analysis of concrete and masonry dam incidents world-wide, with a view to developing
methods of assessing the risk of failure of existing dam structures and as a consequence of
this, identifying possibilities for improvement in design. The results from this work
showed that a large proportion of dams had incidents associated with the strength of their
foundations, which indicated a need for a better understanding of the strength of rock
masses.
The second and major objective of this thesis was to provide a detailed assessment of the
applicability of the Hoek-Brown criterion to estimating the shear strength of jointed rock
masses and to improve upon any deficiencies found in the criterion. A particular focus
was placed on low stress situations (e.g. dams and slopes) in weak materials. This work
also included the development of methods for assessing the strength of intact rock and
rockfill and new methods for estimating the stability of rock slopes using rock mass rating
systems.
To achieve the first objective the author collated and analysed the largest and most
comprehensive database of world-wide concrete and masonry dam incidents using both
published literature and unpublished records personally obtained from the dam industry.
Results were presented on what factors have led to a higher chance of a dam incident
occurring. These results were then used to develop an approximate method of assessing
probabilities of failure.
The process for assessing the Hoek-Brown criterion had several components. Firstly,
large databases of triaxial tests on intact rock and rockfill were statistically analysed to
assess the applicability of the Hoek-Brown criterion at the limits of intact rock and very
poor rock mass. Secondly, analyses of the failures of large scale rock masses were
carried out to assess how well the Hoek-Brown criterion predicted the insitu rock mass
strength. Finally, high quality triaxial tests on rock mass were obtained from the literature
and used together with the results of the previous steps and plausibility checks to create
new equations for estimating the parameters in the Hoek-Brown criterion.
1.2 THE BACKGROUND TO THIS THESIS
This thesis is divided into two sections: the risk assessment of concrete and masonry dam
failures and accidents and their causes, and the shear strength of rock masses as shown in
Figure 1.1.
The analysis of concrete and masonry dam incidents
Chapter 2
The shear strength of intact rock
Chapter 3
The shear strength of rockfill
Chapter 4
Empirical rock slope design
Chapter 5
The shear strength of rock masses
Chapter 6
The shear strength of rock masses
Thesis

Figure 1.1. Thesis structure
The dams community as part of the Dams Risk Project (together with the ARC and The
Faculty of Engineering at the University of New South Wales) provided initial funding for
this research. Details of the specific contributors are provided in the acknowledgement
section. The first section of the thesis (Chapter 2) was carried out in response to the
specific needs of the research project and its sponsors. The aim of the project was to
provide a guide as to which types of dams were more likely to experience incidents
(failures and accidents) based on a statistical analysis of the historical performance of
dams. The authors role on the project was to study concrete and masonry gravity dams.
The second section of this thesis (Chapters 3-6) had its origins in the results of Chapter 2
and the general interests of the author and the project sponsors. It was found that failure
through the foundation was common in the list of dams analysed. Furthermore, it was
found that information on how to assess the strength of the foundations of dams on rock
masses was limited. For example, the ANCOLD (1991) guidelines on design criteria for
concrete gravity dams suggest using references such as McMahon (1985) and Hoek
(1983) to assess the strength of the foundation. The guidelines also state that in the
absence of more reliable data, preliminary analysis of foundations on sound jointed hard
rock where sub-horizontal joints are not continuous, the following peak effective shear
strength parameters are suggested:
c
peak
= 0.14
c
or 1.4MPa whichever is the lesser, where
c
is the
unconfined compressive strength of the rock substance.
peak
= 45
This approach is very misleading and in many cases would over-estimate the strength.
The authors aim for this section was to assess how good the methods for estimating rock
mass strength were and to suggest possible changes to existing methods or new methods if
required
The work on rock mass strength was extended from looking at the foundations of dams to
looking at the strength of rock masses in slopes and other works. This was mainly in an
attempt to find better case studies to analyse, to cover a wider stress range and to provide
a work on rock mass strength that had applicability wherever an assessment of rock mass
strength was required.
The different sections of the thesis given in Figure 1.1 are described in more detail
below.
1.3 THE CHAPTERS IN THIS THESIS
1.3.1 The Analysis of Concrete and Masonry Dam Incidents
Many attempts have been made at compiling and assessing statistics of dam failures. The
main attempts at assessing dam incidents on a world-wide scale have been by ICOLD
(1974, 1983 and 1995). ICOLD (1974) analysed previous dam failures and accidents
based on questionnaires provided by the National Committees on Large Dams. ICOLD
(1983) attempted to improve the completeness of the information with further
questionnaires. An existing dam population was also developed by ICOLD for
comparison with failures. The population comprised a sample of dams from the ICOLD
World Register of Dams (ICOLD 1973, 1976 and 1979). ICOLD (1995) was an attempt
to update the statistics on failures of dams with particular emphasis on comparisons with
dam types, heights and years commissioned of existing dams. Although an extensive
analysis, the ICOLD attempt lacks depth in some key areas. Most notably in information
on the foundation conditions and the geometry of the dams where failures have occurred.
The accuracy and consistency of the ICOLD data has also come into question during this
current research.
Various other attempts have been made to compile data on failures and accidents, all of
which either suffer from a lack of detail or from a limited data set. Most of the statistical
analyses of failures and accidents and attempts to determine probabilities of failure (Von
Thun (1985), da Silveira (1984, 1990), Fell (1996), Blind (1983), and Schnitter (1993))
tend not to go into much detail, generally assessing only height, year commissioned and
type of dam structure. Most of the emphasis in the analysis of dam incidents has been on
embankment dams.
This section of the thesis (Chapter 2) describes the creation and analysis of a database on
concrete and masonry dam incidents known as CONGDATA. The aim was to carry out as
complete a study of concrete and masonry dam incidents as was practicable, with a
greater emphasis than in other studies on the geology, mode of failure, and the warning
signs that were observed. The study assessed the characteristics of the population of
dams, and compared these with the characteristics of those dams that had experienced
incidents. This helped to provide a guide as to which dams were more likely to
experience incidents.
This analysis was used to develop an approximate method of assessing probabilities of
failure. This can be used in initial risk assessments of large concrete and masonry dams
along with analyses of stability for various annual exceedance probability floods.
1.3.2 The shear strength of rock masses
Methods used for assessing the shear strength of jointed rock masses are based on
empirical criteria (Hoek and Brown, 1980, Yudhbir et al, 1983, Ramamurthy et al, 1994
and Sheorey, 1997). As a general rule such criteria are based on laboratory scale
specimens with very little, and often no, field validation.
The most commonly used strength criterion, having received widespread interest and use
over the last two decades, is the Hoek-Brown empirical rock mass failure criterion, the
most general form of which is given in Equation 1.1. Hoek and Brown (1980) developed
this rock mass criterion as they found that there were really no suitable criteria for the
purpose of underground excavation design (Hoek, 2001). The equation, which has
subsequently been updated by Hoek and Brown (1988), Hoek et al. (1992), Hoek et al.
(1995) and Hoek et al (2002), was based on their criterion for intact rock. The only
rock mass tested and used in the original development of the Hoek-Brown criterion was
152mm core samples of Panguna Andesite from Bougainville in Papua New Guinea
(Hoek and Brown, 1980). Hoek and Brown (1988) later noted that it was likely this
material was in fact disturbed. The validation of the updates of the Hoek-Brown
criterion have been based on experience gained whilst using this criterion. To the
authors knowledge the only data published supporting this experience has been two mine
slopes cited in Hoek et al (2002).

a
c
b c
s m

+ =

3
3 1
(1.1)
This thesis assesses the Hoek-Brown criterion in detail and modifies it into a more
generalised form to account for various inconsistencies in the current version. The
assessment of the criterion is carried out by looking at several of its bounds including
intact rock (Chapter 3) and rockfill (Chapter 4). Case studies of various failures and
highly stressed rock masses are used, together with published laboratory test results on
rock mass samples, to assess the Hoek-Brown criterion and to develop new equations
that can be used to estimate the parameters of the Hoek-Brown equation (Chapter 6).
The individual chapters in this section of the thesis not only provide a basis for modifying
the Hoek-Brown criterion (discussed in Chapter 6) but also have their own individual
results including:
Chapter 3 - A statistical analysis of a database of over 4500 triaxial tests on intact
rock and the subsequent development of new shear strength criterion for intact rock.
Chapter 4 - A review of current criteria for the shear strength of rockfill and the
development of a new shear strength criterion for compacted rockfill based on a
database of over 550 rockfill triaxial tests gathered from the literature and sponsors.
Chapter 5 - An analysis of current empirical slope design methods and the
development of new slope design curves based on the authors database of mine pit
slopes.
1.4 PUBLISHED PAPERS/REPORTS
The following papers and reports were published during the period of this thesis.
Douglas, K.J. (1998) Case studies in the assessment of rock mass criteria. 3rd Young
Geotechnical Professionals Conference, Melbourne.
Douglas, K.J. and Mostyn, G. (1999) Strength of large rock masses field verification.
Rock Mechanics for Industry, Proceedings of the 37
th
U.S. Rock Mechanics Symposium,
Vail, Colorado, USA. 1:271-276. Balkema, Rotterdam, ISBN 90 5809 099 X0.
Douglas, K., Spannagle, M. and Fell, R. (1998a) Estimating the probability of failure of
concrete and masonry gravity dams. 1998 ANCOLD-NZSOLD Conference on Dams,
Sydney.
Douglas, K., Spannagle, M. and Fell, R. (1998b) Report on Analysis of Concrete and
Masonry Dam Incidents. UNICIV, The School of Civil and Environmental Engineering,
The University of New South Wales.
Douglas, K., Spannagle, M. and Fell, R. (1999a) Analysis of Concrete and Masonry Dam
Incidents. The International Journal on Hydropower & Dams. 6(4):108-115.
Aqua~Media, Surrey, ISSN 1352-2523.
Douglas, K., Spannagle, M. and Fell, R. (1999b) Estimating the probability of failure of
concrete and masonry gravity dams. ANCOLD Bulletin. No. 112:53-63. Australian
National Committee on Large Dams, ISSN 0045-0731.
Duran, A. and Douglas, K. (1999) Do slopes designed with empirical rock mass strength
criteria stand up? Proceedings ISRM 9
th
International Congress on Rock Mechanics,
Paris, France, 1, pp. 87-90. Balkema, Rotterdam, ISBN 90 5809 070 1.
Duran, A. & Douglas, K.J. (2000) Experience with empirical rock slope design.
GeoEng2000: An International Conference on Geotechnical & Geological
Engineering, 19-24 November, Melbourne, Australia, 2, pp. 41 and CD-Rom paper no.
SNES1186, Technomic Publishing, Pennsylvania, ISBN 1-58716-068-4.
Glastonbury, J. & Douglas, K.J. (2000) Catastrophic rock slope failures. GeoEng2000:
An International Conference on Geotechnical & Geological Engineering, 19-24
November, Melbourne, Australia, Vol. 2 pp. 21 and CD-Rom paper no. SNES0507,
Technomic Publishing, Pennsylvania, ISBN 1-58716-068-4.
Helgstedt, M.D., Douglas, K.J. and Mostyn, G. (1997) A re-evaluation of in-situ direct
shear tests at Aviemore Dam, New Zealand. Australian Geomechanics, 37 (June), pp.
56-65.
Mostyn, G. & Douglas, K.J. (2000) Issues Lecture: The shear strength of intact rock and
rock masses. GeoEng2000: An International Conference on Geotechnical & Geological
Engineering, 19-24 November, Melbourne, Australia, Vol. 1, pp. 1389-1421, Technomic
Publishing, Pennsylvania, ISBN 1-58716-067-6.
Mostyn, G., M.D. Helgstedt and K.J. Douglas (1997) Towards field bounds on rock
mass failure criteria. International Journal of Rock Mechanics and Mining Sciences,
Vol. 34 (3-4): Paper No. 208.
Analysis of Concrete and Masonry Dam Incidents Page 2.1
2 ANALYSIS OF CONCRETE AND MASONRY DAM INCIDENTS
2.1 OUTLINE OF THIS CHAPTER
This chapter comprises a component of the research project on the risk assessment of
dams. 17 sponsors comprising major dam owners and consultants from Australia and
New Zealand support the project. The United States Bureau of Reclamation (USBR)
and BC Hydro from Canada also assisted with the project.
The aim of this component of the research was to provide a guide as to which dams
were more likely to experience incidents based purely on a statistical analysis of
historical incidents. The chapter describes the analysis of historical concrete and
masonry dam incidents. For comparative purposes a compilation of data from a sample
of existing concrete and masonry dams from Australia, New Zealand, Portugal and the
USA is also presented for comparison with the incident database. A large portion (4168
dams from 22 countries) of the ICOLD (1973 and 1979) World Register was entered
into a computer for further comparative purposes. The source for the analysis was a
database developed by the author on failures and accidents in concrete and masonry
dams known as CONGDATA.
Many attempts have been made at compiling and assessing statistics of dam failures.
The main attempts at assessing dam incidents on a worldwide scale have been by
ICOLD (1974, 1983 and 1995). ICOLD (1974) analysed previous dam failures and
accidents based on questionnaires provided by the National Committees on Large
Dams. ICOLD (1983) attempted to improve the completeness of the information with
further questionnaires. The presentation of the data and analyses was improved with the
use of tables. An existing dam population was also developed for comparison with
failures. The population comprised a sample of dams from the ICOLD World Register
of Dams (ICOLD 1973, 1976 and 1979). ICOLD (1995) was an attempt to update the
statistics on failures of dams with particular emphasis on comparisons with dam types,
heights and years commissioned of existing dams. Although an extensive analysis, the
ICOLD attempt lacks depth in some key areas. Most notably in information on the
foundation conditions and the geometry of the dams where failures have occurred. The
accuracy and consistency of the ICOLD data has also come into question during this
current research (see Section 2.2.5).
Vogel (1980, with updates to 1994) and Babb and Mermel (1968) have compiled lists of
dam incidents with some limited comments and dimensions. Their main value is in
providing a large source of references. USCOLD (1976 and 1988) collated a large
amount of information on incidents in the USA. Other attempts at collecting data on
historical incidents have been made by Jorgensen (1920), Jansen (1980), Varshney and
Raheem (1971), USCOLD (1996) and Rao (1960). All of these either suffer from a lack
of detail or from a limited data set.
Smaller country scale data collections have been made for:
Spanish accidents and failures (Gomez Laa et al, 1979);
the deterioration of Italian Dams (Paolina et al, 1991);
South African dam incidents (Olwage & Oosthuizen, 1984);
Swedish accidents (Graham & Bartsch, 1995);
failures and accidents in the United Kingdom (Charles, 1985);
incidents in Australia (Ingles, 1984); and
failures and accidents in the USA (Hatem, 1985).
Von Thun (1985) made an assessment of USA dams and their probability of failure
based on a calculation of dam years. The parameters assessed were generally similar to
those of ICOLD. Others who have attempted to analyse probabilities of failure include:
da Silveira (1984, 1990); Fell (1996); Blind (1983); and Schnitter (1993) who generally
based their analysis on ICOLD data and experience. Serafim (1981a, 1981b); Tavares
and Serafim (1983); Smith (1972); Biswas and Chatterjee (1971); Gruner (1963, 1967);
Kaloustian (1984) analysed incidents using ICOLD data and their own selected
databases. These analyses tend not to go into much detail, generally assessing only
height, year commissioned and type of dam structure. Most of the emphasis in the
analysis of dam incidents has been on embankment dams.
This study set out to carry out as complete a study of concrete and masonry dam failures
and accidents as was practicable, with a greater emphasis than in other studies on the
geology, mode of failure, and the warning signs that were observed. The study also sets
out to assess the characteristics of the population of dams, and compares the
characteristics of the failures and accidents with the population of dams, so a probability
of failure or accident can be assigned. This data provides the basis for initial risk
assessments of dams.
The basic definitions used in CONGDATA and the subsequent analyses have been taken
from ICOLD and are given in Section 2.2.3.1. The term incident has been used for both
accidents and failures.
Section 2.2 of this chapter describes the methods used in compiling and assessing the
incident statistics. The results have been presented in Section 2.3. A method of first
order probability assessment for gravity dams is provided in Section 2.4.
2.2 STRUCTURE AND ASSEMBLY OF CONGDATA DATABASE
2.2.1 Sources of Data
CONGDATA began with the information from the three ICOLD compilations of failures
and accidents:
ICOLD (1995) Dam Failures Statistical Analyses.
ICOLD (1983) Deterioration of Dams and Reservoirs.
ICOLD (1974) Lessons From Dam Incidents.
Where practicable ICOLD (1995) definitions were used. ICOLD (1995) was the main
reference for the failures whilst accidents were principally from ICOLD (1983). ICOLD
(1974) was used for further details when adding information into CONGDATA.
The information in CONGDATA was then checked and updated using other existing
databases including:
USCOLD (1976) Lessons from dam incidents, USA.
USCOLD (1988) Lessons from dam incidents. USA-II.
Vogel (1980) Bibliography of the History of Dam Failures.
Babb and Mermel (1968) Catalogue of Dam Disasters, Failures and Accidents.
A large literature review was then conducted to gather as much information on dam
failures and accidents as possible. References cited in the databases above were sought
and then further references were obtained from journals; conference proceedings;
reports; theses; and Internet pages. Published and unpublished reports were also
accessed through sponsors and dam organisations. All references were followed to their
origins as far as practically possible. The literature review was far more extensive than
those previously reported for the development of other databases. Table 2.1 shows the
bibliography for the failed dams contained in CONGDATA.
Data from several additional dams was added to the database during the data gathering
process. The additions are described in detail in Section 2.2.4.
The sponsors of the research project, who are listed below, also provided access to
information on their dams.
Australian Water Technologies, Sydney Water Corporation;
Department of Land and Water Conservation;
NSW Department of Public Works and Services;
SA Water Corporation;
ACT Electricity and Water;
Hydro-Electric Commission;
Dams Safety Committee of NSW;
Department of Land and Water Conservation - Dams Safety;
Snowy Mountains Engineering Corporation (SMEC);
Queensland Department of Natural Resources;
Goulburn-Murray Water;
Gutteridge Haskins and Davey;
Melbourne Water;
Pacific Power;
Sydney Water Corporation;
Water Authority of Western Australia;
Electric Corporation of New Zealand;
Snowy Mountains Hydro-Electricity Authority.
The United States Bureau of Reclamation (USBR) in Denver allowed access to the
information on their dams. This information was collected over two, three-week periods
by the author together with Professor Robin Fell and Mark Foster. Other organisations
that allowed access to data included:
BC Hydro; and the
Alberta Dam Safety Association.
The data collected from the sponsors and other assisting organisations was used as a
source of information on failures and more notably to assist in a collation of information
on dam populations.
Page 2.6
Table 2.1. Bibliography for failed dams in CONGDATA
Dam References
Angels, USA Babb and Mermel (1968); ICOLD (1974, 1995); USCOLD (1975); Vogel (1994)
Ashley, USA
Babb and Mermel (1968); Engineering News (1909); ICOLD (1974, 1984, 1995); Jorgensen (1920); Rao (1960); Scott and Von Thun (1993);
Vogel (1994)
Austin (Texas), USA
Babb and Mermel (1968); Blanton (1915); Bowers (1928); Engineering News (1900a-e, 1901, 1902, 1908a-b, 1910b, 1915, 1916a-b);
Engineering News Record (1911, 1918, 1911f); Engineering Record (1900a-b, 1911a-c, 1915a-b); Freeman and Alsop (1941); Hatton (1912);
Hill (1902); Hornaday (1899); ICOLD (1974, 1995); Jorgensen (1920); King and Huber (1993); Leger et al (1997); McDonough (1940);
Parker (1900); Patterson (1900); Rao (1960); Rosenberg (1900); Sawyer (1911); Schuyler (1908); Smith (1972); Taylor (1915); Taylor (1900);
USCOLD (1975)
Bacino di Rutte, Italy Fry (1996); Vogel (1984, 1994)
Bayless, USA
Babb and Mermel (1968); Bartholomew (1989); Bowers (1928); Engineering News (1910b, 1911); Engineering Record (1911d-e); ICOLD
(1974, 1995); Jansen (1980); Leger et al (1997); Rao (1960); Scott and Von Thun (1993); Smith (1972)
Bouzey, France
Babb and Mermel (1968); Baker (1897); Courtney (1897); Engineering News (1897?); Fry (1996); ICOLD (1969, 1974, 1995); Institute of
Civil Engineering (1897?); Jansen (1980); Jorgensen (1920); Leger et al (1997); Mary (1968); Rao (1960); Schuyler (1908); Smith (1972,
1995); The Engineer (1896, 1942); Vogel (1984, 1994); Wegmann (1889)
Cheurfas, Algeria
Babb and Mermel (1968); Benassini and Barona (1962); ICOLD (1969, 1974, 1995); Schuyler (1908); Smith (1972); Vogel (1994); Wegmann
(1889)
Chickahole, India ICOLD (1995); Lemprire et al (1997); Lemprire (1993); Murthy et al (1979); Vogel (1994)
El Gasco, Spain
Babb and Mermel (1968); Berga (1997); Gomez Laa et al (1979); ICOLD (1984); Jorgensen (1920); Schnitter (1994); Schuyler (1908); Smith
(1972)
Elmali I, Turkey Babb and Mermel (1968); ICOLD (1974, 1995); Vogel (1994); Yildiz and zcek (1994)
Elwha River, USA Babb and Mermel (1968); Engineering Record (1912); ICOLD (1974, 1984, 1995); Reineking. (1914); USCOLD (1975); Vogel (1994)
Fergoug I & II, Algeria Babb and Mermel (1968); ICOLD (1969, 1974, 1995); Lemprire (1993)
Gallinas, USA
Babb and Mermel (1968); Engineering News Record (1957); ICOLD (1974, 1984, 1995); Lemprire (1993); Sherman (1910); USCOLD
(1975); Vogel (1994)
Page 2.7
Dam References
Gleno, Italy
Babb and Mermel (1968); Bowers (1928); Coutinho Rodrigues (1987); Engineering News Record (1924a-c); Gruner (1963); ICOLD (1974,
1984, 1995); ITCOLD (1967); Mary (1968); Smith (1972); Vogel (1984, 1994)
Granadillar, Spain Berga (1997); Gomez Laa et al (1979); ICOLD (1984)
Habra, Algeria
Babb and Mermel (1968); Gruner (1963); Gurtu (1925); Jansen (1980); Jorgensen (1920); Leger et al (1997); Rao (1960); Schuyler (1908);
Smith (1972); Vogel (1994); Wegmann (1889)
Hauser Lake II, USA Bowers (1928); ICOLD (1974, 1995); Jorgensen (1920); Rouve et al (1977); Sizer (1908); USCOLD (1975)
Idbar, Yugoslavia ICOLD (1974, 1984, 1995); Mary (1968); Milovanovitch (1958)
Khadakwasla, USA
Babb and Mermel (1968); Biswas and Chatterjee (1971); Gruner (1967); Hunter (1964); ICOLD (1969, 1974, 1995); INCOLD (1967); Inglis
(1962); Jansen (1980); Lemprire (1993); Murthy et al (1979); Murti (1967); Rao (1967); Vogel (1994)
Kohodiar, India ICOLD (1995)
Komoro, Japan Babb and Mermel (1968); ICOLD (1974, 1984, 1995); Vogel (1994)
Kundli, India Babb and Mermel (1968); ICOLD (1969, 1974, 1984, 1995); Rao (1960); Vogel (1994)
Lake Lanier, USA
Babb and Mermel (1968); Bowers (1928); Coutinho Rodrigues (1987); Engineering News Record (1926a-b); Feld (1968); ICOLD (1974,
1984, 1995); Mary (1968); Rao (1960); Schnitter (1994); Scott and Von Thun (1993); USCOLD (1975); Veltrop, J.A. (1988); Vogel (1994)
Leguaseca, Spain Berga, L. (1997); Fry, J. (1996); Guerreiro et al (1991); ICOLD (1995); Lemprire et al (1997)
Lower Idaho Falls, USA ICOLD (1984, 1995); USCOLD (1988); Vogel (1994)
Lynx Creek, USA Babb and Mermel (1968); ICOLD (1974); Vogel (1994)
Malpasset, France
Babb and Mermel (1968); Bellier et al (1976); Biswas and Chatterjee (1971); Carlier (1974); Commission Administrative d'Enqute - France
(1965a-b); Commission de Contre-Expertise - France (1966); Coutinho Rodrigues (1987); Engineering News Record (1959, 1960a-b, 1962,
1963, 1964a-b, 1967); Flagg; Gosselin (1960); Gruner (1963, 1967); ICOLD (1974, 1984, 1995); Jansen (1988); Londe (1977); Mary (1968);
Rao (1960); Scott and Von Thun (1993); Smith (1972); Stapleton (1976); Steger and Unterberger (1990); Terzaghi (1962); Vogel (1984,1994)
Meihua, China Coutinho Rodrigues (1987); ICOLD (1995)
Mohamed V, Morocco ICOLD (1984)
Moyie River, USA
Babb and Mermel (1968); Bowers (1928); Coutinho Rodrigues (1987); Engineering News Record (1926b); Feld (1968); ICOLD (1974, 1984,
Page 2.8
Dam References
1995); Mary (1968); Rao (1960); Schnitter (1994); Scott and Von Thun (1993); USCOLD (1975); Veltrop (1988); Vogel (1994)
Overholser, USA ICOLD (1974, 1984, 1995)
Pagara, India Babb and Mermel (1968); ICOLD (1969, 1974, 1984, 1995); Rao (1960); Vogel (1994)
Puentes, Spain
Anderson and Trigg (1976); Babb and Mermel (1968); Berga (1997); Gomez Laa et al (1979); Hinds (1953); ICOLD (1974, 1984, 1995);
Jansen (1980); Jorgensen (1920); Mary (1968); Schnitter (1994); Schuyler (1908); Smith (1972); Vogel (1984, 1994); Wegmann (1889)
Santa Catalina, Mexico Babb and Mermel (1968); Lemprire (1993); Schuyler (1906); Vogel (1994)
Selsford, Sweden ICOLD (1974, 1984, 1995)
Sig, Algeria Babb and Mermel (1968); ICOLD (1969, 1974, 1995); Vogel (1994)
St Francis, USA
ASCE (1929); Babb and Mermel (1968); Bowers (1928); Clements (1969); David Rogers (1995); David Rogers and McMahon (1993);
Engineering News Record (1928a-j, 1929a-b); Feld (1968); Gruner (1963); Grunsky and Grunsky (1928); ICOLD (1974, 1984, 1995); Jansen
(1980, 1988); Jorgensen (1928); Leger et al (1997); Outland (1963); Ransome (1928); Rao (1960); Scott and Von Thun (1993); Smith (1972);
Stapleton (1976); The Engineer (1928a-b); USCOLD (1975); Veesaert, C. (198?); Vogel (1994)
Stony River, USA
Bowers (1928); Engineering News (1914a-c); Engineering Record (1914); Finch (1914); ICOLD (1974); Jorgensen (1920); Rao (1960)

Tigra, India
Babb and Mermel (1968); Gurtu (1925); ICOLD (1969); ICOLD (1974); ICOLD (1984); ICOLD (1995); Jansen (1980); Lemprire (1993);
Rao (1960); Vogel (1994)
Torrejon Tajo, Spain Gomez Laa et al (1979); ICOLD (1984)
Vega de Tera, Spain
Babb and Mermel (1968); Berga (1997); Bollo (1965); Engineering News Record (1959b, 1960c); Gomez Laa et al (1979); Gruner (1963,
1967); ICOLD (1974, 1984, 1995); Rao (1960); Vogel (1984, 1994)
Xuriguera, Spain Berga (1997); Fry (1996); ICOLD (1995)
Zerbino, Italy
Babb and Mermel (1968); Engineering News Record (1935); Fry, J. (1996); ICOLD (1974, 1984, 1995); ITCOLD (1967); Lemprire (1993);
Vogel (1984, 1994)
2.2.2 CONGDATA Layout
The database was created using Microsoft Access for Windows 95 Version 9.0 (Access).
The information was grouped under the following categories:
General description;
incident Details;
dimensions;
geology;
hydrology; and
references.
A list of the parameters entered into the database is given in Table 2.2.
Queries were developed in Access to analyse the data. Tables were then linked to
Microsoft Excel for Windows 95 (Excel) spreadsheets for further analysis and
interpretation. Excel was used to graph the various parameters.
Table 2.2. CONGDATA parameters
Variable Description Codes
ID Identification number
Significant Incident? (Y/N) Whether incident is significant
Dam Name Name of dam
Country Country dam is in
Alternate Name 1 Other name
Dam Type Type of dam Section 2.2.3.2
Year Commissioned Year dam commissioned
Fail Acc Year of incident
Failure/Accident Incident Category Section 2.2.3.1
Fail-Time Time to incident Section 2.2.3.3
CDR Time Time to incident Section 2.2.4.1
Fail-Type Where incident occurred
Fail-Mode How incident occurred
Cause A-E Causes of incident Tables 2.2-2.6
Detection Method Method of detecting incident
Fail-Comments Comments about incident
Remedial Measures A-D Methods of remediation Table 2.7
Concrete/ Masonry Type of concrete/masonry
Foundation Whether foundation soil/rock Section 2.2.3.5
Geology A-C Types of foundation geology Section 2.2.4.4
Geology Comments Comments about foundation
H
lf
(m) Height above lowest foundation Figure 2.1
H
d
(m) Structural height Figure 2.1
h
wu
(m) Height of water upstream - FSL Figure 2.1
h
wt
(m) Height of water at toe Figure 2.1
H
f
(m) Height to failure plane Figure 2.1
h
wf
(m) Height of water at failure Figure 2.1
W (m) Width of dam base Figure 2.1
W
f
(m) Width at failure plane Figure 2.1
Width of Spillway (m) Length of spillway
Width of Non-Overflow Section (m) Crest length - spillway
Width of Failed Section Length of failed section
Where Failed Location of failure
Upstream (xH:1V) Upstream slope of dam Figure 2.1
Downstream (yH:1V) Downstream slope of dam Figure 2.1
Valley Shape, L1 (m) Crest length Figure 2.2
L2 Left abutment length Figure 2.2
Table 2.2. CONGDATA parameters
Variable Description Codes
L3 Main valley width Figure 2.2
L4 Right abutment length Figure 2.2
Radius of Curvature (m) Radius of curvature of dam
Warning Type 1-3 Type of warning given Section 2.2.4.11
Warning Time (weeks) Time from warning to incident
Post-Tensioned? (Y/N) Whether post-tensioned
Gallery (Y/N) Whether there is a gallery
Gallery Elevation Height to gallery from dam base
Drain Depth (m) Depth of drains into foundation
Drain Spacing (m) Spacing of drains along dam
Shear Key (Y/N) Whether there is a shear key
Grouting Type Type of grouting
Grout Depth Depth of grouting into foundation
No. of victims Number of deaths due to incident
References: Main references
2.2.3 Data Entered into CONGDATA
Details of dam incidents are given in CONGDATA. The following sections describe the
coding used for input into Access.
2.2.3.1 Definitions of Failures/Accidents
ICOLD(1995) define failure as collapse or movement of part of a dam or part of its
foundation, so that the dam cannot retain water. In general, a failure results in the
release of large quantities of water, imposing risks on the people or property
downstream. ICOLD(1974) give the following definitions for failures and accidents.
F1 - A major failure involving the complete abandonment of the dam
F2 - A failure which at the time may have been severe, but yet has permitted the
extent of damage to be successfully repaired and the dam again brought into use
A1 - An accident to a dam which has been in use for some time but which has been
prevented from becoming a failure by immediate remedial measures, including
possibly drawing down the water
A2 - An accident to a dam which has been observed during the initial filling of the
reservoir and which has been prevented from becoming a failure by immediate
remedial measures, including possibly drawing down the water
A3 - An accident to a dam during construction, i.e. by settlement of foundations,
slumping of side slopes etc., which have been noted before any water was
impounded and where the essential remedial measures have been carried out, and
the reservoir safely filled thereafter.
The term incident is used to describe failures, accidents and major repairs.
USCOLD(1988) give the following definitions for other accidents and deteriorations.
These have been adopted for the database.
AR - Accidents or unusual problems encountered in the reservoir upstream of the
dam, which have occurred during operation of the project, but which have not
caused failure or major accident to the dam structure.
MR - Extensive or important repairs to a dam that were required because of
deterioration or to update certain features. Refacing of deteriorated concrete,
repair of deteriorated riprap, or replacement of gates are examples under this
definition.
DDC - Damage to partially constructed dam or to temporary structure required for
construction prior to the dam being essentially completed. Failure of cofferdam or
unplanned overtopping of partially completed dam are examples under this
definition.
Where the exact definition of the failure or accident is uncertain an F or an A has
been used respectively.
The term significant incident has been introduced to describe failures, accidents and
major repairs where the incident has directly affected the dam stability. Cases where the
dam has been repaired due to a theoretical danger, such as the updating of design
standards, or due to minor damage to the dam or spillways have not been included under
this term.
2.2.3.2 Types of Dams
Coding for the types of dams in CONGDATA are:
PG
CB
VA
MV
PG(M)
CB(M)
VA(M)
MV(M)
- Concrete gravity
- Concrete buttress
- Concrete arch
- Concrete multi-arch
- Masonry gravity
- Masonry buttress
- Masonry arch
- Masonry multi-arch
2.2.3.3 Failure Types
Codes for the failure types were obtained from ICOLD(1983) and are:
Ff
Fm
Fb
Fa
Ffb
Ffa
Fba

Fbm
- Failure due to the dam foundation
- Failure due to the dam materials
- Failure due to the structural behaviour of the dam body
- Failure due to the appurtenant works
- Failure due to the foundation and to the structural behaviour of the dam body
- Failure due to the foundation of the dam and to the appurtenant works
- Failure due to the structural behaviour of the dam body and to the appurtenant
works
- Failure due to the structural behaviour of the dam body and to the dam materials
2.2.3.4 Incident Time
The times at which the incident took place (or was detected) are indicated by the codes
below. These codes were obtained from ICOLD(1983). In Section 2.2.4.1 the incident
time is further discussed and T4 and T5 are redefined.
T1
T2
T3
T4
T5
- During construction
- During the first filling
- During the first five years
- After five years
- Not available
2.2.3.5 Type of Foundation
The foundation type was split into two categories as shown below.
R
S
- Rock mass
- Soil mass
This was further differentiated as discussed in Section 2.2.4.6.
2.2.3.6 Dam Height
Where the height of the dam (from lowest foundation) is uncertain the following
definitions from ICOLD(1983) have been used. In other cases the actual height has been
added.
H1 5m H1 < 15m
H2 15m H2 < 30m
H3 30m H3 < 50m
H4 50m H4 < 100m
H5 100m H5
H6 Not available
2.2.3.7 Detection Methods
The methods for detecting incidents and the need for major repairs were obtained from
ICOLD(1983) and are:
D01
D02
D03
D04
D05
D06
D07
D08
D09
D10
D11
D12
D13
- Direct observation
- Sampling and laboratory test
- Water flow measurements
- Phreatic level measurements
- Uplift measurements
- Pore pressure measurements
- Turbidity measurements
- Chemical analysis of water
- Seepage path investigations
- Joint and crack measurements
- Horizontal displacement measurements
- Vertical displacement measurements
- Angular displacement measurements
D14
D15
D16
D17
D18
D19
D20
D21
D22
D23
D24
D25
- Strain measurements
- Stress measurements
- Water level measurements
- Temperature measurements
- Hydrometric measurements
- Rainfall measurements
- Seismicity control
- Sounding investigation
- Water pressure measurements
- Silting measurements
- Design revision (new criteria)
- Not available

2.2.3.8 Classification of Causes of Incidents of Dams And Reservoirs
The following tables show the codes defining the types and causes of incidents and the
need for major repairs that occurred at the dams. The tables were obtained from
ICOLD(1983) with some additions from ICOLD(1995). The codes used are followed in
the database by a letter that determines their origin.
x - ICOLD(1983)
y - Not from ICOLD
- ICOLD(1995)
It will be noted that the causes are an unfortunate mi xture of physical and human
factors. They have been adopted for consistency with ICOLD data.

Table 2.3. Causes of incidents of concrete dams
1.1
1.1.1
1.1.2
1.1.3
1.1.4
1.1.5
1.1.5.1
1.1.5.2
1.1.6
1.1.7
1.1.8
1.1.9
1.1.10
1.1.11

1.1.12
1.1.13

1.1.14
1.2
1.2.1

1.2.2

1.2.3
1.2.4
1.2.5
1.2.6
1.2.7
1.2.8
1.2.9

1.2.10
1.2.11

1.2.12

1.2.13
1.3

1.3.1

- Due to foundation
- Inadequacy of site investigation
- Deformation and land subsidence
- Shear strength
- Seepage
- Internal erosion
- in foundation
- in abutment
- Degradation (including swelling)
- Initial state of stress
- Tensile stresses at the upstream toe
- Preparation of the foundation surface
- Strengthening treatment
- Grout curtains and other watertight
systems
- Drainage systems
- Sealing of galleries, shafts and
boreholes used for investigation
- Leak of drainage system
- Due to concrete
- Reactions of concrete constituents
(including alkali-aggregate reaction)
- Reaction between concrete constituents
and the environment (including
dissolution of calcium hydroxide)
- Resistance to freezing and thawing
- Attack by bacteria
- Compressive strength
- Shear strength
- Tensile strength
- Permeability
- Concreting (including order of casting
of monoliths)
- Cooling
- Structural joints (including watertight
systems)
- Arrangement of reinforcements and
anchorages
- Ageing of concrete
- Due to unforeseen actions or to
actions of exceptional magnitude (as a
principle, when the case does not fall
under other headings)
- Hydrostatic pressure and from
accumulated silt (including pressure and
impact of ice in the reservoir)
1.3.2
1.3.3
1.3.4
1.3.5

1.3.6
1.3.7
1.3.7.2
1.3.7.3
1.3.8
1.4

1.4.1

1.4.2
1.4.3

1.4.4

1.4.5
1.4.6
1.4.7
1.5

1.5.1

1.5.2
1.5.3

1.5.4

1.5.5
1.5.6
1.6
1.6.1
1.7
1.7.1
1.7.2
1.7.3
1.7.4
1.7.5

2.3.9
- Uplift
- Earthquakes (natural or man-made)
- External temperature variation
- Temperature variation due to the heat
of hydration
- Moisture variation
- Overtopping
- of abutment
- of main section
- Deterioration of concrete-rock interface
- Due to structural behaviour of the
arch and multiple arch dams (including
the construction period)
- Shape of the dam and its position in the
valley
- Tensile stresses
- Stress concentration due to shape
discontinuities in the foundation surface
- Stress concentration at openings and
shape discontinuities
- Artificial abutments and foundation
- Distribution and types of joints
- Facings
- Due to structural behaviour of gravity
and buttress dams
- shape of the dam and its position in the
valley
- Tensile stresses
- stress concentration at openings and
shape discontinuities
- Facings
- Due to monitoring
- Inadequacy of instrumentation
- Due to maintenance
- Periodic inspections
- Cleaning of drains
- Control of seepage
- Pumping of seepage water
- Deterioration of instrumentation

- Failure due to an upstream dam
collapse
Table 2.4. Causes of incidents of masonry dams
3.1
3.1.1
3.1.2
3.1.3
3.1.4
3.1.5
3.1.6
3.1.7
3.1.8
3.1.9
3.1.10
3.1.11

3.1.12
3.1.13

3.2
3.2.1

3.2.2

3.2.3
3.2.4
3.2.5
3.2.6
3.2.7
3.2.8
3.2.9

3.2.10

3.3
3.3.1
3.3.2
- Due to foundation
- Inadequacy of site investigation
- Shear strength
- Seepage
- Internal erosion
- Tensile stresses at the upstream toe
- Preparation of the foundation surface
systems
- Drainage systems
- Due to mortar
- Reactions of masonry constituents
- Reaction between masonry constituents
and the environment (including
dissolution of calcium hydroxide)
- Compressive strength
- Shear strength
- Tensile strength
- Permeability
- Masonry construction (including order
of placement)
systems)
- Due to stone
- Weathering
- Joints between stones
3.4

3.4.1

3.4.2
j3.4.3
3.4.4
3.4.5

3.4.6
3.5

3.5.1

3.5.2
3.5.3

3.5.4
3.5.5
3.6
3.6.1
3.7
3.7.1
3.7.2
3.7.3
3.7.4
3.7.5

- Due to unforeseen actions or to actions
of exceptional magnitude (as a
principle, when the case does not fall
under other headings)
- Hydrostatic pressure and from
accumulated silt (including pressure and
impact of ice in the reservoir
- Uplift
- Earthquakes (natural or triggered)
- External temperature variation
- Variations due to changes of moisture
content
- Overtopping
- Due to structural behaviour of
masonry dams (including the
construction period)
- Shape of the dam and its position in the
valley
- Tensile stresses
- Facings
- Due to monitoring
- Deterioration of instrumentation

Table 2.5. Causes of incidents to appurtenant works
4.0
4.0.1
4.1

4.1.1
4.1.2
4.1.3
4.1.4
4.1.5
4.1.6
4.1.7
4.1.8
4.1.9
4.1.10

4.1.11
4.1.12

4.2
4.2.1

4.2.2

4.2.3
4.2.4
4.2.5

4.2.6
4.2.7
4.2.8
4.2.9
4.2.10

4.2.11

4.2.12
4.2.13
4.3
4.3.1
4.3.2
4.4
4.4.1
4.4.2
4.4.3
4.4.4
- I nadequate design
- Tunnels and canals
- Due to foundations (when they dont
have good dam characteristics)
- Inadequacy of site investigations
- Shear strength
- Percolation
- Internal erosion
- Preparation of foundation surface
systems
- Drainage systems
- Due to concrete
- Reactions of concrete constituents
- Reactions between concrete
constituents and the environment
(including dissolution of calcium
hydroxide)
- Mechanical strength (including tensile
strength)
- Permeability
- Concreting (cooling included)
- Cracking
- Surface finishing (facing included)
systems)
- Arrangement of reinforcements and
anchorages
- Erosion by abrasion
- Erosion by cavitation
- Due to riprap
- Disintegration of blocks
- Removal of blocks
- Due to steel and other materials
- Chemical and biological agents
- Erosion by abrasion
- Erosion by cavitation
- Mechanical strength
4.5

4.5.1

4.5.2
4.5.3
4.5.4
4.5.5
4.5.6

4.6
4.6.1
4.6.2
4.6.3
4.6.4
4.6.4.2
4.7

4.7.1
4.7.2
4.7.3
4.7.4
4.7.5
4.7.6
4.7.7
4.7.8
4.7.9
4.7.10
4.8
4.9
4.9.1

4.9.2

4.10
4.10.1
4.11
4.11.1
4.11.2
4.11.3
4.11.4
4.11.5

4.11.6
4.11.7
- Due to unforeseen actions or actions
of exceptional magnitude (when the
case doesnt fall under other headings)
- Hydrostatic pressure and pressure due
to silt accumulation
- Pressure and impact of ice
- Uplift
- Earthquakes (natural or triggered)
- Temperature and moisture variations
- Delay in construction at the time of
flood
- Due to structural behaviour
- Structural behaviour of spillways
- Insufficient capacity of spillway
- Erosion of spillway basement
- Inadequate design of spillway
- of canal or tunnel
- Due to water flow, water level and
water-borne debris (including
construction periods)
- Excessive rates of flow
- Turbulence
- Vortices
- Waves
- Abnormal pressures
- Entrapped air
- Inaccurate discharge curves
- Solid materials carried by water flow
- Discharge of floating materials
- Piping outside inserted conduit
- Due to local scour
- Due to operation
- Sudden opening of the discharge
equipment
- Inadequate instructions for operating
the discharge equipment
- Due to monitoring
- Deterioration of measurement
instrumentation
- Malfunction of discharge equipment
- Debris in stilling basins
Table 2.6. Causes of incidents of reservoirs
5.1

5.2

5.3

5.4

5.5
- Slope sliding

- Overturning of rock blocks

- Permeability

- Silting

- Ecological balance
Table 2.7. Causes of incidents downstream of dam
6.1

6.2

6.3
- Equilibrium of river bed

- Slope stability

- Ecological balance
2.2.3.9 Classification of Remedial Measures
Table 2.8 shows the coding used for remedial measures. The codes used were obtained
from ICOLD(1983).The codes used are followed by a letter which determines their
origin.
x - ICOLD(1983)
y - Not from ICOLD
- ICOLD(1995)
Table 2.8. Classification of remedial measures

R101
R102
R103
R104
R105
R106
R107
R108
R109

R201
R202
R203

R204
R205

R301
R302
R303
R304
R305
R306
R307
R308
R309

R401
R402

R403
R404
R405
R406
R407

R408

- Of a general nature
- Investigation
- Monitoring
- Lowering of reservoir level
- Raising of dam crest
- Overall reconstruction (same design
- Reconstruction with new design
- None
- Not available
- Scheme abandoned
- I n foundations
- Water tightening treatment
- Drain & filter construction or repair
- Strengthening by grouting or other
methods (excluding anchoring)
- Filling in of fractures and cavities
- Anchoring
- I n concrete and masonry dams
- Water tightening treatment
- Drain construction or repair
- Thermal protection (excluding facing)
- Facing
- Reconstruction of deteriorated zones
- Execution of joints
- Strengthening by grouting
- Strengthening by anchoring
- Strengthening by shape correction
- I n earth and rockfill dams
- Impervious core repair
- Construction or repair of other watertight
systems
- Drain & filter construction or repair
- Slope protection construction or repair
- Filling in of cracks and cavities
- Upstream slope flattening, construction of
upstream berm or other stabilisation
methods
- Downstream slope flattening, construction
of downstream berm or other stabilisation
methods

R501
R502

R503

R504

R505

R506

R507

R508
R509
R510
R511

R512

R513

R514

R515

R601
R602
R603
R604

R605
R606
R607

R701
R702

- I n appurture works
- Discharge increase
- Construction of additional
appurtenant work
- Overall reconstruction of
appurtenant works
- Partial reconstruction with
strengthening or structural changes
- Shape correction of surfaces
contacting flow
- Aeration devices: construction or
increase of capacity
- Repair of surfaces contacting flow
(including facings and special
treatments)
- Joint water tightening treatment
- Construction & repair of drains
- Slope protection & stabilisation
- Sediment discharge removal from
surfaces contacting flow
- Construction, modification and
repair of valves and gates
- Establishment and updating of rules
for gate and valve operations
and other correcting measures
- Abandon of appurtenant work
- I n reservoir
- Reforestation
- Torrent training
- Sediment discharge diversion
- Slope regularisation, protection and
strengthening
- Draining
- Water tightening
- Dredging
- Downstream of dam
- Draining
- Slope regularisation, protection and
strengthening
2.2.4 Selection of Additional Variables
As the ICOLD (1974, 1983 and 1996) databases were limited in their scope it was
decided to add additional variables to CONGDATA. The additional variables were
generally proposed by the author and reviewed by the sponsors. Further variables were
added where requested by the sponsors. Some potential variables were rejected due to
limited information in the literature, reports etc. Following is a description of the
additional data variables including a discussion of why each was chosen.
2.2.4.1 Time of Incidents
It is important to understand at what age dams are more likely to fail or experience
accidents. This can give dam owners a guide as to what intensity of monitoring they
need to have throughout the life of a dam. ICOLD (1983) have analysed the time to
failure and grouped their data into categories T1 to T5 as shown in Section 2.2.3.4. The
oldest group is T4, which indicates an incident occurred after five years. It is clear that
this is a large category that cannot adequately indicate potential deterioration effects in
dams. The following grouping was used to allow for a better distribution, and hence
understanding, of times to failure.
T1 - During construction
T2 - During first fill
T3 - 0-5 years
T4 - 5-10 years
T5 - 10-20 years
T6 - 20-30 years
T7 - 30-40 years
T8 - 40-50 years
T9 - >50 years
T10 - >5 years (else unknown)
T11 - Unknown

2.2.4.2 Foundation Incident Mode
Where the foundation has played a part in the incident of the dam further codes have
been added. This allows for a better understanding of the foundation parameters
affecting different incident modes. The codes are:
S - Sliding - where failure has occurred by the dam sliding on the foundation. Sliding
can be along the dam-foundation interface or along a foundation discontinuity.
P - Piping - of materials within soil foundations or rock discontinuities (generally
joints).
SC - Scour - of the foundation or the abutment.
U - Uplift - in the foundation.
D - Deformation - settlement or other movements of the foundation not including
sliding.
L - Leakage - beneath the dam or through the abutments.

2.2.4.3 Dam Incident Mode
Where the incident occurred in the dam the following codes have been used to define
the incident mode.
SH - Shear (sliding) within the dam.
T - Tensile (overturning) within the dam.
C - Compressive failure within the dam.
CR - Cracking (due to concrete hydration etc.)
ST - Structural damage to appurtenant structure such as spillway gates.
LD - Leakage - through dam.
EQ - Earthquake damage.
2.2.4.4 Comments on Incidents
The causes of incidents as given in Section 2.2.3.8 are often too general to explain the
type of incident. A brief description of the incident has been included in the database to
allow for a better understanding of the causes of the incident.
2.2.4.5 Description of the Failure or Accident
Brief descriptions of the failure or accident and warning are included in the database.
2.2.4.6 Additional Geological Information
Previous dam failure databases have only listed the foundation as soil, rock or both. The
dam geology has been included in the database in an attempt to determine whether
certain foundation geology types are more susceptible to incidents and vice-versa. The
geology of each dam was categorised into the following categories:

Foundation Geology Categories - Rock
Sedimentary Metamorphic Igneous
Conglomerate Gneiss Granite
Sandstone Schist Gabbro
Mudstone Phyllite Rhyolite
Shale Slate Andesite
Siltstone Marble Basalt
Claystone Quartzite
Limestone Hornfels
Dolomite
Chalk
Agglomerate
Volcanic Breccia
Tuff
Saline Rocks
Coal
Lignite

Foundation Geology Categories - Soil
Alluvial Aeolian Marine
Lacustrine Colluvial Volcanic (ash)
Glacial Residual

Unfortunately this detail is often not available, so the database is incomplete.

2.2.4.7 Dam Dimensions
The databases that have been developed previously included the height of the dam
(taken as above the lowest foundation for ICOLD) and crest length. These are
insufficient to fully describe the dam. To allow for the determination of gradients and
performing simple analyses of some of the dam incidents, further dimensions were
included in the database. The height to full supply level (FSL), tailwater height and the
water height at failure were included. These are shown in Figure 2.1 and listed below.
All heights, excluding H
lf
, have the general foundation level of the dam as their
reference level.
H
lf
- Height of dam above lowest foundation
h
d
- Structural height
h
wu
- Reservoir height at full supply level (FSL)
h
wt
- Height of the tail water
W - Base width of dam section
h
f
- Height to failure plane (=0 if in foundation)
h
wf
- Reservoir height at failure
W
f
- Width of failure plane
xH:1V - Upstream slope
yH:1V - Downstream slope
The drain depth, gallery height and length of spillway were also included in the
database. The extent of each failure was also seen as important and so the length of the
failed section and where the dam failed (spillway section/non-overflow section/both)
were also included in the database.
2.2.4.8 Valley Shape
Stress concentrations and differential movements can occur at changes of section. This
is particularly important with sharp section changes in the foundation. For this reason a
method was developed to assess the valley shape. The parameters given below are
shown in Figure 2.2.
L1 - Crest length
L2 - Left abutment length
L3 - Length of valley section
L4 - Right abutment length
2.2.4.9 Radius of Curvature
A dam will have increased stability where there is some curvature in the dam and load is
transferred to the dam abutments. The database includes the radius of curvature of the
dam. For dams with straight axes the radius is shown as straight.


Figure 2.1. Definition of dimensions in CONGDATA
h
f
H
lf
h
w
h
d

W
f
h
wt

Drain
Depth
Gallery
Height
W
1V
yH xH
1V
Full Supply
Level
Dam Crest
Failure Surface
(where applicable)
Gallery
Dam
Foundation
Drain
Holes
Grout
Holes
Normal
Tailwater
Level
Shear
Key
Water Level at Failure
h
wf


Figure 2.2. Definition of dimensions in CONGDATA - section across river
2.2.4.10 Monitoring and Surveillance Data
In some cases there have been signs of displaceme nt, cracking, seepage and other
factors prior to the incident, giving some warning. These have been included in the
database as:
0 - None observed
1 - Foundation piping
2 - Foundation leakage
3 - Dam leakage
4 - Horizontal displacements
5 - Vertical displacements
6 - Cracking
7 - Expansion & cracking
8 - Concrete deterioration
9 - Scour of the foundation
11 - Overtopping
12 - Slide downstream of dam
13 - Abnormal uplift development
14 - Unknown
A brief description of the warning is also included. This allows some quantification of
the warning e.g. the amount of leakage, and time before failure.

L1
L2 L3 L4
Dam Crest
Slope
Change
Slope
Change
Dam Foundation
2.2.4.11 Warning Rating
The following qualitative codes were used to show whether there was sufficient warning
prior to the failure to allow for preventative measures and/or warning of people
downstream.
Y - Yes
M - Maybe
N - No
F - Flood
DF - Dam failure upstream
? - No data
2.2.4.12 Warning Time
The time from when a warning was given to when the dam failed or when an accident
occurred and the dam was remediated was recorded as the warning time.
2.2.4.13 Other Design Factors
(a) Post-Tensioning
Whether the dam was post-tensioned was included to assess the effects of post-
tensioning dams.
(b) Gallery
The presence of a gallery allows for better maintenance and uplift pressure relief and the
provision or otherwise of a gallery is included in the database.
(c) Drain Depth and Spacing
Drain depths and spacing were included in the database to assess the effects of reducing
uplift pressure on dam stability.

(d) Shear Key
A shear key may increase the resistance of a dam to sliding and the presence of the key
is included.
(e) Grouting Type and Depth
Consolidation and/or grout curtains can be used to improve the stability of dam
foundations and to reduce uplift pressures. The presence of, depth and spacing are
included in the database.
(f) Number of Victims of Dam Failures
This was included to crudely assess the hazard of the dam. It is possible that high hazard
dams may have a lower chance of failure as they have better maintenance and higher
factors of safety in design.
2.2.5 Assumptions Made in Assembling the Database
The majority of the information in CONGDATA has been derived from ICOLD
(1974,1983 and 1995). The ICOLD data was collated by sending questionnaires to the
various National Committees. This method of data collection caused several problems
(ICOLD, 1995).
Some failures were not reported due to a lack of response from some National
Committees.
Replies from National Committees were not consistent with each other - some
committees calling incidents failures where others would call them accidents.
Gate failure was included by some committees whilst others did not include them. It
was ICOLD (1995) policy not to include gate failures.
The data from China was inconsistent with the rest of the world. China has the same
amount of dams as the rest of the world put together yet has only reported 3 dam
failures as opposed to 180 for the world. When comparing similar construction
periods (post-1955) this becomes 3 failures as opposed to 50. It was ICOLD (1995)
policy to ignore China when performing their statistical analyses. This policy has
also been adopted here.
When assessing the ICOLD data more specific inconsistencies were found in the
following:
Dam type - Where failures occurred in composite structures (e.g.
embankment/concrete gravity) some National Committees listed the dam as a
composite structure (TE/PG) whereas others listed only the section of the dam that
failed (e.g. TE). It is important when analysing dam failures that the section that
failed be identified so that misleading conclusions are not made. Dams where failure
occurred only through the embankment section were discarded in the preparation of
CONGDATA.
Height - When comparing ICOLD data to that of other reports/papers/drawings etc.
inconsistencies became apparent in the assigning of heights to each dam. Where
possible the data was changed to what was understood to be the accurate height.
Where corroborating information was not available the ICOLD heights were
assumed.
Length - Similar inconsistencies to the height category were found here. Attempts
were also made to determine the crest length of the failed section.
Year - generally the years of construction and incident were found to be accurate.
Some small inconsistencies (1-2 years) were found in old dams. There were some
errors found in the accidents.
Foundation - In ICOLD some dams are noted as having soil/rock foundations. Where
possible it was determined where the failure occurred and which foundation type
played a part. Where there was no other information the ICOLD foundation was
assumed.
Failure type and cause - It appears that most of the ICOLD causes were chosen by
the individual National Committees (and potentially smaller dam owners that the
questionnaires were passed on to). There appears to be a bias as to which failure
categories each country chooses. This has resulted in marked inconsistencies in the
ICOLD causes. It is also often difficult to assess how a dam failed by the failure
category alone. An attempt has been made to assess all the dams in CONGDATA
independently. However, often the ICOLD data is the only available. Failure types
were found to be misleading in several dams and have been corrected.
Remediation measures - Similar problems arise here as for failure type. However,
many of the failed dams have been abandoned and so the effects are minimal.
Many of the causes of incidents in CONGDATA are subjective but they have been
chosen with as much care as possible from the references available.
Where several sources have been found with conflicting information an attempt has
been made to select the most credible source. Most of the dams with most uncertainty
are the older dams (prior-1950s).
It should be remembered that many of the failures occurred a long time ago and hence
information is scarce.

2.2.6 Data on the Population of Dams
The assessment of dam incident statistics is of value to the dam engineering community.
With these statistics engineers can see which dams have had more dam incidents than
other types. This method of analysis however can lead to incorrect conclusions. For
example, from a cursory assessment of the failure statistics for concrete and masonry
and embankment dams it is shown that there is many more embankment failures
compared with concrete and masonry dams. This could lead to the assumption that an
embankment dam is much more likely to fail than a concrete dam. If the analysis is
continued by comparing the failure statistics to the total population of existing dams
then it is shown that the percentage of failures for each dam type is roughly the same
(ICOLD, 1995).
ICOLD (1995) was the first to attempt to produce statistics on failures taking into
account the number and type of existing dams. The population data was taken from the
ICOLD World Register of Dams (1984 edition and 1988 updating). ICOLD (1995)
compared statistics on existing and failed dams for their type, height and year
commissioned. The results of the analyses assisted in qualifying many assumptions that
were made on the basis of incident statistics alone.
The assessment of the incidents in CONGDATA needed to be qualified with dam
population data. ICOLD (1995) used a computerised version of the World Register of
Dams that was unavailable to the author. To overcome this, the populations of dams in
countries where either a failure had occurred or there was a large number of
concrete/masonry dams were entered into a database to use for basic comparisons with
the incident data. The table below shows the breakdown of the 4168 dams from 22
countries that were used.
Population of Dams from World Register of Dams used for Analysis
Country Gravity Arch Buttress Multi-
Arch
Total
Algeria 5 1 4 10
Australia 69 39 10 3 121
Austria 23 15 38
Brazil 86 3 9 4 102
Canada 190 6 19 2 217
France 130 85 11 12 238
Great Britain 95 11 14 1 121
India 146 146
Italy 208 65 24 8 305
Japan 536 44 17 3 600
Mexico 101 6 3 1 111
Morocco 11 4 15
Norway 26 38 42 3 109
New Zealand 13 19 2 34
Portugal 27 19 4 1 51
South Africa 95 59 7 15 176
Spain 546 30 23 4 603
Sweden 12 5 27 44
Switzerland 51 48 3 102
Turkey 12 1 1 14
USA 717 169 46 25 957
Yugoslavia 30 19 1 4 54

CONGDATA included many more variables for each dam incident than is included in
the World Register. A major component of this chapter is an assessment of the
foundation geology type that ICOLD does not assess. It was therefore assessed that the
population of dams needed to come from sources other than the World Register.
The ideal statistical analysis would be made on the total population of dams however
this would be impossible to collect. A compromise was made where large subsets of the
world population were chosen. The populations chosen and the reasons why are given
below.

Australia/New Zealand - This population was chosen for a number of reasons. The
dam population is large and covers numerous geology and topography types. The
sponsors of the project comprised the major dam owners in the two countries and
hence access to data was made easier. It was also important to make sure the project
produced results that could be used by the sponsors in Australia and New Zealand.
Appendix C provides a listing of the dams used.
USBR - The USBR has been involved with a large number of dams that cover the
western half of the USA. This population covers a wide area and hence a wide range
of geology and topography. It was also seen as important to include a population
from the country with the highest number of reported incidents. Another major factor
was the free access to data that the USBR gave the author. Information on the dams
was also available from USBR(1996). The list of dams used for the population is
given in Appendix C.
US National Inventory of Dams - This computerised database comprised 1049 large
concrete and masonry dams. Such statistics as foundation geology were not included.
The inventory instead allowed assessment of the basic variables of dam type, age and
height in the country with the greatest number of reported failures.
Portugal - Due to the easy access to the LNEC(1992) report on the Internet this
population was also assessed. An attractive feature of this population was the
inclusion of foundation geology types in a country with a much different geological
environment (generally igneous and metamorphic). This population is shown in
Appendix C.
The populations of dams from the US National Inventory of Dams and Portugal were
collated directly from the CD-ROM and the Internet respectively. The author also
collected information from the USBR offices in Denver. Further information was taken
from the Internet, personal communication with USBR staff, journal papers and various
dam compilation reports published by the USBR and the United States Committee on
Large Dams (USCOLD). The information on the Australia/New Zealand population
was collected in person by the author and by using questionnaires sent to the sponsors
and several other dam owners. Where required additional information was collected
from journal papers and conference proceedings.
The populations chosen above have several limitations including:
Limited extent - Using subsets of the world population can limit the extent to which
the information is used. The information can be expected to be as accurate as
possible in the areas surveyed but may not be typical of other areas. Countries where
geological environments and dam design and construction methods are different to
those assessed are likely to have led to different results. It is believed that the use of
populations that cover a wide area of land and are located in the areas of most
failures has reduced potential inaccuracies.
Errors/omissions - Where data has been collected second hand there is always a
chance of inconsistencies. Attempts to limit these were made by providing extensive
information with the questionnaires and checking data against other references. This
problem was also limited by personal collection of a large amount of the population
data.
2.3 RESULTS OF ANALYSIS OF THE DATABASE
2.3.1 Summary of Incidents
This chapter describes the main results obtained from the analysis of the database
CONGDATA. A total of 485 dams comprising: 46 failures; 174 accidents; and 265
major repairs were entered into CONGDATA for 29 countries. Table 2.9 shows the
number of incidents by dam type in the database. Table 2.10 shows the number of
significant incidents as defined in Section 2.2.3.1. Figure 2.3 and Table 2.11 show the
distribution of reported incidents by country.
Table 2.9. Number of dam incidents in database by type
Type Failures Accidents Major
Repairs
Total Population
(1)

PG 10 44 165 219 3434
PG(M) 21 17 39 77
VA 3 85 22 110 808
VA(M) 3 0 0 3
CB 4 8 30 42 316
CB(M) 3 1 2 6
MV 2 17 6 25 105
MV(M) 0 2 1 3
Total 46 174 265 485 4663
Note (1) ICOLD (1984) world population excluding China.
Table 2.10. Number of significant dam incidents in database by type
Type Failures Accidents Major
Repairs
Total Population
(1)

PG 10 38 52 100 3434
PG(M) 21 15 19 55
VA 3 85 1 89 808
VA(M) 3 0 0 3
CB 4 8 11 23 316
CB(M) 3 1 0 4
MV 2 15 0 17 105
MV(M) 0 2 0 2
Total 46 164 83 293 4663
Note (1) ICOLD (1984) world population excluding China.
Figure 2.4 shows the dam incidents as a percentage of the total population of dams in
each country. Algeria shows a large proportion of failures to their dams. The population
in Algeria was taken as 14 (those in existence in 1983 plus those that failed). Due to
their small populations Morocco and Turkey show high percentages of failures. India
(5%) is noticeable particularly for its larger population. The USA has a failure rate of
approximately 2%.
Unfortunately many of the variables for each dam remained unknown due to a lack of
published information. This was often due to insufficient reporting of old dam incidents.
To simplify the analysis, and improve the quality, the nature of the accidents and major
repairs to dams were initially assessed to see if the incident was likely to lead to failure
of the dam. These incidents were then denoted as significant, a term which is used in
some of the results in this chapter. It would appear likely that the few numbers of major
repairs in some countries might be due to inadequate data rather than the absence of
major repairs.

Table 2.11. Number of dam incidents reported in each country
Country Failures Accidents Major Repairs Total Population
(1)

(No. of Cases) (No. of Cases) (No. of Cases) (No. of Cases) (No. of Dams)
Algeria 7 1 0 8 14
Australia 0 4 21 25 121
Austria 0 5 3 8 89
Brazil 0 2 2 4 121
Cameroon 0 1 0 1 2
Canada 0 1 9 10 219
Chin
(2)
1 2 1 4 1290
Czechoslovakia 0 0 3 3 47
Finland 0 0 1 1 13
France 2 18 24 44 296
Germany 0 1 2 3 53
Great Britain 0 2 1 3 121
India 6 12 1 19 128
Ireland 0 0 1 1 8
Italy 3 21 57 81 327
Japan 1 4 11 16 703
Mexico 1 0 0 1 159
Morocco 1 0 0 1 18
New Zealand 0 0 1 1 38
Norway 0 1 0 1 108
Portugal 0 4 3 7 47
Rhodesia 0 6 3 9 19
South Africa 0 8 2 10 180
Spain 6 16 11 33 568
Sweden 1 0 0 1 45
Switzerland 0 10 4 14 106
Turkey 1 0 0 1 14
USA 16 56 102 174 754
Yugoslavia 0 1 2 3 58
TOTAL 46 176 265 487 5662
Note (1) Population from ICOLD (1984).
(2) Chinese dams excluded in statistical analysis.

0 5 10 15 20 25 30 35 40
Algeria
Australia
Austria
Brazil
Cameroon
Canada
China
Czechoslovakia
Finland
France
Germany
Great Britain
India
Italy
Japan
Mexico
Morocco
New Zealand
Norway
Portugal
Rhodesia
South Africa
Spain
Sweden
Switzerland
Turkey
USA
Yugoslavia
Percent of Total
Failure
Accident
Major
Repair

Figure 2.3. The distribution of reported dam incidents vs country

0 10 20 30 40 50
Algeria
Austria
Cameroon
China
Finland
Germany
India
Japan
Morocco
Norway
Rhodesia
Spain
Switzerland
USA
Percentage of Total
Failure
Accident
Major Repair

Figure 2.4. Reported incidents as percentage of countrys dam population from
ICOLD (1984)
2.3.2 Year Commissioned of Dams Experiencing Incidents
Figure 2.5 and Figure 2.6 show the year dams were commissioned (broken into
decades) for concrete gravity dam and masonry gravity dam incidents respectively.
There were a total of 10 failures, 44 accidents and 165 major repairs for concrete gravity
dams and 21 failures, 17 accidents and 39 major repairs for masonry gravity dams. Due
to their age, it is considered likely that masonry gravity dam accidents and major repairs
are less likely to have been reported to ICOLD.
Concrete gravity dam failures occurred in dams commissioned in the 1900s through to
the 1920s. No failures occurred in dams commissioned between 1926 and 1963. Three
concrete gravity dam failures occurred in the 1960s. There was a similar lack of
failures in masonry gravity dams commissioned between 1930 and 1966. These periods
of no failures are likely to be a function of the number of dams built and improvement
in the understanding and construction of dams.
Figure 2.7 shows the year commissioned for all dam incidents. This shows failures and
accidents to dams commissioned in the 1930s and 1940s dropping off. This follows a
similar trend to the world population shown in Figure 2.8.
The ICOLD World Register data does not allow for the separation of concrete and
masonry gravity dams. The USA population of dams (FEMA, 1995) was used to give a
rough estimate of this separation. Figure 2.9 shows the year commissioned for concrete
and masonry gravity dams in the USA. It should be noted that the USA data has been
collated from dam owner responses and there is the chance that some dams have been
denoted as concrete where in fact they were masonry. The peak in construction of
masonry dams correlates reasonably with the peak in masonry gravity dam incidents
(Figure 2.6). Peaks in dam commissioning were noted in the 1880s and 1910s. Peaks
in failures of masonry gravity dams are noted in dams commissioned in the 1870s to
1890s and 1910s to 1920s.
The graphs show that there were more incidents to dams commissioned in the 1910s,
1920s, 1950s and 1960s. However this appears to follow the trend in construction of
dams. The incident numbers are likely to be partly a function of the number of dams
built as well as design or construction deficiencies in these periods. The number of
accidents and major repairs drops off prior to 1920 but this is very likely to be due to the
way the data was collected. There is a much higher chance of having details of failures,
from the period prior to 1900, than accidents.
Figure 2.10, Table 2.12 and Table 2.13 compare the failure and accident statistics with
those of the population of dams as at 1983. The percentages refer to each subset (year
commissioned and dam type). Generally there was a reduction in the number of failures
per population with time. A small rise in the failure rate can be seen in the 1950s and
1960s. There are a number of various peaks in the percentage of failures for buttress
and multi-arch dams, but there are too few incidents to make definitive judgements on
this.
To 1983 only
0 5 10 15 20 25 30 35 40
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
Decade
Beginning
Percentage of Total Incidents
Failure
Accident
Major Repair

Figure 2.5. Year commissioned vs concrete gravity dam incidents
To 1983 only
0 5 10 15 20 25 30 35 40 45
<1800
1800-1859
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
Decade
Beginning
Failure
Accident
Major Repair

Figure 2.6. Year commissioned vs masonry gravity dam incidents

To 1983 only
0 5 10 15 20 25 30
<1800
1800-1849
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
Decade
Beginning
Failure
Accident
Major Repair

Figure 2.7. Year commissioned vs all dam incidents

Population to 1978
0 5 10 15 20 25 30 35
<1800
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
Decade
Beginning
Percent of Total for Dam Type
Gravity
Arch
Buttress
Multi-Arch

Figure 2.8. Year commissioned for world population data obtained from ICOLD
(1979)
0
5
10
15
20
25
30
35
<1850 1850s 1860s 1870s 1880s 1890s 1900s 1910s 1920s 1930s 1940s 1950s 1960s 1970s 1980s >1990
Decade
%
Concrete Gravity
Masonry Gravity
All Gravity

Figure 2.9. Year commissioned vs percentage of gravity dams constructed in the
USA

0 5 10 15 20 25
<1900
1900-1909
1910-1919
1920-1929
1930-1939
1940-1949
1950-1959
1960-1969
1970-1979
Failures/Population (%)
All
Gravity
Arch
Buttress
Multi-Arch

Figure 2.10. Year commissioned - failures/population per period

Table 2.12. Year commissioned - failures vs population per period
Years Gravity Arch Buttress Multi-Arch All
NUMBER OF FAILURES
<1900 14 0 0 0 14
1900-1909 4 0 1 0 5
1910-1919 4 1 2 0 7
1920-1929 4 2 2 1 9
1930-1939 1 0 0 0 1
1940-1949 0 0 1 0 1
1950-1959 0 2 1 1 4
1960-1969 4 0 0 0 4
1970-1979 0 0 0 0 0
1980-1983 0 1 0 0 1
FAILURES/POPULATION (%)
<1900 11.2 - - - 10.3
1900-1909 3.5 - 22.4 - 3.6
1910-1919 1.9 3.1 8.9 - 2.5
1920-1929 1.1 2.0 7.2 5.3 1.7
1930-1939 0.3 - - - 0.2
1940-1949 - - 3.4 - 0.2
1950-1959 - 1.2 1.1 5.6 0.4
1960-1969 0.5 - - - 0.4
1970-1979 - - - - -
1980-1989 - 1.8 - - 0.2

Table 2.13. Year commissioned - accidents vs population per period
Years Gravity Arch Buttress Multi-Arch All
NUMBER OF SIGNIFICANT ACCIDENTS
<1900 2 1 1 0 4
1900-1909 3 2 0 0 5
1910-1919 8 6 2 4 20
1920-1929 6 11 1 3 21
1930-1939 4 6 0 2 12
1940-1949 2 6 0 3 11
1950-1959 7 28 3 4 42
1960-1969 15 21 2 0 38
1970-1979 5 4 0 1 10
1980-1983 1 0 0 0 1
ACCIDENTS/POPULATION (%)
<1900 1.6 12.8 no pop - 2.9
1900-1909 2.6 11.2 - - 3.5
1910-1919 3.7 18.5 8.9 35.8 7.1
1920-1929 1.6 11.0 3.6 15.8 4.0
1930-1939 1.2 7.6 - 17.9 2.6
1940-1949 0.6 9.2 - 24.4 2.5
1950-1959 0.9 16.1 3.4 22.3 4.0
1960-1969 1.9 9.6 2.7 - 3.4
1970-1979 1.4 5.7 - 11.2 2.2

2.3.3 Height
Figure 2.11 shows the height range distribution for all the significant incidents in
CONGDATA. The last two ranges were chosen as 150-199m and >200m. The few
dams higher than 150m were spread over a large range of heights.
The failures appear to be more prevalent in the 15-50m height range (a total of 39).
There are seven reported failures for dams of height 50-70m. No reported failures have
occurred in dams higher than 70m. Most accidents occurred in the dams in the height
range 15-60m. The same numbers of major repairs have generally taken place per 10m
height range between 15m and 80m. There is a marked drop off from 80m onwards. The
number of major repairs per 10m peaks at a height range of 40-49m.
Figure 2.12 and Figure 2.13 show the height versus number of dams for significant
incidents in concrete gravity and masonry gravity dams respectively. Proportionally
more failures, accidents and major repairs have occurred in higher dams for concrete
gravity dams than masonry gravity dams but this may simply reflect the fact that there
are fewer high masonry gravity dams.
There were no accidents reported in the range 120-199m for concrete gravity dams. No
incidents were reported for masonry gravity dams higher than 100m.
0 5 10 15 20 25
15-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
100-109
110-119
120-129
130-139
140-149
150-199
>200
Unknown
(m)
Number of Dams
Failures
Accidents
Major Repairs

Figure 2.11. CONGDATA - height ranges for all dam significant incidents
(Insert: Figure 2.14 Failures/Population (%) for comparison)
Note: No failures
for dams 70m or
higher
0 1 2 3 4 5 6
15-19
20-29
30-39
40-49
50-59
60-69
(m)
All
Gravity
Arch
Buttress
Multi-Arch

0 2 4 6 8 10
15-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
100-109
110-119
120-129
130-139
140-149
150-199
>200
Unknown
(m)
Number of Dams
Failures
Accidents
Major Repairs

Figure 2.12. CONGDATA - Height ranges for concrete gravity dam significant
incidents
(Insert: modified Figure 2.14 Failures/Population (%) for comparison)

Note: No
failures for
dams 70m or
higher
0 1 2 3 4
15-19
20-29
30-39
40-49
50-59
60-69
(m)
Gravity

0 1 2 3 4 5 6 7 8
15-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
Unknown
(m)
Number of Dams
Failures
Accidents
Major Repairs

Figure 2.13. CONGDATA - height ranges for masonry gravity dam significant
incidents
Table 2.14, Table 2.15 and Figure 2.14 show the percentage of failures and accidents of
concrete and masonry dams as a percentage of the population created from ICOLD
(1979 and 1984). The database was created from the ICOLD (1979) dam population and
extrapolated to the population in ICOLD (1984).
The data shows the ratio of failures to population does not exhibit any major trend.
There appears to be a higher percentage of failures to population in the 40-49m and 60-
69m height ranges. There is a slight trend of increasing percentage of gravity dam
failures with height. Arch, buttress and multi-arch dams are shown to be more likely to
have failures in the 15-39m height range.

Table 2.14. Percent of concrete & masonry dam fails vs population for height
Dam Height (m)
Type
15-19 20-29 30-39 40-49 50-59 60-69
Unknown
ALL
NUMBER OF FAILURES
PG 3 - 1 2 1 3 - 10
PG(M) 3 6 3 7 - 1 1 21
VA 2 - - - - 1 - 3
VA(M) 1 1 1 - - - - 3
CB 3 1 - - - - - 4
CB(M) 1 1 1 - - - - 3
MV - 1 1 - - - - 2
MV(M) - - - - - - - -
All
concrete
8 2 2 2 1 4 - 19
All
masonry
5 8 5 7 - 1 1 27
NUMBER OF FAILURES/POPULATION
(1)
(%)
Gravity 0.7 0.6 0.7 2.5 0.4 2.6 N/A 0.9
Arch 3.4 0.6 1.0 - - 1.7 N/A 0.8
Buttress 4.6 2.7 1.9 - - - N/A 2.4
Multi-
Arch
- 4.5 5.6 - - - N/A 2.0
All 1.2 0.8 0.9 1.9 0.3 2.2 N/A 1.0
Note (1) Population height ranges from ICOLD(1979) extrapolated to population in ICOLD(1984).
Table 2.15. Percent of concrete & masonry accidents vs population for height
Dam Height (m)
Type
15-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99 100-149 150-199 >200 Unknown
NUMBER OF SIGNIFICANT ACCIDENTS
PG
4 7 3 3 3 2 1 2 2 3 - 2 6
PG(M)
- 4 3 1 3 1 - 2 1 - - - -
VA
- 2 13 9 12 5 6 3 4 16 5 6 4
VA(M)
- - - - - - - - - - - - -
CB
1 1 - - 1 - - 1 - 1 1 - 2
CB(M)
- - - - 1 - - - - - - - -
MV
1 2 2 4 1 1 1 2 - - - - 1
MV(M)
- - 2 - - - - - - - - - -
All concrete
6 12 18 16 17 8 8 8 6 20 6 8 13
All masonry
- 4 5 1 4 1 8 2 1 - - - -
NUMBER OF ACCIDENTS/POPULATION
(1)
(%)
Gravity
0.5 1.1 1.0 1.1 2.5 2.0 1.2 6.7 7.1 4.7 - 35.8 N/A
Arch
- 1.2 12.5 11.8 18.8 8.3 15.3 7.9 12.8 22.5 24.8 76.6 N/A
Buttress
1.1 1.4 - - 8.9 - - 11.2 - 25.0 no pop - N/A
Multi-Arch
3.0 8.9 22.3 35.8 22.3 89.4 29.8 44.7 - - - - N/A
All
0.6 1.3 2.9 3.6 6.3 3.9 6.0 9.0 9.3 14.4 21.5 55.0 N/A
Note (1) Population height ranges from ICOLD(1979) extrapolated to population in ICOLD(1984).

Figure 2.15 shows the height distribution for the world population at 1984. There
appears to be an exponential like drop in numbers of dams per 10m height range. There
were 26.8% of large dams in the range 20-29m; dropping to 0.6% at 110-119m; and to
0.4% at 140-149m. Note that the range 15-19m had 22.7% due to the smaller height
range (5m c.f. 10m).
Note: No
failures for dams
70m or higher
0 1 2 3 4 5 6
15-19
20-29
30-39
40-49
50-59
60-69
(m)
All
Gravity
Arch
Buttress
Multi-Arch

Figure 2.14. Height of failed dams - failures/population (%)

22.7
26.8
16.7
10.2
7.1
4.9
2.9
2.4
1.6
0.9
0.6
0.5
0.6
0.4
0.6
0.3
0.8
0.0 5.0 10.0 15.0 20.0 25.0 30.0
15-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
100-109
110-119
120-129
130-139
140-149
150-199
>200
Unknown
(m)
Percent of Dams

Figure 2.15. World dams - height ranges for all concrete & masonry dams
2.3.4 Age at Failure
Figure 2.16 to Figure 2.18 show the age at failure for all, concrete gravity and masonry
gravity dams. T2 (during first filling) is the most common time for failure to occur.
Concrete gravity dams do appear to have proportionally fewer failures during first
filling than do masonry gravity dams.
The term first filling may be misinterpreted and as such a further analysis was carried
out. All failed dams were assessed to see whether they had failed at their maximum
water level, and whether this was the first time such a level had been reached. The
results of this analysis are given in Table 2.20.
Of the 46 dams assessed 29 had failed at their highest level ever recorded; four were not
at the highest level recorded; and there were 13 cases with insufficient information. For

the unknown cases four were during a flood and can be assumed to be at or near the
highest water level. Of the four that did not fail at their highest recorded water level:
Bayless (B) had been at the same level the year prior when sliding had also occurred
(Bayless (A));
Bouzey (B) had been 0.1m higher for over a year;
Meihua had overtopped by 0.3m previously (0.8m higher than during failure); and
Leguaseca failed at a low reservoir storage. This multi-arch structural failure was due
to concrete deterioration in the acidic reservoir water.
From this analysis it is clear that the majority of failures have occurred when the
reservoir was at its highest recorded level (which could be defined as first fill). Note
however, that several of these dams failed at water levels the same or slightly higher
than those previously recorded. The water levels were often reached during a rapid stage
of first filling or during flood. Of those dams where information was available, most
failed within two days of reaching their final water level and several failed within six
hours.
Table 2.16, Table 2.17 and Figure 2.21 show the age at failure versus year
commissioned for various failure modes. Foundation piping failures generally occurred
in the first three years. Exceptions to this were Puentes, Bacino di Rutte and Austin (A).
Puentes, which was commissioned in 1790, failed in first fill which took 11 years.
Bacino di Rutte failed due to piping in the foundation. During first fill a crack appeared
under the dam which was filled. The dam was emptied and the silt removed 13 years
later. The dam failed during refilling of the reservoir. Austin (A) failed due to a
combination of scour, piping and sliding during overtopping at the highest water level
the dam had experienced.
Foundation sliding occurred in less than five years in all but two cases. Zerbino dam
failed after ten years due to scour and sliding during overtopping. Xuriguera failed after
42 years, unfortunately no further details on the failure were available.
Structural sliding was more evenly distributed with five failures occurring after ten
years. One structural tensile/shear failure occurred after 80 years (Khadakwasla Dam

that failed during overtopping due to an upstream dam failure). There are a number of
dams with unknown failure modes. Most of these failed during overtopping events (see
Figure 2.22). Most of the failures that occurred after five years were due to overtopping
(12 compared to 6 non-overtopping). Prior to five years of age non-overtopping failures
were more prevalent (21 compared to 4).
Figure 2.23 shows the age at failure versus year commissioned for different dam types.
Masonry dams appear to have failed at all ages. Concrete dams, with the exception of
the three below, have failed within ten years of commissioning. There is insufficient
information on the exceptions to determine why they failed at a later time.
Kohodiar (India) - Combined concrete gravity/earthfill dam, unknown failure mode.
Xuriguera (Spain) - Concrete gravity dam apparently failed by foundation sliding.
Hauser Lake II (USA) - Concrete gravity dam with no failure information.
Table 2.16. No. of dam foundation sliding & piping failures vs age at failure
Sliding Piping Age at Failure
Grav
.
Arch Butt. Total Grav
.
Arch Butt. Total
T1 During
construction
- - - - - - - -
T2 During first fill 3 1 - 4 4 1 3 8
T3 0-5 years 1 - 1 2 - - - -
T4 5-10 years 1 - - 1 - - - -
T5 10-20 years - - - - - 1 - 1
T6 40-50 years 1 - - 1 - - - -
ALL 6 1 1 8 4 2 3 9


Table 2.17. No. of structural (shear or tensile) failures vs age at failure
Age at Failure Grav. Arch Butt. Multi-Arch Total
T1 During construction 1 - - - 1
T2 During first fill 1 - 2 1 4
T3 0-5 years - 1 - - 1
T4 5-10 years 2 - - - 2
T5 10-20 years 2 - - - 2
T6 20-30 years - - - 1 1
T7 30-40 years - - - - -
T8 40-50 years 1 - - - 1
T9 >50 years 1 - - - 1
ALL 8 1 2 2 13


0 5 10 15 20 25 30 35 40 45
T1 (during const'n)
T2 (first fill
T3 (<5 yrs)
T4 (5-10yrs)
T5 (10-20yrs)
T6 (20-30yrs)
T7 (30-40yrs)
T8 (40-50yrs)
T9 (>50yrs)
T10 (>5yrs, time
unknown)
T11 (unknown)
Percent of Dams
Failure
Accident
Major Repair

Figure 2.16. Age at incident - all dams

Figure 2.24 and Figure 2.25 show age at significant incident versus year commissioned
for different dam types. As expected the accidents/major repairs tend to occur mostly
after 1915. Older incidents are less likely to be recorded. Accidents and major repairs
appear to occur at a later stage than that of failures. The distribution of ages to incident
for both masonry and concrete dams appears to be similar.
Table 2.18 gives the breakdown of incidents in all types of concrete and masonry dams.
0 5 10 15 20 25 30 35
T1 (during const'n)
T2 (first fill
T3 (<5 yrs)
T4 (5-10yrs)
T5 (10-20yrs)
T6 (20-30yrs)
T7 (30-40yrs)
T8 (40-50yrs)
T9 (>50yrs)
T10 (>5yrs, time
unknown)
T11 (unknown)
Percent of Dams
Failure
Accident
Major Repair

Figure 2.17. Age at incident - concrete gravity dams

0 5 10 15 20 25 30 35 40 45
T1 (during const'n)
T2 (first fill
T3 (<5 yrs)
T4 (5-10yrs)
T5 (10-20yrs)
T6 (20-30yrs)
T7 (30-40yrs)
T8 (40-50yrs)
T9 (>50yrs)
T10 (>5yrs, time unknown)
T11 (unknown)
Percent of Dams
Failure
Accident
Major Repair

Figure 2.18. Age at incident - masonry gravity dams

Figure 2.19, Figure 2.20 and Table 2.18 show the time to significant incidents for dams.
The data is presented as the number of incidents in a time period divided by the
population of dams that had survived that time period. The population was taken from
ICOLD (1979) and extrapolated to 1983 dam numbers.
First filling is still the predominant failure time. There appears to be a slight rise in the
rate of failures with time (ignoring T2). After 40 years of age there is a jump in the
failure rate. It should be noted that the older age groups are represented by a small
population

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11
Time to Significant Incident
I
n
c
i
d
e
n
t
s
/
P
o
p
u
l
a
t
i
o
n

(
%
)
Failures
Accidents
Major Repairs

Figure 2.19. Time to significant incident - gravity dam incidents/population (%)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11
Time to Significant Incident
I
n
c
i
d
e
n
t
s
/
P
o
p
u
l
a
t
i
o
n

(
%
)
Failures
Accidents
Major Repairs

Figure 2.20. Time to significant incident - all dam incidents/population (%)


Table 2.18. Time to significant incident - incident/population of dams surviving
period (%)
Failures
Grav Arch Butt MA Total
Tot 0.90 0.75 2.24 1.87 0.99
T1 0.09 - - - 0.07
T2 0.32 0.25 1.60 0.93 0.41
0-5 0.03 0.13 0.66 - 0.09
5-10 0.16 - - - 0.12
10-20 0.13 0.19 - - 0.13
20-30 0.12 - - 1.52 0.14
30-40 - - - - -
40-50 0.32 0.53 - - 0.32
>50 0.32 - - - 0.24
T10 - - - - -
T11 - 0.13 - - 0.02
Accidents
Tot 1.54 10.69 2.88 14.02 3.47
T1 0.14 0.63 0.64 - 0.26
T2 0.41 2.39 0.32 5.61 0.86
0-5 0.27 1.16 0.99 0.95 0.49
5-10 0.03 - - - 0.02
10-20 0.22 0.19 - - 0.19
20-30 0.25 - - - 0.19
30-40 0.08 0.38 1.09 - 0.18
40-50 0.11 0.53 - - 0.16
>50 0.32 0.53 - - 0.32
T10 0.21 2.06 0.33 3.81 0.62
T11 0.09 4.03 0.32 5.61 0.90
Major Repairs
Tot 2.06 0.13 3.53 - 1.78
T1 0.06 - 0.64 - 0.09
T2 0.12 - 0.32 - 0.11
0-5 0.18 - - - 0.13
5-10 0.10 - - - 0.07
10-20 0.39 - - - 0.29
20-30 0.37 - 0.87 - 0.33
30-40 0.31 - - - 0.24
40-50 0.43 - 1.49 - 0.41
>50 0.64 0.53 1.49 - 0.65
T10 0.72 - 1.32 - 0.62
T11 0.09 - 0.32 - 0.09
T1: During construction; T2: During first fill; T10: >5 years, else unknown; T11: Unknown.
Page 2.70

Table 2.19. Time to significant incident
No. Failures Accidents Major Repairs
PG PG(M) CB CB(M) VA VA(M) MV ALL PG PG(M) CB CB(M) VA MV(M) ALL PG PG(M) CB VA ALL TOTAL
T1 2 1 - - - - - 3 4 1 1 1 5 - 12 2 - 2 - 4 21
T2 3 8 3 2 2 - 1 19 11 3 1 - 19 - 40 2 2 1 - 5 63
T3 1 - 1 1 - 1 - 4 6 3 3 - 9 - 22 6 - - - 6 32
T4 1 4 - - - - - 5 1 - - - - - 1 2 1 - - 3 9
T5 1 2 - - - 1 - 4 4 1 - - 1 - 6 7 2 - - 9 19
T6 - 2 - - - - 1 3 1 3 - - - - 4 5 1 1 - 7 14
T7 - - - - - - - 0 1 - 1 - 1 - 3 3 1 - - 4 7
T8 1 2 - - - 1 - 4 - 1 - - 1 - 2 3 1 1 - 5 11
T9 1 2 - - - - - 3 1 2 - - 1 - 4 3 3 1 1 8 15
T10 - - - - - - - 0 6 1 1 - 16 2 28 16 8 4 - 28 56
T11 - - - - 1 - - 1 3 - 1 - 32 - 42 3 - 1 - 4 47
10 21 4 3 3 3 2 46 38 15 8 - 85 2 164 52 19 11 1 83 294

Page 2.71

Table 2.20. Details of dam failure water levels
Dam Name Dam
Type
Year
Com.
Year
Fail
Fail
Type
Fail
Mode
MWL

(m)
Height
at fail
(m)
Highest
record
level?
Height
above
previous
Time
(hrs)
Comments
Torrejon-Tajo PG 1967 1965 Fa SH DNA
Zerbino PG 1925 1935 Faf S/SC 10 15 Y 5m
>FSL
Large flood caused overtopping.
Mohamed V PG 1966 1963 Fb ? DNA
Elwha River PG 1912 1912 Ff P 31 31 Y 1
st
fill 240 Failure occurred 10 days after pond was first
filled.
Xuriguera PG 1902 1944 Ff S
Bayless (A) PG 1909 1910 Ff S 12.5 12.5 Y 1
st
fill 48 Failed 2 days after spillway began discharging.
Bayless (B) PG 1909 1911 Ff S 12.5 12.5 N <6 Was at this level previous year when failure A
occurred. Failed at 2-2:30 on day the reservoir
filled.
St Francis PG 1926 1928 Ff S 61 61 Y 1
st
fill 170 Gradual first fill.
Hauser Lake II PG 1911 1969 ? ? DNA
Kohodiar PG/TE 1963 1983 ? ? DNA
Fergoug I PG(M) 1871 1881 Fa SC >43 Y Flood due to failure of Habra dam.
Fergoug II PG(M) 1885 1927 Fa SC? >43 Y Flood due to failure of Habra dam.
Sig PG(M) 1858 1885 Fa SC? Y Flood due to failure of Cheurfas dam.
Santa Catalina PG(M) 1900 1906 Fa ?
Cheurfas PG(M) 1884 1885 Fb ? Y 1
st
fill
Granadillar PG(M) 1930 1933 Fb ?
Bouzey PG(M) 1881 1895 Fb T 19.7 19.6 N 0.1m >1 year Had been at 19.7m for over a year previously.
Khadakwasla PG(M) 1879 1961 Fb T/SH 28 32.7 Y* 3.9m 4 Flood due to failure of Panshet dam.
Page 2.72

Dam Name Dam
Type
Year
Com.
Year
Fail
Fail
Type
Fail
Mode
MWL

(m)
Height
at fail
(m)
Highest
record
level?
Height
above
previous
Time
(hrs)
Comments
* Overtopped by 2.7m and failed when
overtopping had receded to 1.8m.
Habra (B) PG(M) 1872 1881 Fba T/SH 33 36.9 Y Overtopping.
Angels PG(M) 1895 1895 Ff P
Tigra PG(M) 1917 1917 Ff S 27.1 26.7 Y 1.1m 0.5 Spillway section overtopped by 1.1m. Whole dam
overtopped by 0.15m.
Austin (A) PG(M) 1893 1900 Ff SC/P/S 20.7 24.1 Y 0.4m Flood overtopped dam by 3.4m.
Puentes PG(M) 1791 1802 Ffb P >47 47 Y 1
st
fill* *1
st
fill took 11yrs. Dam filled from 22-47m in
final 4 mths.
Kundli PG(M) 1924 1925 Fm ? Y 1
st
fill Rapid 1
st
fill due to floods.
Chickahole PG(M) 1966 1972 Fm T 27.4 26 ? Flood rise of 1.5m immediately prior to failure.
Gallinas PG(M) 1910 1957 Fm/Fa ? Y Overtopped by record flood.
Lynx Creek PG(M) 1891 1891 Fm ? Flood.
Pagara PG(M) 1927 1943 Fmb T? 28.7 30 Y 1.3m <12 Overtopped by 0.4m in flood.
Habra (A) PG(M) 1871 1872 Fmb T/SH 33 Y Flood after 1
st
fill.
Habra (C) PG(M) 1881 1927 Fmb T/SH 33 37 Y Flood overtopped, largest since repair.
Elmali I PG(M)/TE 1892 1916 Fa ? Overtopped.
Lower Idaho
Falls
ER/PG(M) 1914 1976 Fa ? Y Overtopped from upstream failure of Teton.
Vaughn Creek VA 1926 1926 Ff P 17 17 Y 1
st
fill 48
Malpasset VA 1954 1959 Ff S 66 65.7 Y* 3 * Just previously exceeded this by 0.1m for 3
hours.
Moyie River VA 1924 1926 Ffa SC 14 16-18 Y 2-4m Storm and upstream dam failure flood overtopped
dam.
Page 2.73

Dam Name Dam
Type
Year
Com.
Year
Fail
Fail
Type
Fail
Mode
MWL

(m)
Height
at fail
(m)
Highest
record
level?
Height
above
previous
Time
(hrs)
Comments
Meihua VA(M) 1981 1981 Fb 21.5 N Previously overtopped by 0.3m (0.8m > than at
failure).
Bacino di Rutte VA(M) 1952 1965 Ff D/P 12 Y* <48 * Highest since sediment removed. Dam had been
filling for 2 days.
Ashley CB 1908 1909 Ff P 17 17 Y 1
st
fill >1 Just spilling when pipe failed.
Stony Creek CB
(Ambursen)
1913 1914 Ff P 13 13 ? Unsure how long at this level or if it had been
higher. Dam in service 6 months.
Komoro CB 1927 1928 Ff S/P No suggestion of high water level.
Overholser CB
(Ambursen)
1920 1923 Ffa SC Y Overtopped in flood.
Austin (B) CB(M) 1915 1915 Fba SH Y 3 Highest since rebuilt.
Vega de Tera CB(M) 1956 1959 Fm T/C 33 33 Y 1.75m <0.5 Previous year was at 31.25m. Flood had just
completed 1
st
fill. The dam reportedly was
breached at the moment of topping of the crest
Selsfors CB/TE 1943 1943 Ff P 20 18.2 Y 1
st
fill 6
Gleno MV 1923 1923 Fb T/C 32 32 Y 1
st
fill 1 month Had been at full supply level for 1 month.
Leguaseca MV 1958 1987 Fb T/C N low reservoir storage

Page 2.74

T3
T4
T5
T6
T7
T8
T9
0
10
20
30
40
50
60
70
80
90
100
1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Year Commissioned
A
g
e

(
y
e
a
r
s
)
Structural
Structural (T1 or T2)
Fndn Sliding
Fndn Sliding (T1 or T2)
Fndn Piping
Fndn Piping (T1 or T2)
Unknown
Unknown (T1 or T2)
T1- During construction
T2- First Fill

Figure 2.21. Failure mode: age at failure versus year commissioned (all dams)

Page 2.75

T9
T8
T7
T6
T5
T4
T3
0
10
20
30
40
50
60
70
80
90
100
1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Year Commissioned
A
g
e

(
y
e
a
r
s
)
?
Over Topping
Non-Over Topping
T2- First Fill

Figure 2.22. Over topping: age at failure versus year commissioned (all dams)

Page 2.76

T9
T8
T7
T6
T5
T4
T3
0
10
20
30
40
50
60
70
80
90
100
1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Year Commissioned
A
g
e

(
y
e
a
r
s
)
Concrete Gravity
Masonry Gravity
Concrete Buttress
Masonry Buttress
Concrete Arch
Masonry Arch
Concrete Multi-Arch
T2- First Fill

Figure 2.23. Dam type: age at failure versus year commissioned
Page 2.77

0
10
20
30
40
50
60
70
80
90
100
1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Year Commissioned
A
g
e

(
y
e
a
r
s
)
Fail - Concrete Gravity
Fail - Masonry Gravity
Fail - Other
Acc/MR - Concrete Gravity
Acc/MR - Masonry Gravity
Acc/MR - Other

Figure 2.24. Age at significant incident versus year commissioned
Page 2.78

0
10
20
30
40
50
60
70
80
90
100
1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Year Commissioned
A
g
e

(
y
e
a
r
s
)
Fail - Concrete Gravity
Fail - Masonry Gravity
Fail - Other
Acc/MR - Concrete Gravity
Acc/MR - Masonry Gravity
Acc/MR - Other

Figure 2.25. Age at significant incident versus year commissioned

2.3.5 Incident Causes
Table 2.21 shows that failures in the foundation are much more common in concrete
dams than in masonry dams (47% compared to 19% and 68% compared to 22% when
combined failure types, F
f
, F
fa
and F
fb
, are included). Failures due to the dam body or
materials are more common in masonry dams.
Table 2.21. Failure types
Dam Type Ff Fb Fa Fm Ffa Ffb Fbm Fba Unknown
PG 5 1 1 1 2
PG(M) 3 4 5 4 1 3 1
VA 1 2
VA(M) 1 1 1
CB 3 1
CB(M) 1 1 1
MV 2
Total
Concrete
9
(47)
3
(16)
1
(5)
- 4
(21)
- - - 2
(11)
Total
Masonry
5
(19)
5
(19)
5
(19)
6
(22)
- 1
(4)
3
(11)
2
(7)
-
Total All
14
(30)
8
(17)
6
(13)
6
(13)
4
(9)
1
(2)
3
(7)
2
(4)
2
(4)
NOTE: Figures in brackets are percentages for each dam type.
Tables D1 to D3 in Appendix D give the incident causes for all dams, concrete gravity
dams and masonry gravity dams respectively. These have been derived from the ICOLD
failure causes terminology shown in Section 2.2.3.8.
Table 2.22 to Table 2.24 show the most common causes of incidents to all dams,
concrete gravity dams and masonry gravity dams respectively. It should be noted that
the percentage of dams column can total more than 100% since there can be more than
one cause for each incident.


Table 2.22. Main causes of incidents in all dams
Rank Code Description No % of Dams
FAILURES
1 3.4.6 Overtopping 10 22
2 3.4.2 Uplift 8 17
3 1.1.4 seepage in the foundation 7 15
4 1.1.5 piping through the foundation 6 13
5 4.7.1 excess rates of flow 6 13
ACCIDENTS
2 4.8 local scour 16 9
5 4.11.6 discharge equipment malfunction 10 6
MAJOR REPAIRS
1 1.2.3 freezing and thawing 53 20
2 1.3.4 external temperature variation 28 11
3 1.2.2 reaction between concrete & environment 22 8
4 1.2.8 concrete permeability 22 8
5 3.2.2 reaction between masonry & environment 22 8

Overtopping, uplift and foundation seepage and piping are the most common causes of
failure for all dams combined. Foundation problems (shear and seepage) are the major
cause of failures for concrete gravity dams. Masonry gravity dams have more failures
due to overtopping. Seepage and flow problems are the main causes of accidents whilst
concrete reactions, temperature and freeze-thaw cause the most major repairs. These
types of major repairs tend to cause only surficial damage.

Table 2.23. Main causes of incidents in concrete gravity dams
FAILURES
1 1.1.3 shear strength in the foundation 4 40
3 1.3.2 Uplift 2 20
ACCIDENTS
3 4.6 due to structural behaviour 7 16
5 4.11.6 discharge equipment malfunction 5 11
MAJOR REPAIRS
2 1.2.2 reaction between concrete & environment 15 9
3 1.2.8 concrete permeability 15 9
4 1.3.2 Uplift 15 9
5 4.2.12 concrete erosion by abrasion 13 8

Table 2.24. Main causes of incidents in masonry gravity dams
FAILURES
1 3.4.6 Overtopping 8 38
2 3.4.2 Uplift 7 33
3 2.3.9 upstream dam collapse 5 24
4 3.5.2 tensile stresses 5 24
ACCIDENTS
1 3.4.2 Uplift 6 35
2 3.5.2 tensile stresses 3 18
3 3.1.4 seepage in foundation 2 12
4 3.2.8 mortar permeability 2 12
5 3.2.9 masonry construction (including order) 2 12
MAJOR REPAIRS
1 3.2.2 reaction between masonry & environment 20 51
2 3.2.8 mortar permeability 10 26
3 3.1.4 seepage in foundation 8 21
5 3.4.1 hydrostatic, silt and ice pressure 5 13


When all the dams were analysed together, similar incident codes were grouped
together to better distinguish the main causes of incidents. Only significant incidents
were included. The incidents were separated into those with soil foundations and those
with rock or unknown foundations. Figure 2.26, Table 2.25 and Table 2.26 show the
results of this analysis. Table 2.25 only shows results for failures of dams with soil
foundations. There were only three dam accidents where the dam was known to have a
soil foundation. The results show that piping was the predominant cause of failure for
dams with soil foundations.
For dams with rock or unknown foundations, the major cause of failure was overtopping
followed by shear strength of the foundation. Piping was the sixth most common cause
of failure with five cases noted. The causes of accidents for dams with rock or unknown
foundations were seepage, scour, piping, permeability in the concrete and tensile
stresses in the dam body. Major repairs were caused by reactions of the
masonry/concrete with the environment, concrete/masonry permeability and
construction methods.
Table 2.25. Main failure causes for dams with soil foundations
Rank ICOLD Codes Description No % of
Dams
1 1.1.5, 3.1.5, 4.1.5 internal erosion in the foundation
(piping)
6 67
2 1.1.4, 3.1.4, 4.1.4 seepage in the foundation 2 22
2 1.1.9, 3.1.9, 4.1.8 foundation preparation 2 22

Table 2.26. Main significant incident causes for dams with rock or unknown
foundations
Rank ICOLD Codes Description No % of
Dams
FAILURES
1 1.3.7, 3.4.6 overtopping 12 32
2 1.1.3, 3.1.3, 4.1.3 shear strength in the foundation 8 22
3 1.4.2, 1.5.2, 3.5.2 tensile stresses in the concrete/masonry 7 19
5 4.7.1 excess rates of flow (3 due to overtopping) 6 16
6 1.1.5, 3.1.5, 4.1.5 internal erosion in the foundation (piping) 5 14
6 1.2.6, 1.3.6 shear strength of concrete/masonry 5 14
ACCIDENTS
2 4.8 local scour 15 9
3 1.1.5, 3.1.5, 4.1.5 internal erosion in the foundation (piping) 13 8
3 1.2.8, 3.2.8, 4.2.6 permeability in the concrete/masonry 13 8
3 1.4.2, 1.5.2, 3.5.2 tensile stresses in the concrete/masonry 13 8
MAJOR REPAIRS
1 1.2.2, 3.2.2,
4.2.2, 4.4.1
reaction between concrete/masonry &
environment
21 26
2 1.2.8, 3.2.8, 4.2.6 permeability in the concrete/masonry 16 20
2 1.2.9, 1.2.10,
3.2.9, 4.2.7
method of construction (including cooling) 16 20
4 1.2.3, 3.2.3, 4.2.3 freezing and thawing 12 15
5 1.2.11, 3.2.10,
3.3.2, 4.2.10
structural joints in concrete/masonry 10 12


Failure modes for incidents involving the foundation will be discussed further in
Section 2.3.8. For the failures in the dam structure the following was noted:
The numbers of structural failures attributed to poor construction and design
flaws were similar. There was difficulty in separating design and construction
problems as, in many cases, both contributed to the failure.
Only one concrete gravity dam failed due to an inadequacy in the structure. The dam
(Torrejon-Tajo, Spain) failure cause was traced to organics present in the aggregate,
and filling of the dam by a flood during construction before the concrete had fully
set.
Overtopping preceded 5 of the failures.
The majority of the failure cases were masonry gravity dams, probably reflecting the
quality of construction and materials.


0 5 10 15 20 25 30 35
fndn shear strength
fndn seepage
fndn piping
overtopping
excess flow rates
local scour
upstream dam fail
tensile stresses
structural joints
mat shear strength
mat tensile strength
compressive strength
construction method
mat permeability
enviro & mat reaction
freeze/thaw
Percent of Incidents
Failure
Accident
Major Repair

Figure 2.26. Causes of significant incidents (rock & unknown foundations)

0 1 2 3 4 5 6 7
fndn seepage
fndn piping
fndn preparation
uplift
overtopping
shape of dam
structural joints
mechanical strength
discharge malfunction
Number of Incidents
Failure
Accident

Figure 2.27. Causes of significant incidents (soil foundations)

2.3.6 Monitoring and Surveillance Data
2.3.6.1 Using ICOLD Terms
Overtopping was the most common failure warning type followed by dam leakage and
no warning. However, as can be seen from Table 2.27 it is masonry dams which are
most susceptible to overtopping. Dam leakages followed by cracking were the most
prevalent warning in accidents. Major repairs tended to have been prompted by dam
leakage or concrete deterioration. It appears that the accidents and major repairs tend to
have a structural warning that can be noticed, whereas the failure warnings are more
difficult to notice. Figure 2.28 and Figure 2.29 show the warning types for concrete
gravity dam and all dam incidents respectively.

Table 2.27 to Table 2.29 show the warning types for failures, accidents and major
repairs for each dam type. It should be noted here that there can be more than one
warning type per dam failure.
From Section 2.3.5, it appears that the accidents and major repairs generally occur
where there has been obvious signs of distress (e.g. surficial damage, uplift records,
seepage monitoring). Whether these problems may signal potential instability in the
dam is questionable. For example, it is unlikely that cavitation damage in a spillway will
lead to failure of the dam. Failures have occurred where it is likely that little warning
was given or where the least amount was known, that is, in the foundation.
0 5 10 15 20 25 30 35 40 45
None
Foundation piping
Foundation leak
Dam leak
Move, horizontal
Move, vertical
Cracking
AAR
Conc. deterioration
Scour
Overtopping
Downstream slide
Uplift develop
Failure
Accident
Major Repair

Figure 2.28. Warning types - gravity dams

0 5 10 15 20 25 30 35 40
None
Foundation piping
Foundation leak
Dam leak
Move, horizontal
Move, vertical
Cracking
AAR
Conc. deterioration
Scour
Overtopping
Downstream slide
Uplift develop
Failure
Accident
Major Repair

Figure 2.29. Warning Types - All Dams

Table 2.27. Warning types vs dam type - failures
Warning Type PG

PG(M) CB CB(M) VA VA(M) MV Total
None 1 3 - 1 - - 1 6
Foundation piping - 2 1 - 2 - - 5
Foundation leak 1 1 1 - 3 - - 6
Dam leak 3 2 2 - - 1 1 9
Move, horizontal 2 2 - - 1 - - 5
Move, vertical - - 1 - - - - 1
Cracking 2 - 1 - 1 1 1 6
AAR - - - - - 1 - 1
Conc deteriorate - - - - - - 2 2
Scour - 2 - - - - - 2
Overtopping 1 11 1 1 1 1 - 16
Downstream slide - - - - - - - -
Uplift develop - - - - - 1 - 1
Unknown 4 2 - 1 - - - 7
Total 14 25 7 3 8 5 5 67

Table 2.28. Warning types vs dam type - accidents
Warning Type PG

PG(M)

CB

CB(M) VA

MV MV(M) Total

None 1 - 1 - 2 - - 4
Foundation piping 5 - 1 - 1 - - 7
Foundation leak 7 1 1 - - - - 9
Dam leak 6 9 1 - 2 - - 18
Move, horizontal 3 1 - - 3 - - 7
Move, vertical 2 2 - - 4 - - 8
Cracking 3 5 2 - 4 - - 14
AAR - - - - - - - 0
Conc deteriorate - - 1 - - - - 1
Scour 3 1 - - - - - 4
Overtopping 2 2 - 1 - - - 5
Downstream slide 3 - - - - - - 3
Uplift develop 2 1 - - 1 - - 4
Unknown 20 4 4 - 77 17 2 124
Total 57 26 11 1 94 17 2 208

Table 2.29. Warning types vs dam type - major repairs
Warning Type PG

PG(M)

CB

CB(M) MV MV(M) Total

None - - - - - - -
Foundation piping 1 1 - - - - 2
Foundation leak 7 4 - - - - 11
Dam leak 17 10 4 - - - 31
Move, horizontal 1 - 1 - - - 2
Move, vertical - - - - - - -
Cracking 9 1 6 - - - 16
AAR 4 - 1 - - - 5
Conc deteriorate 18 7 2 - - - 27
Scour 2 - - - - - 2
Overtopping 2 - - - - - 2
Downstream slide 4 - - - - - 4
Uplift develop 3 1 - - - - 4
Unknown 110 21 21 2 6 1 161
Total 178 45 35 2 6 1 267

2.3.6.2 Details of Warnings
Warnings prior to dam failures are very important as they allow for the possibility of
either preventing the failure if detected early enough or, importantly, they allow time for
people downstream to be notified and evacuated. A warning, even a few hours prior to
failure, can have a major effect on loss of life. Table 2.31 was created to describe each
of the failures and their warnings. A subjective warning rating was given to each failure.
The ratings were taken as to whether a dam failure had a sufficient warning which could
have led to people downstream being advised of the impending failure. Table 2.31 also
includes information on the failure type and failure mode.
Many of the dam failures had limited information and as such could not be given a
warning rating. Table 2.30 shows the results for this analysis.


Table 2.30. Warning ratings for failed dams
Warning Rating Number of Dams
Yes 10
No 1
Maybe 9
Dam failure upstream 5
Flood 5
Unknown 16

Ashley Dam was the only dam where the failure signs were deemed insufficient to
allow a warning to be given. There was some seepage 1.5 to 2 hours prior to failure but,
the time was insufficient to allow for a warning to be given. Failure occurred through
the alluvial foundation. Selsfors Dam had a small seepage 4.25 hours prior to collapse.
The seepage increased slowly for 0.5 hour and then rapidly. The signs may have been
enough to give a limited warning.
An important note to come from this analysis is that most of the warnings comprised a
rapid increase in flow prior to failure. Quantity of flow appears not to be as critical.
Table 2.32 shows a similar analysis for significant accidents. Most of the accidents gave
signs of problems developing. Blackbrook II did not, as the accident was caused by an
earthquake. Bhandardara Dam which went close to failure through the dam body had
insufficient warning. Cracking occurred quickly at a flood level slightly higher than that
recorded previously.
Page 2.92

Table 2.31. Details of dam failures and descriptions of warnings
Dam Name Dam
Type
Fail
Type
Fail
Mode
Failure Description Warning Description Warning
Rating
Torrejon-Tajo PG Fa SH Failure of outlet gate. No details available ?
Zerbino PG Faf S/SC Scour due to overtopping, followed by foundation failure
(sliding or overturning)
No details available, but some warning issued -
Despite warnings, the flood drowned 100
people. Large scour occurred in power plant
tunnel in 1928 (same rock).
M
Mohamed V PG Fb ? No details available No details available ?
Elwha River PG Ff P Piping through alluvial sand and gravel during construction of
cutoff. Dam was completed and reservoir part filled before
cutoff construction.
Leakage into caisson for cutoff. Failed in 1.5
hours.
M
Xuriguera PG Ff S No details available No details available ?
Bayless (A) PG Ff S Left half of dam moved 450mm downstream at base, 790mm at
top by sliding on foundation.
Landslide left abutment downstream, large
leakage 4.5-15m downstream of dam 12 days
before dam failed.
Y
Bayless (B) PG Ff S Rapid failure of most of dam by sliding and overturning. Previous failure 8 months before. No remedial
action taken.
Y
St Francis PG Ff S Sudden failure in foundation due to softening of conglomerate
or sliding on existing landslide or foliation surface in schist.
Foundation seepage measured as reservoir rose to
1-2ft
3
/sec (30-60l/sec) or 6-9hrs before failure
water level recorder dropped 0.1m in hour
before failure. (There is a suggestion that this was
due to the tilting of the dam, but in any case it
would have acted as a warning.) Some evidence of
cracking in foundation 2 months before failure
(due to landslide in abutment).
Y/M
Hauser Lake
II
PG ? ? No details available. No details available. ?
Kohodiar PG/TE ? ? No details available. No details available. ?
Fergoug I PG(M) Fa SC 50m spillway section scoured by flood and failed. Flood caused by failure of Habra dam upstream. DF
Page 2.93

Dam Name Dam
Type
Fail
Type
Fail
Mode
Rating
Fergoug II PG(M) Fa SC? 125m section failed during flood. Flood caused by failure of Habra dam upstream. DF
Sig PG(M) Fa SC? Overtopped in flood. Founded on gravel. Flood caused by failure of Cheurfas dam upstream
2 hours before.
DF
Santa Catalina PG(M) Fa ? Overtopping. No details available. No details available. ?
Cheurfas PG(M) Fb ? No details available. ICOLD cite piping in foundation as cause,
but failure in dam, which is inconsistent.
No details available. ?
Granadillar PG(M) Fb ? Failure of dam due to inadequate cross section. No data available. ?
Bouzey (B) PG(M) Fb T Sudden tensile/compressive and overturning failure. Failure
surface slope gently 3.5m from upstream face, then steeply.
Crush and shear marks near downstream face.
Dam had leaked badly in foundation and moved
up to 0.34m downstream 11 years before failure.
Repairs had been carried out 3 years before, and
crest deflection 25mm observed. No warning
immediately before failure.
Y
Khadakwasla PG(M) Fb T/SH Failure in masonry. Tensile/compression probably enhanced by
stress concentration due to sudden change in foundation
elevation.
Dam was overtopped for 4 hours prior to failure,
and was vibrating. Flood due to failure of Panshet
dam upstream 7 hours prior to breach.
DF
Habra (B) PG(M) Fba T/SH Sudden failure in masonry during flood. No warning immediately before failure (flood). F
Angels PG(M) Ff P Piping in (soil?) foundation. No data available. ?
Puentes PG(M) Ffb P Piping failure through alluvium in foundation. Leakage from fndn noted just over 0.5hr prior to
failure. Just prior to failure there was a large
explosion from the discharge wells and a large
increase in leakage. It is said that the dam emptied
in 1hr. A messenger was sent to warn the town of
Lorca when the leakage was first noted (by bike)
but was overtaken by the flood wave.
M
Tigra PG(M) Ff S Sliding on weak shale (?) seam in foundation under flood level.
Dam overtopped by 0.15m only so overtopping itself unlikely to
be critical re scour, but may have affected uplift inside dam.
Dam went overtopped hour before failure. F
Austin (A) PG(M) Ff SC/P/
S
Sliding on weak seam in foundation of two 80m long sections of
the spillway, moved downstream 20m.
Whirlpools in storage 1 year before, 2m scour at
toe of spillway section of dam. Failed in 3 minutes
Y
Page 2.94

Dam Name Dam
Type
Fail
Type
Fail
Mode
Rating
during flood of record.
Kundli PG(M) Fm ? Failure attributed to green uncured lime mortar masonry. No data available. ?
Chickahole PG(M) Fm T Sudden tensile/overturning failure. Flood rise of 1.5m
immediately prior to failure.
No warning immediately prior to failure. Cracking
of dam occurred during consolidation grouting of
foundation.
M
Gallinas PG(M) Fm/Fa ? Overtopped and washed out (no details). No data available. Early warnings credited
with preventing loss of life.
M
Lynx Creek PG(M) Fm ? Failure in masonry in flood. No details available. ?
Pagara PG(M) Fmb T? Overtopping. No additional details. No data available. ?
Habra (A) PG(M) Fmb T/SH Sudden failure in foundation or masonry during overtopping by
flood.
No warning immediately before failure (large
leakage in dam on first filling but had reduced).
Y
Habra (C) PG(M) Fmb T/SH Sudden failure in masonry during flood. Flood no details about any warnings but,
reportedly did not result in a loss of human lives
because of adequate advance warnings.
F
Elmali I PG(M)/T
E
Fa ? Overtopped. No data available. No data available. ?
Lower Idaho
Falls
ER/PG(
M)
Fa ? Overtopped due to failure of Teton dam upstream. Failure of Teton dam 96km upstream. DF
Vaughn Creek VA Ff P Foundation piping and arch concrete failure. Considerable flow
below west abutment, followed by settlement and sliding of
abutment and in a short time, its overturning.
Some seepage in abutment on first filling. Very
large and serious leakage just before failure
through abutment.
Y
Malpasset VA Ff S Sudden shear failure in foundation controlled by geology and
uplift.
Failure very rapid. Seepage in abutment on first
filling 15 days, and more 2 days before failure.
17mm displacement of dam base compared to
estimated 10mm.
M
Moyie River VA Ffa SC Spillway scoured and undermined left abutment dam left
standing.
No details available, but scour should have been
evident.
F
Meihua VA(M) Fb ?
Page 2.95

Dam Name Dam
Type
Fail
Type
Fail
Mode
Rating
Bacino di
Rutte
VA(M) Ff D/P Foundation seepage and movement causing crack to open
upstream of dam on first filling. Crack sealed. Dam operated for
13 years, but failed when sediment removed from reservoir and
dam refilled. Failure was piping initiated along crack, giving
breach 12m by 2m into which dam collapsed.
Prior seepage, observation of cracks in foundation,
and displacement.
Y
Ashley CB Ff P Piping failure in fine sand with a little clay and gravel, 6m deep
below cutoff.
Seepage in foundation noted 1.5-2 hours before
piping failure.
N
Stony Creek CB Ff P Piping in foundation followed by settling of dam, cracking and
collapse of dam.
Large leakage through weep holes in floor of the
dam 24 hours before flow developed rapidly in
last 20 minutes before failure.
Y
Komoro CB Ff S/P Failure due to softening of volcanic ash in foundation. Unclear
whether piping, sliding or both.
Overholser CB Ffa SC Overtopping. Scour of abutment. No details available. ?
Austin (B) CB(M) Fba SH Flood destroyed 20 gates of masonry dam, and filled tailrace and
draft tubes with debris.
Flooding F
Vega de Tera CB(M) Fm T/C Structural failure of masonry buttress. No warning noted. Heavy leakage occurred
through masonry but may have been unrelated.
M
Selsfors CB/TE Ff P Piping in foundations fluvioglacial sand, followed by collapse of
dam into void.
Small seepage into abutment 4.25 hours prior to
failure, increased slowly for 0.5 hour, then rapidly.
M/N
Gleno MV Fb T/C Rapid structural failure of multiple buttress arch dam attributed
to weakness in poor quality supporting masonry.
Leakage through dam and on the cut off between
dam and foundation during and after construction.
Leakage increased markedly in the days before
failure up to 50l/sec.
Y
Leguaseca MV Fb T/C Structural failure of an arch due to concrete deterioration in
acidic reservoir water.
No details available (concrete deterioration). M
Page 2.96

Table 2.32. Details of dam significant accidents and descriptions of warnings
Dam Name Dam
Type
Fail
Type
Fail
Mode
H
d
/
W
Rating
Bingham PG Fa SC ? Spillway failed by overturning due to piping and erosion of
the weathered foundation rock.
Wilbur PG Fa ST ? Overtopping of dam caused damage to power station
downstream. Dam was not damaged.
Flood F
Upper
Glendevon
PG Ff/Fb P 1.32 Leakage of 25l/sec on first filling, giving high uplift. Leakage through foundation of 25l/sec on
first filling.
Y
Mequinenza PG Ff S 0.6 Weak bedding surfaces in limestone, lignite and marl
exposed during construction, led to strengthening works
being built before the dam was completed.
Horizontal and vertical movements
anticipated but did not occur because dam
was strengthened before completion.
Y
Aguilar PG Ff P 1.25 Piping of clay filled joint in limestone foundation giving
leakage of 50l/sec.
Leakage in joint in foundation of 50l/sec. Y
Villagarcia PG Ff P ? Leakage and piping through rock foundation of up to
100l/sec on first filling.
Leakage up to 100l/sec in foundations Y
Hales Bar PG Ff P 1.61 Leakage through karst limestone foundation, reaching
47600l/sec (47.6m
3
/sec) 27 years after construction. Many
attempts to stop leakage failed, dam abandoned 51 years after
construction.
Leakage up to 47600l/sec, whirlpools in
reservoir, boils downstream.
Y
Woodbridge
(A)
PG Ff P ? Piping of alluvial foundation. No details available. ?
Zardezas PG Ff S ? Foundation slide during construction. No details available. ?
Don Marco PG Ff S 1.44 Scour of foundations due to spillway, and sliding of dam on
weak zone in foundation rock.
Sliding of dam, erosion of downstream
foundation.
Y
Castrelo PG Ff S ? Landslide from abutment onto power station outlet. Landslide in abutment. Y
Burrinjuck
(C)
PG Ffa S 1.57 Rock slide in spillway channel partly damaged outlet works. No details available. ?
Page 2.97

Dam Name Dam
Type
Fail
Type
Fail
Mode
H
d
/
W
Rating
Great Falls
Generating
Station (A)
PG Ff P ? Leakage through a narrow ridge in reservoir which increased
from 560l/sec on first filling, to 12600l/sec over 20 years.
Leakage was through limestone interbedded with shale.
Leakage began at 560l/sec, increasing
steadily each year at 640l/sec/year to 14
years after filling, and 840l/sec/year to
12600l/sec, 24 years after filling. Leakage
was from 19 areas.
Y
Dworshak PG Fm CR/L 1.25 Thermal cracking which developed to give up to 380l/sec
leakage into the drainage gallery.
Cracking prior to initial filling, remained
small for 9 years, then suddenly opened to
give 380l/sec leakage.
M
Jandula PG(M) Fa T/SH ? Overtopped by flood to a depth of 0.15m. Flood overtopping. F
New Croton PG(M) Fa CR 1.5 Cracking of spillway concrete due to vibration by floodwater
over flashboards on top of the dam.
Flood, cracking in spillway, vibration and
leakage up to 9l/sec.
F
Blackbrook II PG(M) Fb CR 1.25 Earthquake caused cracking of parapit wall. Temporary
increase in foundation seepage.
Cracking of dam, increased foundation
seepage and earthquake itself.
N
Mulshi PG(M) Fb L/SH ? Leakage through dam increased to 42l/sec, analysis showed
inadequate stability. Mortar quality was an issue.
Leakage through dam increased from
3.6l/sec 28 years after construction to
42l/sec 11 years later.
Y
Thokarwadi PG(M) Fb L/SH ? Fine cracks right through dam on abutments. 200l/sec
leakage from weep holes drilled low down on downstream
face.
Fine cracks through dam on abutments.
200l/sec flow from weep holes drilled in
downstream face near foundation.
Y
Walman PG(M) Fb L/SH ? Leakage at many places, maximum 280l/sec. Leakage at many places up to 280l/sec. Y
Bhandardara PG(M) Fb T/SH 1.15 Cracking of dam due to tensile failure of masonry under
slightly higher flood level from previously. Also greatly
increased leakage in dam. Dam must have gone very close to
collapse.
Leakage through dam for 43 years less than
1.8l/sec. Suddenly increased to 870l/sec at
dam/foundation interface and 150mm
diameter hole in dam as a powerful jet (1
day after flood level reached). Cracking of
dam located from upstream to downstream
face.
N
Gela (A) PG(M) Fb L/SH ? Considerable seepage through dam into inspection gallery. Considerable seepage through dam into
inspection gallery.
Y
Page 2.98

Dam Name Dam
Type
Fail
Type
Fail
Mode
H
d
/
W
Rating
El Gasco (A) PG(M) Fba ? Flood overtopped dam, saturated clay and rock filling
between two outer masonry walls.
No data available. ?
Bouzey (A) PG(M) Ff S 1.66 135m length of dam slid up to 0.34m downstream.
Foundations disturbed up to 3m below dam.
Spring discharges in foundation 50-75l/sec
2 years before accident, increasing to
230l/sec after accident.
Y
Shirawata PG(M) Fmb L/SH ? Leakage through dam increased to 600l/sec 10 years after
construction. Mortar quality was an issue.
Leakage through dam increased from first
filling to 600l/sec 10 years after
construction.
Y
Olef CB Fbm CR Tensile cracking of buttress dam during curing of concrete in
construction.
Cracking of concrete. Y
Estremera CB Ff P High leakage through alluvial foundation with solution of
gypsum.
High leakage through foundation. Y
Ayers Islands CB Fm CR/L Concrete deterioration by freeze-thaw until a hole formed in
buttress slab concrete.
Concrete deterioration, hole formed,
leakage of dam.
M
Austin (C) CB(M)/
PG(M)
Fa ST Spillway piers destroyed during flood and hollow concrete
dam section partly destroyed.
Flood. Dam previously damaged and
foundations scoured.
F
Austin (D) CB(M)/
PG(M)
Faf SC/P Scour and piping of foundation of hollow concrete dam
caused collapse of 60m of dam.
Flood. Dam previously damaged and
foundations scoured.
Y
Umberumba VA Fa SC/L Overtopped by flood, scour of downstream toe, leakage
under dam.
No details available. (Flood) F
Idbar VA Ff P High seepage and piping of limestone foundation which had
not been grouted. Dam was abandoned.
Leakage, piping of foundation. Y
Vajont VA Fa S Massive landslide in reservoir caused overtopping of dam by
many metres (>100m). Dam remained intact.
Movements in landslide accelerating with
time.
M

2.3.7 Remedial Measures
Abandonment of the dam and reconstruction with a new design were the most
common remedial measures for failures. For accidents, reconstruction of deteriorated
zones in appurtenant works and water tightening treatment in the foundations were the
most common. Repairing concrete/masonry facing or reconstructing the deteriorated
concrete/masonry was the most frequent remedial method for major repairs. Figure 2.30
shows the most common remedial measures vs incident type. shows the number of
dams within each remedial measure category. Table 2.33 shows the number of dams
within each remedial measure category.

0 5 10 15 20 25 30 35
Reconstruction
(same design)
Reconstruction
(same design)
Not available
Scheme abandoned
Foundation
watertightening
Drain and filter
construction
Concrete
watertightening
Concrete facing
Reconstruction
(deteriorated zones)
Grouting
Dam shape correction
Appurtenant surface
repair
Appurtertenant Reconst.
(deteriorated zones)
Percent of Incidents with Remedial Measure
Failure
Accident
Major Repair

Figure 2.30. Most common remedial measures - all dam incidents


Table 2.33. Remedial measures - all dam incidents
Failures Accidents Major Repairs All Incidents
Remedial Measure
Number %* Number %* Number %* Number %*
Of a general nature:
Investigation 1 2 8 5 19 7 28 6
Monitoring 0 0 9 5 14 5 23 5
Lowering of reservoir level 1 2 10 6 11 4 22 5
Overall reconstruction (same design)_ 5 11 0 0 0 0 5 1
Reconstruction with new design 10 22 3 2 2 1 15 3
None 1 2 8 5 7 3 16 3
Not available 1 2 11 6 8 3 20 4
Scheme abandoned 15 33 5 3 1 0 21 4
In foundations:
Water tightening treatment 0 0 20 11 24 9 44 9
Drain & filter construction or repair 0 0 15 9 23 9 38 8
Strengthening by grouting or other methods 0 0 10 6 6 2 16 3
Filling in of fractures & cavities 1 2 2 1 1 0 4 1
Anchoring 1 2 0 0 2 1 3 1
In concrete and masonry dams:
Water tightening treatment 1 2 15 9 28 11 44 9
Drain construction or repair 1 2 1 1 14 5 16 3
Thermal protection (exc. facing) 0 0 0 0 7 3 7 1
Facing 0 0 9 5 56 21 65 13
Reconstruction of deteriorated zones 5 11 10 6 35 13 50 10
Execution of joints 0 0 3 2 5 2 8 2
Strengthening by grouting 0 0 9 5 21 8 30 6
Strengthening by anchoring 2 4 7 4 11 4 20 4
Strengthening by shape correction 4 9 8 5 3 1 15 3
In appurtenant works:
Discharge increase 1 2 8 5 7 3 16 3
Construction of additional appurtenant work 1 2 2 1 1 0 4 1
Overall reconstruction of appurtenant works 0 0 2 1 6 2 8 2
Partial reconstruction with strengthening 0 0 5 3 6 2 11 2
Shape correction of surfaces contacting flow 0 0 3 2 5 2 8 2
Aeration devices: construction or capacity inc. 0 0 0 0 2 1 2 0
Repair of surfaces contacting flow 0 0 8 5 19 7 27 6
Slope protection & stabilisation 0 0 2 1 1 0 3 1
Const., modification & repair of valves & gates 0 0 10 6 4 2 14 3
Establish. & update rules for gate & valve ops 0 0 0 0 2 1 2 0
Reconstruction of deteriorated zones 1 2 19 11 8 3 28 6
Abandonment of appurtenant work 0 0 0 0 1 0 1 0
In reservoir:
Reforestation 0 0 2 1 0 0 2 0
Torrent training 0 0 5 3 0 0 5 1
Sediment discharge diversion 0 0 1 1 0 0 1 0
Slope regularisation, protection & strengthening 0 0 2 1 0 0 2 0
Water tightening 0 0 4 2 1 0 5 1
Dredging 0 0 2 1 2 1 4 1
Downstream of Dam:
Draining 0 0 2 1 0 0 2 0
Slope regularisation, protection & strengthening 0 0 2 1 2 1 4 1
TOTAL 52 242 365 659
Note: (*) Percent of dams (of particular incident type) with particular remedial measure

2.3.8 Geology
2.3.8.1 Geology of Dam Foundations Experiencing Incidents
In previous databases and analyses the dam foundation geology has been simply
described using categories of soil and/or rock. Since a large proportion of failures have
occurred due to deficiencies in the foundation, an improved analysis would be to
classify what type of soil or rock the dam was founded on, and then assess whether
certain foundation types are more susceptible to failure.
The aim of this section is to assess the geology of the foundations of dams that have
failed with particular reference to those that have undergone failure due to sliding or
piping in the foundation. There are 65 dams in the database that have experienced
foundation incidents, of which there are 19 failures 25 accidents and 25 (16 of which
were significant) major repairs. Table 2.34 and Table 2.35 list the dams that have had
failures or accidents (respectively) due to deficiencies in the foundation.
Figure 2.31 shows the age to failure for dams with failure in the foundation. Times to
failure and accidents in the foundation tend to be confined to less than five years. Major
repairs have occurred up to 45 years after commissioning. Failures due to the foundation
have occurred mainly in dams constructed prior to 1940.
Figure 2.32 shows the foundation geology types for incidents occurring in the
foundation. Limestone, shale, granite and alluvium are the most common foundation
geology types for dam foundations that have had accidents. Shale, limestone, sandstone
and alluvium are the most common for major repairs. However, there are a large
number of foundation major repairs (27%) with unknown foundation geologies.
The two main foundation failure modes are:
(a) Sliding on/in the Foundation
Table 2.34 and Table 2.35 show that sliding is most prevalent in interbedded
sedimentary sequences particularly with shale, and in schistose metamorphic where
weaknesses could be expected. The tuff and conglomerate (and shale for Bayless Dam)
were noted to have softened when wet. In the case of Malpasset the rock type played

some role in the failure but it was predominately due to uplift pressure and a fault zone.
In the case of Zerbino Dam the failure occurred along the foliations of the schist. There
are no cases of dam sliding associated with igneous rocks. Overtopping preceded four of
the foundation sliding cases.
Table 2.34. Geology for dams with failure in the foundation
Dam Name Dam
Type
Year
Failed
Failure
Mode
*

Fndn
Material
**

Geology
FAILURES
Bayless (A) PG 1910 Slide R Shale Sandstone
Bayless (B) PG 1911 Slide R Shale Sandstone
St Francis PG 1928 Slide R
Conglomerate
Schist
Xuriguera PG 1944 Slide R Unknown
Austin (A) PG(M) 1900 Slide R Shale Limestone Dolomite
Tigra PG(M) 1917 Slide R Shale Sandstone
Malpasset VA 1959 Slide/Uplift R Gneiss
Komoro CB(M) 1928 Slide/Piping R Tuff
Elwha River PG 1912 Piping S/R Fluvioglacial
Conglomerate

Angels PG(M) 1895 Piping S Unknown
Puentes PG(M) 1802 Piping S Alluvium Sandstone
Vaughn Creek VA 1926 Piping (abt) S/R Residual
Conglomerate

Ashley CB 1909 Piping S Fluvioglacial
Selsfors CB 1943 Piping S/R Fluvioglacial
Stony River CB 1914 Piping S Alluvial Shale
Bacino di Rutte VA(M) 1965 Deformation/Piping R Dolomite
Zerbino PG 1935 Scour/Slide R Schist Hornfeld
Moyie River VA 1926 Scour R Unknown
Overholser CB 1923 Scour R Unknown

*
Piping failure through abutment denoted by (abt).

**
Note: S= Soil; R= Rock

Table 2.35. Geology for dams with accidents in the foundation
Dam Name Dam
Type
Year
Failed
Failure
Mode
*

Fndn
Material
**

Geology
Castrelo PG - Slide R Granite
Don Marco PG 1975 Slide R Unknown
Mequinenza PG 1966 Slide R Limestone Lignite
Zardezas PG 1932 Slide R Sandstone Limestone
Conglomerate
Bouzey (A) PG(M) 1884 Slide R Sandstone
Dobra VA 1954 Slide R Unknown
Aguilar PG 1963 Piping R Limestone
Great Falls (A) PG 1945 Piping R Shale Limestone
Hales Bar PG 1964 Piping R Limestone Shale
Kawamata PG 1966 Piping ? Unknown
Upper Glendevon PG 1956 Piping R Andesite Agglomerate Siltstone
Villagarcia PG 1961 Piping R Granite
Woodbridge (A) PG - Piping S Alluvial
Idbar VA 1959 Piping R Limestone Schist
Estremera CB 1955 Piping S Alluvial
Logan Martin PG/TE 1964 Piping R Dolomite Limestone
Koshibu PG 1969 Piping/Leakage R Granite
Bingham PG - Piping/Scour R Unknown
Austin (D) CB(M) 1937 Scour/Piping R Limestone Shale Dolomite
Saulspoort PG 1988 Scour R Sandstone Siltstone Dolerite
Albigna PG - Deformation R Granite
Santa Maria VA 1968 Deformation R Granite
Gerlos VA 1964 Deformation R Unknown
Kariba VA 1958 Leakage R Unknown
Kolnbrein VA 1978
Uplift/Tension/Leakage
R Gneiss

*
Piping failure through abutment denoted by (abt).

**
Note: S= Soil; R= Rock


(b) Piping through the Foundation
Piping has tended to occur in soils namely alluvium, fluvioglacial and residual.
Although large concrete dams are generally not built on soil foundations, smaller
structures such as weirs are. Where foundations were rock, piping failure was through
the abutment of the dam. The abutment is defined by ICOLD(1978) as that part of the
valley side against which the dam is constructed (i.e. zones L2 and L4 of Figure 2.2 in
Section 2.2.4.9). A disproportionately high number of piping failures occurred in
buttress and arch dams. This is likely to be due to the high hydraulic gradients in the
foundations/abutments of these types of dams. Note, the scouring associated with
Overholser Dam was also through the abutment. When accidents are included limestone
becomes notably more prevalent.
0
10
20
30
40
50
60
1880 1900 1920 1940 1960 1980 2000
Year Commissioned
Age
Fail - Slide
Fail - Piping
Fail - Other
Acc - Slide
Acc - Piping
Acc - Other
Xuriguera
Austin (A)
Bacino di Rutte
Zerbino

Figure 2.31. Foundation incidents age, type and year commissioned - all dams


(45%)
0 5 10 15 20 25
Sandstone
Shale
Siltstone
Conglomerate
Sedimentary
Limestone
Agglomerate
Gneiss
Schist
Hornfeld
Lignite
Dolomite
Dolerite
Andesite
Basalt
Granite
Volcanic Ash
Alluvial
Residual
Unknown
Percent of Dams
FAIL (19)
ACC (25)
MR* (16)
Note(*) Significant incidents only
Figure 2.32. Foundation incidents geology - all incidents


2.3.8.2 Geology of the Population of Dams
As discussed above a large proportion of concrete dam failures have occurred in the
foundation. ICOLD (1974, 1983 and 1995) and USCOLD (1975 and 1988) have only
assessed the foundation of dams as soil or rock. Little work has been done in attempting
to compare foundation geology to likelihood of failure. This would allow comparison of
the geology of those dams experiencing incidents to the geology of the population of
dams allowing identification of those with disproportionately high or low number of
incidents.
To gain a better understanding of which foundation geology is likely to cause problems
a population of dams was required. The difficulty in doing this was finding populations
of concrete and masonry dams where the geology of dams could reasonably be attained.
The following populations were chosen:
USBR;
Australia/New Zealand; and
Portugal.
Descriptions of the populations are given below. The results of the analysis are shown
in Table 2.36 and Table 2.37. It should be noted that where a dam has two foundation
geology types both are included in the tables. This results in the total number of dams
being less than the total number of foundation geology types in Table 2.36 and Table
2.37. The percentage figures are calculated as the number of occurrences of a particular
geology type divided by the number of dams (and not the total number of geology
types). The figures therefore represent the percentage of dams with a particular geology
type.

(a) USBR Large Concrete Dams
The USBR large concrete dam population was chosen for its good information on
geologies. The main sources being:
USBR (1996) Large Concrete Dams Online Database;
USBR SEED Reports;
USBR database Dam Safety Information System; and
personal communication with USBR personnel.
The results of the analysis on the dams are shown in Table 2.36. The results are in
percent per dam type. The number of dams is given in italics. The predominant
foundation types were granite (25%) and sandstone (22%). The total number of
unknowns was six.
(b) Australian and New Zealand Dams
The Australia/New Zealand population of dams was taken primarily from the ANCOLD
dam register with more detailed information provided by the sponsors of the research
project. Other information was taken from ICOLD Congresses, the ANCOLD Bulletin
and other journals. The major New Zealand dam owners (besides ECNZ who were a
sponsor) were contacted and the following companies provided information:
Contact Energy Ltd
Central Electric Ltd
Egmont Electricity Ltd
Marlborough Electric Ltd
Table 2.36 gives the breakdown of foundation geology types. The most common
foundation geology types were sandstone (26%) and granite (14%).

(c) Portuguese Dams
The Portuguese population was taken from LNEC (1996). The results are given in Table
2.36 in a similar method to above. There were 52 dams on rock foundations; 1 on a
soil/rock foundation and 1 unknown. The most common geology types were granite
(50%), schist (30%) and sandstone (19%).
The populations from Australia, New Zealand, the USBR and the Portuguese population
have been added into one population, which is presented in Table 2.37. Sandstone
(24%) and granite (24%) were the most common foundation geology types. 2% of the
dam population had soil, namely alluvium, foundations.
Page 2.110

Table 2.36. Foundation geology for Australia, New Zealand, Portugal and USBR (percent and number for each group)
AUSTRALIA/NEW ZEALAND PORTUGAL USBR
Grav Arch Butt MA ALL Grav Arch Butt MA ALL Grav Arch Butt ALL
Total Dams 97 42 10 3 152 28 20 4 2 54 21 7 31 59
Sandstones 24 23 36 15 20 2 26 40 21 6 15 3 25 1 19 10 24 5 43 3 16 5 22 13
Shale 8 8 5 2 7 10 0 10 2 14 1 3 1 7 4
Siltstone 5 5 14 6 20 2 9 13 0 0
Conglomerate 10 4 3 4 0 5 1 14 1 10 3 8 5
Limestone 5 2 1 2 50 1 2 1 5 1 13 4 8 5
Claystone 3 3 7 3 4 6 0 0
Mudstone 4 4 10 1 3 5 0 0
Chert 2 2 2 1 2 3 0 0
Breccia 2 2 1 2 0 0
Dolomite 5 1 3 1 3 2
Tillite 2 1 1 1 0 0
Marl 5 1 2 1

Schist 7 7 7 3 7 10 21 6 30 6 75 3 50 1 30 16 14 3 6 2 8 5
Quartzite 7 7 12 5 10 1 33 1 9 14 4 1 5 1 4 2 14 1 6 2 5 3
Gneiss 7 7 5 7 5 1 2 1 3 1 2 1
Phylitte 2 2 2 1 2 3 7 2 5 1 25 1 7 4 0
Slate 3 3 7 3 4 6 0 0
Hornfels 0 5 1 2 1 5 1 2 1
Argillite 1 1 1 1 0 0

Granite 21 20 5 2 14 22 43 12 65 13 50 2 50 27 24 5 14 1 29 9 25 15
Basalt 4 4 5 2 10 1 5 7 50 1 2 1 14 3 29 2 10 3 14 8
Tuff 9 9 5 2 20 2 9 13 0 5 1 3 1 3 2
Page 2.111

AUSTRALIA/NEW ZEALAND PORTUGAL USBR
Grav Arch Butt MA ALL Grav Arch Butt MA ALL Grav Arch Butt ALL
Dolerite 8 8 5 2 7 10 0 0
Rhyolite 4 4 5 2 4 6 0 5 1 10 3 7 4
Andesite 3 3 5 2 10 1 4 6 0 0
Porphyry 2 2 1 2 0 6 2 3 2
Diorite 1 1 2 1 1 2 0 5 1 3 1 3 2
Granodiorite 2 2 1 2 4 1 2 1 0
Greenstone 1 1 1 1 0 5 1 3 1 3 2
Agglomerate 1 1 1 1 0 3 1 2 1
Pumice 1 1 1 1 0 0

Volcanic Ash 0 0 0
Alluvium 2 2 1 2 0 5 1 14 1 3 2
Glacial 5 1 2 1
Residual 1 1 1 1 0 0
Unknown 20 19 14 6 50 5 67 2 21 14 18 5 9 5 0
Grav - Gravity; Butt - Buttress; MA - Multi-Arch


Table 2.37. Foundation geology for Australia, New Zealand, Portugal & USBR
dams - totalled Figures
Gravity Arch Buttress Multi-Arch ALL
Total Dams 125 93 21 5 265
Sandstone 27 34 25 23 29 6 24 63
Shale 8 10 3 3 5 1 5 14
Siltstone 4 5 6 6 10 2 5 13
Conglomerate 1 1 8 7 5 1 3 9
Limestone 1 1 6 6 20 1 3 8
Claystone 2 3 3 3 2 6
Mudstone 3 4 5 1 2 5
Chert 2 2 1 1 1 3
Breccia 2 2 1 2
Dolomite 1 1 1 1 1 2
Tillite 1 1 0 1
Marl 1 1 0 1

Schist 13 16 12 11 14 3 20 1 12 31
Quartzite 6 8 9 8 10 2 20 1 7 19
Gneiss 6 7 2 2 3 9
Phylitte 3 4 2 2 5 1 3 7
Slate 2 3 3 3 2 6
Hornfels 1 1 1 1 1 2
Argillite 1 1 0 1

Granite 30 37 26 24 14 3 24 64
Basalt 6 7 5 5 14 3 20 1 6 16
Tuff 8 10 3 3 10 2 6 15
Dolerite 6 8 2 2 4 10
Rhyolite 4 5 5 5 4 10
Andesite 2 3 2 2 5 1 2 6
Porphyry 2 2 2 2 2 4
Diorite 2 2 2 2 2 4
Granodiorite 2 3 1 3
Greenstone 2 2 1 1 1 3
Agglomerate 1 1 1 1 1 2
Pumice 1 1 0 1

Alluvium 2 3 5 1 2 4
Glacial 1 1 0 1
Residual 1 1 0 1
Unknown 19 24 6 6 24 5 40 2 14 37


2.3.9.3 Geology - Comparison Between Incidents and Population
The following assesses the foundation geology more likely to cause foundation piping
and stability problems. This has been based on the statistics of failures and accidents
and the population assumed in Table 2.37. Due to the limited number of foundation
failures that have occurred and the potential inaccuracies introduced by adopting Table
2.37 as a world population, care should be exercised here and the information taken as
qualitative only.
Figure 2.33 to Figure 2.35 gives the number of incidents in each geology type for: all
dams; concrete gravity dams; and masonry gravity dams respectively. From these
figures it becomes evident that soil foundations - most particularly alluvial soils are over
represented in the foundation incidents. The alluvial soils have a tendency to pipe under
the high gradients imposed. No dam has been reported to have failed by sliding on
alluvial soils. Normally a large concrete or masonry dam would not be built on a soil
foundation. It is interesting that sandstone does not appear to be over represented when
the population is taken into account. Failures tend not to occur in sandstone alone but
only when the sandstone is interbedded with shales. Shale and limestone (often
interbedded) have a high incidence for failing. The limestone has a high proportion of
accidents generally due to excessive leakage through dissolution. Another point of note
is that no incidents have occurred in basalt foundations.
Figure 2.36 to Figure 2.39 give the number of incidents in each geology type over the
population of dams in the same geology. The population was estimated using the figures
from Table 2.37 and the estimated world population of dams at 1983 (the available
ICOLD world population data cutoff). For gravity dams conglomerate, limestone,
dolomite and alluvium foundations stand out. Dams with limestone foundations appear
to be very susceptible to accidents. The figures for arch and buttress dams are based on
small failure populations and should therefore be looked at with caution. Dolomite and
gneiss stands out for arch dams whilst alluvium and shale are notable in buttress dams.
Figure 2.40 gives a clear indication of which foundation geology types have a tendency
to slide or pipe fail. Soils (particularly alluvial and fluvioglacial) and limestones are
more likely to have piping problems. Shale (interbedded with other sedime ntary units)
has a greater tendency to be involved with sliding failure because of the likely presence

of weaknesses in the bedding such as bedding surface shears. These conclusions agree
with the general knowledge regarding the geology types (e.g. as described in Fell et al,
1992).
Page 2.115

0
5
10
15
20
25
30
S
a
n
d
s
t
o
n
e
S
h
a
l
e
S
i
l
t
s
t
o
n
e
C
o
n
g
l
o
m
e
r
a
t
e
L
i
m
e
s
t
o
n
e
D
o
l
o
m
i
t
e
S
c
h
i
s
t
G
n
e
i
s
s
H
o
r
n
f
e
l
s
G
r
a
n
i
t
e
B
a
s
a
l
t
D
o
l
e
r
i
t
e
A
n
d
e
s
i
t
e
A
g
g
l
o
m
e
r
a
t
e
A
l
l
u
v
i
u
m
R
e
s
i
d
u
a
l
U
n
k
n
o
w
n
P
e
r
c
e
n
t

o
f

P
o
p
u
l
a
t
i
o
n
.

0
1
2
3
4
5
6
7
8
9
N
u
m
b
e
r

o
f

D
a
m
s
.
ALL - FAIL
ALL - ACC
ALL - POP

Note: Population based on Australia, New Zealand, Portugal and USBR
Figure 2.33. Geology for incidents in the foundation and dam population all dams
Page 2.116

0
5
10
15
20
25
30
35
S
a
n
d
s
t
o
n
e
S
h
a
l
e
S
i
l
t
s
t
o
n
e
C
o
n
g
l
o
m
e
r
a
t
e
L
i
m
e
s
t
o
n
e
D
o
l
o
m
i
t
e
S
c
h
i
s
t
G
n
e
i
s
s
H
o
r
n
f
e
l
s
G
r
a
n
i
t
e
B
a
s
a
l
t
D
o
l
e
r
i
t
e
A
n
d
e
s
i
t
e
A
g
g
l
o
m
e
r
a
t
e
A
l
l
u
v
i
u
m
R
e
s
i
d
u
a
l
U
n
k
n
o
w
n
P
e
r
c
e
n
t

o
f

P
o
p
u
l
a
t
i
o
n
.
0
1
2
3
4
5
6
7
N
u
m
b
e
r

o
f

D
a
m
s
.
PG - FAIL
PG - ACC
Gravity - POP

Figure 2.34. Geology for incidents in the foundation and dam population concrete gravity dams
Page 2.117

0
5
10
15
20
25
30
35
S
a
n
d
s
t
o
n
e
S
h
a
l
e
S
i
l
t
s
t
o
n
e
C
o
n
g
l
o
m
e
r
a
t
e
L
i
m
e
s
t
o
n
e
D
o
l
o
m
i
t
e
S
c
h
i
s
t
G
n
e
i
s
s
H
o
r
n
f
e
l
s
G
r
a
n
i
t
e
B
a
s
a
l
t
D
o
l
e
r
i
t
e
A
n
d
e
s
i
t
e
A
g
g
l
o
m
e
r
a
t
e
A
l
l
u
v
i
u
m
R
e
s
i
d
u
a
l
U
n
k
n
o
w
n
P
e
r
c
e
n
t

o
f

P
o
p
u
l
a
t
i
o
n
.
0
1
2
3
4
N
u
m
b
e
r

o
f

D
a
m
s
.
PG(M) - FAIL
PG(M) - ACC
Gravity - POP

Figure 2.35. Geology for incidents in the foundation and dam population masonry gravity dams
Page 2.118

0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
S
a
n
d
s
t
o
n
e
S
h
a
l
e
S
i
l
t
s
t
o
n
e
C
o
n
g
l
o
m
e
r
a
t
e
L
i
m
e
s
t
o
n
e
D
o
l
o
m
i
t
e
S
c
h
i
s
t
G
n
e
i
s
s
H
o
r
n
f
e
l
s
G
r
a
n
i
t
e
B
a
s
a
l
t
D
o
l
e
r
i
t
e
A
n
d
e
s
i
t
e
A
g
g
l
o
m
e
r
a
t
e
A
l
l
u
v
i
u
m
R
e
s
i
d
u
a
l
U
n
k
n
o
w
n
I
n
c
i
d
e
n
t
s
/
P
o
p
u
l
a
t
i
o
n

(
%
)
.
ALL - Fail
ALL - Acc

Figure 2.36. Foundation geology type as a percentage of the geology population all dams
Page 2.119

0.0
5.0
10.0
15.0
20.0
25.0
30.0
S
a
n
d
s
t
o
n
e
S
h
a
l
e
S
i
l
t
s
t
o
n
e
C
o
n
g
l
o
m
e
r
a
t
e
L
i
m
e
s
t
o
n
e
D
o
l
o
m
i
t
e
S
c
h
i
s
t
G
n
e
i
s
s
H
o
r
n
f
e
l
s
G
r
a
n
i
t
e
B
a
s
a
l
t
D
o
l
e
r
i
t
e
A
n
d
e
s
i
t
e
A
g
g
l
o
m
e
r
a
t
e
A
l
l
u
v
i
u
m
R
e
s
i
d
u
a
l
U
n
k
n
o
w
n
I
n
c
i
d
e
n
t
s
/
P
o
p
u
l
a
t
i
o
n

(
%
)
Gravity - Fail
Gravity - Acc

Figure 2.37. Foundation geology type as a percentage of the geology population gravity dams
Page 2.120

0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
S
a
n
d
s
t
o
n
e
S
h
a
l
e
S
i
l
t
s
t
o
n
e
C
o
n
g
l
o
m
e
r
a
t
e
L
i
m
e
s
t
o
n
e
D
o
l
o
m
i
t
e
S
c
h
i
s
t
G
n
e
i
s
s
H
o
r
n
f
e
l
s
G
r
a
n
i
t
e
B
a
s
a
l
t
D
o
l
e
r
i
t
e
A
n
d
e
s
i
t
e
A
g
g
l
o
m
e
r
a
t
e
A
l
l
u
v
i
u
m
R
e
s
i
d
u
a
l
U
n
k
n
o
w
n
I
n
c
i
d
e
n
t
s
/
P
o
p
u
l
a
t
i
o
n

(
%
)
.
Arch - Fail
Arch - Acc

Figure 2.38. Foundation geology type as a percentage of geology population arch dams
Page 2.121

0.0
5.0
10.0
15.0
20.0
25.0
S
a
n
d
s
t
o
n
e
S
h
a
l
e
S
i
l
t
s
t
o
n
e
C
o
n
g
l
o
m
e
r
a
t
e
L
i
m
e
s
t
o
n
e
D
o
l
o
m
i
t
e
S
c
h
i
s
t
G
n
e
i
s
s
H
o
r
n
f
e
l
s
G
r
a
n
i
t
e
B
a
s
a
l
t
D
o
l
e
r
i
t
e
A
n
d
e
s
i
t
e
A
g
g
l
o
m
e
r
a
t
e
A
l
l
u
v
i
u
m
R
e
s
i
d
u
a
l
U
n
k
n
o
w
n
I
n
c
i
d
e
n
t
s
/
P
o
p
u
l
a
t
i
o
n

(
%
)
.
Buttress - Fail
Buttress - Acc

Figure 2.39. Foundation geology type as a percentage of geology population buttress dams
Page 2.122

0
1
2
3
4
5
6
7
G
r
a
n
i
t
e
S
a
n
d
s
t
o
n
e
S
c
h
i
s
t
B
a
s
a
l
t
S
h
a
l
e
S
i
l
t
s
t
o
n
e
C
o
n
g
l
o
m
e
r
a
t
e
G
n
e
i
s
s
L
i
m
e
s
t
o
n
e
A
n
d
e
s
i
t
e
D
o
l
o
m
i
t
e
H
o
r
n
f
e
l
s
A
g
g
l
o
m
e
r
a
t
e
A
l
l
u
v
i
a
l
R
e
s
i
d
u
a
l
U
n
k
n
o
w
n
N
u
m
b
e
r

o
f

D
a
m
s

0
5
10
15
20
25
30
P
e
r
c
e
n
t

o
f

P
o
p
u
l
a
t
i
o
n

o
f

D
a
m
s
Fail - Piping
Fail - Sliding
Acc - Piping
Acc - Sliding
Population

Figure 2.40. Foundation incident geology and population mode of failure/accident

2.3.9 Other Design Factors in Failed Dams
Due to the limited information no conclusions could be drawn for the following factors.
(a) Post-Tensioning
No dam that failed was found to have been post-tensioned. The dams where there is no
information tend to be older dams (generally masonry) where post-tensioning is
unlikely.
(b) Gallery and Drains
Of the 46 dam failures, information could be found on the gallery and drains for 21
dams. Of these, only Zerbino Dam had drains present. The gallery was 4m above the
base of the dam with drains to the concrete-rock interface. Zerbino overtopped by 3m
causing erosion of the weak foundation rock at the toe, which resulted in foundation
sliding. It is not known what effect, if any, the drains had on the failure.
(c) Foundation Grouting
Of the 46 dam failures, information could be found on the foundation grouting for 20
dams. Of these, 2 dams had curtain grouting; three dams had consolidation grouting;
and one (Vega de Tera) had both. These dams are shown in
Table 2.38. The foundations of the other 14 failed dams were not grouted.
Table 2.38. Failed dams with grouted foundation
Dam
Name
Dam
Type
Grout
Type
Foundation
Geology
Failure Comments
Cheurfas PG(M) curtain limestone Failed in dam body
Austin (B) PG(M) curtain limestone/shale/
dolomite
Seepage softened fndn prior to
sliding - grouting inadequate
Zerbino PG consolidation hornfeld/schist Overtopped by 3m with toe
erosion then sliding
Chickahole PG(M) consolidation gneiss Failed in dam body
Bacino di
Rutte
VA(M) consolidation dolerite Concrete failure due to fndn
movement
Vega de Tera CB(M) both gneiss/schist Failed in dam body

(d) Shear Key
Bouzey Dam was the only failed dam found to have a shear key. The failure occurred
within the body of the dam.
(e) Radius of Curvature
Where information on the radius of curvature for failed gravity dams was available (15),
all but two dams had straight sections. Tigra Dam and St Francis Dam had radii of
curvature of 1000m and 152m respectively.
(f) Valley Shape
18 failed dams were found with information on the valley shape. The gradient of the
valley sides ranged from 0.06 to 2.0 (H/L) for gravity dams and 0.6 to 1.3 for arch
dams. The averages were 0.72 and 0.84 respectively.
Table 2.39 shows the ratio of crest length to dam height for both failed dams (where
information was available) and the population of dams. The structural height, H
d
, was
used for the failed dams. The population from ICOLD as described in Section 2.2.6 was
used for the comparison. Dams with composite embankment sections were omitted from
both the failure and population analyses. The data shows that the failures were in
relatively wide valleys (L
1
/H
d
3.1 for gravity dams) where three-dimensional effects
are unlikely to make a significant impact to the strength of the dams.
Elwha River Dam, a gravity dam which pipe failed, had a very narrow valley (11m)
with reasonably steep sides. However the failure was likely to be mainly due to the
alluvial foundation. No conclusive results were attained from this analysis.

Table 2.39. Crest length/height for failed dams and population
DAM FAILURES POPULATION
TYPE Number Range Mean Number Range Mean
Gravity 27 3.1-53 13.2 2887 0.3-182
(1)
10.1
Arch 5 2.9-4.3 3.6 663 0.2-29 3.8
Buttress 6 6.0-26 13.2 232 1.0-131 10.1
Multi-Arch 2 3.5-6.8 5.1 82 2.0-47 9.2
Note (1) 80% of the population of gravity dams has a crest length/height greater than 3.1.

(g) Upstream/Downstream Slopes
Table 2.40 shows the upstream and downstream slopes for the failed dams where the
information was available. Of the 15 gravity dams in the table, 13 had vertical or near
vertical upstream slopes. On the downstream face the concrete gravity dams ranged
from 0.55:1 (H:V) to 1:1. The masonry gravity dams ranged from 0.38:1 to 3:1. The
arch dams ranged from near vertical to 0.32:1.
Table 2.40. Upstream and downstream slopes for failed dams
Failure Mode Dam Name Dam Type Upstream
(xH:1V)
Downstream
(yH:1V)
Foundation Dam
Bayless (A) PG 0 1 S
Bayless (B) PG 0 1 S
Elwha River PG 0 0.75 P
St. Francis PG 0 1 S
Zerbino PG 0.05 0.55 S/SC
Angels PG(M) 0 0.6 P
Austin (A) PG(M) 0 0.38 SC/P/S
Bouzey PG(M) 0 1 T
Chickahole PG(M) 0.1 0.7 T
Habra (A) PG(M) 0.3 0.8 T/SH
Habra (B) PG(M) 3 1 T/SH
Habra (C) PG(M) 3 1 T/SH
Khadakwasla PG(M) 0.05 0.4 T/SH
Puentes PG(M) 0 0.6 P
Tigra PG(M) 0 0.67 S
Malpasset VA 0 0 S
Moyie River VA 0 0.06 SC
Vaughn Creek VA 0 0.2 P
Bacino di Rutte VA(M) 0.12 0.12 D/P
Gallinas VA(M) 0 0.32 ?
Meihua VA(M) 0 0 SH
Ashley CB 1 0.5 P
Stony Creek CB 1 0.15 P
Vega de Tera CB(M) 0.05 0.75 T/C
Austin (B) CB(M) 0 1 SH
Gleno MV 0.85 0.1 T/C


(h) Dam Height/Base Width (H
d
/W)
Table 2.41 shows the dam structural height and height of water at failure over base
width (H
d
/W and h
wf
/W respectively) for the failed dams where the information was
available. Figure 2.1 in Section 2.2.4.9 shows the definition of these terms. The H
d
/W
and/or h
wf
/W ratios give an indication of the stability and hydraulic gradient of the
dams. A high H
d
/W or h
wf
/W indicates a slender dam with potentially a high hydraulic
gradient. These are common for arch dams. The definitions for the failure modes are
given in Sections 2.4.2 and 2.4.3.
Dams that failed by piping generally had soil foundations. Those with alluvial
foundations had h
wf
/W ratios of 0.6 to 1.1. Vaughn Creek, an arch dam which pipe
failed through its extremely to highly weathered conglomerate abutment, had a ratio of
3.0. Austin (A), the only dam to have pipe failure through rock (weathered) had a h
wf
/W
of 1.2. Gravity dams that failed by sliding had h
wf
/W ratios of 1.2 to 2.1. Of these,
Zerbino Dam (h
wf
/W=2.1) was the only dam known to have drainage. Malpasset Dam,
an arch dam, had a h
wf
/W of 5.8.

Table 2.41. H
d
/W for failed dams
Dam Name Dam Type H
d
/W h
wf
/W Failure Mode
Foundation Dam
Bayless (A) PG 1.6 1.6 S
Bayless (B) PG 1.6 1.6 S
Elwha River PG 1.4 0.6 P
St. Francis PG 1.2 1.2 S
Zerbino PG 1.7 2.1 S/SC
Austin (A) PG(M) 1.0 1.2 SC/P/S
Bouzey PG(M) 1.7 1.7 T
Cheurfas PG(M) 1.0 ?
Chickahole PG(M) 1.3 1.0 ?
Fergoug I PG(M) 1.3 ?
Fergoug II PG(M) 1.3
Habra (A) PG(M) 1.3 T/SH
Habra (B) PG(M) 1.3 1.2 T/SH
Habra (C) PG(M) 1.3 1.4 T/SH
Khadakwasla PG(M) 1.8 2.0 T/SH
Puentes PG(M) 1.1 1.1 P
Tigra PG(M) 1.5 1.5 S
Malpasset VA 6.0 5.8 S
Moyie River VA 7.0 SC
Vaughn Creek VA 4.3 3.0 P/D
Gallinas VA(M) 3.1 3.2 ?
Meihua VA(M) 18.3 17.5 SH
Ashley CB 1.2 1.1 P
Stony River CB 1.0 0.9 P
Vega de Tera CB(M) 2.0 1.8 T/C
Austin (B) CB(M) 0.7 1.2 SH
Gleno MV 1.1 1.1 T/C

(i) Stability Analyses
Gulan (1995) and Rich (1995) collated information for 13 concrete gravity dams that
had failed by either sliding or overturning through their foundations or the concrete
mass. Of the 13 cases, nine failures were back analysed to determine the shear strength
properties of either the foundation or concrete. Table 2.42 shows the results from the
analyses, which have been checked and some adjustments to the cohesion results made.
The results are quoted as c=0, or c, =0. Actual strengths are between these limits.
The results for Khadakwasla Dam have been omitted as the analysis technique was not

valid for the failure mode. The failure plane for Khadakwasla Dam was 6m below the
base of the dam.
An additional analysis was carried out for Bhandardara Dam, an 82m high gravity dam
in India. The dam suffered extensive cracking, from an elevation of 39m at the upstream
face to the toe, and came close to failure. The dam has been extensively investigated
and several papers describe the accident including: Murthy et. al. (1976 & 1979); and
Kulkarni & Kulkarni (1994). Two simple analyses were carried out: the first assuming a
horizontal failure at the elevation where the cracking initiated; and the second assuming
an angled crack from the location of crack initiation to the toe. The results from the
analyses have been included in the tables and figures below.
Table 2.43 shows the reanalysed stresses along the failure planes. As can be seen seven
of the dams had tensile stresses, up to -280KPa at the heel of the dam. Bhandardara
Dam, a concrete gravity structure, experienced up to -440kPa tension.
Table 2.42. Back analysed shear strengths for failed dams (mod. from Rich, 1995)
Name Dam
Type
Failure

()
C
(KPa)
Foundation Concrete
Austin (A) PG(M) Foundation
sliding
49
0
0
120
limestone rubble limestone in
portland cement-
mortar
Bouzey
(1
st
)
PG(M) Foundation
sliding
40
0
0
110
sandstone &
schist
masonry in lime-
mortar
Bouzey
(2
nd
)
PG(M) Through
concrete
34
0
0
75
sandstone &
schist
masonry in lime-
mortar
El Habra
(3
rd
)
PG(M) Foundation
sliding
46
0
0
605
int. sandstone &
clay
rubble masonry in
lime-mortar
Tigra PG(M) Foundation
sliding
48
0
0
195
stratified
sandstone
rubble masonry in
lime-mortar
Bayless PG Foundation
sliding
43
0
0
300
int. sandstone &
shale
cyclopean concrete
St. Francis PG Foundation
sliding
41
0
0
155
mica schist &
conglomerate
portland cement
Bhandardara
(horizontal)
PG Severe
cracking -
tension &
shear
>46
0
0
>1015
basalt rubble masonry
Bhandardara
(angled)
PG severe
cracking -
tension &
shear
>71
0
0
>480
basalt rubble masonry

Table 2.43. Calculated normal stresses along the failure plane of back analysed
gravity dams
Name Dam
Type
n
Upstream
(KPa)
n
Downstream
(KPa)
Austin PG(M) -20 +210
Bouzey (1
st
) PG(M) -20 +265
Bouzey (2
nd
) PG(M) -10 +220
El Habra (3
rd
) PG(M) -280 +735
Tigra PG(M) +25 +355
Bayless PG -155 +425
St. Francis PG +35 +355
Bhandardara (horizontal) PG -440 +1085
Bhandardara (angled) PG -50 +320

The average stresses acting along the failure planes have been calculated using the
forces on each dam provided by Rich (1995). Figure 2.41 and Figure 2.42 compare the
ANCOLD guidelines (ANCOLD, 1991) to the failure stresses of the nine failure cases.
It was assumed that shear strength only acted in the region of compression along the
failure plane. The figures show that the failure stresses were much lower than those
recommended by ANCOLD for initial assessments. The likely reason for this is the
existence of continuous defects through or below the dam. The friction angle and
cohesion suggested by ANCOLD assumes no continuous defects. The results show the
importance of having a good geotechnical model for the dam and a good bond at the
dam/foundation interface.


0
200
400
600
800
1000
1200
1400
1600
1800
0 100 200 300 400 500 600
n
(kPa)
(kPa)
ANCOLD Bayless Austin
Bouzey I Habra St Francis Tigra

Figure 2.41. Average failure stresses for dams with failure through the foundation
0
500
1000
1500
2000
2500
0 200 400 600 800 1000
n
(kPa)
(kPa)
ANCOLD
Bhandardara (horizontal) Bhandardara (angled)

Figure 2.42. Average failure stresses for Bhandardara Dam


2.4 METHOD OF FIRST ORDER PROBABILITY ASSESSMENT
2.4.1 Probability of Failure
2.4.1.1 Introduction
This section describes an attempt to develop a first estimate of the annual probability
of failure of concrete and masonry dams based on the history of dam failures. Average
annual probabilities of failure have been assessed for all concrete and masonry dam
types. These probabilities have been further refined for concrete and masonry gravity
dams.
The initial or average annual probability of failure was calculated as the number of
dam failures, using the history of failures, over an estimate of the population of dams.
The cut off year for the population of dams was taken as 1992 as the latest ICOLD
statistics on failures (ICOLD, 1995) go up to this time. Dams were separated using the
following categories:
(a) Dam type: gravity, arch, buttress, multi-arch;
(b) year commissioned;
(c) age at failure (0-5 years and >5 years); and
(d) Concrete or masonry (gravity dams).

2.4.1.2 Population of Dams
The total number of concrete and masonry dams as at 1992 (excluding China) is shown
in Table 2.44. Since the ICOLD world population data for post 1983 was not available,
the population for the period 1983-1992 was estimated as shown in the table below.
Table 2.44. Number of dams as at 1992
Year
Commissioned
Number
of Dams
Reference
1700-1799 37 ICOLD (1983)
1800-1899 167 ICOLD (1983)
up to 1977 4446 ICOLD (1984)
1978-1982 217 ICOLD (1984)
1983-1992 434 estimated as 2 x 1978-82
Total 5097

Dams were divided into gravity, arch, buttress and multi-arch dams. Where a dam was
described as a composite section an assessment of the category best describing the dam
was made. The population was also split according to age (year commissioned) to
account for progress in the methods used for dam construction. The breakdown of the
population of dams into dam types and year commissioned was performed using a
computer database created by the author using ICOLD (1979). The database comprised
the concrete and masonry dams from the 26 countries with the largest dam populations.
These countries included all those that had experienced failures (excluding China).
Table 2.45 shows the percentage split for population of dams according to dam type and
year commissioned.


Table 2.45. Population of dams by dam type and year commissioned
Year
Commissioned
Gravity
(%)
Arch
(%)
Buttress
(%)
Multi-Arch
(%)
<1900 2.7 0.2 0.0 0.1
1900-1909 2.5 0.4 0.1 0.1
1910-1919 4.7 0.7 0.5 0.2
1920-1929 8.3 2.2 0.6 0.4
1930-1939 7.5 1.7 0.5 0.2
1940-1949 7.4 1.4 0.6 0.3
1950-1959 16.9 3.8 1.9 0.4
1960-1969 17.3 4.8 1.6 0.3
1970-1977 7.8 1.5 0.4 0.2
1977-1983
(1)
81.1 12.0 6.0 0.9
Note (1) Data from ICOLD (1983)
For dams commissioned during the period 1978 to 1982 the distribution of concrete and
masonry dam types was taken from ICOLD (1983). Dams commissioned between 1983
and 1992 were assumed to have a similar distribution of dam types. Table 2.46 shows
the number of dams as at 1992 calculated from Table 2.44 and Table 2.45.
Table 2.46. Number of dams (excluding China) in the population
Year
Commissioned
Gravity Arch Buttress Multi-Arch Total
1700-1799 34 2 0 1 37
1800-1899 152 10 1 4 167
1900-1909 109 17 4 3 133
1910-1919 205 31 21 11 267
1920-1929 362 94 27 18 501
1930-1939 327 75 20 11 433
1940-1949 321 62 28 12 422
1950-1959 738 164 85 17 1004
1960-1969 757 208 71 13 1049
1970-1977 339 67 18 8 433
1978-1982 176 26 13 2 217
1983-1992* 352 52 26 4 434
Total 3872 808 314 103 5097
Note (1) Estimated as 2 x 1977-1982


2.4.1.3 Dam Year
As most failures occur prior to five years after commissioning (Douglas et al, 1998) the
failure probabilities were broken into: less than or equal to five years of age; and greater
than five years of age. Equations 1 and 2 were used to calculate the number of dam
years for dams less than or equal to five years, and for dams greater than five years of
age respectively.
Y n
=
5
5 (2.1)
( ) Y y
i >
=
5
5 (2.2)
where,
n = total number of dams
y
i
= age of individual dam in years
2.4.1.4 Probabilities of Failure
Annual probabilities (number of failures/number of dam years) and straight probabilities
of failure (number of failures/number of dams) were calculated from the database of
failures and the population of dams.
A distinction was made between dams commissioned prior to, and those commissioned
after 1930. This represents the historical change to a better understanding of uplift
pressures and materials properties for dams. Categories without failures have been
denoted as NF.
The probabilities were recalculated for the various failure modes. The following failure
modes were used:
All modes (Table 2.47 and Table 2.48)
Sliding (Table 2.49 and Table 2.50)
Piping (Table 2.51 and Table 2.52)
Through the dam body (Table 2.53 and Table 2.54)
Table 2.55 and Table 2.56 show the number of failures with unknown failure modes.
Table 2.55 shows those unknowns where failure during overtopping was known to have
occurred.

Table 2.47. Annual probability of failure (1992, exc. China) - all failure types
Year Gravity Arch
Comm. 0-5 years
(1)
>5 years Total 0-5 years
(1)
>5 years Total
1700-1799 5.9E-03 NF 1.2E-04 NF NF NF
1800-1899 6.6E-03 3.8E-04 6.0E-04 NF NF NF
1900-1909 3.7E-03 2.2E-04 4.2E-04 NF NF NF
1910-1919 2.0E-03 1.4E-04 2.5E-04 NF 4.5E-04 4.2E-04
1920-1929 1.1E-03 8.9E-05 1.7E-04 4.2E-03 NF 3.2E-04
1930-1939 6.1E-04 NF 5.4E-05 NF NF NF
1940-1949 NF NF NF NF NF NF
1950-1959 NF NF NF 1.2E-03 1.9E-04 3.3E-04
1960-1969 5.3E-04 1.2E-04 2.0E-04 NF NF NF
1978-1982 NF NF NF 7.7E-03 NF 3.2E-03
1983-1992
(3)
NF NF NF NF NF NF
1700-1929 2.8E-03 1.9E-04 3.3E-04 2.6E-03 8.9E-05 2.5E-04
1930-1992
(3)
2.0E-04 2.6E-05 5.5E-05 6.2E-04 5.8E-05 1.5E-04
Total
(3)
7.9E-04 1.1E-04 1.8E-04 1.0E-03 7.0E-05 1.8E-04

Year Buttress Multi-Arch
Comm. 0-5 years
(1)
(1)
>5 years Total
1900-1909 4.7E-02 NF 2.7E-03 NF NF NF
1910-1919 1.9E-02 NF 1.2E-03 NF NF NF
1920-1929 1.5E-02 NF 1.1E-03 1.1E-02 NF 8.3E-04
1940-1949 7.3E-03 NF 7.7E-04 NF NF NF
1950-1959 2.4E-03 NF 3.2E-04 NF 1.8E-03 1.6E-03
1983-1992
(3)
NF NF NF NF NF NF
1700-1929 1.9E-02 NF 1.2E-03 5.4E-03 NF 3.2E-04
1930-1992
(3)
1.6E-03 NF 2.5E-04 NF 5.1E-04 4.3E-04
Total
(3)
4.5E-03 NF 5.8E-04 2.0E-03 2.0E-04 3.7E-04

Year All Concrete & Masonry
Comm. 0-5 years
(1)
>5 years Total
1700-1799 5.4E-03 NF 1.1E-04
1800-1899 6.0E-03 3.5E-04 5.5E-04
1900-1909 4.5E-03 1.8E-04 4.3E-04
1910-1919 3.0E-03 1.6E-04 3.4E-04
1920-1929 2.8E-03 6.4E-05 2.7E-04
1930-1939 4.6E-04 NF 4.1E-05
1940-1949 4.7E-04 NF 5.0E-05
1950-1959 4.0E-04 6.2E-05 1.1E-04
1960-1969 3.8E-04 8.7E-05 1.4E-04
1970-1977 NF NF NF
1978-1982 9.2E-04 NF 3.8E-04
1983-1992
(3)
NF NF NF
1700-1929 3.6E-03 1.6E-04 3.6E-04
1930-1992
(3)
3.6E-04 3.9E-05 9.0E-05
Total
(3)
1.1E-03 9.7E-05 2.1E-04
Notes

(1) Assumes dam years =
number of dams * five years
life
(2) NF - No Failure
(3) Assumes number of dams
constructed in 1983-1992 =
2 * number of dams in
1978-1982


Table 2.48. Probability of failure (as at 1992, exc. China, non-annualised) -
all failure types
Year Gravity Arch
Comm. 0-5 years >5 years Total 0-5 years >5 years Total
1700-1799 3.0E-02 NF

3.0E-02 NF NF NF
1800-1899 3.3E-02 5.3E-02 8.5E-02 NF NF NF
1900-1909 1.8E-02 1.8E-02 3.7E-02 NF NF NF
1910-1919 9.8E-03 9.8E-03 2.0E-02 NF 3.3E-02 3.3E-02
1920-1929 5.5E-03 5.5E-03 1.1E-02 2.1E-02 NF 2.1E-02
1930-1939 3.1E-03 NF 3.1E-03 NF NF NF
1950-1959 NF NF NF 6.1E-03 6.1E-03 1.2E-02
1960-1969 2.6E-03 2.6E-03 5.3E-03 NF NF NF
1978-1982 NF NF NF 3.8E-02 NF 3.8E-02
1983-1992

NF NF NF NF NF NF
1700-1929 1.4E-02 1.6E-02 3.0E-02 1.3E-02 6.5E-03 2.0E-02
1930-1992

1.0E-03 6.6E-04 1.7E-03 3.1E-03 1.5E-03 4.6E-03
Total

3.9E-03 4.1E-03 8.0E-03 5.0E-03 2.3E-03 7.3E-03

1900-1909 2.4E-01 NF 2.4E-01 NF NF NF
1910-1919 9.4E-02 NF 9.4E-02 NF NF NF
1920-1929 7.5E-02 NF 7.5E-02 5.5E-02 NF 5.5E-02
1940-1949 3.6E-02 NF 3.6E-02 NF NF NF
1950-1959 1.2E-02 NF 1.2E-02 NF 5.9E-02 5.9E-02
1983-1992

NF NF NF NF NF NF
1700-1929 9.3E-02 NF 9.3E-02 2.7E-02 NF 2.7E-02
1930-1992

7.7E-03 NF 7.7E-03 NF 1.5E-02 1.5E-02
Total

2.2E-02 NF 2.2E-02 9.7E-03 9.7E-03 1.9E-02

Comm. 0-5 years >5 years Total
1700-1799 2.7E-02 NF 2.7E-02
1800-1899 3.0E-02 4.8E-02 7.8E-02
1900-1909 2.2E-02 1.5E-02 3.7E-02
1910-1919 1.5E-02 1.1E-02 2.6E-02
1920-1929 1.4E-02 4.0E-03 1.8E-02
1930-1939 2.3E-03 NF 2.3E-03
1940-1949 2.4E-03 NF 2.4E-03
1950-1959 2.0E-03 2.0E-03 4.0E-03
1960-1969 1.9E-03 1.9E-03 3.8E-03
1970-1977 NF NF NF
1978-1982 4.6E-03 NF 4.6E-03
1983-1992

NF NF NF
1700-1929 1.8E-02 1.4E-02 3.2E-02
1930-1992

1.8E-03 1.0E-03 2.8E-03
Total

5.3E-03 3.7E-03 9.0E-03


Table 2.49. Annual probability of failure (as at 1992, excluding China) - sliding
failures
Year Gravity Arch
Comm. 0-5 years
(1)
(1)
>5 years Total
1800-1899 1.3E-03 NF 4.6E-05 NF NF NF
1900-1909 3.7E-03 1.1E-04 3.2E-04 NF NF NF
1910-1919 9.8E-04 NF 6.3E-05 NF NF NF
1920-1929 5.5E-04 4.5E-05 8.3E-05 NF NF NF
1950-1959 NF NF NF 1.2E-03 NF 1.6E-04
1983-1992
(3)
NF NF NF NF NF NF
1700-1929 1.2E-03 2.7E-05 8.8E-05 NF NF NF
1930-1992
(3)
NF NF NF 3.1E-04 NF 4.9E-05
Total
(3)
2.6E-04 1.3E-05 4.1E-05 2.5E-04 NF 3.1E-05

Comm. 0-5 years
(1)
(1)
>5 years Total
1920-1929 7.5E-03 NF 5.6E-04 NF NF NF
1983-1992
(3)
NF NF NF NF NF NF
1700-1929 3.7E-03 NF 2.5E-04 NF NF NF
1930-1992
(3)
NF NF NF NF NF NF
Total
(3)
6.5E-04 NF 8.3E-05 NF NF NF

Comm. 0-5 years
(1)
>5 years Total
1700-1799 NF NF NF
1800-1899 1.2E-03 NF 4.2E-05
1900-1909 3.0E-03 9.1E-05 2.6E-04
1910-1919 7.5E-04 NF 4.9E-05
1920-1929 8.0E-04 3.2E-05 8.9E-05
1930-1939 NF NF NF
1940-1949 NF NF NF
1950-1959 2.0E-04 NF 2.7E-05
1960-1969 NF NF NF
1970-1977 NF NF NF
1978-1982 NF NF NF
1983-1992
(3)
NF NF NF
1700-1929 1.1E-03 2.2E-05 8.1E-05
1930-1992
(3)
5.1E-05 NF 8.2E-06
Total
(3)
2.8E-04 1.0E-05 4.1E-05
Notes

life
(2) NF - No Failure
2 * number of dams in 1978-
1982


sliding failures
Year Gravity Arch
1700-1799 NF NF

NF NF NF NF
1800-1899 6.6E-03 NF 6.6E-03 NF NF NF
1900-1909 1.8E-02 9.2E-03 2.7E-02 NF NF NF
1910-1919 4.9E-03 NF 4.9E-03 NF NF NF
1920-1929 2.8E-03 2.8E-03 5.5E-03 NF NF NF
1950-1959 NF NF NF 6.1E-03 NF 6.1E-03
1983-1992

NF NF NF NF NF NF
1700-1929 5.8E-03 2.3E-03 8.1E-03 NF NF NF
1930-1992

NF NF NF 1.5E-03 NF 1.5E-03
Total 1.3E-03 5.2E-04 1.8E-03 1.2E-03 NF 1.2E-03

Comm.
0-5 years >5 years Total 0-5 years >5 years Total
1700-1799
NF NF NF NF NF NF
1800-1899
NF NF NF NF NF NF
1900-1909
NF NF NF NF NF NF
1910-1919
NF NF NF NF NF NF
1920-1929
3.8E-02 NF 3.8E-02 NF NF NF
1930-1939
NF NF NF NF NF NF
1940-1949
NF NF NF NF NF NF
1950-1959
NF NF NF NF NF NF
1960-1969
NF NF NF NF NF NF
1970-1977
NF NF NF NF NF NF
1978-1982
NF NF NF NF NF NF
1983-1992

NF NF NF NF NF NF
1700-1929
1.9E-02 NF 1.9E-02 NF NF NF
1930-1992

NF NF NF NF NF NF
Total

3.2E-03 NF 3.2E-03 NF NF NF

Comm.
0-5 years >5 years Total
1700-1799
NF NF NF
1800-1899
6.0E-03 NF 6.0E-03
1900-1909
1.5E-02 7.5E-03 2.2E-02
1910-1919
3.7E-03 NF 3.7E-03
1920-1929
4.0E-03 2.0E-03 6.0E-03
1930-1939
NF NF NF
1940-1949
NF NF NF
1950-1959
1.0E-03 NF 1.0E-03
1960-1969
NF NF NF
1970-1977
NF NF NF
1978-1982
NF NF NF
1983-1992

NF NF NF
1700-1929
5.4E-03 1.8E-03 7.2E-03
1930-1992

2.5E-04 NF 2.5E-04
Total

1.4E-03 3.9E-04 1.8E-03


Table 2.51. Annual probability of failure (as at 1992, excluding China) - piping
failures
Year Gravity Arch
Comm. 0-5 years
(1)
(1)
>5 years Total
1700-1799 5.9E-03 NF 1.2E-04 NF NF NF
1800-1899 1.3E-03 NF 4.6E-05 NF NF NF
1910-1919 9.8E-04 NF 6.3E-05 NF NF NF
1920-1929 NF NF NF 2.1E-03 NF 1.6E-04
1950-1959 NF NF NF NF 1.9E-04 1.6E-04
1983-1992
(3)
NF NF NF NF NF NF
1700-1929 7.0E-04 NF 3.8E-05 1.3E-03 NF 8.3E-05
1930-1992
(3)
NF NF NF NF 5.8E-05 4.9E-05
Total
(3)
1.6E-04 NF 1.8E-05 2.5E-04 3.5E-05 6.1E-05

Comm. 0-5 years
(1)
(1)
>5 years Total
1700-1799
NF NF NF NF NF NF
1800-1899
NF NF NF NF NF NF
1900-1909
4.7E-02 NF 2.7E-03 NF NF NF
1910-1919
9.4E-03 NF 6.1E-04 NF NF NF
1920-1929
NF NF NF NF NF NF
1930-1939
NF NF NF NF NF NF
1940-1949
7.3E-03 NF 7.7E-04 NF NF NF
1950-1959
NF NF NF NF NF NF
1960-1969
NF NF NF NF NF NF
1970-1977
NF NF NF NF NF NF
1978-1982
NF NF NF NF NF NF
1983-1992
(3)
NF NF NF NF NF NF
1700-1929
7.5E-03 NF 4.9E-04 NF NF NF
1930-1992
(3)
7.8E-04 NF 1.2E-04 NF NF NF
Total
(3)
1.9E-03 NF 2.5E-04 NF NF NF

Comm. 0-5 years
(1)
>5 years Total
1700-1799
5.4E-03 NF 1.1E-04
1800-1899
1.2E-03 NF 4.2E-05
1900-1909
1.5E-03 NF 8.6E-05
1910-1919
1.5E-03 NF 9.7E-05
1920-1929
4.0E-04 NF 3.0E-05
1930-1939
NF NF NF
1940-1949
4.7E-04 NF 5.0E-05
1950-1959
NF 3.1E-05 2.7E-05
1960-1969
NF NF NF
1970-1977
NF NF NF
1978-1982
NF NF NF
1983-1992
(3)
NF NF NF
1700-1929
1.1E-03 NF 6.1E-05
1930-1992
(3)
5.1E-05 9.8E-06 1.6E-05
Total
(3)
2.8E-04 5.1E-06 3.6E-05
Notes

life
(2) NF - No Failure
1982


piping failures
Year Gravity Arch
1700-1799 3.0E-02 NF

3.0E-02 NF NF NF
1800-1899 6.6E-03 NF 6.6E-03 NF NF NF
1910-1919 4.9E-03 NF 4.9E-03 NF NF NF
1920-1929 NF NF NF 1.1E-02 NF 1.1E-02
1950-1959 NF NF NF NF 6.1E-03 6.1E-03
1983-1992

NF NF NF NF NF NF
1700-1929 3.5E-03 NF 3.5E-03 6.5E-03 NF 6.5E-03
1930-1992

NF NF NF NF 1.5E-03 1.5E-03
Total 7.7E-04 NF 7.7E-04 1.2E-03 1.2E-03 2.5E-03

Comm.
1700-1799
NF NF NF NF NF NF
1800-1899
NF NF NF NF NF NF
1900-1909
2.4E-01 NF 2.4E-01 NF NF NF
1910-1919
4.7E-02 NF 4.7E-02 NF NF NF
1920-1929
NF NF NF NF NF NF
1930-1939
NF NF NF NF NF NF
1940-1949
3.6E-02 NF 3.6E-02 NF NF NF
1950-1959
NF NF NF NF NF NF
1960-1969
NF NF NF NF NF NF
1970-1977
NF NF NF NF NF NF
1978-1982
NF NF NF NF NF NF
1983-1992

NF NF NF NF NF NF
1700-1929
3.7E-02 NF 3.7E-02 NF NF NF
1930-1992

3.8E-03 NF 3.8E-03 NF NF NF
Total

9.5E-03 NF 9.5E-03 NF NF NF

Comm.
1700-1799
2.7E-02 NF 2.7E-02
1800-1899
6.0E-03 NF 6.0E-03
1900-1909
7.5E-03 NF 7.5E-03
1910-1919
7.5E-03 NF 7.5E-03
1920-1929
2.0E-03 NF 2.0E-03
1930-1939
NF NF NF
1940-1949
2.4E-03 NF 2.4E-03
1950-1959
NF 1.0E-03 1.0E-03
1960-1969
NF NF NF
1970-1977
NF NF NF
1978-1982
NF NF NF
1983-1992

NF NF NF
1700-1929
5.4E-03 NF 5.4E-03
1930-1992

2.5E-04 2.5E-04 5.0E-04
Total

1.4E-03 2.0E-04 1.6E-03


Table 2.53. Annual probability of failure (as at 1992, excluding China) -
tension/shear failures through dam body
Year Gravity Arch
Comm. 0-5 years
(1)
(1)
>5 years Total
1800-1899 2.6E-03 2.4E-04 3.2E-04 NF NF NF
1920-1929 NF 4.5E-05 4.1E-05 NF NF NF
1960-1969 2.6E-04 6.0E-05 9.8E-05 NF NF NF
1978-1982 NF NF NF 7.7E-03 NF 3.2E-03
1983-1992
(3)
NF NF NF NF NF NF
1700-1929 4.6E-04 8.0E-05 1.0E-04 NF NF NF
1930-1992
(3)
6.8E-05 1.3E-05 2.2E-05 3.1E-04 NF 4.9E-05
Total
(3)
1.6E-04 4.6E-05 5.9E-05 2.5E-04 NF 3.1E-05

Comm. 0-5 years
(1)
(1)
>5 years Total
1700-1799
NF NF NF NF NF NF
1800-1899
NF NF NF NF NF NF
1900-1909
NF NF NF NF NF NF
1910-1919
9.4E-03 NF 6.1E-04 NF NF NF
1920-1929
NF NF NF 1.1E-02 NF 8.3E-04
1930-1939
NF NF NF NF NF NF
1940-1949
NF NF NF NF NF NF
1950-1959
2.4E-03 NF 3.2E-04 NF 1.8E-03 1.6E-03
1960-1969
NF NF NF NF NF NF
1970-1977
NF NF NF NF NF NF
1978-1982
NF NF NF NF NF NF
1983-1992
(3)
NF NF NF NF NF NF
1700-1929
3.7E-03 NF 2.5E-04 5.4E-03 NF 3.2E-04
1930-1992
(3)
7.8E-04 NF 1.2E-04 NF 5.1E-04 4.3E-04
Total
(3)
1.3E-03 NF 1.7E-04 2.0E-03 2.0E-04 3.7E-04

Comm. 0-5 years
(1)
>5 years Total
1700-1799
NF NF NF
1800-1899
2.4E-03 2.2E-04 3.0E-04
1900-1909
NF NF NF
1910-1919
7.5E-04 NF 4.9E-05
1920-1929
4.0E-04 3.2E-05 6.0E-05
1930-1939
NF NF NF
1940-1949
NF NF NF
1950-1959
2.0E-04 3.1E-05 5.4E-05
1960-1969
1.9E-04 4.3E-05 7.1E-05
1970-1977
NF NF NF
1978-1982
9.2E-04 NF 3.8E-04
1983-1992
(3)
NF NF NF
1700-1929
7.2E-04 6.5E-05 1.0E-04
1930-1992
(3)
1.5E-04 2.0E-05 4.1E-05
Total
(3)
2.9E-04 4.0E-05 6.8E-05
Notes

life
(2) NF - No Failure
1982


tension/shear failures through dam body
Year Gravity Arch
1700-1799 NF NF

NF NF NF NF
1800-1899 1.3E-02 3.3E-02 4.6E-02 NF NF NF
1920-1929 NF 2.8E-03 2.8E-03 NF NF NF
1960-1969 1.3E-03 1.3E-03 2.6E-03 NF NF NF
1978-1982 NF NF NF 3.8E-02 NF 3.8E-02
1983-1992

NF NF NF NF NF NF
1700-1929 2.3E-03 7.0E-03 9.3E-03 NF NF NF
1930-1992

3.3E-04 3.3E-04 6.6E-04 1.5E-03 NF 1.5E-03
Total 7.7E-04 1.8E-03 2.6E-03 1.2E-03 NF 1.2E-03

Comm.
1700-1799
NF NF NF NF NF NF
1800-1899
NF NF NF NF NF NF
1900-1909
NF NF NF NF NF NF
1910-1919
4.7E-02 NF 4.7E-02 NF NF NF
1920-1929
NF NF NF 5.5E-02 NF 5.5E-02
1930-1939
NF NF NF NF NF NF
1940-1949
NF NF NF NF NF NF
1950-1959
1.2E-02 NF 1.2E-02 NF 5.9E-02 5.9E-02
1960-1969
NF NF NF NF NF NF
1970-1977
NF NF NF NF NF NF
1978-1982
NF NF NF NF NF NF
1983-1992

NF NF NF NF NF NF
1700-1929
1.9E-02 NF 1.9E-02 2.7E-02 NF 2.7E-02
1930-1992

3.8E-03 NF 3.8E-03 NF 1.5E-02 1.5E-02
Total

6.4E-03 NF 6.4E-03 9.7E-03 9.7E-03 1.9E-02

Comm.
1700-1799
NF NF NF
1800-1899
1.2E-02 3.0E-02 4.2E-02
1900-1909
NF NF NF
1910-1919
3.7E-03 NF 3.7E-03
1920-1929
2.0E-03 2.0E-03 4.0E-03
1930-1939
NF NF NF
1940-1949
NF NF NF
1950-1959
1.0E-03 1.0E-04 2.0E-03
1960-1969
9.5E-04 9.5E-04 1.9E-03
1970-1977
NF NF NF
1978-1982
4.6E-03 NF 4.6E-03
1983-1992

NF NF NF
1700-1929
3.6E-03 5.4E-03 9.0E-03
1930-1992

7.5E-04 5.0E-04 1.3E-03
Total

1.4E-03 1.6E-03 2.9E-03


Table 2.55. Number of failures during overtopping where the failure mode was
unknown
Year Gravity Arch
1800-1899 NF 3 3 NF NF NF
1900-1909 NF 1 1 NF NF NF
1910-1919 NF 1 1 NF 1 1
1920-1929 NF NF NF 1 NF 1
1700-1929 NF 5 5 1 1 2
Total NF 5 5 1 1 2

Comm.
1700-1799
NF NF NF NF NF NF
1800-1899
NF NF NF NF NF NF
1900-1909
NF NF NF NF NF NF
1910-1919
NF NF NF NF NF NF
1920-1929
1 NF 1 NF NF NF
1930-1939
NF NF NF NF NF NF
1940-1949
NF NF NF NF NF NF
1950-1959
NF NF NF NF NF NF
1960-1969
NF NF NF NF NF NF
1970-1977
NF NF NF NF NF NF
1978-1982
NF NF NF NF NF NF
1983-1992

NF NF NF NF NF NF
1700-1929
1 NF 1 NF NF NF
1930-1992

NF NF NF NF NF NF
Total

1 NF 1 NF NF NF

Comm.
1700-1799
NF NF NF
1800-1899
NF 3 3
1900-1909
NF 1 1
1910-1919
NF 2 2
1920-1929
2 NF 2
1930-1939
NF NF NF
1940-1949
NF NF NF
1950-1959
NF NF NF
1960-1969
NF NF NF
1970-1977
NF NF NF
1978-1982
NF NF NF
1983-1992

NF NF NF
1700-1929
2 6 7
1930-1992

NF NF NF
Total

2 6 8


Table 2.56. No. of failures where the failure mode was unknown (no overtopping)
Year Gravity Arch
1800-1899 1 NF 1 NF NF NF
1910-1919 NF 1 1 NF NF NF
1920-1929 1 NF 1 NF NF NF
1930-1939 1 NF 1 NF NF NF
1960-1969 1 1 2 NF NF NF
1960-1992 1 1 2 NF NF NF
1700-1929 2 1 3 NF NF NF
1930-1992 2 1 3 NF NF NF
Total 4 2 6 NF NF NF

Comm.
Total NF NF NF NF NF NF

Comm.
1700-1799 NF NF NF
1800-1899 1 NF 1
1900-1909 NF NF NF
1910-1919 NF 1 1
1920-1929 1 NF 1
1930-1939 1 NF 1
1940-1949 NF NF NF
1950-1959 NF NF NF
1960-1969 1 1 2
1970-1977 NF NF NF
1978-1982 NF NF NF
1983-1992 NF NF NF
1960-1982 NF NF NF
1960-1992 1 1 2
1700-1929 2 1 4
1930-1992 2 1 2
Total 4 2 6


2.4.1.5 Gravity Dams - Separation of Concrete and Masonry Dams
The ICOLD(1984) population for gravity dams does not distinguish between dams
made of concrete and those made of masonry. An estimate was made for the population
taking into account the history of dam building and the USA population of dams. Dams
of cyclopean concrete construction were assumed to be concrete.
According to Smith (1972), Schnitter (1994) and Lewis (1988) the first concrete dams
were completed in the 1870s in Australia and the USA; the 1890s in India; and the
1900s in Great Britain. The distribution of concrete and masonry gravity dams in the
USA was taken from the 567 concrete and masonry dams in the US Inventory of dams
(1994) and is presented in Table 2.57.
Table 2.57. Distribution of concrete and masonry gravity dams in the USA
Year
Commissioned
Concrete
(%)
Masonry
(%)
Pre 1900 68.4 31.6
1900-1909 76.5 23.5
1910-1919 93.7 6.3
1920-1929 96.3 3.7
1930-1939 98.3 1.7
1940-1949 100 0
1950-1959 98.9 1.1
1960-1969 100 0
1970-1979 100 0
1980-1989 100 0
1990-1992 100 0

Table 2.57 was not used directly as this was likely to be biased towards concrete dams
due to the modern nature of USA dams compared to much of the rest of the world. It is
also possible that some dams denoted as gravity, and therefore assumed to be
concrete, in the US database are masonry. Some countries such as India, which has
approximately 3.2% of the world concrete and masonry dam population (ICOLD,
1994), commonly use masonry to construct their dams due to material availability and
expense. Table 2.58 shows the distribution chosen for the analysis. It was found that the
probabilities of failure were not sensitive to the assumptions in the concrete/masonry
distribution for the post 1960 period.

Table 2.61 to Table 2.64 show the annualised probabilities of failure for concrete and
masonry dams for the various failure modes. Table 2.65 and Table 2.66 show the
number of failures with unknown failure modes. Table 2.65 shows those unknowns
where failure during overtopping was known to have occurred. A distinction was made
between dams commissioned prior to, and those commissioned after 1930. This
represents the historical change to a better understanding of uplift pressures and
materials properties for gravity dams. Table 2.59 summarises the annualised
probabilities of failure using this distinction. As there were a number of categories
without failures (denoted NF) a maximum annual probability (assuming one failure
to have occurred over the number of dam years) has been calculated and included in the
last row of Table 2.59.
Table 2.51 gives suggested average annualised probabilities of failure for concrete and
masonry gravity dams based on Table 2.59. Unknowns were accounted for by
distributing them evenly through the three dam failure modes (foundation sliding and
piping and failure within the dam body). This allowed for the total probability to be
equal to the sum of the three modes. The probabilities have been rounded down (to one
decimal place) to account for the assumptions in the analysis. In particular, the
population used was that in existence as at 1992 and many dams are likely to have been
decommissioned prior to this time or omitted from the ICOLD database and hence not
included in the population. A larger population would result in lower probabilities of
failure. This was checked for validity by assuming a larger population and re-running
the analysis. Where no failures have occurred the suggested value is lower than that for
the case where one failure had occurred.

Table 2.58. Distribution of concrete and masonry gravity dams chosen for analysis
Year
Commissioned
Concrete
(%)
Masonry
(%)
Pre 1900 0/30 100/70
1900-1909 60 40
1910-1919 75 25
1920-1929 90 10
1930-1939 90 10
1940-1949 95 5
1950-1959 95 5
1960-1969 97.5 2.5
1970-1979 97.5 2.5
1980-1989 97.5 2.5
1990-1992 97.5 2.5
Note (1) 1700-1799/1800-1899
Table 2.59. Summary of annualised probabilities of failure for gravity dams (exc.
China)
Concrete Gravity Masonry Gravity Failure
Mode
Year
Commissioned
1700-1929 1.0E-03 9.3E-05 1.5E-04 5.2E-03 3.4E-04 5.4E-04 All
Modes 1930-1992 1.4E-04 1.4E-05 3.5E-05 1.6E-03 2.4E-04 4.2E-04
1700-1929 6.7E-04 7.0E-05 1.1E-04 1.5E-03 NF 6.0E-05 Foundation
Sliding 1930-1992 NF NF NF NF NF NF
1700-1929 3.4E-04 NF 2.2E-05 1.5E-03 NF 6.0E-05 Foundation
Piping
1700-1929 NF NF NF 7.3E-04 1.6E-04 1.8E-04 Within Dam
Body 1930-1992 7.1E-05 NF 1.1E-05 NF 2.4E-04 2.1E-04
1700-1929 3.3E-04 2.3E-05 2.2E-05 7.3E-04 3.1E-05 3.0E-05 Max.
No Fails
(1)
1930-1992 7.0E-05 1.4E-05 1.1E-05 1.6E-03 2.4E-04 2.1E-04
1700-1929 - - - - 6 6 Unknown
(O/T)
1930-1992 - - - - - -
1700-1929 - 1 1 3 - 3 Unknown
1930-1992 2 2 4 2 - 2
Note (1) Assuming 1 failure (for where no failures have occurred)

Table 2.60. Suggested values for annualised probabilities of failure for gravity
dams (excluding China)
Concrete Gravity Masonry Gravity Failure
Mode
Year
Commissioned 0-5 years >5 years 0-5 years >5 years
pre 1930 N/A 6.4E-05
2
N/A 3.2E-04
2
All Failures
1930-present 1.3E-04
2
1.2E-05
2
1.5E-03
2
2.4E-04
2

pre 1930 N/A 5.0E-05
2
N/A 6.0E-05
1
Foundation Sliding
P
SA
1
4.0E-06
1
5.0E-04
1
2.0E-05
1

pre 1930 N/A 7.0E-06
1
N/A 6.0E-05
2
Foundation Piping
P
PA
1
4.0E-06
1
5.0E-04
1
2.0E-05
1

pre 1930 N/A 7.0E-06
1
N/A 2.0E-04
2
Within Dam Body
P
BA
2
4.0E-06
1
5.0E-04
1
2.0E-04
2

Note: (1) No failures, probability estimated lower than that for one failure.
(2) Probability rounded down to account for smaller than actual population used in the analysis.

Table 2.61. Annualised probabilities of failure for gravity dams - all failures
Concrete Gravity Masonry Gravity Year
Commissioned 0-5 years >5 years Total 0-5 years >5 years Total
1700-1799 NF NF NF 5.9E-03 NF 1.2E-04
1800-1899 NF NF NF 7.5E-03 5.5E-04 7.9E-04
1900-1909 3.0E-03 3.7E-04 5.2E-04 NF 2.7E-04 2.6E-04
1910-1919 1.3E-03 8.9E-05 1.7E-04 3.8E-03 2.7E-04 5.0E-04
1920-1929 6.0E-04 4.9E-05 9.0E-05 5.4E-03 4.4E-04 8.1E-04
1930-1939 NF NF NF 6.0E-03 NF 5.3E-04
1960-1969 5.3E-04 6.1E-05 1.5E-04 NF 2.4E-03 1.9E-03
1700-1929 1.0E-03 9.2E-05 1.5E-04 5.1E-03 3.4E-04 5.4E-04
1930-1992 1.4E-04 1.4E-05 3.4E-05 1.6E-03 2.4E-04 4.2E-04
Total 2.9E-04 4.3E-05 7.5E-05 4.0E-03 3.3E-04 5.2E-04
Table 2.62. Annualised probabilities of failure for gravity dams - sliding failures
1800-1899 NF NF NF 1.9E-03 NF 6.6E-05
1900-1909 3.0E-03 3.7E-04 5.2E-04 NF NF NF
1910-1919 NF NF NF 3.8E-03 NF 2.5E-04
1920-1929 6.0E-04 4.9E-05 9.0E-05 NF NF NF
1700-1929 6.7E-04 6.9E-05 1.1E-04 1.5E-03 NF 6.0E-05
Total 1.2E-04 2.6E-05 3.7E-05 1.0E-03 NF 5.2E-05


Table 2.63. Annualised probabilities of failure for gravity dams - piping failures
1700-1799 NF NF NF 5.9E-03 NF 1.2E-04
1800-1899 NF NF NF 1.9E-03 NF 6.6E-05
1910-1919 1.3E-03 NF 8.3E-05 NF NF NF
1700-1929 3.3E-04 NF 2.2E-05 1.5E-03 NF 6.0E-05
Total 5.8E-05 NF 7.5E-06 1.0E-03 NF 5.2E-05

Table 2.64. Annualised probabilities of failure for gravity dams - dam body
tension/shear failures
1800-1899 NF NF NF 1.9E-03 2.7E-04 3.3E-04
1920-1929 NF NF NF NF 4.4E-04 4.1E-04
1960-1969 2.7E-04 NF 4.9E-05 NF 2.4E-03 1.9E-03
1960-1982 1.6E-04 NF 3.6E-05 NF 1.8E-03 1.4E-03
1960-1992 1.3E-04 NF 3.4E-05 NF 1.8E-03 1.3E-03
1700-1929 NF NF NF 7.3E-04 1.6E-04 1.8E-04
1930-1992 7.0E-05 NF 1.1E-05 NF 2.4E-04 2.1E-04
Total 5.8E-05 NF 7.5E-06 5.0E-04 1.6E-04 1.8E-04


Table 2.65. Number of failures during overtopping where failure mode was
unknown
1800-1899 NF NF NF NF 4 4
1900-1909 NF NF NF NF 1 1
1910-1919 NF NF NF NF 1 1
Total NF NF NF NF 6 6
Table 2.66. Number of failures where failure mode was unknown
1800-1899 NF NF NF 2 NF 2
1910-1919 NF 1 1 NF NF NF
1920-1929 NF NF NF 1 NF 1
1930-1939 NF NF NF 1 NF 1
1960-1969 1 1 2 NF NF NF
1930-1992 1 1 2 1 NF 1
Total 1 2 3 4 NF 4

2.4.2 General Approach for Estimating the Probability of Failure for
Individual Gravity Dams
Not all dams can be considered as average. Corrections can be made to the average
probabilities so that they can be used for particular dams. The following describes a
method to assess multiplication factors for concrete and masonry gravity dams that can
be applied to the average probabilities from the previous section for better or worse
than average dams. The method is for gravity dams that have a straight axis (no
curvature) and are not post-tensioned.
Where a dam is constructed of masonry but can be shown to be of a quality comparable
to that of a good concrete gravity dam, the average annual probability may be taken as
somewhere between that for masonry and that for concrete.
Where a dam has been raised and the full supply level (FSL) increased, the dam should
be treated as a new dam and the age of the dam calculated from this time. That is, the
dam should fall back into the 0-5 years category. This stems from the Section 2.3.4 that
showed that dams have generally failed at or just above their highest recorded water
level.
If the dam is of good design, is very well drained, has good uplift monitoring AND the
dam foundation has been assessed by a suitably qualified rock mechanics practitioner
and found to easily satisfy present day standards then a reduction factor, f
red
, of between
0.9 and 0.1 can be used. This factor should be applied to the annual probability of
failure in Equation 2.6. This factor can NOT be applied to dams with soil foundations
and should NOT be used for initial dam screening assessments where the data available
and the level of investigation and analysis are limited.

2.4.3 Details of the Method for Estimating the Probability of Failure for
Individual Gravity Dams
The following summarises the suggested procedure for estimating the annual probability
of failure of a concrete or masonry gravity dam. The annual probability of failure of the
dam, P, should be calculated as the sum of the probabilities of failure for sliding, piping
and through the dam body.
Sliding through the foundation:
Step (1) Determine the average annual probability of failure, P
SA
, from Table 2.51 in
Section 2.4.1.5.
(2) Determine the multiplication factor for sliding on a soil or rock foundation,
f
SF
, from Table 2.72 in Section 2.4.4.1.
(3) If the foundation is rock go to Step (4), if it is soil go to Step (5).
(4) Determine the geology type factor, f
SG,
from

Table 2.75 in Section 2.4.4.2, then go to Step (6)
(5) f
SG
= 1.0
(6) Determine the structural height/width factor, f
H/W
, from Table 2.77 in Section
2.4.4.4.
(7) Determine the other observations factor, f
O
, from Section 2.4.4.5.
(8) Determine the surveillance factor, f
S
(9) Calculate the probability of a foundation sliding failure as:
P P f f f f f
S SA SF SG H W O S
=
/
(2.3)

Piping through the foundation:
Step (1) Determine the average annual probability of failure, P
PA
, from Table 2.51 in
Section 2.4.1.5.
(2) Determine the multiplication factor for piping on a soil or rock foundation,
f
PF
(3) If the foundation is rock go to Step (4), if it is soil go to Step (5).
(4) Determine the geological environment, f
GE,
factor from Section 2.4.4.3, then
go to Step (6).
(5) f
GE
= 1.0
H/W
2.4.4.4.
O
S
(9) Calculate the probability of a foundation piping failure as:
P P f f f f f
P PA PF GE H W O S
=
/
(2.4)

Failure through the dam body:
Step (1) Determine the average annual probability, P
BA,
of failure from Table 2.51 in
Section 2.4.1.5.
H/W
2.4.4.4.
O
S
(5) Calculate the probability of a failure through the dam body as:
P P f f f
B BA H W O S
=
/
(2.5)
Total annual probability of failure:
( )
B P S red
P P P f P + + = (2.6)
where,
f
red
= Reduction factor, only applied when conditions described in Section
2.4.2 are satisfied.

2.4.4 Gravity Dam Probability Multiplication Factors
The following outlines the basis for assigning the multiplication factors. The factors,
where possible, have been based on the failure statistics in the previous sections. Where
necessary the accident statistics have been used to assist in developing the
multiplication factors. It should be noted however, that most of the dam accidents were
theoretical (eg. a calculation was performed that indicated the dam was unsafe and it
was anchored) and as such of little value to this exercise.
2.4.4.1 Soil/Rock Foundation Factor, f
SF
and f
PF

The probability of a dam failing through the foundation is highly dependent on whether
the foundation is soil and/or rock. An estimation of the multiplication factors for sliding
and piping of gravity dams on soil and rock foundations is outlined below.
The percentage of soil and rock foundations in the world population was estimated from
the USBR, Australia/New Zealand, and Portugal populations (Table 2.67 to Table 2.69)
and is shown in Table 2.70. It is recognised that this may be a somewhat biased sample
but there was no way of practically obtaining data for a larger population.
Table 2.67. Foundation types - USBR
Foundation Gravity Arch Buttress Multi-Arch Total
Rock 18 31 6 - 55
Soil 1 - 1 - 2
Soil and Rock 2 - - - 2
Total 21 31 7 - 59

Table 2.68. Foundation types - Australia/New Zealand
Rock 84 40 6 1 131
Soil - - - - 0
Total 87 40 6 1 134


Table 2.69. Foundation types - Portugal
Rock 26 20 4 2 52
Soil - - - - 0
Unknown 1 - - - 1
Total 28 20 4 2 54

Table 2.70. Gravity dam foundation types - combined
Foundation Number %
Rock 128 94.1
Soil 1 0.7
Soil and Rock 6 4.4
Unknown 1 0.7
Total 136 100

The number of failures (both sliding and piping) in a particular foundation type is
shown below.
Table 2.71. Foundations for gravity dam failures by sliding or piping
Foundation Piping Sliding
PG PG(M) PG PG(M)
Rock 5 2
Soil 2
Soil and Rock 1
Total 1 2 5 2

To determine the factors for soil and rock the following assumptions were made:
All piping failures occurred through the soil section of the foundation.
Combined soil/rock (S/R) foundations are taken as soil.
Unknown foundation types are rock.

The factors were calculated as:
f
percent of failures
percent of population
= (2.7)
For example, the factor for piping through soil foundations is:
f
PF
= =
100%
51%
196
.
. (2.8)
Where there are no failures (0%) the factor is zero. To overcome this problem it was
assumed that 1% of all foundation failures would occur on this particular foundation
type. The results of this analysis are shown in Table 2.72. Table 2.72 shows that the
factor for sliding on rock is greater than that for soil. This can be justified by the fact
that no sliding failures have occurred on soil. It is likely that engineers have taken the
soil into account in the dam design whereas, there may be defects which drastically
reduce the foundation strength, in a rock foundation that may be overlooked in the
design. Historically no gravity dam piping failures have occurred in rock foundations. A
number of accidents have occurred as shown in Table 2.73. It is likely that there is
sufficient warning of the progression of piping through rock foundations to allow for
action to be taken to prevent failure.
Table 2.72. Gravity dam factors for piping and sliding failure on soil and rock, f
SF

and f
PF

Foundation Piping, f
PF
Sliding, f
SF

Rock 0.01* 1.1
Soil or Soil and Rock 19.6 0.2*
* These values were derived by assuming 1% failures.
Table 2.73. Foundation types - accidents
Piping Sliding Foundation
PG PG(M) PG PG(M)
Rock 9 1 5 1
Soil 1
Soil and Rock
Unknown 1
Total 11 1 5 1

2.4.4.2 Geology Types - Sliding on Rock, f
SG

Some rock types are more likely to have weaknesses in the foundation (Fell et al, 1992),
so a geology type factor has been included. The geology population was calculated from
a weighted average of the representative populations from the USBR, Australia/New
Zealand and Portugal. The population for the whole of the USA was assessed by
considering the overall geology map of the USA and comparing the distribution of
geology types west of longitude 100W (where the USBR population lies) with that east
of longitude 100W. A weighted average population was created using the number of
gravity dams in the respective countries as given in ICOLD (1984) as weighting factors.
Equation 2.9 shows the method used. Table 2.74 gives the weighting factors used in the
analysis. Table 2.76 shows the weighted population and the number of sliding failures
in each foundation. The calculated and adopted sliding factors are also included. Table
2.75 shows a summary of the factors adopted. Where there is a high chance of a through
going defect beneath the dam a factor of 3 should be used. The following points should
be noted:
There were three failures in sandstone/shale foundations and none in sandstone
alone.
There was one failure in a combined limestone/dolomite foundation.
G
G G G
G G G
=
+ +
1 1 2 2 3 3
1 2 3

(2.9)
where,
G is the weighted geology type population
G
1
, G
2
and G
3
are the geology type populations for each region

1
,
2
and
3
are the weighting factors
Table 2.74. Weighting factors used for weighted average (ICOLD (1984) dam
population)
Population

Australia/ New Zealand 81
Portugal 27
USA 528

Table 2.75. Adopted gravity dam factors for sliding on a rock foundation, f
SG

Geology Type Multiplication
Factor
Comments
shale, claystone
sandstone with shale
interbeds
limestone with shale
interbeds
default for sandstone

3
Where a dam is known to be on
sandstone but it can not be proved
that no shale/claystone exists then
the default of 3 should be taken.
Mudstone, siltstone,
Conglomerate
Schist, gneiss, phyllite,
slate
Hornfels, limestone,
dolomite

1.5
Mudstone and siltstone represent a
transition from shale to sandstone.
Others based on failure statistics.
Granite
Granodiorite

0.3
A low factor has be deemed
appropriate as there have been no
sliding failures on granite yet there
exists a large population of dams.
Others

0.9 Where it can be proved that the dam
foundation comprises ONLY
sandstone 0.9 can be used else, a
factor of 3 should be taken

Table 2.76. Gravity dam factors for sliding on a rock foundation
Failures Factors Geology
Type
Population
%
No. % Calculated Adopted
Comments
Total 100 13 100
Sandstones 21.1 3 23.1 1.1 0.9 No sandstone only failures
so treated as no failures
Shale 8.0 4 30.8 3.9 3 Based on failure data, includes
shale & sandstone
Siltstone 0.5 0.0 0.0 1.5 Assumed transitional between
shale & sandstone
Conglomerate 3.6 1 7.7 2.1 1.5 Based on failure data
Limestone 3.6 1 7.7 2.1 1.5 Based on failure data
Claystone 0.3 0.0 0.0 3 Similar properties to shale
Mudstone 0.4 0.0 0.0 1.5 Assumed transitional between
shale & sandstone
Chert 0.2 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Breccia 0.2 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Dolomite 3.6 1 7.7 2.1 1.5 Based on failure data
Marl 3.6 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Schist 11.4 2 15.4 1.4 1.5 Based on failure data
Quartzite 0.8 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Gneiss 0.7 0.0 0.0 1.5 Similar properties to schist
Phylitte 0.5 0.0 0.0 1.5 Similar properties to schist
Slate 0.3 0.0 0.0 1.5 Similar properties to schist
Hornfels 3.3 1 7.7 2.4 1.5 Based on failure data
Argillite 0.1 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Granite 20.1 0.0 0.0 0.3 Factor would be 0.4
assuming 1 failure
Basalt 5.0 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Tuff 2.4 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Dolerite 0.8 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Rhyolite 1.9 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Andesite 0.3 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Porphyry 0.2 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Diorite 3.4 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Granodiorite 0.3 0.0 0.0 0.3 Similar properties to granite
Greenstone 3.4 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Agglomerate 0.1 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1
Pumice 0.1 0.0 0.0 0.9 No fails so adjusted such that
(Pop Factor) 1

2.4.4.3 Geology Type - Piping on Rock, f
GE

As there have been no piping failures of concrete dams on rock foundations it was
decided to take all factors as unity. Another factor, which takes account of problem
geological environments, was used as a better indicator of variations of likelihood of
failure from the average. The environments considered for this were those that allowed
for the possibility of open joints and include:
Granitic foundations with sheet joints;
very steep sided narrow valleys with likely stress relief joints parallel to the ground
surface;
sedimentary sequences with stress relief effects;
very weak erodible volcanics; and
limestone or dolomite. (Reference Fell, MacGregor and Stapledon, 1992)
The factor should be chosen on a site by site basis but should not exceed 2. The
minimum multiplication factor should be 1. The default value (where the environment is
unknown) should also be taken as 1.
2.4.4.4 Height on Width Ratio, f
H/W

The structural height to width ratio (h
d
/W) is used to take account of the
stockiness/slenderness of the gravity dam. Hence the h
d
/W ratio offers a first order
guide to the relative likelihood of failure by sliding and within the body of the dams. A
database of h
d
/W ratios was collected from the Australia/New Zealand, USBR
populations and from selected ICOLD international conferences (Questions 26, 30, 45,
52, 56, 59, 65). Where found, dams with any curvature were excluded. Figure 2.43 and
Figure 2.44 show scatter plots of h
d
/W versus year commissioned and h
d
respectively
for the population and failed dams. Failures are scattered amongst the population,
although the majority appear to be more concentrated above the average h
d
/W ratio. The
h
d
/W population does not show any correlation with year commissioned. However, as h
d

increases h
d
/W approaches approximately 1.2. It was decided to apply factors as shown
in Table 2.77 and Figure 2.45. These factors have been derived by dividing the
percentage of failures (due to sliding or in the dam body) by the percentage of the

population in each h
d
/W range, in a similar manner to those for sliding and piping in
Section 2.4.4.1.
Table 2.77. Multiplication factors for structural height/width ratio of gravity dams,
f
H/W

h
d
/W <1.0 1.0-1.19 1.2-1.39 1.4-1.59 1.6-1.79 1.8-1.99
Factor 0.1 0.5 0.9 1.3 3 6

Page 2.166

0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Year Commissioned
h
d
/
W
.
Population - Concrete Gravity
Population - Masonry Gravity
Population - Unknown Gravity
Failure - Sliding
Failure - Body

Figure 2.43. h
d
/W versus year commissioned
Page 2.167

0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
0 20 40 60 80 100 120 140 160 180 200
hd
h
d
/
W
.
Population - Concrete Gravity
Population - Masonry Gravity
Population - Unknown Gravity
Failure - Sliding
Failure - Body

Figure 2.44. h
d
/W versus h
d

Page 2.168

0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
hd/W
F
a
c
t
o
r

(
%

F
a
i
l
u
r
e
s
/
%

P
o
p
u
l
a
t
i
o
n
)
.
Sliding
Body
Combined
Suggested

Figure 2.45. h
d
/W factors

2.4.4.5 Other Observations, f
O

This section allows for a multiplication factor to be applied for observed conditions or
special features of the dam. These features will vary with each dam and must be
assessed on a dam by dam basis. The minimum value for any dam should be 0.9 (no
signs of distress, no or very little leakage). The default value should be 1.0. Conditions,
which would warrant a higher multiplication factor (up to a maximum of 10), include:
Sudden increases in seepage through the dam or foundation
Cracking (of a nature that could effect the dams stability)
High or non-linear uplift pressures (also blocked drains)
Alkali aggregate reaction (AAR) or alkali silica reaction (ASR)
Extensive calcite deposits
Large/non-linear dam movements
2.4.4.6 Surveillance, f
S

Historically, unlike embankment dams, most gravity dams have failed with only a short
amount of warning. This warning may be enough to warn people downstream but, is
usually insufficient to enable the dam to be saved from failure. However dams will
sometimes begin to show some signs of problems developing, allowing intervention (eg
by controlling the water level, or by remedial works). Hence it is considered reasonable
to apply a factor to allow for the quality of monitoring and surveillance. Table 2.78
shows the multiplication factors recommended. The multiplication factors have been
modified from those given by Foster et al (1998) for embankment dams.
Table 2.78. Monitoring and surveillance multiplication factors, f
S

Surveillance Embankment
Dam Factor
Factor
f
S

Inspections annually 2.0 1.5
Inspections monthly 1.2 1.1
Irregular seepage observations, inspections weekly 1.0 1.0
weekly seepage monitoring, weekly inspections 0.8 0.9
Daily monitoring of seepage, daily inspections 0.5 0.8

2.4.5 Results
Figure 2.46 and Figure 2.47 show the potential ranges for the annual probabilities of
failure for various cases. The average dam has been taken as a dam that has: a geology
type factor, f
SG
, of 0.9; a height to width ratio, h
d
/W, of 1.3; and the remaining
multiplication factors as unity. A number of dams that have failed have been plotted
together with dams from the Australia/New Zealand population and USBR population
of concrete and masonry gravity dams. None of the dams plotted have the f
red
reduction
factor, as it could not be proven that they satisfied the criteria described above. Where
unknown, multiplication factors were taken as their default or unity. For the most
common type of dam (commissioned after 1930, greater than five years in age and on a
rock foundation) the potential average probabilities using the method range from 4 x 10
-
8
to 1 x 10
-3
.
ANZ dam failed dam min USBR dam maximum 'average'
1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 1E-2 1E-1 1E+0
post 1930/<5 years/soil fndn
post 1930/<5 years/rock fndn
post 1930/>5 years/soil fndn
post 1930/>5 years/rock fndn
pre1930/>5 years/soil fndn
pre1930/>5 years/rock fndn
Probability of Failure

Figure 2.46. Range of annual probability of failure for concrete gravity dams

1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 1E-2 1E-1 1E+0
post 1930/<5 years/soil fndn
post 1930/<5 years/rock fndn
post 1930/>5 years/soil fndn
post 1930/>5 years/rock fndn
pre1930/>5 years/soil fndn
pre1930/>5 years/rock fndn
min
'average'
failed dams
maximum
Probability of Failure

Figure 2.47. Range of annual probability of failure for masonry gravity dams

2.5 DISCUSSION AND CONCLUSIONS
Most of the results have been discussed in Section 2.3. Below is a brief discussion of
the major components of CONGDATA.
Year commissioned
The data shows a distinct increase in incidents for all concrete/masonry dam types in the
1920s. Another peak occurred in the 1960s for concrete gravity and concrete arch
dams. There were no failures in gravity dams commissioned between 1930 and 1963.
Based on the population data there appears to be a reduction in the failure rate with
time. Buttress and multi-arch dams show some various peaks due to their limited
populations.
Height
No concrete or masonry dams greater than 70m are reported to have failed. Failures in
masonry gravity dams appear to be concentrated below 50m in height. Concrete gravity
dams tend to be spread out more. Accidents and more particularly major repairs are
more evident in greater height concrete dams than masonry gravity dams. This however,
is likely to be due to the lower height at which masonry dams are constructed.
The ratio of failures to population does not exhibit any major trend. There appears to be
a higher percentage of failures to population in the 40-49m and 60-69m height ranges.
Arch, buttress and multi-arch dams are shown to be more likely to have failures in the
15-39m range.
Age at failure
There is a large proportion of dams that have failed during first filling. An analysis of
the water levels at failure show most dams failed at their highest recorded water level.
Several of these were only slightly higher than that recorded previously. There appears
to be a slight rise in the rate of failures with time (ignoring first filling). After 40 years
of age there is a jump in the failure rate. It should be noted that the older age groups are

represented by a small population. Concrete gravity dams of less than five years age
appear to have a greater chance of failure compared to older concrete gravity dams.
Masonry gravity dams are more evenly distributed throughout the ages.
A noticeable problem with the accident/major repair data is its bias to the post 1920s
whereas failures occur much further back. This can be put down to a lack of detailed
dam information in the period prior to 1920. Large dam failures would still have been
published during these times.
Piping tends to occur early in a dams life (<5 years, with one exception). Sliding of the
foundation also tends to occur early but is not as restricted as piping. Structural
problems seem to be more likely than foundation problems with age. Concrete dams
have a tendency to fail at younger ages than masonry dams. Most older (masonry) dam
failures have overtopping as a component. Unfortunately there is usually little
information as to the actual mode of failure.
Incident causes
Foundation problems (sliding, leakage and piping) are the main causes of failure to
concrete dams, overtopping tends to play a bigger part in the failure of masonry dams.
Accidents and major repairs are more likely to come about due to surficial damage to
the dam structure or noticeable uplift or leakage in the foundation.
Piping is the main cause of failure for dams with soil foundations. Overtopping and
foundation shear strength are the main causes of failure for dams with rock or unknown
foundations.
Warning types
Overtopping was the most common failure warning type. This was mainly due to the
masonry gravity dams which are more susceptible to overtopping failure. This could be
due to the poorer quality downstream face of masonry dams which can be eroded during
overtopping events and/or the higher permeability of masonry dams which results in a

more rapid increase in uplift pressures. For accidents, and even more so for major
repairs, the warning signs tend to be visual damage of the dam or excessive leakage.
An analysis of all dam failures showed that, where information was available, most had
some warning which could have resulted in the warning and evacuation of residents
downstream. Often the warning was a sudden increase in the amount and rate of
leakage.
Remedial measures
Where a dam has failed it is usually abandoned or reconstructed with a new design. In
the case of accidents and major repairs it is most common that the damaged section is
replaced with no effect to the dam structure as a whole.
Geology
Soils and limestones are more likely to have piping problems. The alluvial soils have a
tendency to pipe under the high gradients imposed. No dam has been reported to have
failed by sliding on alluvial soils. Normally a large concrete or masonry dam would not
be built on a soil foundation.
Shale (interbedded with other sedimentary units) has a greater tendency to be involved
with sliding failure because of the likely presence of weaknesses in the bedding such as
bedding surface shears. It is interesting that sandstone does not appear to be over
represented when the population is taken into account. Failures tend not to occur in
sandstone alone but only when the sandstone is interbedded with shale. Shale and
limestone (often interbedded) have a high incidence for failing. The limestone has a
high proportion of accidents generally due to excessive leakage through dissolution.
Another point of note is that no incidents have occurred in basalt foundations. These
conclusions agree with the general knowledge regarding the geology types (e.g. as
described in Fell et al, 1992
11
).

Other design factors
The factors in Section 2.3.9 suffered from a lack of information. Generally these could
only be obtained if a dams cross section was available. From the data collected it
appears that the failed dams suffered from a lack of good engineering. Very few dams
were found with galleries (1 dam); drainage (1 dam); grout curtains (4 dams);and shear
keys (1 dam). The downstream slopes appeared to be too steep. Six gravity failures had
downstream slopes of 0.6:1 (H:V) or less. Failed dams, particularly gravity dams, were
usually located in relatively wide valleys or were composite sections with earthfill
dams. Three dimensional effects are unlikely to have contributed any strength in these
cases. h
wf
/W ratios ranged from 0.6 to 2.1 with an average of 1.35.
Generally, unlike embankment dams, concrete and masonry dams are analysable and
hence can readily be checked for stability. The major unknowns for these dams lie in the
foundation where sliding and piping failures can occur.
Section 2.4 gives a method for assessing the first order probability of failure of masonry
or concrete gravity dams. The method accounts for dam age, year commissioned and
type; failure mode; foundation geology; height to width ratio; and monitoring and
surveillance. General probabilities of failure for arch, buttress and multi-arch dams,
based on failure and population statistics, are included.
The author cautions that this approach should only be used as a first order
approximation of the annual probabilities of failure. It is clearly very approximate, and
suffers from being based on small numbers of failures, and limited quality data. Where
significant decisions on dam safety are being made, detailed deterministic and/or
probabilistic methods should be used.
The results from the analysis of CONGDATA are subject to the limitations mentioned
earlier. Whilst all care has been taken in compiling data, it should be remembered that
the information in CONGDATA has come from numerous sources, not all of which
could be validated. The analysis of dams in CONGDATA does not take into account
such things as: surveillance; quality of construction; and quality of geological

description. It is therefore recommended that this work be used in a qualitative sense
only.
The Strength of Intact Rock Page 3.1

3 THE SHEAR STRENGTH OF INTACT ROCK
3.1 INTRODUCTION
The application of the Hoek-Brown criterion (Hoek & Brown, 1980) to intact rock and
rock masses is common in slope engineering, so it is important that it is validated in
both applications, and that the uncertainty of its predictions are well understood. This
chapter presents a detailed analysis of the application of the criterion to laboratory test
data on intact rock and suggests modifications that provide improved predictions of
triaxial strength based on easily measured material properties.
The data consists of 4507 test results from 511 sets obtained from the literature and
original laboratory test reports.

3.2 FAILURE CRITERIA FOR INTACT ROCK
Equation 3.1 presents a generalised criterion where a
1
, a
2
, etc are material properties.
Figure 3.1 presents a generalised failure criterion in the
1
-
3
plane and shows the
locations of the unconfined compressive strength,
c
, Brazillian strength,
Bt
, uniaxial
tensile strength,
ut
and pure tensile strength,
pt
.
,....) , , , , ( fn 0
2 1 3 2 1
a a (3.1)
Sigma 3 / Sigma C
S
i
g
m
a

1

/

S
i
g
m
a

C
-1
0
1
2
3
-0.25 0.00 0.25 0.50
pure t
uniaxial t
Brazillian t

Figure 3.1. Generalised failure criterion
There are two approaches to the selection of a failure criterion for intact rock,
theoretical and empirical. The base of the most commonly adopted theoretical
approaches are those of Coulomb:
tan
n
c + (3.2)
or Griffiths (1924) criterion for fracture initiation:
( ) ( )
3 1
2
3 1
8 +
t
(3.3)
3
/
c
1
/
c

Most practical engineering relies on a linear Mohrs envelope being fitted to
experimental data or to the relevant portion of a theoretical or empirical criterion.
Notwithstanding this it is becoming increasingly common for computer software to be
able to deal directly with one or more non-linear criteria. Thus while the limits and
pitfalls of linearisation are well understood, it is now important to assess the accuracy of
non-linear criteria.
It is well known that the theoretical criteria do not accurately predict the failure strength
of rock and often rely on parameters that are difficult to measure (Ramamurthy et al,
1988, Andreev, 1995, Parry, 1995). Figure 3.2 by Johnston & Chiu (1984) shows that
equations based on the work by Griffith provide poor fits to Melbourne mudstone. For
these reasons many criteria have been developed that seek to capture the important
elements of measured rock strengths or seek to modify theoretical approaches to
accommodate experimental evidence, several of these empirical criteria are listed in
Table 3.1. Most of these share a reasonably similar structure and all have elements that
are likely to fail at the extremes.

Figure 3.2. Comparison of test results with theoretically based failure criteria
(Johnston & Chiu, 1984)
The requirements of a good rock shear strength criterion are: that the criterion is a good
predictor of strength at the stresses of interest (if not all confining stresses); that it is
mathematically simple and preferably based on dimensionless parameters; and that it

can also be used for rock mass (Hoek, 1983). Many of these criteria have been
developed with high (underground) stresses and hard rocks in mind. As such many do
not acurately predict shear strengths at low (or negative) stresses. For example, Mogi
(1966) states Equation 3.6 is applicable to the middle part of curves, but it deviates
from observed values at an initial part and later ductile part of curves. This is a
problem where estimates of the strength of rock at low stresses (say slopes or dams) are
required.
Figure 3.3 shows that the Hoek-Brown shear strength criterion provides a poor fit to
Melbourne mudstone. This is mainly due to the exponent being set at 0.5 in the Hoek-
Brown equation. This is likely a direct result of the equation being created for hard
rocks.

Figure 3.3. Comparison of Hoek-Brown criterion (solid) and Johnston criterion
(dashed) for Melbourne mudstone (Johnston, 1985)
An inspection of Table 3.1 shows that equations (3.6), (3.9), (3.12), (3.14) and (3.15) do
not algebraically reduce to the unconfined compressive strength,
c
, when the minor
principal stress,
1
, is zero. Similarly, (3.6), (3.7), (3.9), (3.10), (3.14) and (3.15) fail to
predict strengths in the tensile range (some completely and others at particular
reasonably expected constant values) whilst (3.5) recommends a tensile cutoff due to it
overestimating
t
.

Equations (3.6), (3.7), (3.10) and (3.14) indicate
0 3 1
3
d d which implies a
friction angle of 90 at
3
= 0.
Equation 3.8 represents straight line (Coulomb type) fits to lab data. As the rock
strength envelope is known to be curved this equation could only be expected to predict
the correct shear strengths over very narrow ranges.
As is discussed elsewhere in this thesis, (3.4) is unsuitable for rock mass as it does not
allow for a tensile strength of zero and hence can not be sensibly extended into the
strength of rock mass or rockfill.
The unconfined compressive strength,
c
, is widely used and is a useful index for rock
strength. Only Equations (3.8), (3.10), (3.11) and (3.15) can be non-dimensionalised by
c
. This aids in the presentation and grouping of data with various unconfined
compressive strengths.
Table 3.2 shows the various exponents suggested (or obtained by regression) by the
authors for their particular criterion. A power function of unity would imply a straight
line (Coulomb). Smaller power functions will generally imply that the criterion has a
higher curvature at very low stresses.
Page 3.6

Table 3.1. Various intact rock failure criteria

Reference Equation Constants Development UCS Tensile strength
3.4
Balmer (1952)
Sheorey et al (1989)
b
t
c

,
_

3
1
1
b = 0.411 to 0.828
(ignoring samples controlled
by defects orientated 15-75
to the load axis)
Sheorey et al used tests on
sandstone and shale plus
literature. 23 sets in total of
which 11 had no UCS
recorded, 6 of which were
also defect controlled.
c
-
t

3.5 Fairhurst (1964) ( ) ( )
3 1
2
3 1
+ + B A
1
1
]
1

,
_

1
2
1
2
2
m
A
t

( )
2
1 2 m B
t

1 +
t c
m
Based on an empirical
generalisation of Griffith
allowing for all n =
c
/
t

possibilities.
b b a t +
2
4
2

b b a t +
2
4
2

Note a tensile cut-off is
recommended
3.6
Hobbs (1966)
Mogi (1966)
n
c 3 3 1
+ +
- f(rock type)
n 0.38 to 0.73 (Mogi)
Developed using triaxial
tests on intact rock and
literature results.
c
(n>0) undefined
3.7
Murrel (1965)
Hobbs (1970)
A
c
F
3 1
+ F, A
Non-linear empirical
adjustment to Coulomb.
Hobbs (1970) found that his
earlier equation (3.6) did not
fit new test data.
c
(A>0)

_
,
c
A
F
1

(undefined for most A)
Page 3.7

3.8 Bodonyi (1970)
3 1
+ a
c

or
c c
a
3 1
1

+

A
Straight line fit to lab test
data
c
a

3.9 Franklin (1971) ( )
b
a
3 1 3 1
+ + a, b 0 Undefined
Page 3.8

3.10
Bieniawski (1974)
Yudhbir et al (1983)
,
_
c c
k
3 1
1
412 triaxial samples used to
determine constants:
0.75 (Bieniawski)
0.65 (Yudhbir et al)
k - varies with rock
type from 2 to 5(see
Table 3.2. Empirical
estimates of exponents
for the equations in
Table 3.1
Equation Parameter Exponent for rock
3.4 b 0.411 to 0.828
3.6 n 0.38 to 0.73
3.8 N/A 1
3.10 0.75 (Bieniawski, 1974)
0.65 (Yudhbir et al., 1983)
3.11 0.5
3.12 0.38 (tuff only)
3.13 B 0.5 0.81 (Johnston & Chiu, 1984)
( )
2
log 0172 . 0 1
c
(Johnston, 1985)
3.14 0.8
Non-dimensional form of
Murrell (1965). Note that k
has become dependent on
c

as
( ) 1
A
c
F k
c
(>0)
1
1
,
_

k
c

(undefined for most )
Page 3.9

3.11 Hoek & Brown (1980a)
( )
2
1
2
3 3 1 c c i
s m + +
or
2
1
3 3 1
,
_
s m
c
i
c c

m
i
depends on rock type
s = 1
Developed using curve
fitting to extensive triaxial
data on hard rock.
s
0.5
c

=
c
(s=1)
2
4
2
s m m
i c c i
+

3.12 Adachi et al (1981)
q
p
p
p

_
,
0 0

or
( )

,
_
+

3
2
3 1 1
0 3 1
p
q = (
1
-
3
)
p = (
1
+2
3
)/3
p
o
= 0.1MPa (unit stress)
, = strength parameters
tuff: = 1.76 and = 0.38.
Based on triaxial test results
plotted on a log-log graph.
The equation is from Hobbs
(1966).
Porous tuff was used to
represent soft rock.
,
_
1
1
3
1 . 0

(MPa)

01
2
3
1
1
.

_
,
_
,

(MPa)

3.13
Johnston & Chiu (1984)
Johnston (1985)
B
c c
B
M
1
]
1
1
3 1

M = f(rock type and
c
)
B = f(rock type) 1984
see Table 3.4.
( )
2
log 0172 . 0 1
c
B
(Johnston, 1985)
Developed empirically for
soft rocks after Hoek &
Brown was shown to give a
poor fit.
c

B
M
c

Page 3.10

3.14
Ramamurthy et al (1985)
Ramamurthy et al (1988)
Rao et al (1988)
,
_
+
3
3 3 1
c
B
B - function of rock
type/quality 1.8 to 3.54
- slope of log-log curve of
(
1
-
3
)/
3
and
c
/
3

0.75 to 0.85 (0.8 usually
taken)
Modified Mohr-Coulomb to
become non-linear.
Empirical work done using
testing on 4 sandstones plus
100 sets from literature.
Extended for anisotropy and
rock mass in 1988.
0 (<1)
B
c
(=1)
Undefined (>1)
0 (<1)

( )
1
B
c

( ) K
7
1
5
1
3
1
, , 1,
else undefined
3.15 Yoshida et al (1990)
b
c
c
s a
1
3
3 1

,
_

or
b
c c c
s a
1
3 3 1
,
_

a = 1.3 to 8.3
s = 0.01 to 0.78
b = 1 to 3.3 (tuff = 7.2)
1/b = 0.3 to 1 (tuff = 0.14)
Proposed for geologic
materials exhibiting time
dependent softening e.g. soft
rocks.
Data came from 16 rock sets
from the literature.
a
c
s
b


Table 3.2. Empirical estimates of exponents for the equations in Table 3.1
Equation Parameter Exponent for rock
3.4 b 0.411 to 0.828
3.6 n 0.38 to 0.73
3.8 N/A 1
3.10

0.75 (Bieniawski, 1974)
0.65 (Yudhbir et al., 1983)
3.11

0.5
3.12

0.38 (tuff only)
3.13 B 0.5 0.81 (Johnston & Chiu, 1984)
( )
2
log 0172 . 0 1
c
(Johnston, 1985)
3.14

0.8
3.15 1/b 0.3 1.0 (tuff = 0.14)

Table 3.3. Suggested values of constant k (Yudhbir et al, 1983 and Bieniawsi,
1974)
Rock Type k
Norite, granite, quartzdiorite, chert 5
Quartzite, sandstone, dolerite 4
Siltstone, mudstone 3
Tuff, shale, limestone 2

Table 3.4. Values of M and B for a range of materials (Johnston, 1991)
Material UCS
(MPa)
M B
Soft clay 0.02 2.5 0.97
Stiff clay 0.2 3.5 0.91
Soft rock 2.0 5.1 0.81
Soft rock 20 7.2 0.68
Hard rock 100 9.0 0.57
Hard rock 200 9.8 0.52

Given the variability typical of rock test results it is likely that any one criteria is as
suitable overall as any of the alternatives. The Hoek-Brown empirical failure criterion
(Hoek & Brown, 1980) was developed in the early 1980s for intact rock and rock
masses, it has been subject to continual refinement for rock masses. For intact rock its
form has not changed and is given in Equation 3.16.

5 . 0
3
3 1
1
,
_
+
c
i
c
m
(3.16)
In common with most of the empirical failure criteria, the Hoek-Brown criterion is
formulated in terms of
1
and
3
and is independent of
2
. The author does not
consider this a major impediment for practical purposes.
It is the authors experience that the Hoek-Brown criterion forms the basis of virtually
all the non-linear criteria now used by practising engineers. Further it forms the basis
for the almost universal extension into rock mass strength. It thus has been adopted as
the basis of examination for the rest of this Chapter. Further, for these reasons it is
important to establish that the criterion does accurately represent actual intact rock
behaviour.

3.3 LABORATORY TEST DATABASE FOR INTACT ROCK
A large database of test results has been assembled for a wide variety of rocks; tests
include uniaxial tensile strength, Brazilian tensile strength, unconfined compressive
strength and triaxial compression and tension. Many of the results were sourced from
Sheorey (1997), Hoek and Brown (1980), Shah (1992) and Johnston (1985) and
checked against original sources where possible. Further test information from other
sources was obtained. Full details of the data are contained in the Appendix. At present,
the data consists of 4507 test results forming 511 sets.
In addition to the principal stresses, and to aid in the examination of the data, the
information in Table 3.5, about the rock samples, was obtained where possible.


Table 3.5. Intact rock database descriptors
Database heading Description
Set Set number (1-511)
Test Test number within set
D?
"*" denotes
1
<4.4
3
from Sheorey (1997) method of denoting
ductile behaviour
R
G good range of
3
(
c
3 . 0
min 3 max 3
);
P poor range
3
(
c
3 . 0
min 3 max 3
< )
s3
Minimum principal stress,
3
(MPa)
s1
Maximum principal stress,
1
(MPa)
Name/location Specific rock name or location of source
Rock type General geological group term
m
i98
Published Hoek et al (1998) m
i
value for the rock type
orientation: BP/weak Orientation relative to bedding or planes of weakness (if any)
moisture Sample moisture condition
diam Sample diameter (mm)
h/d Sample height to diameter ratio
poros Porosity of sample (%)
grain size Sample mean grain size (mm)
B/T/D Comment on behaviour of sample under load based on an observation of
the stress-strain curves, options were:.Brittle/Transitional/Ductile
strain rate Test strain rate (%/second)
comments Other comments on test, sample or test type
Reference Quoted original reference for data
Source Source of data for database::
- original - original reference used
- SH_A.1. - Sheorey (1997) Table A1
- SH_A.2.- Sheorey (1997) Table A2
- Doruk - Doruk (1991)
- Shah - Shah (1992)


3.4 AN ANALYSIS OF THE ANALYSIS OF DATA
When confronted with a set of data, there are a number of questions that have to be
addressed:
What data should be included in the analysis?
What equation should be fitted?
What method of fitting should be adopted?
It is of little value undertaking a comprehensive test program or detailed analysis of the
results if the methodology is flawed. In fact, parameters determined can vary from very
conservative to quite the opposite, both situations have consequences in analysis and
design.
Turning to the first question, the data included is that described in the previous section.
It has been common practice for researchers fitting empirical failure criterion to intact
rock to exclude results thought to exhibit ductile behaviour; this approach has been
adopted by Hoek and Brown (1981), Shah (1992), Johnston (1985) and Sheorey (1997).
In general these researchers have adopted the brittle-ductile transition suggested by
Mogi (1966). The Equation is given by:

3
4 . 3 C (3.17)
There is some confusion in the literature regarding Equation 3.17. Sheorey (1997) has
assumed C to be the difference in the maximum and minimum principal stresses (i.e.
3 1
) whilst the other authors have assumed C to be the maximum principal stress,
1
. Mogi (1966) defined C as the compressive strength but did not define it in relation
to principal stresses. Mogi (1973) however, defines C as
3 1
. Therefore it is
assumed that the brittle-ductile transition, given by Mogi (1966), is:

3 3 1
4 . 3 or
3 1
4 . 4 (3.18)

The exclusion of ductile data would be appropriate if (i) only brittle behaviour was of
interest, (ii) the boundary was clear and (iii) the failure criterion was disjoint across the
transition. In the case of a criterion based solely on Griffiths theory, exclusion of
ductile tests results would be appropriate. It is not necessary, is counter productive and
is arbitrary, for an empirical criterion. Research (Evans et al., 1990, Mogi, 1966 and
1973, Scott & Nielsen, 1991, Rutter & Hadizadeh, 1991, Hoshino et al, 1972) shows
that the transition is not well defined for all rocks, is often curved and certainly occurs
over a wide range of stresses. Gustkiewicz (1985) found that the value of [the brittle
ductile transition] pressure cannot be seen distinctly on the curve of strength versus
confining pressure. Research by others (Evans et al., 1990) also showed that the failure
envelope is not necessarily disjoint at the brittle-ductile transition. Thus an appropriate
criterion can model strength on both sides of the transition.
Triaxial tests with stresses around the brittle-ductile transition should be checked to
ascertain that they actually reached their maximum stress. Often the triaxial apparatus
(or sample) will not allow sufficient strain for this to happen and thus the result should
be ignored.
Sheorey (1997) only uses data points where
1
>4.4
3
and data sets where there are at
least 5 data points.
Doruk (1991) uses the following criteria to decide whether a data set is acceptable:
Each data set must have major and minor principal stresses at failure.
Only the results of effective or drained uniaxial and triaxial compression tests at
room temperature are used.
Each data set must contain at least three triaxial test results.
Only results in the brittle range (defined as
1
3.4
3
and
3

c
) were used.
Doruk (1991) divides the data sets used into the following classes:

Class 1: 5 well fitted data points with
c
3 . 0
min 3 max 3

Class 2: 5 scattered data points with
c
3 . 0
min 3 max 3

Class 3: 5 well fitted data points with
c
3 . 0
min 3 max 3
<
Class 4: < 5 well fitted data points with
c
3 . 0
min 3 max 3

Class 5: < 5 scattered data points with
c
3 . 0
min 3 max 3

Class 6: < 5 well fitted data points with
c
3 . 0
min 3 max 3
<
Class 7: Data sets which contain very scattered data points
In the current work as many test results as possible have been included and only results
for which there is true ductile behaviour (no ma ximum definable
1
or where there is
significant doubt as to their accuracy) have been excluded.
Columns in the rock strength database show the number of points and whether the data
has a good range of
3
(i.e.
c
3 . 0
min 3 max 3
) or a poor range of
3
(i.e.
c
3 . 0
min 3 max 3
< ). The effect of the number of tests in a data set is investigated in
the analysis.
There are several forms of the Hoek-Brown criterion that can be adopted for data
analysis, these include Equation 3.16 and:

>
,
_
+
i c
i c
c
i
c
m
m
m

3 3 1
3
5 . 0
3
3 1
for
for 1
(3.19)
( )
2
3
2
3 1 c c i
m + (3.20)
( ) ( )
2
3 3 1
log 5 . 0 log
c c i
m + (3.21)
Equation 3.16 is strictly the Hoek-Brown criterion, but is undefined for
3
less than
approximately -
c
/m
i
. Equation 3.19 ensures that the criterion is defined over the full
range of
3
. Equations 3.20 and 3.21 are linearisations of the criterion. The impact of
adopting these different forms is discussed at the end of this section.

The method of least squares is very widely adopted in fitting models to data; there are
often very sound statistical reasons to so do. Shah (1992) suggests that the simplex
method with the function (observed-predicted)
2
is a better method than least squares. In
fact, the method presented by Shah is least squares, the simplex is purely a numerical
method to optimise some function, in this case minimising the sum of squared
differences (ie errors). It has been verified that the resulting parameter estimates are the
same as those from other robust least squares procedures.
If the departure of the measured
1
from the predicted
1
(ie the error) is normally
distributed with a variance that is independent of the predictor variables (here
3
), then
the predictions obtained with least squares, either with a simplex or otherwise, will be
uniform minimum variance unbiased estimators; this is highly desirable. But
consideration of data with multiple measurements of
ut
or
Bt
will indicate that straight
least squares is not appropriate for fitting the Hoek-Brown criterion.
Consider an experimental program with multiple measurements of
ut
, it is clear that if a
failure criterion is to be fitted to the test data it is desirable that the estimated tensile
strength should be the average of these measurements (ie the fitted curved should pass
through the middle of the measured values). Equation 3.16 is not defined for measured
values of
ut
less than the fitted value (ie larger tensile strengths) and this forces many
fitting methods to fit the maximum measured (ie most negative) tensile strength as the
estimated tensile strength. Equation 3.19 overcomes this problem, but reference to
Figure 3.1 shows that the slope of the equation to the left of the estimated
ut
is much
less than that to the right; the figure is drawn for an m
i
of 8 and the slope to the right is
much steeper for higher m
i
. Given that a general least squares approach assesses the
error as the observed
1
(ie zero) minus the predicted
1
, then data a given distance to
the right of the estimated
ut
will have a much larger error than data the same distance
to the left. Thus a standard least squares procedure will result in a very poor fit at low
stresses and force a small
ut
and high m
i
, ie the opposite effect to adopting Equation
3.16.
A resolution of the above problem comes about by recognising that in a uniaxial tensile
strength test, the controlled variable is
1
and the measured variable is
3
, thus the real

error is observed
3
minus the predicted
3
. But this error is scaled in
3
and needs to
be adjusted if it is to have equal status with measurements in
1
. It is suggested that
scaling by m
i
is a convenient and accurate approach. Given this it is recommended that a
least squares procedure be used where the error is defined as:

( )
( )

;
>
3 1 3 3
3 1 1 1
3 for predicted measured

i
m
(3.22)
This has been found to provide very good fits for a wide variety of data.
It is the authors experience that the method of parameter estimation can, and often
does, have a large impact on parameters derived from experimental data but the effect is
often camoflaged by the variability of test data. Table 3.6 and Figure 3.4 show the
results of analysis of a simulated test program with results generated for a material with
a Hoek-Brown failure criterion,
c
and m
i
are both normally distributed with
mean/standard deviation of 10/2 MPa and 12/2 respectively. Results generated were 10
uniaxial tensile strength tests, 20 unconfined compressive strength tests, and 4 each
triaxial strength tests at confining pressures of 1, 2, 5, 10, 20, 40 and 80 MPa. Thus
there were 58 data points in all, simulating a very comprehensive test program from
which it should be possible to determine accurate estimates of material properties.
The entire generated data and selected fits are shown on Figure 3.4a. It can be seen that,
with two exceptions, the methods provide a reasonable fit for the majority of the data.
But reference to Figure 3.4b shows that most methods provide a very poor fit to the data
at low stresses, that is over the stress range of interest in slope analysis.

Page 3.20

Table 3.6. Results of different regression methods on artificial data
Case Equation Fitting method Number
c
(MPa) m
i
r
2
(%)
1 Actual data 58 10.0 12.0 na
2 Normal equation 3.16 Least squares 58 14.9 7.75 97.88
3 Extended equation 3.19 Least squares 58 8.46 15.6 99.12
4 Extended equation 3.19 Modified least squares, Eqn 3.22 58 10.7 12.0 99.00
5 Adopting known
c
and normal equation 3.16 Least squares 58 na 5.21 91.69
6 Excluding
t
results and normal equation 3.16 Least squares 48 9.53 13.7 99.06
7 Excluding
c
&
t
results and normal equation 3.16 Least squares 28 6.20 21.4 98.80
8 Stress difference squared 3.20 Least squares 58 3.97 35.2 95.53
9 Stress difference squared and known
c
3.20 Least squares 58 na 13.8 95.47
10 Stress difference squared 3.20 Least sum of absolute differences 58 9.18 15.4 95.45
11 Logarithms 3.21 Least squares 58 8.09 4.12 55.97
12 Logarithms and excluding
t
results 3.21 Least squares 48 9.67 12.2 95.00


Sigma 3 (MPa)
S
i
g
m
a

1

(
M
P
a
)
0
50
100
150
200
250
-10 10 30 50 70
UCS mi
Artificial data 10.012.0
Normal eqn & LS14.97.75
Extended eqn & LS 8.4615.5
Ext eqn & mod LS 10.712.0
Fix UCS & LS 10.05.21
Excl Sc or St & LS 6.1921.4
DS^2 & LS 3.9735.2
Log & LS 8.094.12
Not shown
Excl St & LS 9.5213.7
DS^2 with UCS fixed 10.013.8
DS^2 & Least abs sum 9.1815.4
Log with excl St 9.6712.2
Sigma 3 (MPa)
S
i
g
m
a

1

(
M
P
a
)
0
10
20
30
40
-2 0 2 4 6
UCS mi
Artificial data 10.012.0
Normal eqn & LS14.97.75
Extended eqn & LS 8.4615.5
Ext eqn & mod LS 10.712.0
Fix UCS & LS 10.05.21
Excl Sc or St & LS 6.1921.4
DS^2 & LS 3.9735.2
Log & LS 8.094.12
Not shown
Excl St & LS 9.5213.7
DS^2 with UCS fixed 10.013.8
DS^2 & Least abs sum 9.1815.4
Log with excl St 9.6712.2

Figure 3.4. Fits to artificial data (a) full range (b) low stress range
UCS m
i

Artificial data 10.0 12.0
Normal eqn & least squares 14.9 7.75
Extended eqn & least squares 8.46 15.5
Extended eqn & modified least squares 10.7 12.0
Fixed UCS and least squares 10.0 5.21
Excluding
c
or
t
and least squares 6.2 21.4
(
1
-
3
)
2
Logarithm and least squares 8.09 4.12
UCS m
i

Artificial data 10.0 12.0
Normal eqn & least squares 14.9 7.75
Extended eqn & least squares 8.46 15.5
Extended eqn & modified least squares 10.7 12.0
Fixed UCS and least squares 10.0 5.21
Excluding
c
or
t
(
1
-
3
)
2
Logarithm and least squares 8.09 4.12

The following comments are offered on the various analyses undertaken, listed in the
same order as in Table 3.6.
1. The generated data, the author considers that this is a reasonable representation
of a comprehensive test program in a moderately variable unit.
2. The strict application of least squares to Equation 3.16, ie the usual form of the
Hoek-Brown criterion, results in the uppermost curve in Figure 3.4b. The criterion
cannot be evaluated for
3
less than the estimated tensile strength. This results in large
estimates of
ut
and
c
and thus a low m
i
. From 5 to 80 MPa the curve passes through
the middle of the data. Below 1 MPa the estimated strength is nearly 50% higher than
the true strength even though the regression r
2
is nearly 98%. This problem could be
partially fixed by including only the average measured tensile strength in the analysis
but this ignores considerable readily obtained and economic data and disguises the true
variability.
3. Least squares applied to Equation 3.19 results in vastly improved parameter
estimation but the lower slope to the left of the estimated
ut
produces a low estimate of
ut
and thus somewhat low estimated
c
and high estimated m
i
. A good fit overall with
the highest r
2
, but approximately 15% underestimate of true strength for low
3
.
4. Modified least squares, Equation 3.22, applied to Equation 3.19 results in
accurate estimation of the parameters and does so in almost all circumstances. The fact
that r
2
is slightly less than for method 3 is a necessary consequence of the treatment of
variability of the measured tensile strengths.
5. Least squares applied to Equation 3.16 with
c
fixed at the average of the test
results. It might be thought that knowing one property should help with estimating a
second unknown property, this is not the case here. The problem in 2 above is now
magnified to produce almost the worst fit imaginable. It shows that an r
2
of over 90%
can be obtained with a fit that bears only a passing relationship to the data.

6. Least squares applied to either Equation 3.16 or 3.19, with the tensile strength
test results excluded, results in a good fit. Again the problem is that good economic data
is ignored and the fit at low stress will be more variable.
7. Least squares applied to either Equation 3.16 or 3.19 with both the tensile and
unconfined compression test results excluded. In this case more than half the data is
ignored and, in the present case, the fit at low
3
is more than 30% out. This is a
random error and the fit could be low or high. The problem with this approach is that it
is poorly controlled at the stresses of interest in slope analysis.
8. Least squares applied to Equation 3.20. This is a common form of fitting the
Hoek-Brown criterion to data and estimating
c
and m
i
. This method virtually
minimises errors to the fourth power, hence the lowish r
2
, and dramatically
overweights the larger values of
1
. Errors in parameter estimates are not predictable,
but in this example, estimated
c
and m
i
are 40% and 300% of the true values
respectively, even though the corrupted r
2
is over 95%. Over most of the range of the
test results it is a very good fit but not over that portion of interest in slope design. It is
not recommended.
9. As for 8 above but with
c
fixed at the mean value, in contrast to 5 above this
results in a good fit across the range but relies on a good estimate of
c
and increased
faith that this accurately represents triaxial behaviour.
10. Least sum of absolute differences applied to Equation 3.20. This in large
measure compensates for the overweighting of large
1
values of method 8. The
resulting estimates are good.
11. Least squares applied to Equation 3.21, again a common form of fitting the
Hoek-Brown criterion. As for Equation 3.16 this equation is not defined for
3
less than
pt
. This method has major problems fitting any data which includes a moderate spread
of tensile testing.

12. Least squares applied to Equation 3.21 with the tensile strength test results
excluded. A robust method weighted to low stress results and good for slope analysis
but unable to take advantage of economic and readily available data.
From Table 3.6 it can be seen that r
2
is not a useful indicator of accuracy of estimates of
the parameters and that these estimates can vary widely depending on the method of
analysis. Methods with r
2
in excess of 95% and that model the data very well over most
of the range have estimates of
c
varying from 3.97 to 14.9 MPa and m
i
from 7.75 to
35.2 and this is for artificial data that follows exactly the criterion with only test
variability. Thus many of these methods are very poor estimators of strength in the low
stress region that is of interest in slope analysis.

3.5 HOEK-BROWN CRITERION FOR INTACT ROCK
Modified least squares, Equation 3.22, was combined with the extended formulation of
the Hoek-Brown criterion, Equation 3.19, to estimate
c
and m
i
for all test data in the
database. Discussion in the previous section indicates that the fit is poorly controlled at
low stresses for sets with little data, particularly
c
and
t
. Small changes in the data can
lead to wildly varying estimates of both
c
and m
i
, in general with
c
becoming very
small and m
i
very high but with the fit being almost identical over the range of the test
results. In fact for many data sets
c
and m
i
are not independent but
c
0 as m
i
.
The best solution to this issue is to place plausibility limits on the parameters. A number
of limits were considered and the following ones adopted:
As all the test results were taken from materials described as rock,
c
was limited
to be not less than 1 MPa.
Published values of m
i
fall in the range of 4 to 33 (Hoek & Brown, 1998). As will
be discussed later m
i
is very closely related to the ratio -
c
/
ut
, reference to the figures
in Lade (1993) indicates that this ratio varies from less than 2 to over 50. This limits m
i

to the range 1 to 50. Further m
i
is related to the angle of friction at
3
=0 (ie
0
), which
is of great interest in slope analysis. It was considered that
0
should be limited to the
range of 15 to 65, which for the Hoek-Brown criterion further limits m
i
to the range 1.4
to 40.
The process was completed for 475 data sets involving 3779 test results. The results of
the analysis are provided in Figure 3.5 to Figure 3.8. Figure 3.5 shows a box and
whisker plot of the values of m
i
estimated from the data, m
itest
, against the values of m
i

provided in Hoek & Brown (1998) and Hoek et al (1995), m
ipub
. Several such figures are
presented in this chapter, the whiskers show the range of test results, the box shows the
upper and lower quartiles and the bar the median value. Also shown on this figure is a
linear regression between published m
ipub
and m
itest
weighted for the number of data
points supporting each estimate. The regression equation is:

ipub itest
m m 441 . 0 58 . 7 + (3.23)

This is a very poor relation, r
2
=16.4%, between m
i
determined on the basis of actual
testing and that obtained from the literature.
Page 3.27

m
i
from literature, m
ipub
m
i

f
r
o
m

f
i
t
t
i
n
g

H
B

e
q
u
a
t
i
o
n
,

m
i
t
e
s
t
0
10
20
30
40
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

Figure 3.5. m
i
i
from test results and Hoek-Brown Equation

Figure 3.6 presents m
itest
against rock type ordered in increasing m
ipub
, it can be seen that
it is difficult to ascribe a single or even small range to m
i
on the basis of rock type. It
should be remembered that 50% of the test data falls outside the range indicated by the
box for each rock type, thus for example 50% of the values for sandstone fall below 11
or above 19, and for granite below 19 or above 31.
Page 3.29

m
i
t
e
s
t
0
10
20
30
40
C
l
a
y
s
t
o
n
F
i
r
e
c
l
a
y
G
r
e
e
n
s
t
o
M
u
d
s
t
o
n
e
S
e
r
p
e
n
t
i
S
c
h
i
s
t
S
h
a
l
e
C
h
a
l
k
C
h
l
o
r
i
t
i
L
i
m
e
s
t
o
n
M
a
r
b
l
e
S
i
l
t
s
t
o
n
S
l
a
t
e
B
i
o
c
a
l
c
a
D
o
l
o
m
i
t
e
A
n
h
y
d
r
i
t
S
a
l
t
C
o
a
l
T
u
f
f
P
y
r
o
c
l
a
s
R
h
y
o
l
i
t
e
A
p
l
i
t
e
B
a
s
a
l
t
L
a
m
p
r
o
p
h
T
r
a
c
h
i
t
e
A
g
g

t
u
f
f
G
r
e
y
w
a
c
k
W
h
i
n
s
t
o
n
A
n
d
e
s
i
t
e
D
i
a
b
a
s
e
D
o
l
e
r
i
t
e
Q
u
a
r
t
z
d
o
S
a
n
d
s
t
o
n
G
r
a
n
i
t
e
1
N
o
r
i
t
e
Q
u
a
r
t
z
i
t
D
u
n
i
t
e
E
c
l
o
g
i
t
e
G
a
b
b
r
o
P
e
r
i
d
o
t
i
A
m
p
h
i
b
o
l
D
i
o
r
i
t
e
Q
u
a
r
t
z
d
i
G
r
a
n
o
d
i
o
G
n
e
i
s
s
G
r
a
n
i
t
e

i
from test results and Hoek-Brown equation

Figure 3.7 presents the
c
determined from fitting the Hoek-Brown equation against the
c
determined from
c
testing or, at least, as reported in the literature from which the
data was obtained. Several figures in this chapter are presented in this style. The upper
and lower dashed lines represent 1.5 and 0.67 times the reported
c
values. Further, the
symbols represent the number of test results used to determine the fit, a small cross is 4
or less data points, a small circle is 7 or less, large circle is 12 or less and a square is
more than 12 data points. It can be seen that virtually all the data lies in a very narrow
band, such that the fitted
c
is quite close to the reported
c
.
Figure 3.8 is a similar presentation to Figure 3.7 except that it presents fitted tensile
strength versus reported tensile strength. It can be seen that the fitting method adopted
provides a very good estimate of
t
for those data sets which do have reported tensile
strengths. Most of the other methods fail for such data, so much so that often
practitioners are forced to ignore the valuable information available from inexpensive
tensile testing. This is particularly a problem as such data forms a good control on the
failure envelope over the low stress range (Lade, 1993).
In summary the proposed method results in good fits of the Hoek-Brown criterion to the
data and, in particular, results in good fits in the low stress region. It appears that
published values of the parameter m
i
might be quite misleading as m
i
does not appear to
be related to rock type.
Page 3.31

Unconfined compressive strength (MPa)
U
C
S

f
r
o
m

H
B

e
q
u
a
t
i
o
n

(
M
P
a
)
1
4
7
10
40
70
100
400
700
1000
1 4 7 10 40 70 100 400 7001000

Figure 3.7. Unconfined compressive strength against that predicted by the Hoek-Brown equation
Page 3.32

Tensile strength (MPa)
T
e
n
s
i
l
e

s
t
r
e
n
g
t
h

f
r
o
m

H
B

e
q
u
a
t
i
o
n

(
M
P
a
)
0.1
0.4
0.7
1.0
4.0
7.0
10.0
40.0
70.0
100.0
0.1
0.4
0.7
1.0
4.0
7.0
10.0
40.0
70.0
100.0

Figure 3.8. Uniaxial tensile strength against that predicted by the Hoek-Brown equation

3.6 GENERALISED CRITERION FOR INTACT ROCK
There are a number of concerns regarding the formulation of the Hoek-Brown criterion:
Several authors, including Johnston (1985), note that soil, soft rock, and brittle
rock form a continuum and thus a failure criterion should be able to accommodate the
linear or near linear behaviour observed in soils and soft rocks. Fixing the exponent at a
half means that at best the criterion is a poor model of soft rocks. This is not surprising
as it was developed for brittle rocks but it is a limitation which is often overlooked by
practitioners who apply it to all rocks. Further it is a severe limitation on the extension
of the criterion to rock mass strength.
Lade (1993) in comparing the theories and the evidence regarding rock strength
criteria finds that an appropriate criterion should have three independent characteristics
the opening angle, the curvature and the tensile strength. The fixed exponent on the
Hoek-Brown criterion limits it to modelling only two of these characteristics. In fact as
often used, m
i
is varied to model the curvature over the stress range of the test results
and neither the opening angle nor the tensile strength are modelled. Lade also states that
it may be an advantage to include the tensile strength in determination of material
parameters as it stabilises the fit at low stresses. This is particularly important for slope
analysis.
If the exponent, , is allowed to vary the Hoek-Brown criterion can model widely
varying curvatures and opening angles. It is also able to include an accurate
representation of the tensile strength. This generalised Hoek-Brown criterion for
intact rock has been applied to the full data set. As would be expected, adding an extra
parameter or property always improves the fit but has many other benefits as well.
The equation becomes:

>
,
_
+
i c
i c
c
i
c
m
m
m

3 3 1
3
3
3 1
for
for 1
(3.24)

The modified least squares, Equation 3.22, is adopted.
The limits, given above for fitting the Hoek-Brown criterion, are also placed on the
parameters. For the generalised criterion these become,
c
>1, m
i
in the range 1 to 50,
and m
i
in the range 0.7 to 20 (this is the equivalent limit on
0
). In addition, is
limited to the range 0.2 to 1.
Allowing to vary provides the ability to obtain a much better fit over the low stress
range which is of greatest interest in slope analysis.
The results of the analysis are presented in a series of figures. Figure 3.9 presents a box
and whisker plot of m
i
determined from the data against the published values of m
i
.
Again it can be seen that there is little relationship between the two. Likewise, there was
found to be no relationship between m
i
and rock type.
Page 3.35

m
i
from literature
m
i

f
r
o
m

f
i
t
t
i
n
g

e
q
u
a
t
i
o
n
0
10
20
30
40
50
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

Figure 3.9. m
i
i
from test results and generalised equation

The slope of the generalised criterion at
3
=0 is 1+ m
i
, and is related to
0
by:
( ) ( ) ( ) 45 1 atan 2
5 . 0
0
+
i
m (3.25)
If a classification of samples, say by m
i
or rock type, is predictive of the triaxial
envelope at low stresses then it will be apparent on a plot of that classification against
m
i
. Figure 3.10 and Figure 3.11 present plots of m
i
against published m
i
and rock type
respectively. From Figure 3.10 it can be seen that published m
i
is not a good predictor of
the triaxial envelope at low stress. Examination of Figure 3.11 shows that there is a
weak correlation of rock type with m
i
in that fine grained rocks tend to have the
lowest values, medium to coarse grained higher and rocks with tightly interlocked
crystals the highest. It is not believed that the relationship is strong enough to be used
predictively.
Page 3.37

m
i
from literature
a
l
p
h
a
*
m
i

f
r
o
m

f
i
t
t
i
n
g

e
q
u
a
t
i
o
n
0
5
10
15
20
25
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

Figure 3.10. m
i
i
Page 3.38

A
l
p
h
a
*
m
i
0
5
10
15
20
C
l
a
y
s
t
o
n
F
i
r
e
c
l
a
y
G
r
e
e
n
s
t
o
M
u
d
s
t
o
n
e
S
e
r
p
e
n
t
i
S
c
h
i
s
t
S
h
a
l
e
C
h
a
l
k
C
h
l
o
r
i
t
i
L
i
m
e
s
t
o
n
M
a
r
b
l
e
S
i
l
t
s
t
o
n
S
l
a
t
e
B
i
o
c
a
l
c
a
D
o
l
o
m
i
t
e
A
n
h
y
d
r
i
t
S
a
l
t
C
o
a
l
T
u
f
f
P
y
r
o
c
l
a
s
R
h
y
o
l
i
t
e
A
p
l
i
t
e
B
a
s
a
l
t
L
a
m
p
r
o
p
h
T
r
a
c
h
i
t
e
A
g
g

t
u
f
f
G
r
e
y
w
a
c
k
W
h
i
n
s
t
o
n
A
n
d
e
s
i
t
e
D
i
a
b
a
s
e
D
o
l
e
r
i
t
e
Q
u
a
r
t
z
d
o
S
a
n
d
s
t
o
n
G
r
a
n
i
t
e
1
N
o
r
i
t
e
Q
u
a
r
t
z
i
t
D
u
n
i
t
e
E
c
l
o
g
i
t
e
G
a
b
b
r
o
P
e
r
i
d
o
t
i
A
m
p
h
i
b
o
l
D
i
o
r
i
t
e
Q
u
a
r
t
z
d
i
G
r
a
n
o
d
i
o
G
n
e
i
s
s
G
r
a
n
i
t
e

i

Figure 3.12 presents the
c
obtained from fitting the generalised equation against the
reported
c
. It can be seen that virtually all the data lies in a very narrow band, such that
the fitted
c
is quite close to the reported
c
. As would be expected the overall
correlation is better than that shown on Figure 3.7.
Page 3.40

Unconfined compressive strength (MPa)
U
C
S

f
r
o
m

f
i
t
t
i
n
g

e
q
u
a
t
i
o
n

(
M
P
a
)
1
4
7
10
40
70
100
400
700
1000
1 4 7 10 40 70 100 400 7001000

Figure 3.12. Unconfined compressive strength against that predicted by generalised equation

Figure 3.13 presents the fitted tensile strength versus reported tensile strength. It can be
shown that the uniaxial tensile strength
ut
is bound as:

( ) 1 +
<
i
c
ut
i
c
m m

(3.26)
Page 3.42

Tensile strength (MPa)
T
e
n
s
i
l
e

s
t
r
e
n
g
t
h

f
r
o
m

H
B

e
q
u
a
t
i
o
n

(
M
P
a
)
0.1
0.4
0.7
1.0
4.0
7.0
10.0
40.0
70.0
100.0
0.1
0.4
0.7
1.0
4.0
7.0
10.0
40.0
70.0
100.0

Figure 3.13. Uniaxial tensile strength against that predicted by generalised equation

Table 3.7 presents the errors involved in adopting
c
/(m
i
+1) as
t
. For simplicity the
lower bound has been adopted in plotting Figure 3.8 and Figure 3.13, the error in doing
this is quite small. It should be noted that it is likely that many of the reported
t
are
likely to be Brazilian tensile strengths.
Table 3.7. Error in approximating
ut
as -
c
/(m
i
+1)

m
i
Error (%)
1 All 0
0.8 All <7.6
0.5 1 19
0.5 >8 <10
0.4 1 23
0.4 >9 <10

The fit in Figure 3.13 is extremely good. Figure 3.12 and Figure 3.13 provide
considerable confidence that the fitted curves provide a very good model of triaxial
strength at low stresses. In both cases the unexplained variance of the generalised fits is
about half that of the Hoek-Brown fits.
Figure 3.14 presents a plot of against m
i
as determined for each data set from the
generalised criterion. Also shown on the figure are hyperbolae showing lines of constant
m
i
, ie
0
, for 15 to 65. Inspection of the figure shows that the constraints on m
i
,
and m
i
did not often limit the regression procedure.
Page 3.44

mi
A
l
p
h
a
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50
0
15
25
35 45 55 65
0

i

An interesting and useful observation from Figure 3.14 is that there appears to be a
relationship between and m
i
. Such a relationship is derived in the next section and is
shown on Figure 3.14. It is often thought that the curvature of the strength envelope, ie
against m
i
in the current context, should be greater for strong rocks than weak rocks.
The data set and analysis do not support this contention. Figure 3.15 shows the
relationship between and m
i
plotted for the data divided into four categories
depending on the
c
. It can be seen that the relationship is independent of strength. High
strength rocks can have linear flat failure envelopes and low strength rocks can have
steep curved envelopes. Thus
c
is a truly independent parameter in a rock failure
criterion.
Page 3.46

mi
a
l
p
h
a
UCS<= 40
0.0
0.5
1.0
0 20 40
40<UCS<=100
0 20 40
100<UCS<=200
0.0
0.5
1.0
0 20 40
200<UCS
0 20 40

i
categorised by
c

A consequence of allowing to vary at all, is that a failure envelope with a high
0
can
have a low at high stresses and thus failure envelopes for different rocks normalised
on
c
, can cross at high stresses. This cannot happen with fixed at 0.5 in which case
all envelopes cross only once at
c
. Figure 3.16 shows a family of curves, normalised by
c
, for various m
i
and the typical of the relationship shown on Figure 3.14. It can be
seen that the m
i
equals 40 curve crosses the m
i
equal 10 and 3 curves at 1.4 and 2.5
times
c
respectively. This implies that high frictional strength at low stresses is often
associated with low frictional strength at higher stresses. Figure 3.17 shows some
examples of test data that illustrate this point.
Sigma 3 / Sigma C
S
i
g
m
a

1

/

S
i
g
m
a

C
-2
0
2
4
6
8
10
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
mi 40
10
3
1

Figure 3.16. Family of failure envelopes
Page 3.48

Sigma 3 / Sigma C
S
i
g
m
a

1

/

S
i
g
m
a

C
0
2
4
6
8
10
12
14
16
18
20
-1 0 1 2 3 4 5
Set 382 Sandstone
Set 305 Granite
Set 425 Gabbro

Figure 3.17. Results showing failure envelopes crossing

A relationship between and m
i
implies that the triaxial failure envelope can be
estimated from
c
and either m
i
or
t
, with being determined by the relationship. Thus
there are no more parameters to be determined than for the usual Hoek-Brown criterion.
The parameters can be based on simple testing and provide a more accurate prediction
of strength than published m
i
values, particularly in the low stress region typical of
slopes.

3.7 GLOBAL PREDICTION
A single equation could be fitted to the entire database on the following assumptions:
The value of
c
obtained from fitting the generalised Hoek-Brown criterion is the
best estimate of
c
for each data set.
A reasonable estimate of |
t
| is obtained for each data set by dividing
c
by the
value of m
i
obtained by fitting the generalised Hoek-Brown criterion.
Figure 3.12 and Figure 3.13 show that the above assumptions are quite reasonable for
those cases where there is data to confirm them. On this basis m
i
can be set to
c
/
t
and
Equations 3.19 and 3.24 can be rewritten as:

,
_
t c c
3 3 1
1 (3.27)

,
_
,
_
,
_
,
_
+ +
,
_
c m b a
t c c
t
c
0
exp 1
3 3 1
1 (3.28)
Equation 3.27 is equivalent to the Hoek-Brown criterion but with an exponent not
necessarily equal to 0.5. Equation 3.28 is equivalent to the generalised Hoek-Brown
criterion. The exponent of Equation 3.28 is a general function that varies from a to b
with a midpoint at m
0
and a variable length of step. These equations can be fitted to the
entire data set using Equation 3.22.
Two of the data sets produced extremely large residuals and were ignored in reanalysis.
Fitting Equation 3.27, ie a constant exponent, resulted in being estimated as 0.439 and
an r
2
of 83.5%. This is a reasonable fit when the range of rocks to which it applies is
considered. Examination of the residuals reveals that a better fit will be possible as the
residual is a function of m
i
. This is illustrated on Figure 3.18, for m
i
<10 the residuals are
positive and increase with
1
. The residuals gradually reduce until for m
i
greater than
40 they are predominantly negative. This is strong evidence that is not constant.
Page 3.51

Sigma 1 / Sigma c
R
e
s
i
d
u
a
l

f
o
r

g
l
o
b
a
l

r
e
g
r
e
s
s
i
o
n

w
i
t
h

c
o
n
s
t
a
n
t

a
l
p
h
a
<= 5
-10
-5
0
5
10
0 5 10 15 20
(5,10]
0 5 10 15 20
(10,20]
0 5 10 15 20
(20,30]
-10
-5
0
5
10
0 5 10 15 20
(30,40]
0 5 10 15 20
> 40
0 5 10 15 20
Number under graph is estimated ratio of
-Sigma c / Sigma t

Figure 3.18. Residuals for global fit with constant against
3
/
c
categorised by -
c
/
t

Fitting Equation 3.28 resulted in the following estimate for the exponent:
( ) ( ) 455 . 7 exp 1 08585 . 1 4032 . 0
i
m + + (3.29)
This equation is shown on Figure 3.14 and models the results of the analysis of the
individual data sets very well. This analysis resulted in an r
2
of 94.8%, which is
extremely good for such a global fit. The residuals are plotted on Figure 3.19, there is
no or little trend with m
i
or
1
and it can be seen that this is a much better fit than
Equation 3.27 and Figure 3.18.
Page 3.53

Sigma 1 / Sigma c
R
e
s
i
d
u
a
l

f
o
r

g
l
o
b
a
l

r
e
g
r
e
s
s
i
o
n

w
i
t
h

v
a
r
i
a
b
l
e

a
l
p
h
a
<= 5
-10
-5
0
5
10
0 5 10 15 20
(5,10]
0 5 10 15 20
(10,20]
0 5 10 15 20
(20,30]
-10
-5
0
5
10
0 5 10 15 20
(30,40]
0 5 10 15 20
> 40
0 5 10 15 20
Number under graph is estimated ratio of
-Sigma c / Sigma t

Figure 3.19. Residuals for global fit with variable against
3
/
c
categorised by -
c
/
t


Figure 3.20 shows a three dimensional plot of the failure criterion described by
Equations 3.28 and 3.29, ie
1
as a function of
3
and m
i
(ie
c
/
t
). It can be seen that
for m
i
<8 the failure envelope is close to linear and then becomes more curved. The ridge
at m
i
equals 8 is well supported in the data and may reflect more or less than
Griffith behaviour.
Page 3.55

Figure 3.20. Three dimensional plot of global fit

Figure 3.21 and Figure 3.22 show slices through the model for high and low stress
ranges respectively with the equation for the midrange of each slice also shown. Thus
the upper left subgraph on Figure 3.21 presents all the
1
versus
3
data, normalised by
c
, for sets with
c
/|
t
|<=5 together with the equation for
c
/|
t
|=3. Figure 3.21 provides
the data for
3
up to three times
c
and Figure 3.22 for
3
to half of
c
. It can be seen
that the fits are very good. The ridge at high stress and m
i
=8 are apparent on Figure 3.21
with uniform behaviour at low stress seen on Figure 3.22.
Page 3.57

Sigma 3 / Sigma c
S
i
g
m
a

1

/

S
i
g
m
a

c
<= 5
0
5
10
15
0 1 2 3
(5,10]
0 1 2 3
(10,20]
0 1 2 3
(20,30]
0
5
10
15
0 1 2 3
(30,40]
0 1 2 3
> 40
0 1 2 3
Number under graph is
estimated ratio of
-Sigma c / Sigma t

Figure 3.21.
1
/
c
3
/
c
categorised by -
c
/
t
for high stress
Page 3.58

Sigma 3 / Sigma c
S
i
g
m
a

1

/

S
i
g
m
a

c
<= 5
0
1
2
3
4
5
0.0 0.1 0.2 0.3 0.4 0.5
(5,10]
0.0 0.1 0.2 0.3 0.4 0.5
(10,20]
0.0 0.1 0.2 0.3 0.4 0.5
(20,30]
0
1
2
3
4
5
0.0 0.1 0.2 0.3 0.4 0.5
(30,40]
0.0 0.1 0.2 0.3 0.4 0.5
> 40
0.0 0.1 0.2 0.3 0.4 0.5
Number under graph is
estimated ratio of
-Sigma c / Sigma t

Figure 3.22.
1
/
c
3
/
c
categorised by -
c
/
t
for low stress

Figure 3.23 presents against m
i
including Equation 3.29 and showing those data for
which there is actual, not estimated, values for both
c
and
t
. It can be seen that these
sets are distributed similarly to those for which at least one of these parameters has been
estimated by fitting the generalised Hoek-Brown criterion.
Page 3.60

mi
A
l
p
h
a
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50
Data with compressive and tensile strengths
Other data

i
showing cases with measured or reported
3
and
t


3.8 COMPARISON OF CRITERIA
A comparison of the various criterion as fitted to the database is provided in Table 3.8.
The variance explained approximates r
2
. As would be expected, the generalised Hoek-
Brown criterion provides by far the best fit, r
2
of 99.5%, as it has three parameters and
is fitted to the individual data sets. None the less, the fit obtained is considerably better
than the fit obtained by the Hoek-Brown criterion (ie with fixed at 0.5) , r
2
of 98.9%.
The unexplained variance for the generalised criterion is less than half that of the Hoek-
Brown criterion with of 0.5.
Table 3.8. Comparison of predictions
Variable/prediction Variance
Variance
explained
1
/
c

16.93 0
Global regression with constant
2.78 83.6
Hoek-Brown with published m
i
2.00 88.2
Global regression with variable
0.846 95.0
Hoek-Brown fitted to individual sets 0.186 98.9
Generalised Hoek-Brown fitted to individual sets 0.077 99.5

The above methods compare different ways of fitting triaxial data, ie different criteria
applied to actual triaxial data. Table 3.8 also allows a comparison of three methods of
prediction of triaxial strength that are not based on having actual data but are based on
parameters estimated in some other manner. The methods are discussed in the following
points:
Prediction based on global equation with variable . This method is based on
Equations 3.28 and 3.29 and estimates of
c
and
t
. It has an r
2
of 95.0% when used to
predict the test results in the database. The accuracy of the predictions are illustrated on
Figure 3.21 and Figure 3.22.
Prediction based on global equation with constant . This method, based on
Equation 3.27, is the least accurate of the three and is not discussed further.

Prediction based on published values of m
i
. This method is based on Equation
3.19 with values of m
i
estimated from those widely published in the literature. The
method has an r
2
of 88.2% when used to predict the test results in the database. On
average this method predicts the strengths well but with considerably more scatter than
that from the global equation. Figure 3.24 and Figure 3.25 present the data categorised
by published m
i
, these figures are in a similar form and can be compared to Figure 3.21
and Figure 3.22. It is clear from the figures that at low published m
i
the triaxial strength
is under predicted and at high published m
i
it is over predicted. In effect what this means
is that triaxial strength is poorly predicted by published m
i
values and the method is
predicting the average strength for all tests.
Page 3.63

Sigma 3 / Sigma c
S
i
g
m
a

1

/

S
i
g
m
a

c
<= 7
0
5
10
15
0 1 2 3
(7,9]
0 1 2 3
(9,18]
0 1 2 3
(18,19]
0
5
10
15
0 1 2 3
(19,24]
0 1 2 3
> 24
0 1 2 3
Categorised by published
values of mi

Figure 3.24.
1
/
c
I
against
3
/
c
categorised by m
I
for high stress
Page 3.64

Sigma 3 / Sigma c
S
i
g
m
a

1

/

S
i
g
m
a

c
<= 7
0
1
2
3
4
5
0.0 0.1 0.2 0.3 0.4 0.5
(7,9]
0.0 0.1 0.2 0.3 0.4 0.5
(9,18]
0.0 0.1 0.2 0.3 0.4 0.5
(18,19]
0
1
2
3
4
5
0.0 0.1 0.2 0.3 0.4 0.5
(19,24]
0.0 0.1 0.2 0.3 0.4 0.5
> 24
0.0 0.1 0.2 0.3 0.4 0.5
Categorised by
published values
of mi

Figure 3.25.
1
/
c
I
against
3
/
c
categorised by m
I
for low stress

3.9 SYSTEMATIC ERROR IN HOEK-BROWN CRITERION
If for a particular rock is not equal to 0.5 then there is a systematic error in fitting the
Hoek-Brown criterion to any triaxial test results obtained on that rock. The error is
illustrated on Figure 3.26, two data sets are shown, the upper one is for
c
, m
i
and of
30 MPa, 24 and 0.4 respectively and the lower one for 8 MPa, 5 and 0.8. The different
c
were chosen to separate the curves, and the m
i
and are typical combinations
determined in the analysis of the entire database. The solid lines represent the Hoek-
Brown fits to these data. The residuals are shown on the bottom graph, if is less than
0.5 then there are negative residuals at both the low and high end of the range of
3

tested with positive residuals in the middle range. The sign of the residuals is reversed if
is greater than 0.5. While the fits in the upper graph might look satisfactory for
engineering purposes, the errors in estimates of
c
and m
i
can be significant. Table 3.9
shows the parameters estimated from fitting the Hoek-Brown criterion to these data, it
can be seen that errors in the estimates vary from one half to five times the correct
values. Thus the parameters of this model cannot be considered material properties.
These errors are discussed in more detail below.
Sigma 3 (MPa)
S
i
g
m
a

1

(
M
P
a
)
0
20
40
60
80
100
120
140
160
0 5 10 15 20 25 30 35
R
e
s
i
d
u
a
l

(
M
P
a
)
-20
-10
0
10
20
0 5 10 15 20 25 30 35

Figure 3.26. Pattern of residuals for Hoek-Brown fits

Table 3.9. Errors in fitting Hoek-Brown criterion to materials with 0.5
Actual parameters
Material
c

m
i

r
2
(%)
Upper 30.0 24.0 0.4 100.0
Lower 8.0 5.0 0.8 100.0
Parameters determined by fitting Hoek-Brown criterion
Material Estimated
c

Estimated m
i
r
2
(%)
Upper 33.0 11.7 99.65
Lower 3.94 48.5 97.47

Consider programs of triaxial testing on rocks with
c
equal to 3 MPa, and m
i
and of
(a) 40 and 0.4 and (b) 4 and 0.8. Both are weak rocks and their triaxial strength could be
of interest in design of a large rock slope. Further consider that different test programs
are undertaken in which the maximum
3
is determined by the capacity of the triaxial
apparatus. The following test programs could result:
Program A - 12 stages to a maximum
3
of 10 MPa,
Program B - 10 stages to 5 MPa (ie omitting the last two stages),
Program C - 8 stages to 3 MPa,
Program D - 6 stages to 1.5 MPa, and
Program E - 4 stages, including UCS, to 0.7 MPa.
If there was no variability in the test apparatus or material, and measurement was
perfect, the test results would be as shown on Figure 3.27, that is there is no sample or
test error. Quite different envelopes result if the Hoek-Brown criterion is fitted to these
test programs. The estimated
c
and m
i
are given in Table 3.10. Estimated
c
varies
from 1.66 to 4.20 MPa and m
i
from 8.0 to 29.6 with very high r
2
.

c
and m
i
with
3max
for exact simulated results
Material (a)
c
= 3 MPa, m
i
= 40 and = 0.4
Program
3max

(MPa)
Stages Estimated
c
Estimated m
i
r
2
(%)
A 10 12 4.20 11.5 99.43
B 5 10 3.77 14.2 99.41
C 3 8 3.47 16.9 99.38
D 1.5 6 3.22 20.3 99.52
E 0.7 4 3.07 23.4 99.74
Material (b)
c
= 3 MPa, m
i
= 4 and = 0.8
Program
3max

(MPa)
Stages Estimated
c
Estimated m
i
r
2
(%)
A 10 12 1.66 29.6 97.96
B 5 10 2.23 17.5 98.65
C 3 8 2.59 12.6 98.96
D 1.5 6 2.85 9.60 99.46
E 0.7 4 2.96 8.01 99.85
Page 3.68

Sigma 3
S
i
g
m
a

1
0
5
10
15
20
25
30
35
40
-2 0 2 4 6 8 10 12
Artificial data for
sc=3 MPa, mi=40 and alpha=0.4
Sigma 3
S
i
g
m
a

1
0
5
10
15
20
25
30
35
40
-2 0 2 4 6 8 10 12
Artificial data for
sc=3 MPa, mi=4 and alpha=0.8

Figure 3.27. Hoek-Brown fits to artificial data

The dashed lines on Figure 3.27 show the envelopes fitted to cases A, C and E. It is
emphasised that while the upper line for material (a) and the lower line for material (b)
(ie Case E) do not look like good fits, they are in fact very good fits for the 4 test results,
below
3
equal 0.7 MPa, that form their basis with r
2
of 99.74% and 99.85%
respectively. Such results might erroneously be taken to support the contention that the
material was well modelled by the Hoek-Brown criterion. It can be concluded that the
estimated parameters are as much a function of the test program as of the material
tested. These errors would generally be obscured by the material variability but they are
still present.
Figure 3.28 and Table 3.11 present the results of analysis of data set 434, a sandstone, in
which the analysis has assumed different maximum possible
3
. This further illustrates
the errors that can occur if a Hoek-Brown envelope is fitted to material for which
does not equal 0.5. Depending on the test program, estimates of
c
obtained by fitting
the generalised criterion vary from 85.4 to 57.7 MPa and of m
i
from 6.35 to 13.1,
variations of 150% and 205%. If the Hoek-Brown criterion is fitted, the estimates vary
from 23.8 to 55.5 MPa (230%) and 31 to 138 (445%). Again the r
2
determined for the
fits are very good.
c
and m
i
with
3max
for data set 434
For generalised Hoek-Brown criterion
3max
(MPa) N Estimated
c

Estimated m
i

Estimated
r
2
(%)
All data 20 85.4 6.71 0.75 99.70
400 16 59.2 13.1 0.65 99.63
200 9 57.7 13.0 0.66 99.50
100 5 64.1 6.35 0.87 98.18
For Hoek-Brown criterion
3max
(MPa) N Estimated
c

Estimated m
i
r
2
(%)
All data 20 23.8 138 97.00
400 16 35.1 75.2 98.49
200 9 44.9 49.7 98.20
100 5 55.5 31.0 93.99
Page 3.70

Sigma 3
S
i
g
m
a

1
0
500
1000
1500
2000
2500
0 100 200 300 400 500 600 700
Sigma 3
S
i
g
m
a

1
0
500
1000
1500
2000
2500
0 100 200 300 400 500 600 700

Figure 3.28. Hoek-Brown fits to actual data

Figure 3.29 shows the residuals from fitting the Hoek-Brown criterion to data sets with
c
less than 20 MPa plotted against
3
divided by the maximum test
3
. There are four
graphs showing cases where the estimated from fitting the generalised criterion is (a)
less than 0.4, (b) between 0.4 and 0.6, (c) between 0.6 and 0.8 and (d) greater than 0.8.
On each graph the residuals have been fitted with a quadratic relationship. It can be seen
that these residuals conform almost perfectly to those predicted on Figure 3.26. This is
very strong evidence that the Hoek-Brown model is not appropriate. Figure 3.30 shows
the residuals obtained from fitting the generalised Hoek-Brown criterion. It can be seen
that these show little or no trend.

Data with estimated UCS less than 20 MPa
Sigma 3 / Maximum test sigma 3
R
e
s
i
d
u
a
l

f
r
o
m

f
i
t
t
i
n
g

H
B

e
q
u
a
t
i
o
n
Alpha <= .4
-30
-20
-10
0
10
20
30
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
Alpha (.4,.6]
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
Alpha (.6,.8]
-30
-20
-10
0
10
20
30
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
Alpha > .8
-0.2 0.0 0.2 0.4 0.6 0.8 1.0

Figure 3.29. Residuals for Hoek-Brown fits for weak rock against
3
/
3max

categorised by


Data with estimated UCS less than 20 MPa
Sigma 3 / Maximum test sigma 3
R
e
s
i
d
u
a
l

f
r
o
m

f
i
t
t
i
n
g

g
e
n
e
r
a
l

e
q
u
a
t
i
o
n
Alpha <= .4
-30
-20
-10
0
10
20
30
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
Alpha (.4,.6]
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
Alpha (.6,.8]
-30
-20
-10
0
10
20
30
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
Alpha > .8
-0.2 0.0 0.2 0.4 0.6 0.8 1.0

Figure 3.30. Residuals for generalised fits for weak rock against
3
/
3max

categorised by

3.10 APPLICATION TO SLOPE ENGINEERING
In general the triaxial strength of intact rock is not particularly important in the analysis
or design of rock slopes, even large rock slopes. The maximum depth of the failure
surface for even a 500 m high slope is generally only 100 to 150 m deep, and thus the
maximum
3
of interest is around 4 MPa. The relative contribution of the triaxial
component of strength for various rocks (ie
c
, m
i
and ) and overburden stresses is
given in Table 3.12. It can be seen that for high
0
rocks triaxial strength is generally
significant, even for quite low slopes and high strength, but for most of these situations
intact rock strength will not be critical except in forming the basis of rock mass strength.
For low
0
rocks triaxial strength is important for low strength or high stress conditions
and it is these situations for which intact strength may be critical in design.
Table 3.12. Triaxial component of strength
c

3

High
0
case
m
i
= 50, = 0.4
%
Low
0
case
m
i
= 0.8, = 0.9
%
300 4.0 24 2.3
100 4.0 59 6.9
30 4.0 140 23
10 4.0 280 68
3 4.0 570 230
300 0.5 3.4 0.3
100 0.5 9.8 0.9
30 0.5 29 2.9
10 0.5 70 8.6
3 0.5 160 29

Table 3.13 and Figure 3.31 provide a comparison of different methods of predicting the
triaxial strength of low strength rocks at low stress. The predicted strengths are
compared with the measured strengths for all cases in the database for which
c
is less
than 20 MPa and
3
is between 0 and 5 MPa (excluding UCS test results). The
variances of the residuals scaled on
c
are given in Table 3.13 and the scaled residuals

plotted on Figure 3.31. In general the order of accuracy of the various prediction
methods is the same as discussed for the overall predictions. It is of interest to note that
the global generalised equation (r
2
of 88.6%) is almost as accurate a prediction of the
triaxial strength of these rocks as that obtained by fitting the Hoek-Brown criterion
directly to triaxial test data (r
2
of 90.3%). This reflects the fact that there is abundant
evidence that does not equal 0.5 for a large proportion of the rocks tested see Figure
3.14 and thus there will be systematic errors at low stress.
Table 3.13. Comparison of predictions for weak rocks at low stress
Variable/prediction Variance
Variance
explained %
1
/
c

7.69 0
Global regression with constant
1.91 75.1
Hoek-Brown with published m
i
1.99 74.1
Global regression with variable
0.88 88.6
Hoek-Brown fitted to individual sets 0.74 90.3
Generalised Hoek-Brown fitted to individual sets 0.12 98.5

The above situation arises because the range of
3
over which the tests were performed
hardly ever corresponded to 5 MPa and thus there were almost always systematic errors
at the lower stresses tested. It is sometimes argued that the solution to this is to test over
a range of
3
that represents the field conditions, but this is hardly ever possible as
generally the one set of testing is used to design low and high slopes. Further for a given
slope different portions of the failure surface are at different stresses. Another solution
is to determine the parameters as a function of stress, but this virtually defeats the
purpose of adopting a non-linear failure envelope and confirms they are not a material
property.

Sigma 3
G
e
n
e
r
a
l
i
s
e
d

H
B
-5
0
5
10
0 2 4
Sigma 3
H
B
-5
0
5
10
0 2 4
Sigma 3
G
l
o
b
a
l
-5
0
5
10
0 2 4
Sigma 3
G
l
o
b
a
l

w
i
t
h

f
i
x
e
d

a
l
p
h
a
-5
0
5
10
0 2 4
Sigma 3
H
B

a
n
d

p
u
b
l
i
s
h
e
d

m
i
-5
0
5
10
0 2 4

Figure 3.31. Residuals against
3
for various fits

A useful approximation of the effective stress parameters, c
0
and
0
, at low stress can be
obtained in the following manner. Equation 3.26 can be rearranged to provide an
estimate of m
i
based on
c
and
t
as:

t
c
i
t
c
m
< 1 (3.30)
In addition, can be estimated from Equation 3.29,
o
is given by Equation 3.25 and:
( ) ( )
5 . 0
0
1 2
i c
m c + (3.31)
The writers do not argue that this approximation is a substitute for triaxial testing but, in
the absence of such testing, the approximation should be more accurate than other
methods of estimation such as using
c
and published values of m
i
. Further it results in a
linear failure envelope which is exact at the origin and as such is convenient to use in
many slope stability programs.

3.11 CONCLUSION
This chapter presented an overview of the strength of intact rock. It was demonstrated
that the method of fitting the criterion to the test data has a major effect on the estimates
obtained of the material properties. The results of a recent analysis of a large database of
test results demonstrated that there are inadequacies in the Hoek-Brown empirical
failure criterion as currently proposed for intact rock and, by inference, as extended to
rock mass strength. The parameters m
i
and
c
are not material properties if the exponent
is fixed at 0.5. Published values of m
i
can be misleading as m
i
did not appear to be
related to rock type. The Hoek-Brown criterion can be generalised by allowing the
exponent to vary. This change resulted in a better model of the experimental data. The
most accurate method of estimating m
i
and is through using triaxial tests on intact
rock. The recommended method for regression of the data is modified least squares,
Equation 3.22, combined with the extended formulation of the generalised criterion,
Equation 3.24. The equations are repeated below.

( )
( )

;
>
3 1 3 3
3 1 1 1

i
m
(3.22)

>
,
_
+
i c
i c
c
i
c
m
m
m

3 3 1
3
3
3 1
for
for 1
(3.24)
Analysis of individual data sets indicated that the exponent, , is a function of m
i
which
is, in turn, closely related to the ratio of
c
/
t
. A regression analysis of the entire
database provided a model to allow the triaxial strength of an intact rock to be estimated
from a reliable measurement of its uniaxial tensile and compressive strengths. The
method proposed is the most accurate of those methods that do not require triaxial
testing and is adequate for preliminary analysis. An analysis was presented that showed
applying the Hoek-Brown criterion to most rocks results in systematic errors. Simple
relationships for triaxial strength that are adequate for slope design were presented.
The Shear Strength of Rockfill Page 4.1

4 THE SHEAR STRENGTH OF ROCKFILL
4.1 OUTLINE OF THIS CHAPTER
The work by Marsal (1973) on the shear strength of rockfill showed that the strength of
rockfill may vary directly with normal effective stress, dry density, particle roughness,
particle crushing strength and inversely with grain size, uniformity of grading, and
particle shape.
The aim of this Chapter is to verify and extend this work using an extensive literature
review and an analysis of a triaxial shear strength database which has been collected
from the literature, and from organisations who have carried out testing for dams and
other projects.
In the context of this Chapter the term rockfill encompasses both material of alluvial
origin and that of blasted quarry origin. The differentiation between the two can be
made by looking at parameters such as angularity.
This Chapter also comprises part of the larger study that is investigating rock mass
strength. The author believes that it is reasonable to assume that a compacted rockfill is
representative of a poor quality rock mass as defined by Hoek and Brown (1980). The
author has therefore used the data collected in this report to create a lower bound for the
Hoek-Brown criterion.
This Chapter presents a summary of the main factors affecting rockfill strength from the
literature; the triaxial shear strength database; statistical results from the database; and
several shear-strength criteria for rockfill using Hoek-Brown and other methods.


4.2 FACTORS AFFECTING THE SHEAR STRENGTH OF
ROCKFILL
The author believes that the shear strength of rockfill at a particular confining stress
may be seen as the combination of a basic friction angle (shearing between rock
surfaces),
b
, plus a dilation component less an amount caused by asperity or particle
crushing/shearing and reorientation of particles. In terms of friction angle:

crush
+ = i
b
(4.1)
Dilation ceases to occur at the critical confining pressure as defined by Seed and Lee
(1967).
Marsal (1973) found that the shear strength of rockfill may vary directly with normal
effective stress, dry density, particle roughness, particle crushing strength and inversely
with grain size, uniformity of grading, and particle shape. The following discussion
centres on what factors have been found to affect the shear strength of rockfill. An
attempt is made to relate these factors to the equation above.
4.2.1 Confining Pressure
Several authors have shown the shear strength curve for rockfill is non-linear,
particularly at low confining pressures (triaxial: Leslie, 1963, Marachi et al, 1969, Leps,
1970, Bertacchi & Bellotti, 1970, Penman et al., 1982 and Indraratna et al., 1993; direct
shear: Dobr & Rozsypal, 1974 and Anagnosti & Popovic, 1982; plane strain: Al-
Hussaini, 1983). This would be consistent with the theory that as normal stress is
increased dilation is suppressed and therefore shear strength increase is reduced. Indeed,
Boughton (1970) found some indication that the limiting value of as
3
increases
is the value for the surfaces of the individual rock pieces or
b
in Equation 4.1.
The effect of a curved strength envelope has a large impact on the stability analysis of
rockfill dams where shallow slip surfaces are shown by the analysis to be critical using
traditional average c- analysis. These failure surfaces are not found in practice and this

is likely to be due to the high frictional strength of rockfill at the low confining
pressures acting near the surface.
Figure 4.1 shows a curved envelope together with various c- representations. The
tangent friction angle,
t
,

is defined as the slope of the tangent to the curved shear
strength envelope at a given normal stress. The secant friction angle,
sec
, is defined as
the slope of the tangent from the origin to the Mohrs circle for a particular normal
stress (Equation 4.2). It is this secant friction angle that is most commonly quoted when
test results are published.

=

3 1
3 1 1
sec
sin

(4.2)
Where the Mohr-Coulomb parameters are taken as c = 0 and =
sec
the shear strength
will be underestimated for normal stresses less than
n2
and over estimated for stresses
greater than
n2
. Where the Mohr-Coulomb parameters are taken as c = c
t
and =
t
the
strength will be overestimated at confining pressures less than the normal stress at the
tangent point. Where c
t
is taken as zero, and =
t
, as sometimes is done by designers,
the strength will be significantly underestimated. This shows the importance of using a
non-linear strength criterion or stress dependent parameters e.g. a bi-linear strength
envelope.

sec
n2
0
<
env
<
t
Failure
envelope
t
=
env
<
0
env
<
t
<
0
Assumes =
t
and c = 0
Tangent to circle
through origin =
secant to failure
envelope
Tangent to failure
envelope

Figure 4.1. Methods for representing the shear strength envelope
As is typical in the literature, this report uses =
sec
in its discussion unless otherwise
stated.
Marachi et al (1969) carried out large scale triaxial tests on highly angular argillite,
crushed basalt and rounded amphibolite and found that does not appear to decrease
significantly beyond
n
= 4.5MPa. Note that they did not perform tests beyond a
confining pressure of 4.5MPa. Figure 4.2 and Figure 4.3 show curved strength
envelopes from triaxial and direct shear tests respectively which support this.
Indraratna and his co-workers (Indraratna et al., 1993, 1998, Indraratna, 1994)
performed tests on greywacke rockfill and basalt ballast. They found that the shear
strength of the failure envelope was highly curved for confining stresses of less than
500kPa. Penman et al (1982) found a similar result for confining stresses below 400kPa.
Indraratna found that the shear strength could be approximated by a linear Mohr-
Coulomb criterion at stresses higher than 1.5MPa (as compared to the Marachi value of
4.5MPa above). Figure 4.4 shows the variation of
sec
with
n
on a log scale presented
by Indraratna et al. (1993). The figure also shows a lower limit to the shear strength of
rockfill proposed by Indraratna et al (1993).

sec
, with respect to cell confining
stress,
3
, for (a) dense and (b) medium dense crushed basalt from triaxial tests
(Al-Hussaini, 1983)
0.5
1.0
1.5
2.0
2.5
0.5 1.0 1.5 2.0
n
(MPa)
/
n
1
1
2a
2b
4
5
6
7
3
2b
2b
crushed quarry rockfill
gravel rockfill
1 marble
2a limestone (0.15 x 0.15m)
2b limestone (0.07 x 0.07m)
3 mica & phyllite
4 sandstone-claystone-marl flysch
5 claystone-marl flysch
6 crystalline schist
7 sandstone

Figure 4.3. Average strength of rockfills from large-scale direct shear tests
(Anagnosti & Popovic, 1982)


sec
, with normal stress
n

(Indraratna et al., 1993)
4.2.2 Particle Strength
The strength of the individual particles could be expected to have some effect on the
shear strength of rockfill. Where the confining pressure,
3
, is low compared to the
particle (or unconfined compressive) strength,
c
, it could be expected that variations in
particle strength would have minimal effect on the shear strength. The only exception to
this is if it could be shown that
c
affects
b
. It has been shown however that
b

generally falls within a narrow range for defects in rocks which have not been
tectonically sheared. As
3
/
c
decreases, dilation will be the main contributor to
increased shear strength. As
3
/
c
increases, failure of particles could be expected
through crushing shearing and/or splitting, and thus a reduction in dilation would take
place leading to lower shear strength.
Particle strength should therefore play a major role with higher strength particles
generally leading to higher strength rockfill. Hirshfeld et al (1973) found the curvature
of the shear strength envelope to be more distinct for weak particles. This would be due
to the rapid reduction in dilation with increase in confining stress. A reduction in
dilation for high strength particles could be expected at much higher stresses than weak
particles.

Anagnosti and Popovic (1982) found for crushed rocks, there is a large drop in shear
strength up to normal stresses of 1MPa for strong rockfills (Figure 4.5) with a much
smaller drop for weaker flysch materials (Figure 4.6). It is likely that tests were not
carried out at low enough confining pressures to capture the curvature in the flysch
material.

Figure 4.5. Shear strength and grain size curves for crushed (a) limestone and (b)
marble (Anagnosti and Popovic, 1982)

Figure 4.6. Shear strength and grain size curves for crushed flysch sandstone-marl
rockfill (Anagnosti and Popovic, 1982)

4.2.3 Uniformity Coefficient
Generally, it could be expected that a poorly-graded rockfill (low uniformity coefficient,
c
u
, Equation 4.3), would have a higher strength than a well-graded rockfill assuming a
constant void ratio for both. A well-graded material would be more likely to reduce the
amount of dilation required due to the gaps in the gravel matrix being filled with
smaller particles. However, Marachi et al (1969) claim that if both rockfills were
compacted to their maximum density then the well graded material could be expected to
be stronger as it would have the greater density.

10
60
d
d
c
u
= (4.3)
Where, d
60
is the particle size for which 60% is finer.
Chiu (1994) using data by Marsal (1967) found that the effect of c
u
on the shear strength
of rockfill was more pronounced for low
3
, with little effect at high
3
. c
u
affects
dilation however, at higher
3
that dilation may have been retarded. Sarac & Popovic
(1985) found that the failure envelope was more curved for narrowly graded materials.
Al-Hussaini (1983) found strength increased with c
u
but felt it may also have been
affected by maximum particle size. Anagnosti & Popovic (1982) found for gravels there
is a marked drop in strength for the high strength limestone below
3
=1MPa possibly
due to dilational effects below
3
=0.4MPa caused by the uniform grading (Figure 4.7).
The well-graded and weaker material does not show this effect (Figure 4.8).

Figure 4.7. Shear strength and grain size curves for different gradings of
limestone gravel (Anagnosti and Popovic, 1982)

Figure 4.8. Shear strength and grain size curves for (a) crystalline schist and (b)
sandstone gravels (Anagnosti and Popovic, 1982)

4.2.4 Density
When comparing the degree of compaction of rockfill relative density or void ratio can
be used. Where materials of the same rock type and grading characteristics are tested
dry density may be used as it will be related to relative density. Dry density or void ratio
will not be a good parameter to use where different types and gradings of rockfills are
compared.
Rockfill with a higher relative density has been found to have a higher shear strength
(Boughton, 1970, Leps, 1970, Marsal, 1973, Williams & Walker, 1984, Sarac &
Popovic, 1985, Tosun et al., 1999). The shape of the Mohr-Coulomb failure envelope is
also affected. Initially dense rockfill shows a marked curvature showing a distinct drop
in the friction angle whereas an initially loose sample shows minimal curvature and
drop in friction. The two curves tend to merge at very high confining pressures. This is
consistent with the behaviour of granular soils which reach the critical state or constant
volume/void ratio at large strains. Marachi et al (1969) indicate that the curves continue
as a straight line whose projection passes through the origin. This is likely due to the
dense material requiring dilation at low stresses to fail. At higher confining stresses
dilation is restrained and particle shearing/crushing occurs resulting in a lower angle of
friction. Initially loose material will require much less dilation as particles have more
freedom to move (e.g. rotation) during shearing. Thus without the dilation there will be
a minimal drop in the friction angle.
Al-Hussaini (1983) found that dense crushed basalt has a higher strength than medium
dense crushed basalt. Dobr & Rozsypal (1974) tested basalt at different dry densities in
direct shear. They found that the strength increases with density but, the strength
difference reduces with confining pressure. Chui (1994) also found little difference at
high confining pressures. Alva-Hurtado et al. (1981) found the opposite with the
difference in strengths increasing with confining pressure.
Zeller et al (1957), as reported in Marachi et al (1969), performed triaxial tests on
scalped rockfill samples for Goscheoenalp Dam in Switzerland. The confining pressure
for all tests was 88kPa. It should be noted that the scalping method used did not result in
parallel gradations. Zellers results (Figure 4.9 and Figure 4.10) show a decrease in

friction angle of approximately 50 to 40 for an initial porosity increase of 8% for all
dry rockfill samples (d
max
of 10mm to 100mm).

Figure 4.9. Scalped rockfill gradings (Marachi et al, 1969)

Figure 4.10. Strength porosity relationships with
3
= 88kPa (Marachi et al, 1969)
Marachi et al (1969) tested rockfill material from Pyramid Dam. They found a drop in
friction angle of approximately 4 for an increase in void ratio of 0.3.
Nakayama et al. (1982) found that the friction angle was reduced with an increase in the
void ratio (Figure 4.11). There is no indication in the paper as to what the confining
pressure or method of assessing was. Boughton (1970) and Anagnostic & Popovic
(1985) found a similar result although, as might be expected, Anagnostic & Popovic
(1985) found the effect was more noticeable at low confining pressures and low C
u
.

40
41
42
43
44
45
0.25 0.3 0.35 0.4 0.45 0.5
e
triaxial - green schist - 12% -4.76mm triaxial - green schist - 18% -4.76mm
triaxial - quartz schist - 13% -4.76mm triaxial - quartz schist - 20% -4.76mm
direct shear - quartz schist - 18% -4.76mm direct shear - quartz schist - 25% -4.76mm
direct shear - green schist - 17% -4.76mm

Figure 4.11. Void ratio vs angle of friction (modified from Nakayama et al., 1982)
4.2.5 Maximum Particle Size
There is some conjecture as to the effect of the maximum particle size on the shear
strength of rockfill. It is generally accepted that the shear strength decreases with
particle size (Marachi et al., 1969 & 1972, Marsal, 1973, Chui, 1994). However, some
researches claim no effect (Charles et al., 1980) or the opposite effect (Anagnosti &
Popovic, 1981). Some data from Marsal is tabulated below.
The effect of maximum particle size should be considered as two issues:
(I) the effect of increasing maximum particle diameter, d
max
with constant sample
diameter, D; and
(II) the effect of increasing d
max
with constant d
max
/D ratio.
(both assuming constant sample height to sample diameter, H/D ratio)

max
with Constant D
Marsal (1973) performed tests on basalt rockfill (the results of the tests are tabulated
below). The results show an increase in friction angle of 3-4% for a change in d
max
/D
from 0.07 to 0.18. At high normal stresses (3.9MPa) the effect is more limited (0.3).
Table 4.1. Increase in with d
max
/D from Marsal (1973) data for different
n

Source Rock Type D
(mm)
C
u

C
c

n

(MPa)
d
max1
/D

d
max2
/D (
2
-
1
)/
2

(%)
Marsal (1973) Basalt 1130 A 11 - 18 5.21 0.55 0.8 0.07 0.18 3
Marsal (1973) Basalt 1130 A 11 - 18 5.21 0.55 1.6 0.07 0.18 4
Marsal (1973) Basalt 1130 A 11 - 18 5.21 0.55 3.9 0.07 0.18 0.3
Note: d
max1
,
1
and d
max2
,
2
are the maxi mum particle sizes and secant friction angle for sample one and sample two respectively.
max
with Constant d
max
/D
Thiers & Donovan (1981) present Figure 4.12 to estimate the shear strength of rockfill
at field scale. The figure plots the angle of internal friction versus maximum particle
size. It should be noted that all the samples tested had a constant d
max
/D ratio.
Table 4.2. Reduction in with particle size, d
max
, from Marsal (1973) data
Source Rock Type d
max
/D C
u

C
c

n

(MPa)
d
max1

(mm)
d
max2

(mm)
(
1
-
2
)/
2

(%)
Marsal (1973) Silicified conglomerate .18 A 10 1.72 3.92 38.1 200 7
Marsal (1973) Granitic-gneiss .18 SA 14 1.30 3.92 38.1 200 20
Note: d
max1
,
1
and d
max2
,
2
are the maximum particle sizes and secant friction angle for sample one and sample two respectively.

Figure 4.12. Friction angle vs maximum particle size (Thiers & Donovan, 1981)

4.2.6 Silt and Sand Fines versus Gravel and Larger Particle Content
The percent of material passing 2mm is believed to affect the shear strength of rockfill.
At the point where the failure shear surface does not incorporate gravel to gravel contact
it could be expected that the shear strength of the finer particles would be the sole
control on the shear strength of the material. Large triaxial tests for silty gravels were
carried out by the USBR (1966). Tests were carried out at 0, 35, 50 and 65% gravel
content. Shear strengths were found to be similar for the 0 and 35% gravel content tests.
Shear strengths increased considerably at 50% and 65% gravel content. Further tests on
clayey gravels found that the shear strength increased considerably between 42 and 50%
gravel content. Other conclusions that were reached included:
Clayey and silty gravel show the shear strength is unaffected by gravel contents
below 35%.
Sandy gravel shows an increase in gravel gives an increase in strength up to a
maximum of 50% gravel
At 50% gravel content, silty gravel and clayey gravel strengths significantly increase.
At 65% gravel content the silty gravel has similar shear strength to the sandy gravel.
Data from USBR (1966) and USBR (1961) for silty gravel and clayey gravel are plotted
on Figure 4.13 and Figure 4.14 respectively. Trend lines (logarithmic) have been plotted
on the curves to show the general trends of the data. They should not be used for design.
Marsal (1976) reports two triaxial tests each on rockfill-silt and rockfill-sand mixtures
and compares these to a test on rockfill only. The clean rockfill and 10% sand-rockfill
mixtures had a of 34.1 whilst the 30% sand-rockfill had a of 39. Marsal (1976)
attributes the difference to the lower initial void ratio in the 30% sand-rockfill mixture.
The 10% silt-rockfill mixture showed a decrease in to 28.8 whilst the 30% silt-
rockfill mixture had the strength properties of the silt.
Nakayama et al. (1982) performed triaxial and direct shear tests on schist gravels
(quartz schist and green schist) with maximum particle size of 63.5mm at Inamura
Rockfill Dam. It was found that the friction angle was reduced with an increase in the
percent of 4.76mm particles (Figure 4.15).

30
35
40
45
50
55
60
65
70
75
80
0.0 0.2 0.4 0.6 0.8 1.0
3
(MPa)
38% gravel
53% gravel
67% gravel

Figure 4.13. Effect of gravel content on for silty gravel based on USBR (1966)
30
32
34
36
38
40
42
44
46
48
50
0.0 0.1 0.2 0.3 0.4
3
(MPa)
26% gravel
42% gravel
55% gravel
68% gravel
Log. (26% gravel)
Log. (42% gravel)
Log. (55% gravel)
Log. (68% gravel)

Figure 4.14. Effect of gravel content on for clayey gravel based on USBR (1961)

35
40
45
50
10 15 20 25 30
-4.76mm content (%)
direct shear - quartz schist - e=0.28

direct shear - quartz schist - e=0.39

Figure 4.15. 4.76mm content vs secant angle of friction (Nakayama et al., 1982)
4.2.7 Particle Angularity
The strength of highly angular rockfill could be expected to be higher than that for a
rounded rockfill at similar relative density and low confining pressures. This is due in
some part to the interlocking effect of the angular particles and also increased dilation.
At high stresses the effect may not be as prominent and may be the opposite. An angular
rockfill may allow for stress concentrations that cause breakage of the particles at high
confining pressures reducing dilation and leading to a lower overall rockfill strength
than the rounded particles with less stress concentrations.
The shape co-efficient, C
f
, is often used to describe the shape of particles and is given in
Equation 4.4. The shape coefficient is the volume of n particles of gravel over the
equivalent volume of n spheres of diameter D.

=
n
n
f
D
V
C
0
3
0
6
(4.4)
Bertacchi & Berlotti (1970) found that serpentinite (UCS = 100-150MPa, laminar
surfaces, sharp edges, C
f
= 0.11) gives higher friction angles than tonalite (UCS =
150MPa, irregular rough surfaces, C
f
= 0.17) due to the irregularity in the shape of the
serpentinite particles. Sarac & Popovic (1985) found a similar result but the effect was
less with higher confining pressures.

4.2.8 Other Factors
The type of shear test is believed to have an influence over results. Generally plane
strain tests in the literature (Al-Hussaini, 1983 dense crushed basalt, Barton &
Kjrnsli, 1981, Charles & Watts, 1980) give higher friction angles,
ps
, than triaxial
shear tests,
ax
. Tests by Marsal et al (1967) and Marsal (1973) discussed by Charles
(1991) showed at failure is greater in the plane strain tests by up to 8 for the same
minor principal stress. The Danish Code of Practice (Steensen-Bach, 1989) gave:

ax ps
= 1 . 1 (4.5)
Whilst Wroth (1984) gives:

ax ps
= 9 8 (4.6)
McWilliam (2001), working with the author, used published data to show that the ratio
between
ps
and
ax
varied from 1.02 to 1.1 (c.f. 1.125 from equation 4.6). The ratio
varied with material type and confining pressure.
The triaxial test provides a correct value of as it incorporates the intermediate
principal stress,
2
(=
3
). In a plain strain test, strain is prevented in the direction of
2
. Where the popular Mohr-Coulomb criterion is used the intermediate stress is
ignored which leads to a miscalculation of . It is a common misconception that using
the Mohr-Coulomb derived from plain strain tests gives a better correlation to field
values and should therefore be used in preference to the from triaxial tests. The
backanalysis of failures have traditionally been carried out in two dimensions. This
procedure ignores
2
in the same manner as in the derivation of from the plane strain
lab tests. Thus, the plain strain and the back calculated are both incorrect but, as
they have similar errors in their derivation they tend to give similar results. The author
recommends the use of the derived from triaxial tests for design.
The shear strength of rockfill may decrease with moisture content (Frassoni et al., 1982,
Chui, 1994, Marsal, 1967). This is probably related to the drop in unconfined
compressive strength for some intact rocks.

4.2.9 Summary of Factors Affecting the Secant Friction Angle
Table 4.3 shows a summary of the factors affecting the secant friction angle discussed
in this section. The most significant effects on the secant friction angle appear to be
caused by confining pressures, density and maximum particle size. There is also a
substantial effect on
sec
at high percentages of material finer than gravel. Minor effects
on the secant friction angle are caused by angularity and the uniformity coefficient.
Table 4.3. Summary of factors affecting the secant friction angle
Parameter Effect on
sec

with increase
in parameter
Comment
Confining pressure Decrease Significant effect. The rate of decrease in
sec
will drop with increasing confining
pressure
Unconfined compressive strength of
intact rock
Increase Effect will depend on the ratio of
confining stress to compressive strength
Uniformity coefficient Decrease Minor effect and may reverse if samples
are compacted to their maximum density
prior to testing
Density Increase
Maximum particle size (assuming the
ratio of maximum particle size to
sample diameter is constant)
Decrease
Ratio of maximum particle size to
sample diameter
Increase
Angularity Increase The effect will be most noticeable with
highly angular material
Percent finer than gravel size in sample Decrease The effect will not be significant at low
percentages. At high percentages strength
will approach that of the finer material.


4.3 SHEAR STRENGTH CRITERIA
Table 4.5 shows a number of rockfill shear strength equations from the literature. The
equations are numbered in the first column of the table. The criteria are empirical
having been based on laboratory tests of rockfill.
De Mello (1977) proposed Equation 4.8 as suitable for representing the curved strength
envelope. Charles & Watts (1980) also used this form of equation. Indraratna et al.
(1993) developed a non-dimensional form of Equation 4.8 for both shear stress and
principal stress plots (Equations 4.9 and 4.11). Charles (1991) gives values for the
constants A and B in the De Mello (1977) equation (Table 4.4).
Table 4.4. Parameters obtained using De Mello (1977) (Charles, 1991)
I
D
A B
Sandy gravel .95 4.4 .81
Soft rockfill .95 4.2 .75
Soft rockfill .70 1.4 .90

Indraratna et al. (1998) also uses Equation 4.12 to describe the shear strength of basalt
ballast.
Doruk (1991) created a modified version of the Hoek-Brown criterion by removing the
s component and thus rendering the unconfined compressive strength of the rock mass
to zero. (Equation 4.13)
Sarac & Popovic (1985) analysed a large number of large scale (0.7x0.7x0.4 to
1.9x2.9x1.5m) direct shear tests carried out at the Institute for Geotechnics and
Foundation Engineering, Sarajevo (Equation 4.10). The materials tested were generally
limestone, sandstone, serpentinite and slate and were from a mixture of sources
including: natural; rock debris; and quarry rockfill. The tests were generally carried out
for materials for embankment dams over stress ranges between 0.05 to 2.0MPa. Figure
4.17 shows the test results.
Page 4.20

Table 4.5. Various shear strength criteria for rockfill
Eq Reference Equation Parameters Development
4.8
De Mello (1977)
Charles & Watts (1980)
B
n
A = A, B see Table 4.4 Empirical curved envelope.
4.9
Indraratna et al (1993)
Indraratna (1994)
b
c
n
c
a

a,b = 0.25,0.83 (lower bound,
n
=0.1-1MPa)
a,b = 0.71,0.84 (upper bound,
n
=0.1-1MPa)
a,b = 0.75,0.98 (lower bound,
n
=1-7MPa)
a,b = 1.80,0.99 (upper bound,
n
=1-7MPa)
Non-dimensionalised form of (4.8). Note that if A in (4.8) is
independent of
c
then a is not independent of
c
.
4.10 Sarac & Popovic (1985) ( )
B
n
A
0 max
=
A increases with
c
, C
u
, , d
50

0.7 to 1.5 (Figure 4.19,Figure 4.20)
B increases with
c
, C
u
,
0.419 to 0.911 ( Figure 4.18)
0
= 1MPa
Developed from large-scale direct shear tests (up to
n
=
2MPa) on quarry rockfill and natural gravels. Various
sedimentary rocks were used.(Figure 4.17)
4.11
Indraratna et al (1993)
Indraratna (1994)
c c
3 1

, = 0.4,0.62 (lower bound, 0.1 to 1MPa)
, = 0.78,0.65 (upper bound, 0.1 to 1MPa)
, = 2.71,0.96 (lower bound, 1 to 7MPa)
, = 3.58,0.90 (upper bound, 1 to 7MPa)
Alternative of (4.9) for principal stresses
4.12 Indraratna et al (1998)
b
a
3
3
1

a, b = 84.98,-0.49 (gradation A)
a, b = 125.17,-0.56 (gradation B)
Developed empirically for two gradations of ballast
Undefined at
3
= 0
4.13 Doruk (1991)
c
a
c
m

+ =
3
3 1
m, a

Developed from Hoek-Brown by setting the rockfill
compressive strength to zero (i.e. s = 0).

4.4 DATABASE OF TRIAXIAL SHEAR TESTS
The author has collated a database of large-scale triaxial shear tests. The database has
over 989 individual triaxial shear tests from 307 sets of data. The information in the
database has been collected from selected quality published papers and reports and from
unpublished laboratory reports from organisations that have carried out high quality
testing for dams and other projects. There is an assumption inherent in collating a large
amount of data that all data sets are of equal quality. The dangers of making this
assumption were mitigated to some degree by using the quality sources stated above.
Although virtually all tests were accepted, there were some that were rejected. Tests that
were carried out at a very high strain rate were rejected. As were results where the
strength of the sample was estimated due to the test reaching the strain limit of the
testing apparatus prior to attaining the maximum strength of the sample. Table 4.6
shows the parameters recorded in the database. The complete database is contained in
Appendix F on a CD-ROM that is appended to this thesis.
Table 4.7 summarises the basic statistics of the data. The triaxial cells used for tests
ranged from a diameter of 50.8mm to a diameter of 1130mm. The maximum particle
size in the tests ranged from 4.8mm to 200mm. The minimum particle size in the tests
ranged from 0.0035mm to 40mm. Note that many tests recorded a minimum particle
size of less than a particular diameter.
40% of tests had zero fines (% passing 0.075mm) content whilst 80% of the tests had a
fines content of no more than 5%. 10% of the triaxial tests had fines contents in excess
of 20%. The largest fines content for a test in the database was 45%. Note that the
uniformity coefficient and the coefficient of curvature are very high for tests where
fines have been added to a large diameter rockfill.
18.7% of the triaxial tests were on basalt material whilst 14.9% were on granite, 8.2%
sandstone and 3.9% limestone. Many tests did not give the rock type of the material
tested.
49% of the samples tested were considered angular (with an angularity rating = 8).

Table 4.6. List of parameters in triaxial shear strength database
Parameter Description
Case Test number within database
Reference Data source
Data set Identifying number given for each set of tests performed. Taken from
reference where given.
Test Test number within data set
Site Place of origin of material or location (e.g. dam name)
Origin Source of gravel size particles (e.g. crushed, blasted, alluvial etc.)
Type Rock particle type (e.g. sandstone etc)
UCS (MPa) Unconfined compressive strength of individual particles
UCS
*
(MPa) Includes UCS estimated where not given (see Section 4.5.1)
Ang Angularity of particles from: angular; sub-angular; sub-rounded; rounded
Ang rat A value from 1 (rounded) to 8 (angular) based on a description of the
rockfill given in reference. (0 = not given)
Ds (mm) Sample diameter
Hs (mm) Sample height
Fines (%) Percent of material finer than 0.075mm
Fines
*
(%) Includes estimated fines where not given (see Section 4.5.1)
d
min
(mm) Minimum particle size
d
10
(mm) Diameter corresponding to 10% finer
d
30
d
50
d
60
d
90
d
max
(mm) Maximum particle size
c
u

Coefficient of uniformity:
10
60
d
d
c
u
=
c
c

Coefficient of curvature:
60 10
2
30
d d
d
c
c
=
Grav (%) Gravel content (defined as percent of material greater than 2mm
diameter)
r
d
Relative density
d
(kN/m
3
) Dry unit weight
e
min
Minimum void ratio
e
max
Maximum void ratio
e
i
Initial void ratio
e
i
*
Includes estimated void ratio where not given (see Section 4.5.1)
n Porosity
B
g
Breakage co-efficient (Marsal)
G
S
Specific gravity
w (%) Moisture content
S (%) Degree of saturation
3
Minimum principal effective stress
1
Maximum principal effective stress
af
Axial strain at failure
vf
Volumetric strain at failure
Brit Type of failure (B-brittle, T-transitional, D-ductile). Qualitative term
based on the stress-strain curves (if given) presented in the reference.
sec
Secant friction angle (calculated)
n
Normal stress on failure plane (calculated)
1
-
3
Principal stress difference (calculated)
1
/
3
Principal stress ratio (calculated)


Table 4.7. Summary of basic statistics from the rockfill database
Parameter Minimum Average Maximum
Sample diameter (mm) 50.8 332 1130.3
Minimum particle size (mm) 0.0035 3.6 40
d
10
(mm) 0.001 4.3 53
d
30
(mm) 0.01 9 90
d
50
(mm) 0.09 16 100
d
60
(mm) 0.22 20 110
Maximum particle size (mm) 4.8 59 200
Fines content (%) 0 4.1 45
Uniformity coefficient, c
u
1.3 81 3243
Coefficient of curvature, c
c
0.04 2 16.9
Unconfined compressive strength of
intact rock particles (MPa)
25 153 761


4.5 DATABASE ANALYSIS
This section describes the qualitative and quantitative (using statistical methods)
analysis of the rockfill database. Analyses using both the secant friction angle and
principal stresses were carried out. A final analysis using the Hoek-Brown criterion was
performed as part of the development of a criterion for rock mass strength.
4.5.1 Analysis Methodology
The database contained missing information for some of the test results. If a parameter
used in the statistical analysis was missing for a particular test, that test result was
generally ignored. However, to increase the number of tests, certain assumptions were
made prior to the analysis. These are:
Where an unconfined compressive strength of the intact rock was not given and
reasonable estimates could be made, it was estimated into three groupings of 40, 100
and 200MPa rocks based on a description of the rock type, degree of weathering and
also the origin of the material. These samples were placed. The effect of including these
data sets was evaluated during the statistical analysis.
Where the minimum particle size, d
min
, was recorded as less than a particular diameter
the following table was used to estimate d
min
for the analysis.
Minimum particle size given, d
min
Minimum particle size used, d
minhat

<0.075 (and d
10
>0.075) 0.01
<0.075 (and d
10
0.075) 0.005
<0.02 0.005
<0.005 (and d
10
>0.003) 0.001
<0.005 (and d
10
0.003) 0.0005

Where an initial void ratio was not given it was calculated from one of the following
equations (in order of preference):

n
n
e
=
1
(4.17)

( )
min max max
e e D e e
r
= (4.18)
1 =
d
w s
G
e
(4.19)
where
e = void ratio
n = porosity
e
max
= maximum void ratio
e
min
= minimum void ratio
D
r
= relative density (%)

w
= unit weight of water (kN/m
3
)

d
= dry unit weight (kN/m
3
)
G
s
= specific gravity (assumed = 2.7 if not given)
4.5.2 Secant Friction Angle,
sec
, Versus Normal Stress,
n

Relationships between secant friction angle,
sec
, and normal stress,
n
, were
investigated using the rockfill shear strength database. This relationship was chosen as it
is commonly used in the literature and practice to present and analyse data. The results
are discussed in detail below however, it should be noted at the outset that the fits
obtained using this statistical analysis were inferior to those obtained using a
relationship of maximum principal stress,
1
, versus minor principal stress,
3
. This is
of particular importance at both low and high confining stress,
3
.
4.5.2.1 General Assessment of Database
Figure 4.21 to Figure 4.27 show plots of
sec
vs
n
with
n
plotted on a logarithmic
scale. These plots are discussed in turn below. Note that these analyses were carried out
for the complete data sets. No parameter (e.g.
c
) was estimated for these plots. Where a
specific parameter has been investigated a small + symbol on a figure represents a test
result where the parameter is unknown (except for Figure 4.21).

q Figure 4.21
sec
vs
n
: This figure shows that
sec
has a range of approximately 25
for any given value of
n
. This shows that the bounds given by Indraratna et al
(1993) (equating to a range of about 15) for this type of plot do not indicate the full
range of
sec
possible. Their upper bound appears to be too conservative whilst the
lower bound, although providing a better fit, has several test results below it. The
gradient of the bounds does not appear to be steep enough to adequately describe the
variation in
sec
relative to
n
.
q Figure 4.22
sec
vs
n
sorted on angularity rating: The addition of angularity rating
does not assist greatly with improving the prediction of
sec
vs
n
. The figure shows
that angular rockfill (angrat = 7 or 8) generally lies above sub-angular material
(angrat = 5 or 6). However, rounded material (angrat = 1 or 2) lies roughly in the
centre of the data with angular material lying both above and below the data.
q Figure 4.23
sec
vs
n
sorted on rock type = basalt: This was plotted to see the
amount of variability there was within one class of rock type. Basalt is used as
railway ballast, road base and dam rockfill. A basalt rockfill might be expected to lie
at the higher strength end of all rockfills. Although the figure shows this to be true
there is still a 10-15 range in
sec
for the material for any given
n
.
q Figure 4.24
sec
vs
n
sorted on coefficient of uniformity, c
u
: The figure shows that
c
u
does not appear to be a very good additional predictor with data for each c
u
range
generally being well and evenly spread. It could be argued that the lower strength
rockfills predominately have high c
u
values (>12). However, these types of rockfills
are spread throughout the full strength range.
q Figure 4.25
sec
vs
n
sorted on maximum particle size, d
max
: This data is also quite
scattered although there appears to be a tendancy for high d
max
rockfills to result in a
higher strength than small d
max
rockfills.
q Figure 4.26
sec
vs
n
sorted on percent fines content: The rockfills with a high fines
content (>20%) generally have a lower
sec
. Much of the other data, particularly
with fines less than 10% is well scattered.

q Figure 4.27
sec
vs
n
sorted on unconfined compressive strength,
c
: Rockfills with
a
c
of less than 100MPa appear to have lower strengths than those greater than
100MPa. It is difficult to discern any real difference in the results for rockfills with
strengths of 100-200MPa and greater than 200MPa.
A general conclusion from the analysis of the graphs may be made that the factors
described above do have some effect on the shear strength of rockfill, although none of
them stands out as being substantially better than the others as a predictor. It is unwise
to estimate strength based solely on these graphs due to the interrelationship between
the various factors in the tests. A better statistical analysis is to perform a global
analysis of the data using all these factors and assessing whether they are statistically
significant. This is performed in the next section.

Page 4.33

Sigma n (kPa)
P
h
i

s
e
c
a
n
t

(
d
e
g
)
20
30
40
50
60
70
80
10 40 70 100 400 700 1000 4000 7000

sec
vs normal stress,
n

Approximate location of bounds from
Indraratna et al (1993) - see Figure 4.4.
Page 4.34

Sigma N (kPa)
P
h
i

s
e
c

(
d
e
g
)
20
30
40
50
60
70
80
1
0
4
0
7
0
1
0
0
4
0
0
7
0
0
1
0
0
0
4
0
0
0
7
0
0
0
1
0
0
0
0
angrat=1 or angrat=2

sec
vs normal stress,
n
, sorted on angularity rating
Note a + symbol indicates the
angularity rating is unknown
for that particular test
Page 4.35

Sigma n (kPa)
P
h
i

s
e
c

(
d
e
g
)
20
30
40
50
60
70
80
1
0
4
0
7
0
1
0
0
4
0
0
7
0
0
1
0
0
0
4
0
0
0
7
0
0
0
1
0
0
0
0
type="basalt"
type="basalt r"
type="basalt1"

sec
vs normal stress,
n
, sorted on rock type = basalt
material is not basalt or is
unknown for that particular test
Page 4.36

Sigma n (kPa)
P
h
i

s
e
c

(
d
e
g
)
20
30
40
50
60
70
80
10
40
70
100
400
700
1000
4000
7000
10000
cu<=2
cu<=6
cu<=12
cu>12

sec
vs normal stress,
n
, sorted on coefficient of uniformity, c
u

coefficient of uniformity is
Page 4.37

Sigma N (kPa)
P
h
i

s
e
c

(
d
e
g
)
20
30
40
50
60
70
80
1
0
4
0
7
0
1
0
0
4
0
0
7
0
0
1
0
0
0
4
0
0
0
7
0
0
0
1
0
0
0
0
dmax<=20
dmax<=50
dmax<=100
dmax>100

sec
vs normal stress,
n
, sorted on maximum particle size, d
max
Note a + symbol
indicates d
max
is unknown
for that particular test
Page 4.38

Sigma n (kPa)
P
h
i

S
e
c

(
d
e
g
)
20
30
40
50
60
70
80
1
0
4
0
7
0
1
0
0
4
0
0
7
0
0
1
0
0
0
4
0
0
0
7
0
0
0
1
0
0
0
0
fines<=0
fines<=5
fines<=10
fines<=20
fines>20

sec
vs normal stress,
n
, sorted on percent fines (passing 0.075mm) content
fines content is unknown for
that particular test
Page 4.39

Sigma n (kPa)
P
h
i

s
e
c

(
d
e
g
)
20
30
40
50
60
70
80
1
0
4
0
7
0
1
0
0
4
0
0
7
0
0
1
0
0
0
4
0
0
0
7
0
0
0
1
0
0
0
0
ucs>=200
ucs>=100
ucs>0

sec
vs normal stress,
n
, sorted on unconfined compressive strength of the rock substance, UCS (MPa)
UCS of the intact particles is

4.5.2.2 Statistical Analysis of Database
A non-linear statistical analysis was carried out on the database. The estimation method
used was least squares. An analysis of the form shown below was found to be the most
effective for relating to
n
.

c
n
b a + = (4.20)
The analysis resulted in the following constants:
( ) ( ) 150 0459 . 0 2 756 . 0 172 . 0 267 . 0 43 . 36 + + = UCS c FINES ANG a
c
(4.21)
( ) ( ) 150 408 . 0 2 105 . 5 549 . 0 27 . 10 51 . 69 + + = UCS c FINES ANG b
c
-0.408 (4.22)
3974 . 0 = c (4.23)
where,
( ) 5 = rating angularity ANG for angularity rating >5.5; otherwise 0
FINES = percentage of fines passing 0.075mm (%)
c
c
= coefficient of curvature
UCS = unconfined compressive strength of the rock substance (MPa)
This function resulted in a variance explained of just 61.7%. It should be noted that the
addition of the uniformity coefficient, c
u
, and the maximum and minimum particle sizes,
d
max
and d
min
, did not result in a better fit to the data. As the vs
n
curve is
unconstrained at both low and high
n
it is unlikely that a reasonable fit will be found
regardless of the function used. Due to the relatively poor statistical results it was
decided to proceed with an analysis using principal stresses.


4.5.3 Maximum Principal Stress,
1
, versus Minimum Principal Stress,
3

The analysis using
sec
vs
n
gave poor results. In this section principal stresses are
used. These are constrained at
3
=0,
1
=0. A non-linear statistical analysis was carried
out on the database. The estimation method used was least squares. An analysis of the
form shown in Equation 4.24 was found to be the most effective for relating
1
to
3
.

3 1
= RFI (4.24)
An initial analysis on the whole database (988 data sets) found (Variance explained =
97.13%):
RFI = 4.7002
= 0.8972
To determine which parameters affected RFI, the analysis was carried out iteratively.
The analysis was carried out using a parameter in the database as a predictor of RFI
(e.g.
c
). After each analysis the residuals were plotted against the parameters in the
database. Where a trend was noted for a particular parameter another analysis was
carried out using that additional parameter. The analysis resulted in the following
equations (based on 869 data sets and with a variance explained = 98.82%):
8726 . 0 =
UCS FINES d ANG e
RFI RFI RFI RFI RFI RFI 30598 . 0 1568 . 1 0027 . 0 48763 . 0 3491 . 6
max
+ + =
(4.25)
i
e
e
RFI
+
=
1
1
(4.26)
0 otherwise angular, if 1 = =
ANG
RFI (4.27)
max
max
d RFI
d
= (mm) (4.28)
( )
( ) 20 Fines
20 Fines
1

+
=
e
e
RFI
FINES
where fines is in % (4.29)
( )
( ) 110 UCS
110 UCS
1

+
=
e
e
RFI
UCS
where UCS is in MPa (4.30)


Equation 4.29 and 4.30 are effectively smooth step functions that are ameniable to
equation solving routines.
Figure 4.28 shows the variation of RFI
e
with initial void ratio. This shows that an
increase in initial void ratio will lead to a decrease in
1
and shear strength.
Highly angular rockfill (angularity rating = 8) showed an increase in strength. There
was no noticeable trend for less angular rockfill (angularity rating 7) and hence
Equation 4.27 was chosen.
The strength of the rockfill (
1
) was found to be proportional to the maximum particle
size (Equation 4.28).
The unconfined compressive strength,
c
, was found to have a limited effect on the
strength of the rockfill. Generally strengths below a
c
of 100MPa had similar strengths,
as did those with
c
greater than 120MPa. Equation 4.30 acts as a switch at a
c
around
110MPa. Below a
c
of 110MPa the function rapidly approaches zero and above a
c
of
110MPa the function rapidly approaches unity (Figure 4.29).
The fines content was found to only affect the strength of the rockfill where the fines
content was greater than approximately 20%. Equation 4.29 acts as a switch at a fines
content around 20%. Below a fines content of 20% the function rapidly approaches zero
and above a fines content of 20% the function rapidly approaches unity (Figure 4.30).

0.5
0.6
0.7
0.8
0.9
1
0 0.2 0.4 0.6 0.8 1
Void Ratio
RFI
e

Figure 4.28. RFI
e
versus void ratio
0
0.2
0.4
0.6
0.8
1
50 60 70 80 90 100 110 120 130 140 150
UCS (MPa)
RFI
UCS

Figure 4.29. RFI
UCS
versus unconfined compressive strength

0
0.2
0.4
0.6
0.8
1
0 5 10 15 20 25 30 35 40
FINES (%)
RFI
FINES

Figure 4.30. RFI
FI NES
versus percent fines

The residuals versus unconfined compressive strength, maximum particle size, initial
void ratio, angularity and fines content are shown on Figure 4.31 to Figure 4.35
respectively. These figures shown no discernible trend from which it can be assumed
that the equations provided above are reasonable over the range of the data.
Figure 4.36 shows the residuals plotted against sample diameter. The lack of any major
trend in the data suggests that sample diameter does not have a large effect on rockfill
strength at least up to diameters of about 1.2m. This suggests that the process of scaling
using parallel grading lines when testing rockfill for use in the field is valid and should
not affect the results. Note however, that if fines are increased due to the scaling
strength (
1
) may be reduced.
Figure 4.37 to Figure 4.41 show the effect that a change in unconfined compressive
strength, angularity, fines content, maximum particle size and initial void ratio have on
1
respectively.

Figure 4.42 shows the data used in the analysis together with lines representing RFI=7
(approximate upper bound), RFI=4.7 (approximate best estimate) and RFI=3
(approximate lower bound). Figure 4.43 shows the same plot with all the data in the
database plotted. It shows that the bounds are still reasonable and that the data left out
of the analysis (due to insufficient information) would not have changed the results
significantly. Figure 4.44 shows the same plot for a stress range (
3
) up to 1.5MPa.

Page 4.46

-3
-2
-1
0
1
2
3
0 100 200 300 400 500 600 700 800
UCS (MPa)
R
e
s
i
d
u
a
l
s

(
M
P
a
)

Figure 4.31 Residuals versus unconfined compressive strength of intact rock
Page 4.47

-3
-2
-1
0
1
2
3
0 50 100 150 200 250
Dmax (mm)
R
e
s
i
d
u
a
l
s

(
M
P
a
)

Figure 4.32. Residuals versus d
max
Page 4.48

-3
-2
-1
0
1
2
3
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Void Ratio
R
e
s
i
d
u
a
l
s

(
M
P
a
)

Figure 4.33. Residuals versus void ratio
Page 4.49

-3
-2
-1
0
1
2
3
0 1 2 3 4 5 6 7 8 9
Angularity Rating
R
e
s
i
d
u
a
l
s

(
M
P
a
)
2 - Rounded
4 - Sub-rounded
6 - Sub-angular
8 - Angular

Figure 4.34. Residuals versus angularity rating
Page 4.50

-3
-2
-1
0
1
2
3
0 5 10 15 20 25 30 35 40
Fines (%)
R
e
s
i
d
u
a
l
s

(
M
P
a
)

Figure 4.35. Residuals versus fines content
Page 4.51

-3
-2
-1
0
1
2
3
0 200 400 600 800 1000 1200
Sample diameter (mm)
R
e
s
i
d
u
a
l
s

(
M
P
a
)

Figure 4.36. Residuals versus sample diameter

0
5
10
15
20
0 1 2 3 4 5
3
(MPa)
1
(MPa)
UCS = 50MPa
UCS = 110MPa
UCS = 200MPa
Angularity = 7
Fines = 0%
d
max
= 60
Void ratio = 0.4

1
0
5
10
15
20
0 1 2 3 4 5
3
(MPa)
1
(MPa)
Angularity = 7
Angularity = 8
UCS = 100 MPa
Fines = 0%
d
max
= 60
Void ratio = 0.4

1
(7 = sub-angular to angular; 8 = angular)

0
5
10
15
20
0 1 2 3 4 5
3
(MPa)
1
(MPa)
Fines = 40%
Fines = 20%
Fines = 0%
UCS = 100 MPa
Angularity = 7
d
max
= 60
Void ratio = 0.4

1
0
5
10
15
20
0 1 2 3 4 5
3
(MPa)
1
(MPa)
dmax = 200
dmax = 100
dmax = 5
UCS = 100 MPa
Angularity = 7
Fines = 0%
Void ratio = 0.4

1

0
5
10
15
20
0 1 2 3 4 5
3
(MPa)
1
(MPa)
e = 0.8
e = 0.5
e = 0.2
UCS = 100 MPa
Angularity = 7
Fines = 0%
d
max
= 60

1
Page 4.55

0
5
10
15
20
25
30
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
3
(MPa)
1
(MPa)
RFI = 7
RFI = 3
RFI = 4.7

Figure 4.42.
1
vs
3
showing data used in analysis and RFI relationship
Page 4.56

0
5
10
15
20
25
30
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
3
(MPa)
1
(MPa)
RFI = 7
RFI = 3
RFI = 4.7

Figure 4.43.
1
vs
3
showing all data and RFI relationship
Page 4.57

0
2
4
6
8
10
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
3
(MPa)
1
(MPa)
RFI = 7
RFI = 3
RFI = 4.7

Figure 4.44.
1
vs
3
showing all data and RFI relationship (
3
up to 1.5MPa)

4.5.3.1 Secant Friction Angle Versus Normal Stress
Figure 4.37 to Figure 4.41 are replotted as curves for
sec
vs
n
by using the following
equations:

=

3 3
3 3 1
sec
sin

RFI
RFI
(4.31)

1
3 sec
cos
+
=

RFI
n
(4.32)
Table 4.8 shows the variation in
sec
due to changes in various parameters for
n
=1MPa
and 0.5MPa. These give some idea of the influence of different parameters on the secant
friction angle. Note that these friction angles are based on assumptions as shown on
Figure 4.45 to Figure 4.49. The values chosen for the parameters are realistic upper and
lower bounds from the database.
Table 4.8. Changes in
sec
on Figure 4.45 to Figure 4.49 for
n
=1MPa and
n
=
0.5MPa
n
= 0.1MPa
n
= 0.5MPa
n
= 1MPa Parameter Value
sec

()
Change in
with
increase in
parameter
sec

()
Change in
with
increase in
parameter
sec

()
Change in
with
increase in
parameter
50 46.8 42.2 40.3
c
(MPa)
200 48.1
+1.3
43.6
+1.4
41.8
+1.5
1-7 46.7 42.2 40.3 Angularity
rating
8 48.8
+2.1
44.4
+2.2
42.6
+2.3
0 46.7 42.2 40.3 Fines content
(%)
20 43.9
-2.8
39.1
-3.1
36.9
-3.4
50 47.0 42.4 40.4 Maximum
particle size
(mm)
200 44.8
-2.2
40.3
-2.1
38.3
-2.1
0.2 49.9 45.5 43.7 Initial void ratio
0.6 43.5
-6.4
36.4
-9.1
34.2
-9.5


30
35
40
45
50
0 2 4 6 8
n
(MPa)
sec
UCS = 50MPa
UCS = 200MPa
Angularity = 7
Fines = 0%
d
max
= 60
Void ratio = 0.4

sec
30
35
40
45
50
0 2 4 6 8
n
(MPa)
sec
Angularity = 7
Angularity = 8
UCS = 100MPa
Fines = 0%
d
max
= 60
Void ratio = 0.4

sec
(7 = sub-angular to angular; 8 = angular)

20
25
30
35
40
45
50
0 2 4 6 8
n
(MPa)
Fines = 0%
Fines = 20%
Fines = 40%
UCS = 100MPa
Angularity = 7
d
max
= 60
Void ratio = 0.4

sec
30
35
40
45
50
0 2 4 6 8
n
(MPa)
dmax = 5
dmax = 100
dmax = 200
UCS = 100MPa
Angularity = 7
Fines = 0%
Void ratio = 0.4

sec

20
25
30
35
40
45
50
0 2 4 6 8
n
(MPa)
e = 0.2
e = 0.4
e = 0.6
UCS = 100MPa
Angularity = 7
Fines = 0%
d
max
= 60

sec
4.5.4 Hoek-Brown Criterion
The author considers that rockfill could represent a very poor quality rockmass. As such
an analysis of the database using the Hoek-Brown criterion was performed. The results
from this analysis could then be used as a lower bound rock mass strength estimator.
The statistical analysis was carried out for data where the UCS was known. This
resulted in a sample of 409 triaxial tests. Two approaches were used: one where the
variables are
3
and the dependent variable
1
; and the other where the variables are
3
/
c
and the dependent variable
1
/
c
. Similar analyses to those discussed previously
were carried out i.e. non-linear, quasi-Newton with a loss function of (observed
predicted)
2
. A further analysis was carried out where the estimated values of the UCS of
the intact rock were used where no UCS was recorded for the sample.
The ability of the Hoek-Brown criterion to estimate the shear strength of rockfill (and
subsequently poor quality rock mass) was performed by assuming typical Hoek-Brown
parameters of s = 0 and a = 0.6. This was compared with a further analyses where a was
defined as a parameter. The results from the analyses are shown in Table 4.9.

Table 4.9 and Figure 4.50 show that if the recommended Hoek-Brown parameters are
used a poor fit results. Due to the enforced curvature (a = 0.6),
1
is overpredicted at
low
3
and under predicted at high
3
. Much better fits (Table 4.9 and Figure 4.50) are
obtained where a was a parameter. These analyses showed that an a of 0.90 to 0.95 was
much more suitable for rockfill. Similarly to intact rock, the interrelation between a and
m
b
is also illustrated by the results in Table 4.9.
The conclusion from the point of view of rock masses is that a is too constrained in the
current Hoek-Brown criterion. Modifications need to be made such that the exponent, a,
towards a value of 0.90-0.95 for poor quality rockfill.
A look at the forms of the equations used shows that the method of fitting is very
important to the results obtained. In this case, non-dimensionalising the equations
resulted in a better variance explained.
Table 4.9. Results from the statistical analysis of the rockfill database using the
Hoek-Brown equation
Form of equation Data sets m
b
a Variance explained
6 . 0
3
3 1
0
+ =
c
b
c
m

405
ci
known
0.42929
0.6
(input)
88.95%
a
c
b
c
m
+ = 0
3
3 1

405
ci
known
2.40706 0.89999 97.40%
6 . 0
3 3 1
0
c
b
c c
m

405
ci
known
0.75486
0.6
(input)
90.93%
a
c
b
c c
m

0
3 3 1

405
ci
known
2.71833 0.94877 98.73%
a
c
b
c
m
+ = 0
3
3 1

988
ci
estimated
where
unknown
2.39551 0.8954 97.42%
Page 4.63

0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
3
/
c
1
/
c
6 . 0
3 3 1
75486 . 0

+ =
c c c

94877 . 0
3 3 1
71833 . 2

+ =
c c c

Figure 4.50. Statistical analysis results using Hoek-Brown formula and for a = 0.6 and a = 0.95
Page 4.64

4.6 CONCLUSION
This chapter presented an overview of the strength of rockfill. An analysis of a large
database of test results was used to develop two new shear strength equations, one
relating the secant friction angle and normal stress and the other the principal stresses.
The equation for principal stresses provided a much better fit to the data and is
recommended. Of the parameters statistically investigated, the unconfined compressive
strength, particle angularity, fines content, maximum particle size and void ratio were
found to have the most significant effect on the shear strength of rockfill.
An analysis using the database and the Hoek-Brown criterion demonstrated that the
criterion provided a poor fit to the rockfill data due to the restrictions placed on the
exponent a of approximately 0.6. A good fit was obtained where a was free to vary up
to unity (a value of 0.9 to 0.95 was obtained). This has important implications for the
use of the Hoek-Brown criterion for very poor quality rock masses.

Empirical Rock Slope Design Page 5.1
5 EMPIRICAL ROCK SLOPE DESIGN
5.1 INTRODUCTION
Due to the complexity of rock masses, a number of researchers have attempted to
correlate rock slope design with rock mass parameters. Many of these methods have been
subsequently modified by others and are now currently being used in practice for
preliminary and sometimes final design. This chapter presents a review of the more
commonly used empirical rock slope design methods. The historical development of these
methods and the data used to support them is discussed. A number of case studies are
assessed using the different methods. Finally, a new relationship between the Geological
Strength Index, GSI, slope height and stable slope angle is presented for use in the design
of cuts in rock masses in dry conditions or under moderate water pressures.
5.2 REVIEW OF THE ROCK MASS RATING SYSTEMS
There are several empirical rock mass rating techniques that can be utilised in the design
of slopes. These include:
RMR - Rock mass rating (Bieniawski, 1976 & 1989)
MRMR - Mining rock mass rating (Laubscher, 1977 & 1990)
RMS - Rock mass strength (Selby, 1980)
SMR - Slope mass rating (Romana, 1985)
SRMR - Slope rock mass rating (Robertson, 1988)
CSMR - Chinese system for SMR (Chen, 1995)
GSI - Geological strength index (Hoek et al. 1995)
M-RMR - Modified rock mass classification (nal, 1996)
BQ - Index of rock mass basic quality (Lin, 1998)

The majority of methods require the determination of a basic rock mass rating. The rating
is usually calculated as the summation of a number of rating values that account for intact
rock strength, block size, defect condition and possibly groundwater. A number of the
methods then adjust this value based on such factors as defect orientation, excavation
method, weathering, induced stresses and the presence of major planes of weakness.
Table 5.1 compares a number of these methods. The numbers show the range of
weightings possible for each component of the rating system, whilst an asterix, *, shows
which parameters are taken into account in each method. It should be noted that the
different rock mass rating systems use varying methods to account for each parameter.
The MRMR and M-RMR adjustment factors are multipliers whilst the adjustment factors
for the other methods are added to the basic rock mass rating. The maximum value of 141
for CSMR assumes a slope height of 50m. The numbers shown in brackets, ( ), in Table
5.1 represent negative values.
Table 5.1. Comparison of weightings for various rock mass rating methods
Method RMR
76
RMR
89
MRMR RMS SMR CSMR M-RMR SRMR GSI
Intact strength 0-15 0-15 0-20 5-20 0-15 0-15 0-15 0-30 0-15
Block size 8-50 8-40 0-40 8-30 8-40 8-40 0-40 8-40 8-50
- Spacing * * * * * * * * *
-
RQD * * * * * * * *
Defect condition 0-25 0-30 0-40 3-14 0-30 0-30 0-30 0-30 0-25
- Persistence * * * * * * * * *
- Aperture * * * * * * * * *
- Roughness * * * * * * * *
- Infilling * * * * * * * *
- Weathering * * * * * * *
Ground water 0-10 0-15 * 1-6 0-15 0-15 0-15 - 10
Defect orientation (60)-0 (60)-0 63-100% 5-20 (60)-0 (60)-0 (12)-(5)
- Strike * * * * *
- Dip * * * * * *
- Slope dip defect dip * *
Excavation method - - 80-100% - (8)-15 (8)-15 80-100% - -
Weathering - - 30-100% 3-10 - - 60-115% - -
Induced stresses - - 60-120% - - - - - -
Major plane of weakness - - - - - - 70-100% - -
TOTAL RANGE (52)-100 (52)-100 0-120 25-100 (60)-115 (63)-141 (7)-105 8-100 18-100
B
A
S
I
C

R
O
C
K

M
A
S
S

R
A
T
I
N
G
A
D
J
U
S
T
M
E
N
T
S
- -

Note values in brackets are negative
5.2.1 Methods for Estimating the Basic Rock Mass Rating
The basic rock mass rating attempts to capture the main features of a rock mass that, in the
context of this report, affect the shear strength of the rock mass and subsequently the
stability of slopes in that rock mass.
5.2.1.1 The Rock Mass Rating, RMR, and Geological Strength Index, GSI
Bieniawskis (1973, 1975, 1976, 1989) rock mass rating (RMR) is probably the most
commonly used rock mass rating system for estimating rock mass strength. Initially
created to assess the stability and support requirements of tunnels, it has been found to be
useful in assessing the strength of rock masses for slope stability. Table 5.2 shows the
rating method of Bieniawski (1989). It should be noted that the weighting of the
parameters has changed slightly over the years since its development. Wherever the RMR
is referred to in this document, the subscript will refer to the year of publication of that
version. For example, RMR
76
, refers to the RMR published by Bieniawski (1976).
Hoek et al (1995) modified the RMR so as to make it more applicable to assessing the
strength of rock masses. The result of this was the Geological Strength Index (GSI). Table
5.3 shows the components and ratings from the GSI.
The GSI is based on RMR
76
and is calculated by summing the ratings for each parameter
and adding 10. A rating value of 10 is added as the GSI assumes water conditions to be
dry. No corrections are made for joint orientation as it is assumed to be favourable. Hoek
et al. (1995) believe that joint orientation and water conditions should be assessed during
the analysis.
Hoek et al. (1995) allow for the use of RMR
89
to estimate the GSI by using GSI = RMR
89
5. The author warns that RMR
89
should be used with caution as it can lead to a GSI
difference of up to 10 when compared with the GSI derived as above. Hoek (2000)
provides another method of estimating GSI using Figure 5.1.
This chapter is primarily concerned with the estimate of the strength of large rock masses
and the subsequent correlation with slope angles. The GSI is made up of several
components: intact rock strength; rock quality designation (RQD); defect spacing; and
joint condition (water is already set to dry). All these components can be affected by
scale and thus must be considered carefully when designing for very large rock masses. A
larger discussion of the applicability of using the GSI for rock slopes is contained in
Chapter 6.

Table 5.2. Rock mass rating (Bieniawski, 1989)

Table 5.3. Geological strength index, GSI (Hoek et al, 1995)

5.2.1.2 Mining Rock Mass Rating, MRMR
Laubscher (1977, 1984 and 1990) developed a classification system, based on
Bieniawski (1973), that gave a basic rating from 0 (very poor) to 100 (very good) (Table
5.4 and Table 5.5). The system changes some of the weightings of parameters and alters
the method of determining joint spacing and condition compared to Bieniawski (1973).
Laubscher assesses the rock joint spacing for one, two or three continuous joint
systems. A joint is continuous if its length is greater than one diameter of the excavation
or 3m. It is also continuous if it is less than 3m but it is displaced by another joint that
is, the joints are features that define blocks of ground. Joints logged from boreholes are
placed in three 30 dip ranges. Experience is used to divide these ranges into further
joint sets.
MRMR was specifically developed for use in the assessment of support for underground
openings and there is ambiguity in the assessment of MRMR when dealing with slopes.
For poor and very poor quality rock masses (MRMR<40) the MRMR can be largely
influenced by the evaluation of joint spacing and joint/water conditions. The appropriate
evaluations are very difficult to assess when only borehole data is available and clearly
requires a large degree of judgement even when exposures are available.

Figure 5.1. Estimate of GSI based on geological descriptions. (Hoek, 2000)

Table 5.4. MRMR (Laubscher, 1977)
RQD (%) 100-91 90-76 75-66 65-56 55-46 45-36 35-26 25-16 15-6 5-0
Rating 20 18 15 13 11 9 7 5 3 0
IRS (MPa) 141-136 135-126 125-111 110-96 95-81 80-66 65-51 50-36 35-21 20-6 5-0
Rating 10 9 8 7 6 5 4 3 2 1 0
Defect spacing Depends on number of defect sets and spacing
Rating 30.................................................................................................................................. 0
Defect condition 45 .......................................................See Table ........................................................5
Rating 30.................................................................................................................................. 0

Groundwater
Inflow/10m length 0 25 l/min 25 125 l/min 125 l/min
Joint water pressure/
1
0 0.0 - 0.2 0.2 - 0.5 0.5
Description dry moist Moderate pressure Severe problems
Rating 10 7 4 0
Table 5.5. Defect condition rating for MRMR (Laubscher, 1977)
Parameter Description Percentage adjustment
to maximum rating of
30
Joint expression
(Large-scale)
Wavy unidirectional 90-99
Curved 80-89
Straight 70-79
Joint surface
(Small-scale)
Striated 85-99
Smooth 60-84
Polished 50-59
Alteration zone Softer than wall rock 70-99
Joint filling Coarse hard-sheared 90-99
Fine hard-sheared 80-89
Coarse soft-sheared 70-79
Fine soft-sheared 50-69
Gouge thickness < irregularities 35-49
Gouge thickness > irregularities 12-23
Flowing materials > irregularities 0-11
5.2.1.3 Rock Mass Strength, RMS
Selby (1980, 1982, 1987, 1993 and Moon & Selby, 1983) developed the RMS system
based on correlations between RMS and stable slope angles of natural rock outcrops. The
slopes were located in New Zealand, Antarctica, the Namib Desert and the margins of the
Central plateau of southern Africa. The geology included sedimentary sequences and
metamorphic (quartzite, gneiss, schist and marble) and igneous (dolerite, basalt and
granite) rock masses.
The RMS is calculated in a similar way to RMR with a summation of rating parameters
for intact rock strength, weathering, defect spacing, aperture, defect orientation relative to
the slope, defect continuity and groundwater flow (Table 5.6).
Orr (1992) found an approximate (r
2
= 0.88) correlation between RMS and RMR
88
:
130 2 . 2
88
= RMS RMR (5.1)
Selby uses natural slopes in his database, and thus the slopes have been exposed over
geological time. These slopes could therefore be seen to be conservative when compared
to slopes in a pit with a limited design life. Selby breaks the natural slopes into small
sections (generally bedding layers) and assesses the slope angle of these. The slope
angles of these segments of limited height are generally structurally controlled (slaking
mudstones may be an example of an exception) in practice and thus not applicable to
correlations with rock mass.
Schmidt and Montgomery (1996) modified the RMS for application to deep-seated
bedrock landsliding in sedimentary rocks. They examined a total of 61 slopes, of which
17 were rockslides. The slopes were in the Eocene Chuckanut Formation, which
comprises a fluvial sequence of interbedded sandstone and siltstone/mudstone. Schmidt
and Montgomery note that there were distinct planes of weakness at lithological contacts.
The sandstone and siltstone/mudstone layers showed distinct differences in strength.
Table 5.6. RMS Classification and Ratings (mod. Selby, 1980)
Intact strength
(N-type Schmidt
100-60 60-50 50-40 40-35 35-10
Hammer R)
Rating 20 18 14 10 5
Weathering unweathered slightly
weathered
moderately
weathered
highly
weathered
completely
weathered
Rating 10 9 7 5 3
Defect spacing >3m 3-1m 1-0.3m 0.3-0.05m <0.05m
Rating 30 28 21 15 8
Defect
orientation
very favourable

steep dips into
slope, cross
defects
interlock
favourable

moderate dips
into slope
fair

horizontal dips, or
nearly vertical
(hard rocks only)
unfavourable

moderate dips
out of slope
very unfavourable

steep dips out of
slope
Rating 20 18 14 9 5
Defect aperture <0.1mm 0.1-1mm 1-5mm 5-20mm >20mm
Rating 7 6 5 4 2
Defect continuity none continuous few continuous continuous, no
infill
continuous,
thin infill
continuous, thick
infill
Rating 7 6 5 4 1
Groundwater
outflow
none trace slight
<25
l/min/10m
2
moderate
25-125
l/min/10m
2

great
>125
l/min/10m
2

Rating 6 5 4 3 1

The main changes to the RMS were in the defect orientation parameter. Schmidt and
Montgomery (1996) state that defects with moderate dips into the slope should have a
higher rating than those with steep dips into the slope. Steep dips into the slope are more
likely to cause toppling and so this change appears reasonable. They also give steep dips
out of the slope a higher rating than moderate dips out of the slope. This appears to
contradict what would be expected.
An important point to note is that Schmidt and Montgomerys data appears to be based on
translational slides and hence structurally controlled rather than on rotational rock mass
slides. This would be backed up by their statement that the vast majority of deep-seated
rockslides occur on hillslopes inclined at 15 to 35. Schmidt and Montgomery also
state that the low RMS values associated with rockslides are due to the intact rock and
defect orientation parameters.
Based on their data, Schmidt and Montgomery claim that the RMS successfully
discriminates localised areas of low rock mass strength within a landscape exhibiting
deep-seated rockslides. However, as this data appears defect controlled it is not
considered to be reliable for use with a rock mass rating and hence this author cannot
come to the same conclusion. Also, the method could only apply in bedded rocks given its
database.
5.2.1.4 Slope Rock Mass Rating, SRMR
Robertson et al (1987), using back analysis of slopes at Island Copper Mine in British
Columbia, found that the RMR and MS (Hoek-Brown correlation to RMR) were poor
predictors of the strength of rock masses for weak rock masses. They developed the
Island Copper Rock Mass Rating (ILC-RMR) which modified RMR for RMR values less
than 40. As this method did not allow for consistency in strength assessment (i.e. different
rock mass rating methods above and below RMR = 40), Robertson (1988) proposed the
SRK Geomechanics Classification of rock masses (SRMR), shown in Table 5.7.
Robertson (1988) defines weak rock masses as those with shear strength parameters less
than:
c = 0.2 MPa
= 30
This being equivalent to a jointed specimen having an unconfined compressive strength,
UCS, of less than 0.7MPa. Robertson gives three cases where rock mass strength can be
this low (or combinations of these).
1. Where the intact rock is very weak or soil-like.
2. Where there is an intense number of defects that allow the material to fail along
random stepped surfaces.
3. Where there is sufficient freedom of rotation in the mass to allow intact material to
rotate to allow for the formation of a failure surface. Note, rotation/freedom increases
with equidimensional, rounded intact particles with weak infill or voids.
The SRMR varies from RMR
74
in the following ways:
Groundwater is ignored as it is assumed that groundwater is a destabilising force and
does not influence the rock mass strength. The maximum groundwater factor (15) has
been added to the intact rock factor.
For material in the soil strength range additional classes and ratings have been
added (S1-S5).
The RQD has been replaced with a handled RQD (HRQD). This is in effect a
disturbed RQD where the material has been firmly twisted and bent but without
substantial force or use of any tools or instruments. High RQD values will therefore
not be assigned for weak or weakly cemented rock. Note that the RQD should only be
applied to hard rock masses and as such, if properly recorded, should be equivalent
to HRQD. The HRQD suffers from the same problems as RQD when using it for large
slopes.
Discontinuity spacing is substituted with handled discontinuity spacing in a similar
manner to HRQD.
The discontinuity condition parameters stay unchanged except that the rating is limited
to less than or equal to ten for mat1erial with intact rock strength less than or equal to
R1. This is to stop weak rock being given a high discontinuity condition rating.
Table 5.7. SRK Geomechanics Classification or Slope Rock Mass Rating (SRMR)
PARAMETER RANGES OF VALUES
I
s50

(MPa)
> 10 4 - 10 2 - 4 1 2 For this low range
UCS test is preferred
<1
Strength of
intact rock
material
UCS
(MPa)
R5
>250
R4
100-250
R3
50-100
R2
25-50
R1
5-25
R1
1-5 S5 S4 S3 S2 S1
Rating 30 27 22 19 17 15 10 6 2 1 0
Handled RQD (%) 90-100 75-90 50-75 25-50 <25
Rating 20 17 13 8 3
Handled (mm)
discontinuity spacing
>2000 600-2000 200-600 60-200 <60
Rating 20 15 10 8 5
Condition of
discontinuities
Rock > R1
Very rough surfaces
Not continuous
No separation
Unweathered rock
wall
Rock > R1
Slightly rough
surfaces
Separation < 1mm
Slightly weathered
walls
Rock > R1
Slightly rough
surfaces
Separation < 1mm
Highly weathered
walls
Rock R1
Slickensided surfaces
OR
Gouge < 5mm
OR
Separation 1 5mm
Continuous
Rock < R1
Soft gouge > 5mm thick
OR
Separation > 5mm
Continuous
Rating 30 25 20 10 0

The SRMR system was found to give similar rating values as the ILC-RMR for the Island
Copper Mine. Therefore, Robertson (1988) concludes that Robertson et al (1987)
correlations with rock strength can be used with the SRMR. The SRMR or SRK-RMR
system was also checked using Getchell Mine, Nevada. Table 5.8 shows the correlations
given by Roberston (1988). Figure 5.2 shows these correlations as Mohr-Coulomb
strength curves. It can be seen that for SRMR = 20-25 the results seem invalid for normal
stresses less than about 700kPa (for values less than this it implies that higher SRMR
values give lower strengths). The author does not know the normal stresses acting on the
Island Copper Mine Slopes. Table 5.8 and Figure 5.2 show that the correlations vary a
considerable amount with sites and thus the rating system may need refining. Robertson
cautions that more case histories are required before the data in Table 5.8 can be used
with confidence.

Table 5.8. SRMR strength correlation (Robertson, 1988)
Strength Parameters
Island Copper Mine Getchell Mine
Rock
Mass
Class
SRMR
c
(kPa)
c
(kPa)

IVa 35-40 86 40 - -
30-35 72 36 - -
25-30 69 34 48 30
IVb
20-25 138 30 48 26
Va 15-20 62 27.5 48 24
Vb 5-15 52 24 14 21

0
200
400
600
800
1000
0 200 400 600 800 1000
n
(kPa)

(
k
P
a
)
35-40
30-35
25-30
20-25
15-20
5-15

0
200
400
600
800
1000
0 200 400 600 800 1000
n
(kPa)

(
k
P
a
)
35-40
30-35
25-30
20-25
15-20
5-15

Figure 5.2. SRMR strength correlation (a) Island Copper Mine (b) Getchell Mine
(Robertson, 1988)
5.2.1.5 Modified Rock Mass Classification, M-RMR
nal (1996) developed the M-RMR from the RMR method with additional features for
better characterisation of weak, stratified, anisotropic and clay bearing rock masses. The
method was based on investigations carried out at a borax mine and two coal mines. The
geology at the borax mine comprised laminated and bedded limestone with continuous
beds (varying from 2m to 9m in thickness) of consolidated clay. Stability was affected by
the presence of water. The coal mines consisted of lignite with associated coal measure
rocks (marl, claystone, mudstone) and clayey limestone.
The rating is given below. I
UCS
, I
RQD
, I
JC
, I
JS
, I
GW
and I
JO
are the ratings for
c
, RQD,
joint condition, joint spacing, groundwater and joint orientation respectively. Table 5.9
and Table 5.10 show the ratings for I
JC
. The tables appear very extensive however, as
they are based on a very limited database the author believes these should not be used for
general application. F
c
is the weathering coefficient and A
b
and A
w
are the adjustment
factors for blasting and major planes of weakness respectively. The factors are discussed
further in the rating adjustment section.
( ) ( )
JO GW JS JC RQD UCS c w b
I I I I I I F A A RMR M + + + + + = - (5.2)

515 . 0
856 . 0
c UCS
I = (5.3)
For RQD>10:
RQD I
RQD
173 . 0 7 . 2 + = (5.4)
For RQD10:
RQD I
RQD
443 . 0 = (5.5)

187 . 0
93 . 3
S JS
J I = (5.6)

W
GW
e I
03 . 0
15

= (5.7)
Where, w = Inflow of groundwater per 10m of tunnel length (lt/min)
(Note, damp = 0-10; wet = 10-20; dripping = 20-35; flowing >35)
For ICR5:
12 =
JO
I (5.8)
Where, ICR = Intact core recovery (%)
For 5<ICR<25:
75 . 13 35 . 0 = ICR I
JO
(5.9)
For ICR25:
5 =
JO
I (5.10)
( )
2
00515 . 0
2
6 . 0 0015 . 0

+ =

d
I
d C
e I F (5.11)
Where, I
d-2
= Slake durability index
Table 5.9. Joint condition index I
J C
(nal, 1996)
Intact core recovery Condition J
JC

No filling 10 ICR < 5
Filling 0
No filling, RQD = 0 13
No filling, 0 < RQD < 10 17
No filling, RQD 10 22
Filling 5mm (soft) 0
Filling 5mm (hard) 4
Filling 1 - 5mm (soft) 8
Filling 1 - 5mm (hard) 11
5 ICR 25
Filling < 1mm 14
No filling
( ) D A C R W + +
Filling 5mm (soft) 0
Filling 5mm (hard)
( ) D C + 2
Filling 1 - 5mm (soft)
( ) D C + 4
Filling 1 - 5mm (hard)
( ) D C + 6
ICR > 25
Filling < 1mm
( ) D C + 8

It is interesting to note that I
JO
is calculated from the intact core recovery from boreholes.
nal (1996) indicates that Bieniawskis (1989) adjustments should be used where field
surveys are available.
Table 5.10. Ratings for joint condition parameters (nal, 1996)
Parameter Condition Rating Parameter Condition Rating
Unweathered 8 Very low 3.5
Slightly weathered 7 Low 3
Moderately weathered 6 Medium 2
Highly weathered 4 High 1.5
Very highly weathered 2

Continuity
C
Very high 1

Weathering
W
Decomposed 0 0.0 - 0.01mm 4
Undulating, very rough 8 0.01 - 1.0mm 3
Undulating, rough 6 1.0 - 5.0mm 2
Undulating, slightly rough 4

Aperture
A
>5mm 0
Undulating, smooth 2 None 1
Undulating, slickensided 1 0 - 1mm 4
Planar, very rough 4 1-5mm, hard 3.5
Planar, rough 3 1-5mm, soft 3
Planar, slightly rough 2 >5mm, hard 2
Planar, smooth 1

Filling
F
>5mm, soft 0

Roughness
R
Planar, slickensided 0
5.2.1.6 Basic Quality, BQ
The BQ system was introduced as a forced standard in China in 1995 (Lin, 1998). It
was developed using an extensive database of projects around China. A number of
numerical analysis techniques were used to assess the data including: reliability analysis;
stepwise regression (dynamic cluster analysis and expert system); and stepwise
discriminative analysis (dynamic cluster analysis and expert system). The final equation
chosen to represent a rock mass is shown below.
2 2
9786 . 276 0492 . 1 0064 . 0 1130 . 546 9212 . 1 0451 . 41
v v c c v c
K K K BQ + + + = (5.12)
where,
velocity seismic intact
velocity seismic insitu
=
v
K
c
= unconfined compressive strength of the intact rock
Although this equation is called precise it should be noted that other parameters were
found to have some importance including rock unit weight and average defect spacing.
There is also no statistical data presented to support the formula.
Table 5.11 shows the rock classes. There are no correlations to rock mass properties
provided in the paper.
Table 5.11. The basic quality, BQ, rock mass classes (Lin, 1998)

5.2.2 Adjustment Factors to basic rock mass ratings
Most of the empirical rating methods apply adjustment factors to their basic rock mass
rating. These adjustment factors account for such things as defect orientation, excavation
method, weathering, induced stresses and major planes of weakness.
Bieniawski (1976 and 1989) applies the adjustments by subtracting them from the rock
mass rating.
Table 5.1 shows that the defect orientation adjustment can dominate the RMR. If the
defect orientations are deemed very unfavourable an adjustment of -60 is required to
the basic rock mass rating. Even for defect orientations denoted as fair this adjustment
is -25. There is no guideline as to what very unfavourable means. Bieniawski (1989)
recommends the use of the Romana (1985) SMR corrections for slopes.
Romana used the same basic rock mass rating as RMR
89
but developed new adjustment
factors for joint orientation and blasting to account for the lack of guidelines in the RMR
methods. The equation for SMR is shown below. The joint orientation weighting includes
a factor for the difference between joint dip and slope angle, F
3
. This requires an iterative
approach for design. Table 5.12 and Table 5.13 show the adjustment ratings.

4 3 2 1 89
F F F F RMR SMR + = (5.13)
Romana (1985) developed his factors not only for rock mass failures but also for wedge
and planar failure. A rock mass rating method should not be used for these two cases as
they are defect controlled and can be assessed using such measures as stereographic
projection. Even if the method was applicable, the ratings for planar failure are
questionable. F
2
depends on defect dip and must account for the defect shear strength
however, the method seems to assume that friction angles are quite high. For example,
bedding surface shears may attain strengths of below 12 yet these would be given a
very favourable rating of 0.15.

Table 5.12. Adjustment rating for joints (after Romana, 1985)
Case Very
Favourable
Favourable Fair Unfavourable Very
unfavourable
P
s j

T
o
180
s j

>30

30-20

20-10

10-5

<5
P/T
( )
2
1
sin 1
s j
F =
0.15 0.4 0.7 0.85 1.00
P
j

<20 20-30 30-35 35-45 >45
P
j
F
2
2
tan =

0.15 0.4 0.7 0.85 1.00
T F
2
1.00 1.00 1.00 1.00 1.00
P
s j
>10 10-0 0 0-(-10) <-10
T
s j
<110 110-120 >120 - -
P/T F
3
0 -6 -25 -50 -60
P - Planar failure
s
- Slope dip direction
j
- Defect dip direction
T - Toppling failure
s
- Slope dip
j
- Defect dip

Table 5.13. Adjustment Rating for methods of excavation of slopes (after Romana,
1985)
Method Natural
Slope
Presplitting Smooth
Blasting
Blasting or
Mechanical
Defficient
Blasting
F
4
+15 +10 +8 0 -8

Figure 5.3 shows the problem with attempting to predict structurally controlled failures
with rock mass ratings. The example shows a defect dipping out of the slope at 60. The
dip direction is within 15 of the dip direction of the slope. The intact rock has a high
strength and there are no other defects. The defect shown is unweathered, fairly tight and
slightly rough. By inspection, this is an unstable slope however, the SMR rates it as II
Good (Table 5.14).
Table 5.14. Tentative description of SMR classes (after Romana, 1985)
SMR 0-20 21-40 41-60 61-80 81-100
Class V IV III II I
Description Very Bad Bad Normal Good Very Good
Stability Completely
Unstable
Unstable Partially
Stable
Stable Completely
Stable
Failures Big planar or
soil like
Planar or big
wedges
Some joints
or many
wedges
Some blocks None
Support Reexcavation Important/
Corrective
Systematic Occasional None

Figure 5.3. Example of Planar Failure Case with High SMR

The CSMR method (Chen, 1995) is based on the SMR method. The CSMR applies a
discontinuity condition factor, , that describes the conditions of the controlling
discontinuity on which the ratings F
1
, F
2
and F
3
are based (Table 5.15). This factor ranges
from 0.7 to 1.0. The CSMR method also assumes that the SMR method is applicable for a
slope height of 80m but must be adjusted for other slope heights, H, using the slope height
factor, . The relationship for , based on an extensive survey and rigorous analysis of
slopes in China, is shown in Figure 5.4. With the addition of the two new factors, the
equation for CSMR is defined as:

4 3 2 1 76
F F F F RMR CSMR + = (5.14)
60
UCS rating = 15 (300MPa)
RQD rating = 20 (100%)
Spacing rating = 20 (> 2m)
Condition rating = 25
Groundwater rating = 15 (dry)
F
1
= 0.7 (15)
F
2
= 1.0
F
3
= -60
F
4
= +10 (presplitting)

SMR = 63
H 4 . 34 57 . 0 + = (5.15)
where, H = Slope height in metres
Table 5.15. Discontinuity condition factor (Chen, 1995)

Defect Condition
1.0 Faults, long weak seams filled with clay
0.8 to 0.9 Bedding planes, large scale joints with gouges
0.7 Joints, tightly interlocked bedding planes

0
2
4
6
8
10
1 10 100 1000
H (m)

Figure 5.4. Slope height, H, vs slope height factor, (after Chen, 1995)
The CSMR has been based on the SMR and thus has similar problems. CSMR
acknowledges the affect of slope height. It is the authors view that height should not be
grouped with the rock mass rating (a defacto strength estimate) but should be addressed
during the stability analysis where it will contribute to the stresses acting.
Laubscher (1977) adjusts his MRMR for weathering; field and induced stresses; change
in stress due to mining; orientation and type of excavation with respect to geological
structures; and blasting effects. The multipliers were developed primarily for
underground excavations but are also used for slopes.
nal (1996) uses corrections for weathering, blasting and major planes of weakness.
Table 5.16 and Table 5.17 show the adjustment factors A
b
and A
w
for blasting and major
planes of weakness respectively.
( ) ( )
JO GW JS JC RQD UCS c w b
I I I I I I F A A RMR M + + + + + = - (5.16)
Table 5.16. Blasting adjustment, A
b
(nal, 1996)
No blasting 1.0
Smooth blasting 0.95
Fair blasting 0.90
Poor blasting 0.85
Very poor blasting 0.80
Table 5.17. Major plane of weakness adjustment, A
w
(nal, 1996)
No major weakness zones 1.0
Stiff dykes 0.90
Soft ore zones 0.85
Host rock/ore contact zones 0.80
Folds; synclines; anticlines 0.75
Discrete fault zones 0.70

It is not understood why the RQD rating is adjusted for weathering whilst the joint
spacing rating is not. It is felt that ideally A
w
should not be used in a rock mass
classification system for slopes. If major planes of weakness exist they should be
considered individually during the analysis phase.
MRMR, RMS and M-RMR contain an adjustment for weathering. The author believes that
weathering should not be used as an adjustment factor in the estimation of rock mass
strength. The effect of weathering is to alter the intact strength and defect condition
parameters over geological time. Present day weathering condition should already have
been accounted for in the strength of the rock substance. Where further weathering may be
expected within the design life of a slope, the parameters for intact strength and defect
condition could be adjusted. However, the extent of these effects needs to take into
account the scale of influences versus the scale of the slope. That is, surficial weathering
may not extend into the slope to such a degree as to affect the strength of the rock mass
along the shear failure surface.
The excavation method adjustment (MRMR, SMR, CSMR and M-RMR) was originally
designed for support of underground excavations where it has obvious implications. The
method of excavation may affect low height slopes, however large slopes are unlikely to
be affected by blasting (with regard to rock mass failure). The author believes that the
rock mass involved in the failure is usually remote to the region affected by blasting.
Blasting may affect the stability of benches. However, failures of slopes of this scale can
be expected to be caused by failure along structure and not through rock mass due to the
low stresses acting. Hoek et al (2002) suggest that destressing can have a significant
affect on the strength of rock masses for slopes. Where blasting or destressing is believed
to have affected the rock mass to a large degree, then these affects should be accounted
for in the assigning of weightings for block size (smaller) and joint condition (persistence
and aperture) and not as an adjustment factor.
The author believes that the best currently available rock mass rating for slope design is
the GSI (Hoek et al, 1995). The GSI is relatively simple to use and accounts for the major
factors that affect rock mass strength (block size and strength and defect condition).
Reducing the GSI can incorporate affects due to destressing and blasting. The effect of
groundwater should be included in the analysis as a stress and the effect of major
structures should be analysed separately.
5.3 A REVIEW OF SLOPE DESIGN METHODS WHICH ARE
BASED ON ROCK MASS RATINGS
5.3.1 Correlations with Shear Strength Parameters and Slope Angles
Bieniawski (1976) and Robertson (1988) provided estimates of cohesion and friction
angle values for different RMR and SRMR ranges respectively that could be used for
slope stability analysis. Robertsons (1988) shear strength correlations (Table 5.8) were
based on the back analysis of failed slopes in weak rock masses at two mine sites.
Bieniawskis (1976) shear strength correlations (Table 5.18) were based on experience
working with underground excavations, slopes and foundations. Cited published case
studies included: a cable jacking test in highly weathered to friable gneiss and schist
bedrock (Pells, 1975); a 50m high toppling slope failure induced by base shearing
through weak decomposed amphibolite; and slopes (stable and one failure) consisting of
dolerite, shale and melaphyre (Bieniawski, 1975). The cases reported comprised failures
in very poor quality rock masses.
Table 5.18. Rock mass properties for RMR
76
(Bieniawski, 1976)
RMR
76
<20 21-40 41-60 61-80 81-100
Rock mass cohesion (kPa) <100 100-150 150-200 200-300 >300
Rock mass friction angle () <30 30-35 35-40 40-45 >45

Laubscher (1977) presents a table of stable slope angles versus MRMR independent of
slope height (Table 5.19). These slope angles were based on Laubschers 20 years of
experience working predominately in metamorphic and volcanic rocks of varying quality,
taking in the whole range of possible MRMR values. Laubscher (1977) mentions an
example of a pit slope with MRMR of 12 that failed at a slope angle of 45. A stable
slope angle was found at 35, which corresponds with the value in Table 5.19. No
information with regard to the slopes height, geology or failure mode was provided.

Table 5.19. Stable slope angle versus MRMR (Laubscher, 1977)
MRMR 81-100 61-80 41-60 21-40 0-20
Slope Angle 75 65 55 45 35

Abrahams and Parsons (1987) performed a statistical analysis of Selbys RMS data and
developed the following relationship:
( ) 072 . 141 681 . 2 degrees = RMS Angle Slope Stable (5.17)
The data has a narrow spread of RMS (approximately 55 to 90) and yet, the slope angles
predicted by the equation range from approximately 5 to 90 for their RMS data. The
line of best fit to Selbys data indicates a slope angle of -74 for a very weak rock
mass (RMS=25). This is not meaningful and is likely due to the lack of weak and very
weak rock masses (RMS<50) in the data. Selby (1980) divided the slopes he assessed
into sections of similar geology, generally with slope heights less than 40m. This resulted
in short lengths of slopes under different stresses being compared. Slope angles for low
slope heights are likely to be governed by the dip of defects and less so the strength of
rock mass where the intact rock strength is high enough to prevent intact rock failure.
These aspects may account for some of the scatter in the data and subsequent poor
statistical results.
Orr (1992) proposed the following relationship between RMR (converted from RMS)
and slope angle as the limit of long term stability, using the data on Figure 5.5. Orr (1992)
states that the equation is for slopes up to 50m high and with RMR values of between 20
and 77.
( ) 71 ln 35 angle Slope = RMR (5.18)

Figure 5.5. RMR versus slope angle (Orr, 1996)

Romana (1985) correlated SMR with stability class. Of the 28 slopes (ignoring the six
toppling failures) presented by Romana (1985) only six had failed and only one of those
was a rock mass failure (soil like failure), the others were planar or wedge failures
(Table 5.20). The highest known slopes tested against SMR by Romana were up to 62m
in calcareous slopes and all were stable or partially stable (Jord et al., 1999). Slope
heights less than 40m are invariably controlled by structure (Duran and Douglas, 1999)
and hence there is very little if any published rock mass failure data supporting the
Romana (1985) SMR correlation.
Table 5.20. Case records for SMR (after Romana, 1985)
Rock Excavation method SMR Failures
Limestone Presplitting 85 None
Sandy marl Natural slope 84 None
Limestone Presplitting 77 3 small blocks
Gneiss Presplitting 72-75 Small wedges during construction
Limestone Blasting 74 None
Dolestone Blasting 64-76 Small planes during construction
Limestone Smooth blasting 61-73 None
Marl Smooth blasting 71 None
Limestone Blasting 70 Small blocks
Sandstone/Siltstone Natural slope 68 Small blocks
Limestone Deficient blasting 59 Many blocks
Marl/Limestone Mechanical excavation 55 Local problems
Gypsum rock Natural slope 52 Some wedges (1m
3
)
Claystone/Sandstone Blasting 47 Big wedge (15m
3
)
Claystone Mechanical excavation 46 Surface erosion
Sandstone/Marl Blasting 43 Many wedges
Limestone Deficient blasting 40 Many failures
Gypsum Rock Natural slope 31-43 Big wedge (100m
3
)
Sandy marl Mechanical excavation 32 Blocks, mud flows
Sandstone/Marl Blasting 30 Big planar failure during construction
Limestone Blasting 29 Several wedges (50m
3
)
Marl Smooth blasting 36 Almost total planar failure after
weathering
Volcanic tuff/Diabase Blasting 30 Big planar failure
Marl Blasting 16 Total planar failure after weathering
Marl Blasting 42 Small wedges
Marl Blasting 17 Total planar failure after weathering
Marl Blasting 43 Small wedges
Slate/Greywacke Mechanical excavation 17 Soil like failure
The CSMR uses the same stability class correlation as SMR. Figure 5.6 shows
correlations between observed behaviour (ESMR) and SMR and CSMR for 44 slopes
with heights ranging from 8 to 42m. Chen (1995) uses a similar method proposed by
Collado and Gili (1988) to determine ESMR. The ESMR is derived from estimates of the
factor of safety from field engineers and the following empirical equation:

15 . 0
5 . 52
100
=
F
ESMR
where, F = Factor of safety
All points falling on the 45 line would represent a good correlation. Although the CSMR
has slightly less scatter than the SMR, both methods can be seen to have a poor
correlation.

Figure 5.6. Observed cases (ESMR) vs (a) SMR, (b) CSMR (Chen, 1995)
Moon et al (2001) found that rock mass classification techniques (RMR, SMR and RMS),
used for slope angle estimation, perform poorly for weak rock masses where failure
occurs through intact rock, rather than solely along defects.
Tsiambaos & Telli (1991) compare the RMR and SMR systems for limestone slopes.
They found that the RMR system leads to an underestimation of the stability conditions
(i.e. the actual rock conditions are better than predicted) of limestone cuts whilst the SMR
was a better predictor. It should be noted that the only stability problems the slopes had
were rock falls that were structurally controlled.
Of the methods presented, Robertsons (1988) approach has the most merit as it was
developed specifically for slopes and is based on slope failures in weak rock masses.
Unfortunately, only two slopes have been used and hence c and values are only
available for two specific rock masses. Other methods presented have limited value as
they are based on either long-term natural slopes that are often structurally controlled;
stable slopes; or slopes of only limited height.
5.3.2 Available Slope Performance Curves
Slope performance curves provide a valuable tool in the design process where rock mass
failure plays a strong control in the stability of slopes. The curves are derived from the
performance of stable and unstable slopes plotted on a slope angle versus slope height
plot. The curves are often site specific and take into account the impact of existing
failures, the remaining time frame for mining and the acceptable risks to the mining
operation.
Extending slope design curves from being a site specific tool to a general tool must be
treated with caution. Early attempts at doing this include Lane (1961) and Fleming et al
(1970 for slopes in shale, Coates et al (1963) for incompetent rock slopes, Shuk (1965)
for natural slopes, Lutton (1970) and Hoek (1970) for general rock excavations.
Hoek and Bray (1981) present a collection of data, from mines, quarries, dam foundation
excavations and highway cuts, on stable and unstable slopes in hard rock (Figure 5.7).
The plot also shows a curve representing the highest and steepest slopes that have
successfully been excavated in hard rock (note that many failures have occurred in flatter
slopes). This line can therefore be used as a guide as to the upper bound heights and slope
angles that can be considered in slope design.
McMahon (1976) attempted to group similar rock masses together and came up with
correlations (based on log-log graphs) relating slope length, L, with slope height, H
(Equations 5.19 and 5.20 and Table 5.21).

b
aL H = (5.19)
( ) angle slope tan H L = (5.20)

Figure 5.7. Upper bound slope height versus slope angle curve for rock masses
(Hoek & Bray, 1981)
Table 5.21. Parameters for McMahons (1976) slope relationship
Rock mass type a B
Massive granite with few joints 139 0.28
Horizontally layered sandstone 85 0.42
Strong but jointed granite and gneiss 45 0.47
Jointed partially altered crystalline rocks 16 0.58
Stable shales 8.5 0.62
Swelling shales 2.4 0.75

Figure 5.8 shows the data from McMahon (1976) replotted on a slope height versus slope
angle curve. The relationships from Table 5.21 are also plotted on the graph. It can be
seen from the figure that the relationships provide a poor fit to a lot of the data,
particularly the stronger rock masses. The curves also tend toward about 10 which is not
supported by the data. It should also be noted that only the data for shale was near limit
equilibrium and so the curves for the other rock mass types represent conservative (by an
unknown amount) boundaries.
Haines and Terbrugge (1991) took this technique further and tried to correlate slope
design curves with rock mass ratings. The Haines and Terbrugge (1991) slope design
methodology makes use of the MRMR empirical rock mass strength assessment as
presented by Laubscher (1977 and 1990). The slope design methodology is presented in
Figure 5.9 and again in Figure 5.10, replotted on the basis of slope angle versus slope
height and with contours of MRMR presented. Haines & Terbrugge (1991) divide the
graph into three design zones where: (1) classification alone may be adequate; (2)
marginal on classification alone; and (3) slopes require additional analysis.
0
500
1000
1500
2000
2500
3000
0 10 20 30 40 50 60 70 80 90
Slope angle
H
e
i
g
h
t

(
m
)
.
Massive granite with few joints
Horizontally bedded sandstones
Strong but jointed granite and gneiss
Jointed partially altered rocks
Horizontally bedded stable shales
Swelling shales
Clay shale
Jointed & altered
Sandstone and shale
strong granite and gneiss

Figure 5.8. Slope angle versus slope height with regression curves (modified after
McMahon, 1976)

Case studies were utilised by Haines & Terbrugge (1991) to evaluate the design curves.
It is noted that several of the cases presented related to slopes for feasibility studies that
had at the time not been excavated. The cases of excavated slopes have been presented in
Figure 5.10. The MRMR values, grouped into intervals, are presented by the use of
different symbols. Three aspects of the Haines & Terbrugge (1991) design curves are of
concern:
Firstly, the data does not appear to indicate the validity of the design curves presented.
Vertical design lines could have been equally appropriate, i.e. independent of slope
height. This is in keeping with the use of MRMR for open pit slopes, as originally
proposed by Laubscher (1977), shown in Table 5.19.
Secondly, the shape of the interpreted design curves does not appear valid. The curves
are broadly convex in shape and nearly linear for slope heights of up to 100m. This is at
odds with the wide body of experience, which suggests a concave shape is appropriate.
In addition, the experience presented in Figure 5.11 suggests a predominant curvature
would be expected for slope heights up to 100m. Finally, the curves are not asymptotic
and indicate a continuing reduction in slope angle with slope height is required.
Thirdly, none of the case studies presented by Haines & Terbrugge (1991) related to
unstable slopes. Moreover, a third of the cases related to road cuttings where typically a
high degree of conservatism is utilised. As such the design curves have a large element of
conservatism built into them.

Figure 5.9. Slope height vs slope angle for MRMR (Haines & Terbrugge, 1991)
20
40
60
80
100
0
50
100
150
200
250
20 30 40 50 60 70
Slope angle (deg)
S
l
o
p
e

H
e
i
g
h
t

(
m
)

.
0-10
10-20
20-30
30-40
40-50
50-60
60-70
Classification alone
may be adequate
Marginal on
classification alone
Slopes require
additional analysis
MRMR
MRMR
Values

Figure 5.10. Haines & Terbrugge (1991) slope design replotted on basis of slope
angle versus slope height showing Haines & Terbrugge (1991) slope data.

5.3.3 Pells Sullivan Meynink Slope Performance Curves
Through the utilisation of aggressive designs within interim pits and documentation of
stable and failed slopes, slope performance curves have been established at several
mining operations by Pells Sullivan Meynink Pty Ltd, consulting geotechnical engineers
and engineering geologists, to aid in the slope design process. They indicate that the
methodology has been of enormous benefit in slope design for mining operations where
multiple open pits are developed in similar geotechnical conditions, a poor rock mass is
present and rock mass failures have been the predominant control on slope stability.
Figure 5.11 presents six case studies where slope performance curves were developed on
the basis of the site specific stability.
Pells Sullivan Meynink (pers. com. Alex Duran) indicate that the methodology has been
of particular use in operations where multiple shallow pits have been developed and
where no geotechnical studies were available. Although these curves are not presented
the shapes of the curves are in keeping with those presented in Figure 5.11.
Three key regions can be defined on the slope performance curves in Figure 5.11. The
three regions and the division within the slope angle versus slope height plot are in
keeping with typical rock mass strengths for poor quality rock masses and the concept of
a curved strength envelope.
Region 1: Typically negligible rock mass failures occur for slope heights of less
than 40m. This is in keeping with findings of a survey of 54 mining operations
which indicated generally stable pits for slope heights less than 45m (Swindells,
1990). This is anticipated since for these heights, i.e. low stress levels, rock mass
strengths typically have a high friction angle. Typical slip-circle analyses indicate
steep slopes can be achieved for these slope heights. The experience of the author
and that of Pells Sullivan Meynink personnel indicates unstable slopes in Region 1
are invariably controlled by structure except where very poor rock mass conditions
are encountered.

Figure 5.11. Slope performance curves for case studies (Duran & Douglas, 1999)

0
50
100
150
200
250
20 30 40 50 60 70
Slope angle (deg)
S
l
o
p
e

H
e
i
g
h
t

(
m
)

.
Case 1
Case 2
Case 3
Case 4
Case 6
Case 9
1
2
3
4
6
9
After Figure 7,
Hoek & Bray
(1981)
Numbers refer to
cases in Table 2
Solid symbols
represent unstable
slopes
Upper bound
from Hoek &
Bray (1981)
see Figure 5.7
Slope
performance
curves
Numbers refer to
cases in Table 5.22
Solid symbols
represent unstable
slopes as does the
symbol x
Region 2: For slopes above 40m in height rock mass failures become apparent in
poor quality rock masses. The design curves display a pronounced concave
curvature with a reduction in the overall slope angle for increasing slope height.
This curvature is in keeping with that indicated by Hoek & Bray (1981) and with
typical slip circle analyses. The curvature of the design curves conform to a curved
rock mass strength envelope. Back analyses indicate the use of a fixed friction angle
and cohesion provides a poor fit. Whilst back analyses using a decreasing friction
angle and increasing cohesion, with increasing slope height (i.e. increasing stress
level), provides a better fit to the curves. It should be noted the design curves are
roughly sub-parallel and with curves further to the right representing better rock
mass conditions.
Region 3: For slope heights above 90m the curves show a trend of becoming
asymptotic to a given slope angle. This is in keeping with the fact that at higher
stress levels the curved envelope approaches a straight line and a constant friction
angle is implied.

5.4 ANALYSIS OF CASE STUDY DATA
5.4.1 Case Studies Used
Table 5.22 summarises the case studies that were made available to the author by Pells
Sullivan Meynink Pty Ltd. Each case represents a particular open pit mine. Each mine
may have several stable/unstable slopes in the database. Where this situation exists the
different slopes are denoted a, b, c etc. Case 5 was obtained separately from Glastonbury
(2002).
A total of 13 cases with 37 slopes, of which 21 were stable and 16 had failed, was made
available. All failures were considered to have had a rock mass failure component.
Table 5.22. Summary of slope data from case studies
Geology Mine
Case
Height
(m)
Slope
angle
()
MRMR
interval
GSI
interval
Water Stable
Saprolite/Basalt 1a 70 49 20-30 40-50 none yes
Saprolite/Basalt 1b 41 50 20-30 40-50 none no
Saprolite/Basalt 1c 41 55 20-30 40-50 none no
Saprolite/Basalt 1d 46 55 20-30 40-50 none no
Saprolite/Basalt 1e 57 49 20-30 40-50 none no
Saprolite/Basalt 2a 58 50 30-40 50-60 none yes
Saprolite/Basalt 2b 60 48 30-40 50-60 none yes
Saprolite/Basalt 2c 60 52 30-40 50-60 none yes
Volcanoclastics 3a 20 39 10-20 30-40 mod no
Volcanoclastics 3b 40 32 10-20 30-40 mod yes
Volcanoclastics 3c 60 31 10-20 30-40 mod yes
Talc chlorite schist 4a 70 44 30-40 40-50 mod no
Talc chlorite schist 4b 120 35 30-40 40-50 mod no
Talc chlorite schist 4c 120 38 30-40 40-50 mod no
Talc chlorite schist 4d 150 31 30-40 40-50 mod yes
Talc chlorite schist 4e 150 35 30-40 40-50 mod yes
Argillite 5a 250 42 30-40 50-60 mod no

Table 5.22. Summary of slope data from case studies (cont.)
Geology Mine
Case
Height
(m)
Slope
angle
()
MRMR
interval
GSI
interval
Water Stable
Argillite 5b 107 37 30-40 50-60 mod yes
Argillite 5c 80 38 30-40 50-60 mod yes
Schist 6a 70 45 20-30 40-50 mod no
Schist 6b 95 45 20-30 40-50 mod no
Mudstone/siltstone 7 38 39 40-50 50-60 none yes
Breccia 8 200 65 60-70 70-80 none yes
Faulted breccia 9a 78 32 20-30 40-50 high yes
Faulted breccia 9b 50 34 20-30 40-50 high yes
Faulted breccia 9c 77 37 20-30 40-50 high no
Faulted breccia 9d 60 40 20-30 40-50 high no
Sheared siltstone 10a 97 36 20-30 40-50 mod yes
Siltstone 10b 157 48 50-60 60-70 none yes
Siltstone 10c 60 53 50-60 60-70 none yes
Siltstone 11 110 48 30-40 40-50 none no
Shale 12a 29 39 10-20 30-40 mod yes
Shale 12b 37 28 10-20 30-40 mod yes
Shale 12c 30 40 10-20 30-40 mod no
Shale 12d 45 26 10-20 30-40 mod no
Granodiorite breccia 13a 40 75 50-60 70-80 none yes
Granodiorite breccia 13b 90 80 50-60 70-80 none yes
5.4.2 Correlations of MRMR, SRMR and RMS with GSI
Data from 12 of the case studies have been used to provide correlations between three
rating methods (MRMR, SRMR and RMS) and GSI (Table 5.23 and Figure 5.12). The
data from Selby (1980) has also been used for the correlation between RMS and GSI and
is plotted on Figure 5.12.
Table 5.24 and Table 5.25 show the data used to determine the GSI for the case studies.
Table 5.26, Table 5.27 and Table 5.28 show the data used to determine the MRMR,
SRMR, and RMS for the case studies respectively. Best estimate data was used for all
case studies and interpolation used to choose ratings for each parameter.
Some assumptions were made in assessing SRMR for the authors case studies, based on
the intact strength and character of borehole core, to assess handled RQD and spacing.
Table 5.23. Correlation between rating methods authors case studies
MINE
CASE
Rock Unit GSI MRMR SRMR RMS
1 Saprolite/Basalt 41 22 42 58
2 Saprolite/Basalt 52 36 49 74
3 Volcanoclastics 37 15 35 59
4 Talc chlorite schists 45 30 47 69
6 Schist 44 22 45 51
7 Mudstone/Siltstone 57 43 49 76
8 Breccia 76 65 97 98
9 Faulted breccia 49 24 57 71
10a Sheared siltstone 48 24 51 54
10c Siltstone 68 54 63 80
11 Siltstone 46 35 55 63
12 Shale 39 18 41 52
13 Granodiorite breccia 73 55 83 85

GSI was chosen since it provides a measure of the basic rock mass quality. Correlation
with the other rating systems was not considered appropriate in view of the rating
adjustments required. The correlations exhibit a good fit, even though there is limited data
for the correlations with MRMR and SRMR.
GSI = 0.78MRMR + 25.22
R
2
= 0.94
GSI = 0.67SRMR + 15.10
R
2
= 0.84
GSI = 1.07RMS - 22.39
R
2
= 0.82
0
20
40
60
80
100
0 20 40 60 80 100
Rating
G
S
I
MRMR
SRMR
RMS

Figure 5.12. Correlations of GSI with MRMR, SRMR, RMS rating (mod. Duran and
Douglas, 2002).
Table 5.24. Summary of best estimate GSI data for mine cases
MINE
CASE
Rock Unit UCS RQD Spacing
Defect
Condition
GSI
value 3MPa 46% 0.6m Table 5.25 1
a-e
Saprolite/
Basalt rating 1.3 9.3 11.2 9.8
41
value 5MPa 56% 2m Table 5.25 2
a-c
Saprolite/
Basalt rating 1.5 11 20 9.7
52
a-c
Volcanoclastics
rating 2.2 9.1 6.1 9.6
37
a-e
Talc chlorite
schists rating 3.8 6.1 14.6 11.1
45
a-b
Schist
rating 2.1 12.6 6.1 12.6
44
value 5MPa 75% 1m Table 5.25
7
Mudstone/
Siltstone rating 1.5 14.6 14.6 16.7
57
value 150MPa 98% 5m Table 5.25
8 Breccia
rating 11.4 19.5 11.8 23.3
76
a-d
Faulted
breccia rating 6.4 9.9 10.3 12.8
49
a
Sheared
siltstone rating 3.1 17.7 20 16.7
68
b-c
Siltstone
rating 3.1 13.6 10.3 10.8
48
value
11 Siltstone
rating
RMR from Q = 1.3 and
44 ln 9 + = Q RMR
46
a-d
Shale
rating 2.7 8.3 6.1 12.1
39

Table 5.25. Summary of defect condition for GSI
MINE
CASE
Rock Unit
Length
(m)
Separation
(mm)
Roughness Infilling Weathering
1 Saprolite/Basalt 5 0-1 smooth soft High
2 Saprolite/Basalt 5 0-1 smooth soft High
3 Volcanoclastics 20 0-1 slickensided soft moderate
4 Talc chlorite schists 10 1-5 smooth soft fresh
6 Schist 20 1-5 smooth hard fresh
7 Mudstone/Siltstone 5 0-1 smooth hard fresh
8 Breccia 2 <1 rough hard fresh
9 Faulted breccia 15 1-5 slightly rough soft fresh
10a Sheared siltstone 15 1-5 smooth soft fresh
10c Siltstone 5 0-1 smooth hard fresh
11 Siltstone
12 Shale 10 1-5 slightly rough hard moderate

Page 5.45
Table 5.26. Summary of best estimate of Laubschers MRMR data for mine cases
Adjustments Mine
Case
UCS RQD
Defect set
spacing
Defect condition
*
RMR
Weathering Orientation Blasting
MRMR
3MPa 46% 0.6m/0.8m/5m wavy/smooth undulose/soft medium slight 1 year poor good conventional
1
1 8 15 13 37 0.9 0.75 0.9
22
5MPa 56% 2m/5m/10m curved/smooth undulose/soft medium slight 4 years good good conventional
2
2 10 22 11 45 0.96 0.9 0.92
36
13MPa 45% 0.1m/2m/5m straight/smooth planar/soft medium slight 1 year topple poor
3
3 8 12 8 31 0.9 0.65 0.85
15
30MPa 25% 1m/2m/5m slight undulating/slickensided undulose/soft fine nil poor fair
4
4 12 18 9 43 1 0.8 0.88
30
12MPa 65% 0.1m/2m/2m slight undulating/smooth-rough planar/non soft med nil very poor fair conventional
6
3 10 11 11 35 1 0.7 0.9
22
5MPa 75% 1m/5m/5m straight/rough planar/no filling nil good good conventional
7
1.5 12 21 17 51 1 0.9 0.94
43
150MPa 98% 5m/5m/5m slight undulating/rough undulose/non soft medium nil good good conventional
8
16 15 25 20 76 1 0.9 0.94
65
60MPa 50% 0.5m/2m/5m straight/smooth to rough/gouge nil fair poor
9
6 8 16 3 33 1 0.85 0.85
24

Page 5.46
Table 5.26. Summary of best estimate of Laubschers MRMR data for mine cases (cont.)
Adjustments Mine
Case
UCS RQD
Defect set
spacing
Defect condition
*
RMR
Weathering Orientation Blasting
MRMR
23MPa 70% 0.5m/0.8m/1m curved/slickensided undulose/non soft medium nil very poor poor
10a
3 10 11 16
40
1 0.7 0.85
24
23MPa 90% 2m/2m/5m curved/smooth stepped/non soft coarse nil very good pre-split
10b,c
3.5 14 20 22
59
1 0.95 0.97
54
nil fair fair conventional
11 RMR from Q = 1.3 and 44 ln 9 + = Q RMR 46
1 0.85 0.9
35
18MPa 40% 0.1m/2m/5m slight undulating/smooth-rough, planar/hard med. nil topple poor
12
3 6 12 11
32
1 0.65 0.85
18
*
large scale/small scale/infilling

Page 5.47
Table 5.27. Summary of best estimate of SRMR data for mine cases
Mine
Case
Rock unit UCS
Handled
RQD
Spacing
Defect
Condition
SRMR
value 3MPa 34.50% 0.3 to 0.4m R1
1 Saprolite/Basalt
rating 15 7 10 10
42
value 5MPa 28% 1 to 2m R1
2 Saprolite/Basalt
rating 15 8.5 15 10
49
value 13MPa 22.50% 0.08m schist R1
3 Volcanoclastics
rating 15 3 7 10
35
value 30MPa 25% 1 to 2m R2, Slickensided surfaces
4 Talc chlorite schists
rating 19 3 15 10
47
value 12MPa 48.75% 0.08m schist R1
6 Schist
rating 17 10.5 7 10
45
value 5MPa 37.50% 0.5 to 2m R1
7 Mudstone/Siltstone
rating 16 8 15 10
49
value 150MPa 98% 5 R4, rough, not continuous
8 Breccia
rating 27 20 20 30
97
value 60MPa 40% 0.5 to 2m R3, slightly rough - gouge
9 Faulted breccia
rating 22 8 15 12
57

Page 5.48
Table 5.27. Summary of best estimate of SRMR data for mine cases (cont.)
Mine
Case
Rock unit UCS
Handled
RQD
Spacing
Defect
Condition
SRMR
value 23MPa 52.50% 0.4 to 0.6m R1
10a Sheared siltstone
rating 19 10.5 11 10
51
value 23MPa 90% 2 R1
10b,c Siltstone
rating 19 18.5 15 10
63
value
11 Siltstone
rating
SRMR estimated using MRMR and multiplying by average
SRMR/MRMR ratio for all cases (=1.2)
55
value 18MPa 30% 0.08 to 0.5m R1
12 Shale
rating 17 6 8 10
41

Page 5.49
Table 5.28. Summary of best estimate of RMS data for mine cases
Defect Mine
Case
Rock unit UCS
Weathering Spacing Orientation Aperture Length
Water RMS
value 3MPa high 0.6m unfavourable 0-1mm 5m none
1 Saprolite/Basalt
rating 5 5 21 9 6 6 6
58
value 5MPa high 2m favourable 0-1mm 5m none
2 Saprolite/Basalt
rating 5 5 28 18 6 6 6
74
value 13MPa moderate 0.1m favourable 0-1mm 20m slight
3 Volcanoclastics
rating 5 7 15 18 6 4 4
59
value 30MPa fresh 1m unfavourable 1-5mm 10m slight
4 Talc chlorite schists
rating 10 10 25 9 5 6 4
69
value 12MPa fresh 0.1m unfavourable 1-5mm 20m mod
6 Schist
rating 5 10 15 9 5 4 3
51
value 5MPa fresh 1m very favourable 0-1mm 5m trace
7 Mudstone/Siltstone
rating 5 10 25 20 6 5 5
76
value 150MPa fresh 5m very favourable <1mm 2m none
8 Breccia
rating 18 10 30 20 7 7 6
98
value 60MPa fresh 0.5m fair 1-5mm 15m moderate
9 Faulted breccia
rating 14 10 21 14 5 4 3
71

Page 5.50
Table 5.28. Summary of best estimate of RMS data for mine cases (cont.)
Defect Mine
Case
Rock unit UCS
Weathering Spacing Orientation Aperture Length
Water RMS
value 23MPa fresh 0.5m very unfavourable 1 to 5mm 15m slight
10a Sheared siltstone
rating 5 10 21 5 5 4 4
54
value 23MPa fresh 2m very favourable 0 to 1mm 5m trace
10b,c Siltstone
rating 5 10 28 20 6 6 5
80
value 25MPa fresh 0.05-0.3m fair <1mm few none
11 Siltstone
rating 5 10 15 14 7 6 6
63
value 18MPa moderate 0.1m favourable 1 to 5mm 10m mod
12 Shale
rating 5 7 10 18 5 4 3
52
5.4.3 General Assessment of the Parameters in GSI
The GSI calculated from RMR (GSI
RMR
) contains ratings for UCS, RQD, defect spacing
and defect condition. This section uses the case study data to assess how well these
individual ratings differentiate rock mass strength. Figure 5.13 shows GSI versus slope
height for both failed and stable slopes from the case studies. A general observation
shows that the failed slopes are gathered at the lower end of the ratings. Figure 5.14 to
Figure 5.17 show the same graph as Figure 5.13 with defect spacing rating, defect
condition rating, RQD rating and UCS rating substituted for GSI respectively. The RQD
and spacing ratings (which essentially are both substitutions for block size) both appear
to differentiate between the rock masses well and have the failed slopes toward the lower
rating values. The defect condition rating appears to group most rock masses in the case
studies together. The UCS rating does not appear to differentiate between the failed and
stable slopes. This is not suprising as slope failure will generally occur along defects
unless the intact rock strength is very low due to the low stress environment.
0
50
100
150
200
250
0 10 20 30 40 50 60 70 80 90 100
GSI Rating
H

(
m
)
.
Failed
Stable

Figure 5.13. GSI versus slope height for failed and stable slopes
0
50
100
150
200
250
0 5 10 15 20 25 30
Spacing Rating
H

(
m
)
.
Failed
Stable

Figure 5.14. GSI defect spacing rating versus slope height for failed and stable
slopes
0
50
100
150
200
250
0 5 10 15 20 25
Defect Rating
H

(
m
)
.
Failed
Stable

Figure 5.15. GSI defect condition rating versus slope height for failed and stable
slopes
0
50
100
150
200
250
0 2 4 6 8 10 12 14 16 18 20
RQD Rating
H

(
m
)
.
Failed
Stable

Figure 5.16. GSI RQD rating versus slope height for failed and stable slopes
0
50
100
150
200
250
0 2 4 6 8 10 12 14
UCS Rating
H

(
m
)
Failed
Stable

Figure 5.17. GSI UCS rating versus slope height for failed and stable slopes

5.4.4 Development of Generalised Slope Design Curves
5.4.4.1 Use of MRMR in Haines and Terbrugge Method
Table 5.22 shows the MRMRs that have been evaluated for the case studies. Each case
has been allotted into a MRMR interval owing to the uncertainties in assessing a rigorous
value.
The data presented by Haines & Terbrugge (1991) and the new case studies (Figure 5.10
and Table 5.22 respectively) have been combined into one database.
The author (Duran & Douglas, 1999) presented cases 1-10 and the Haines & Terbrugge
(1991) design methodology in Figure 5.18. Several of the cases do not confirm the Haines
& Terbrugge design curves. For cases 1, 2, 6 and 8 the Haines & Terbrugge curves are
conservative for these mining operations. In cases 4 and 9 however, the design curves
would have indicated the use of steeper slopes. For these two latter cases there are
additional factors which affected the overall stability of slopes. For case 4 structure made
a significant contribution to the failure. Whilst for case 9, groundwater played a critical
role.
5.4.4.2 Revised Method Using MRMR
Figure 5.19 presents design curves that have been defined by the author (Duran &
Douglas, 1999) on the basis of the available data. An upper bound design curve has been
suggested which relates to MRMR values greater than 40. This is based on the experience
of Duran & Douglas (1999) as no rock mass failures have been observed in slopes that
are comprised of good, or better, rock mass quality. This observation was previously
related by Robertson (1988) who indicated that stability was almost exclusively
controlled by structure where MRMR was greater than 40.
It must be stressed that these design curves are based on slopes in mi ning operations,
where some instability is acceptable and slopes need only to stand up over short time
frames of up to two years. Appropriate reductions in slope angles could be utilised where
a conservative design is required.
It should be noted that cases 4 and 9 were treated as exceptions in defining the design
curves. This clearly reinforces that in the design of rock slopes there needs to be a careful
assessment of the influence of structure and groundwater in defining an acceptable design.
5.4.4.3 Method Based on the Use of the Geological Strength Index, GSI
The case studies discussed above together with additional data from Haines and
Terbrugge (1991) and Selby (1980) have been used to create slope design curves based
on GSI. Slope height versus slope angle was plotted for two ground water conditions, dry
and moderate pressures. Moderate water pressures are defined as where the piezometric
surface reaches the surface at a distance of 4 x the slope height. This is taken from Hoek
and Brays (1981) circular failure slope chart No. 3. Figure 5.20 and Figure 5.22 show
the data and the authors proposed design curves for dry and moderate water pressures
respectively. Where failure occurred predominantly through the rock mass the data is
presented as a solid symbol. Figure 5.20 does not contain a curve for GSI = 30 due to a
lack of data. It could be assumed that this curve would lie 10-15 to the right of the curve
for GSI = 40 based on the curves in Figure 5.22.
The slope design curves for various ranges of MRMR presented in the previous section
(Figure 5.19) have been used as the basis for deriving the GSI slope design curves. These
curves have been assessed for GSI values of 40 and 50, utilising the authors correlation.
The authors design curves provide a very good fit of the data.
Slope designs using strength estimates estimated by Bieniawski (1976), assuming no
rating adjustment for orientation, are presented on Figure 5.21 and Figure 5.23 based on
stability charts from Hoek and Bray (1981) and assuming a Factor of Safety of one. As
readily evident, Bieniawskis strength estimates are too high. Robertson (1988) provided
estimates of shear strength for back-analyses of failures. Using the correlation of SRMR
with GSI presented earlier, Robertsons rock mass strengths were assessed for GSI
values of 30 and 40, Figure 5.21 and Figure 5.23. Robertson (1988) suggested rock mass
failure in slopes was unlikely for an SRMR of greater than 35 (GSI40) and this is
confirmed by the data presented in Figure 5.21 and Figure 5.23. The strengths estimated
by Robertson (1988), if correlated to GSI, appear to overestimate slopes angles for dry
slopes. For moderate water pressures the curves are similar to the authors curves for
heights greater than 150m for lower heights, the authors curves predict flatter stable
slope angles.
20
40
60
80
100
0
50
100
150
200
250
20 30 40 50 60 70
Slope angle (deg)
S
l
o
p
e

H
e
i
g
h
t

(
m
)

.
0-10
10-20
20-30
30-40
40-50
50-60
60-70
MRMR
Solid symbols
represent unstable
slopes
Solid symbols
represent unstable
slopes as do the
symbols
+ x
Haines &
Terbrugge
MRMR
curves

Figure 5.18. Haines & Terbrugge (1991) slope design curves & slope data (Figure
5.10) with additional case studies (Duran & Douglas, 1999)
0
50
100
150
200
250
20 30 40 50 60 70
Slope angle (deg)
S
l
o
p
e

H
e
i
g
h
t

(
m
)

.
0-10
10-20
20-30
30-40
40-50
50-60
60-70
MRMR
> 40
30
20
Solid symbols
represent unstable
slopes
MRMR =
Solid symbols
represent unstable
slopes as do the
symbols
+ x
S S
H
H
=Significant contribution
to failure from structure.
=High water pressures in
slope
S
H

Figure 5.19. Suggested slope design curves for MRMR (Duran & Douglas, 1999)

0
50
100
150
200
250
10 20 30 40 50 60 70 80 90
Slope Angle (deg)
S
l
o
p
e

H
e
i
g
h
t

(
m
)
.
20-30
30-40
40-50
50-60
60-70
70-80
80-90
GSI
GSI
40
GSI
50
Solid symbols represent
unstable slopes

design curves for a dry slope

0
50
100
150
200
250
10 20 30 40 50 60 70 80 90
Slope Angle (deg)
S
l
o
p
e

H
e
i
g
h
t

(
m
)
.
20-30
30-40
40-50
50-60
60-70
70-80
80-90
GSI
GSI30
GSI40
RMR30
RMR<20
Robertson (1988)
RMR50
Bieniawski (1979)
Author's curves
GSI40
GSI50
Solid symbols represent
unstable slopes

Figure 5.21. Slope height vs slope angle case study data and a comparison of design
curves for a dry slope
0
50
100
150
200
250
10 20 30 40 50 60 70 80 90
Slope Angle (deg)
S
l
o
p
e

H
e
i
g
h
t

(
m
)
.
20-30
30-40
40-50
50-60
60-70
70-80
80-90
GSI
S
= Solid symbols represent
unstable slopes
= Significant contribution to
failure from structure
= High water pressures in
slope
S
H
H
H
S
GSI
30
GSI
40
GSI
50

design curves for moderate pressures

0
50
100
150
200
250
10 20 30 40 50 60 70 80 90
Slope Angle (deg)
S
l
o
p
e

H
e
i
g
h
t

(
m
)
.
20-30
30-40
40-50
50-60
60-70
GSI
GSI
30 GSI
40
RMR
30
RMR<20
Robertson
RMR
50
Bieniawski
= Solid symbols represent
unstable slopes
= Significant contribution to
failure from structure
= High water pressures in
slope
S
H
H
H
S S
Author's
curves
GSI
40
GSI
50
GSI
30

Figure 5.23. Slope height vs slope angle case study data and a comparison of design
curves for moderate pressures

5.5 CONCLUSION
A rock mass rating system should provide a measure of the basic quality/strength of the
rock mass. Aspects such as ground water, excavation method, slope height and orientation
of structure should not be included in the rock mass rating and should be taken account of
during analysis.
Correlation of GSI with several other rock mass ratings indicates a good correlation and
would suggest GSI is an adequate indicator of basic rock mass quality for rock slopes.
Slope design curves have been developed based on a number of stable and unstable open
pit mine slopes. Shear strength estimates for rock slopes that were proposed by
Bieniawski (1976) are too high for values of GSI below 40. The design curves using
strength estimates proposed by Robertson (1988) predict steeper angles than the authors
design curves.
Most slopes will be structurally controlled and therefore a rock mass rating system will
not be applicable for most slope design. Empirical slope design using rock mass rating
systems should only be considered for slopes in rock masses with GSI values lower than
about 45 and only after any potential structurally controlled failures have been
investigated. The method should only be applied at the preliminary stage or as a site
specific tool to complement detailed mapping and analysis.
The Shear Strength of Rock Masses Page 6.1
6 THE SHEAR STRENGTH OF ROCK MASSES
6.1 INTRODUCTION
This chapter contains a discussion of current methods of predicting rock mass strength
and develops modifications to the Hoek-Brown strength criterion that account for the
results in the previous chapters. The author has concentrated mainly on developing the
criterion for use at low confining stresses (e.g. slopes) however, the resulting criterion
should be applicable for the full stress range.
A rock mass criterion should only be used where there are a sufficient number of
closely spaced discontinuities that isotropic behaviour involving failure on
discontinuities can be assumed (Hoek and Brown, 1997). Such a situation for slopes is
illustrated on Figure 6.1, it should be noted that the concept of closely spaced should be
defined in terms of the scale of the failure surface.
Figure 30 : Heavily jointed rock mass

Figure 6.1. Heavily jointed rock mass
Slope failures in which the failure surface is entirely through the rock mass are not
common. This is due to the low stresses typically acting in a slope. Failure through the
rock mass usually requires large scale (relative to the slope in question) defects
concentrating stresses into regions of weak rock mass. For example a long vertical joint
or subvertical fault may lead to over stressing of weak material at the toe of the slope
(Figure 6.2). In this case a rock mass criterion is only applicable to the region of rock
mass at the toe of the slope.

long
subvertical
joint
Rock mass
failure at toe of
slope

Figure 6.2. Example of shear failure through rock mass at the toe of a slope -
Nattai Escarpment Failure
6.2 ESTIMATING THE SHEAR STRENGTH OF A ROCK MASS
As mine slopes become larger (stresses increasing), the necessity to account for the
strength of rock masses in design increases. This section discusses different techniques
to determine the strength of a rock mass.
6.2.1 Predicting Rock Mass Strength from Discontinuities
Initial methods of determining the shear strength of rock mass relied on using intact and
rock joint strengths. This section briefly describes current methods for estimating rock
joint strengths and the problems with extending these for use with rock masses.
The two most commonly used approaches to estimating discontinuity strength are those
by Patton (1966) and Barton and his co-workers (Barton, 1971a-c, 1973, 1976, Barton
and Bandis, 1982, 1990, Barton and Choubey, 1977). Patton (1966) showed that the bi-
linear equation presented below could be used to estimate the shear strength of
discontinuities.
) tan( i
b n
+ (6.1)
The problem with this equation is that the dilation angle, i, is stress dependent and also
scale dependent. There is also some conjecture as to how the basic friction angle,
b
,
should be measured. The author believes that the basic friction angle should not be
measured on artificially polished or otherwise artificially modified samples. The
reasoning for this is that
b
should represent the minimum friction angle that a
discontinuity could reasonably be expected to attain in the field. Therefore the best way
to determine
b
would be to measure the friction angle at high strain or to use the
approach suggested by Hencher (1995) where the shear test results are corrected for
dilation using the equations below and plotted to find the slope of the curve (
b
).
( ) i i i
n c
cos sin cos (6.2)
( ) i i i
n nc
cos sin cos + (6.3)
where,
c
and
nc
are the corrected shear strength and normal stresses respectively.
There is, at present, no definitive way of measuring the dilation angle, i, for field scale
discontinuities. The approach by McMahon (1985) or similar, where asperities are
measured for a certain wavelength (or interlimb angle) of the defect are used to predict
i, is probably the best current approach.
Bandis et al (1981) developed the following shear strength criterion for discontinuities:

1
]
1
,
_
b
n
n
JCS
JRC

10
log tan (6.4)
Where,
JRC = Joint roughness coefficient
JCS = Joint compressive strength
This equation was developed based on 100mm long discontinuities and as such can be
expected to resonably predict shear strengths on this scale. Barton and Bandis (1982)
presented the following formulae to predict JRC and JCS for field scale samples.

0
02 . 0
0
0
JRC
n
n
L
L
JRC JRC

,
_
(6.5)

0
03 . 0
0
0
JRC
n
n
L
L
JCS JCS

,
_
(6.6)
where,
JRC
n
= Joint roughness coefficient for a discontinuity of length L
n

JCS
n
= Joint compressive strength for a discontinuity of length L
n

JRC
0
= Joint roughness coefficient for a discontinuity of length L
0

JCS
0
= Joint compressive strength for a discontinuity of length L
0
These equations are now widely used to predict the shear strength of discontinuities in
the field. However, there appears to be a problem with the form of the equations. Figure
6.3 shows a plot of JRC
n
vs JRC
0
for discontinuity lengths varying from 1m to 100m. It
is clear from this figure that the general trend of the equation must be incorrect. For
example, for a discontinuity length of 100m a JRC
0
of 8 would imply a JRC
n
of
approximately 2.65 whereas, a JRC
0
of 20 would imply a JRC
n
of 1.26. This implies
that two discontinuities identical in every respect except small-scale roughness would
have reversed larger scale roughness. Clearly this is wrong. It is understood that this
equation was developed on small-scale models and therefore the author suggests that
this equation may only be appropriate for joint lengths in the region of 1m. For larger
joint lengths/block sizes, and high JRC
0
, the scaling equations are clearly unsuitable.
This has important implications where at least part of the failure surface is controlled by
long joints or where the block size of the rock mass is in excess of 1m.
Some early studies (e.g. Bray, 1966) suggested that the strength of a rock mass could be
estimated as the lowest strength envelope of the individual discontinuities in the rock
mass (assuming that the strength of the intact material is relatively high). This could be
true where the failure of the rock mass was purely due to sliding along discontinuities.
However, this theory does not account for any interlocking or rotation of the intact rock
blocks in the rock mass. Any interlocking of the rock mass would require a dilation of
the mass to fail and hence an increase in strength. The shear strength of the individual
discontinuities in the rock mass could therefore be seen only as a lower bound to the
shear strength of the rock mass and not much use in design.
Where failure is partly through intact rock it has been suggested that the strength of the
rock mass could be estimated as a combination of the joint and intact rock strengths
(e.g. John, 1962, 1969, Einstein et al, 1983).
Brown (1970) showed that failure through a rock mass is more complex than the
methods above suggest. Brown (1970) tested models of rock masses and found that:
At low confining stresses collapse was possible due to block movement involving
the opening of joints and dilation of the mass. He found this mode of probable
practical significance in studying the behaviour of large masses of jointed rock in
which the unit rock block is small compared with the dimensions of the mass as a
whole as, for example, in deep open cut mines.
Axial cleavage of the rock blocks (gypsum plaster in the models) occurred at low
confining stresses.
Where failure occurred at low confining pressures through the model material or as
a combined shear and tension failure the strengths were lower than theories
developed from intact rock and rock joints would suggest.
The prediction of the strength of discontinuities on the field scale is, at best,
approximate. When applied, the predicted strengths of these discontinuities can only be
used to give a lower bound envelope to the shear strength of rock masses. Failure modes
for rock masses are more complex than simple shearing along defects and through intact
rock, particularly at low confining stresses. It is for these reasons that researchers have
taken an empirical approach to estimating the shear strength of rock masses.
Page 6.7
0
1
2
3
4
5
6
7
8
9
0 2 4 6 8 10 12 14 16 18 20
JRC
0
(for L
0
= 0.1m)
J
R
C
n
JRC JRC
L
L
n
n
JRC
1
]
1
0
0
0 02
0
.
L
n
= 1m
2m
5m
10m
20m
50m
100m
65 . 2
m 100
8
m 1 . 0
0
0
n
n
JRC
L
JRC
L
26 . 1
m 100
20
m 1 . 0
0
0
n
n
JRC
L
JRC
L
= Example point

Figure 6.3. Assessment of Barton and Bandis (1982) JRC
0
vs JRC
n
6.2.2 Predicting Rock Mass Strength using Empirical Formulae
Yudhbir et al (1983), Ramamurthy et al (1994) and Sheorey (1997) (Equations 6.7 to
6.9 respectively) present rock mass criteria that have been developed as extensions of
strength criteria for intact rock. The modification process has typically been based on
model tests, small sample testing and limited experience. The criteria all assume a non-
zero unconfined compressive strength,
cm
, and hence tensile strength,
tm
, for the rock
mass. These criteria would therefore be expected to overpredict the strength for poor
quality rock masses at the low stresses common to failure surfaces in slopes.

,
_
,
_
c
cm
c
m
c
b b a
3
1
3 1
or (6.7)

m
b
cm
m
a

,
_
+
3
3 3 1
(6.8)

m
b
tm
cm

,
_

3
1
1 (6.9)
where, , a
m
, b
m
are constants
6.2.3 Predicting Rock Mass Strength using the Hoek-Brown Criterion
The most commonly and almost exclusively used strength criterion for rock mass, over
the last two decades, is the Hoek-Brown empirical rock mass failure criterion, the most
general form of which is given in Equation 6.10. Hoek and Brown (1980) developed
this criterion, as there was no suitable alternate empirical strength criterion. The
equation, which has subsequently been updated by Hoek and Brown (1988), Hoek et al.
(1992) and Hoek et al. (1995), was based on their criterion for intact rock discussed
earlier in this paper. Although initially developed for hard rock masses, this criterion is
now used for all types of rock masses and stress regimes. The only rock mass tested
and used in the original development of the Hoek-Brown criterion was 152mm core
samples of Panguna Andesite from Bougainville in Papua New Guinea (Hoek and
Brown, 1980) together with rock mass models (Brown, 1970). Hoek and Brown (1988)
later noted that it was likely this material was in fact disturbed. The validation of the
updates of the Hoek-Brown criterion have been based on experience gained whilst using
this criterion. To the authors knowledge the only data published supporting this
experience has been two slopes cited by Hoek et al (2002).

a
c
b c
s m

,
_

3
3 1
(6.10)
The parameters m
i
and m
b
are intact and mass material parameters; a and s are
parameters that depend on the rock mass characteristics; and
c
is the uniaxial
compressive strength of the intact rock. Estimating the parameters, m
b
, s and a in the
Hoek-Brown criterion is done by correlation with rock mass rating parameters. The
most current of these is the Geological Strength Index (GSI) (Hoek, 1994). The
correlations with GSI were modified by Hoek et al (2002) to allow for a continuous
transition at GSI=25 and to introduce a rock mass disturbance factor, D. These
equations are given in Table 6.1.
Table 6.1. Estimation of Hoek-Brown co-efficients
Parameter Hoek et al (pre 2002) Hoek et al (2002)
m
b

,
_
28
100
exp
GSI
m
m
i
b

,
_
D
GSI
m
m
i
b
14 28
100
exp
GSI>25
,
_
9
100
exp
GSI
s
s
GSI<25 0 s
,
_
D
GSI
s
3 9
100
exp
GSI>25 5 . 0 a
a
GSI<25
200
65 . 0
GSI
a
( )
3 20 15
6
1
2
1

+ e e a
GSI

A summary of all the changes to the equations for estimating the shear strength of rock
masses using the Hoek-Brown criterion was presented by Hoek (2002) and is
reproduced here as Figure 6.6.
Hoek and Brown (1980) introduced the material parameter m
i
. Figure 6.4 gives an
example from Hoek (1999) of how to estimate m
i
based on rock type. Hoek and his co-
workers have presented numerous variations of this table. Chapter 3 on intact rock has
shown that rock type is a poor predictor of m
i
and as such a discussion of these
variations has not been provided here.
Hoek (1997) provides Figure 6.5 to determine the GSI directly. The GSI tables have
been modified several times up to Hoek et al (2002). These modifications will be
discussed in the following sections. Hoek et al (1995) say that the GSI may also be
calculated using Bieniawskis (1976 and 1989) rock mass rating (RMR), GSI
RMR
, or
Bartons (1974) Q-system, GSI
Q
.

Figure 6.4. Values of the parameter m
i
for intact rock (Hoek, 1999)

Figure 6.5. Estimation of GSI (Hoek, 1997)


Figure 6.6. History of the Hoek-Brown criterion (Hoek, 2002)

Figure 6.6. History of the Hoek-Brown criterion (Hoek, 2002) (cont.)

Note 1: These are corrected Balmer equations. The original equations were incorrect.
Figure 6.6. History of the Hoek-Brown criterion (Hoek, 2002) (cont.)
6.3 A DISCUSSION OF THE HOEK-BROWN CRITERION WITH
PARTICULAR REFERENCE TO SLOPES
6.3.1 Calculation of GSI
GSI
RMR
and GSI
Q
are derived from the rating parameters for several rock mass
properties (Equations 6.11 and 6.12).
( )
+ + + condition defect spacing defect strength intact Ratings RQD GSI

RMR
(6.11)
44 log 9 +
,
_
a
r
n
e Q
J
J
J
RQD
GSI (6.12)
where J
r
= joint roughness number
J
n
= joint set number
J
a
= joint alteration number
The RMR was derived for and on the basis of a data set of underground tunnels that
were of the order of 10 to 20m in span (Bieniawski, 1989). The Q-system was
developed in a similar way. It can be expected from this that the RMR and Q-system
could be expected to be reasonable indicators of rock mass properties for underground
tunnels and caverns. Where a slope is in the order of several hundred meters high,
ranges of values for each parameter in the RMR system begin to lose meaning. This is
discussed further below for each component of the GSI.
6.3.1.1 Intact Strength
It is well known that intact rock exhibits a strength scale effect. This scale effect exists
up to block sizes of at least one metre. Figure 6.7 shows this effect. The author suggests,
that this should be considered when assigning a rating for intact rock strength.

Figure 6.7. Scale effect of Intact Rock (Hoek and Brown, 1980)
6.3.1.2 RQD
Since the RQD is based on a fixed length of 100mm of drill core, the ability of the RQD
to give meaningful information reduces as slopes get larger. Roof stability of a 10m
diameter tunnel is likely to be affected by a reduction in RQD. However, for slopes
several hundred meters high the RQD (particularly estimated from borehole data) has
questionable value. On the scale of a large pit slope it is highly unlikely that all the
defects encountered in boreholes would be of significance to the overall rock mass
stability. The defect sets controlling stability for a large rock slope could be expected to
have a spacing far in excess of 100mm because the closely spaced ones are unlikely to
be persistent. If only these defects are taken into account in the GSI then a rating of 20
(the highest possible) will always apply for RQD. There is a general correlation between
RQD and defect spacing and as such the need for both spacing and RQD in the method
is questioned.
6.3.1.3 Defect spacing
Most systems refer to the term joint. The author believes that the term defect is more
appropriate as all types of defects including bedding planes, shears, faults, lithological
contacts (and not just joints) will interact to form blocks in a rock mass.
The defect spacing suffers from a similar problem to that of the RQD. The maximum
spacing interval is greater than 3m and greater than 2m for Bieniawski (1976) and
Bieniawski (1989) respectively. A defect spacing of this size would result in large
blocks which would have a much lower probability of forming blocks that would fall
out of a 10m diameter tunnel roof. When assessing loading of rock mass around the
tunnel, stress concentrations would generally be confined within single blocks and so a
maximum GSI parameter would be applicable. However, for a large rock slope the
region of over stressed rock can be expected to comprise several of these blocks. The
critical defect spacing may be much larger than 3m. Thus, a maximum GSI parameter
may not be suitable. Figure 6.8 shows blocks from a 400m high slope failure. A tunnel
of 15m span is unlikely to have rock mass strength problems with a block size as big as
those in the figure.
6.3.1.4 J oint condition
The analyst must take into account the large scale (i.e. scale of rock mass) defect
characteristics as well as those on the small scale. This is similar to considering both the
basic friction angle,
b
, and the field roughness, i, from Pattons (1966) shear strength of
defects formula. The thickness of defect infilling and defect roughness should be
considered proportionately to the size of the rock blocks in the slope. Figure 6.9 shows
two defects, on the small scale (borehole) defect A would have a high rating and defect
B a low rating. However, when one looks at the large scale (large slope), defect A would
be expected to have a lower strength.

Figure 6.8. Slope failure block size

Very rough
Smooth
Very rough
Smooth
&
infilled
Defect A
Defect B

Figure 6.9. Effect of scale on defect properties
6.3.1.6 GSI from Bartons Q-System
Hoek et al. (1995) also offer a method of using the Q-system developed by Barton et al.
(1974) to estimate GSI. The system includes RQD, joint number, J
n
, joint roughness and
joint alteration. RQD/J
n
represents block size. This system suffers in a similar way to
RMR in its over reliance on RQD for large rock masses.
6.3.1.7 GSI from published figures
Figure 6.5 shows estimates of GSI provided by Hoek (1997). The main components
affecting the strength of the rock mass are covered (ie structure and surface conditions).
It is not clear how scale is to be interpreted on this figure. The author believes that GSI
should be interpreted as being on the scale of the rock mass under assessment. Using
judgement the user can estimate the condition of their rock mass at the scale of their
slope. For example, a blocky rock mass at a scale of 10m is vastly different to a
blocky rock mass at the scale of 500m. Smaller relative block size leads to more
freedom for block rotation and a greater chance for mass failure. Liao & Hencher
(1997) showed that relative block size was critical in deciding the mode of failure. The
smaller the block size (when compared to slope height) the more likely rock mass
failure would be the dominant failure mechanism.
The author recommends the use of Figure 6.5 for calculations of GSI for slopes. The use
of GSI
RMR
and GSI
Q
from boreholes should only be used for preliminary strength
estimates. It should be remembered that the key to the structure column is degree of
interlocking. The degree of interlocking should be assessed on the scale of the slope
under consideration. For example, the rock mass controlling the slopes for an ultimate
pit may be considered interlocked whilst the rock mass may be considered as very well
interlocked on the scale of individual benches of the same slope. It should also be
remembered that where block size is of the same order as that of the structure being
analysed, the Hoek-Brown criterion should not be used. The stability of the structure
should be analysed by considering the behaviour of blocks and wedges defined by
intersecting structural features (Hoek, 1997).
6.3.1.8 A Note on Schistose Rocks
Hoek et al (1998) presented Figure 6.10 as a new version of the GSI estimation table
(Figure 6.5). This new figure allows for the estimation of GSI for sheared rock masses.
The author does not agree with this new extension to GSI. This extension to GSI
contradicts the basic premise of a rock mass as defined in the Hoek-Brown criterion:
The Hoek-Brown failure criterion is only applicable to intact rock or to
heavily jointed rock masses which can be considered homogeneous and
isotropic. In other words the properties of these materials are the same in
all directions.
The criterion should not be applied to highly schistose rocks such as
slatesThe strength of the discontinuities should be analysed in terms of a
shear strength criterion such as that published by Barton (1976).
(Hoek et al, 1995)
One reason that Hoek et al (1998) gave for extending the GSI table was that the
empirical equation for estimating the modulus of deformation of the rock mass, E
d
, from
the GSI and
ci
developed by Hoek and Brown (1997), overestimated E
d
in schistose
rocks.

,
_
40
10
10
10
GSI
ci
d
E

(6.13)
The conclusion by Hoek et al (1998) was that GSI must therefore be lower for schistose
rocks.
The author has, together with his Kung (2001), collated a number of case studies from
the literature where rock mass deformability has been measured. The results from the
case studies, together with data presented by Bieniawski (1978) and Serafim and Pereira
(1983) are shown in Figure 6.11 and Table 6.2. The error bars acknowledge the
variation of E
d
in the test results. The uncertainty in GSI was due both to the variability
of the rock masses and also the conversion from published RMR values to GSI. Figure
6.12 shows a plot of E
d
as measured in the field versus E
d
predicted from the equaton
above using the published unconfined compressive strengths. Figure 6.11 and Figure
6.12 show that there is, not suprisingly, a large amount of scatter in the data. Also, a lot
of the data is over predicted by the above equation.

Figure 6.10. GSI Table (Hoek et al, 1998)

Table 6.2. Rock mass deformability case studies
Reference Rock type GSI Ed (GPa) Type of In-situ Testing
Santiago J.L. (1986) Claystone 34 (30-63) 1.6 Pile Loading Test
Georgiadis M. (1986) Mudstone 49 (40-49) 1.8 Dilatometer Test
Gifford A.B. (1986) Claystone 53 (44-62) 4.40 (0.51-8.28) Plate Bearing Test
Pavlakis M. (1980) Siltstone 39 (23-60) 0.11 (.05-.17) Pressuremeter Test
Poulton M.M. (1986) Sandstone 67 (54-79) 0.3 (0.25-0.36) Plate Loading Test
Schultz R.A. (1995) Basalt 60 (51-79) 25 (10-40) Jointed Block Test
Sandstone 3.9 (2.2-5.6) Flat Jack Test
Cheng Y. (1993)
Sandstone
75 (55-87)
3.7 (2.3-5.1) Plate Loading Test
Giovanni B. (1993) Limestone 79 (63-91) 37.5 (32.5-42.5) Plate Loading Test
Mcdonald P. (1993) Basalt 39 (30-74) 1.39 (.2-4.6) Pressuremeter Test
Granodiorite 52 3.8
Granite 55 2.5
Siltstone 60 14
Tuff 75 11
Marble 76 18
Goodman Jack Test
Sandstone 46 (35-57) 7.5 (5-10)
Granite 48 (41-55) 17.5 (15-20)
Siltstone 68 (64-71) 25 (20-30)
Littlechild B.D. (2000)
Marble 78 (71-84) 15 (10-20)
Cross-Hole Geophysics Test
Sandstone 60 (55-65) 6.6
Shale 63 (58-68) 13.8
Shale 72 (67-77) 13.8
Mudstone 75 (70-80) 5
Bieniawski Z.T. (1990)
Shale 75 (70-80) 12.4
MPBX
35 (30-40) 4
42 (37-47) 8
43 (38-48) 10
45 (40-50) 7
62 (58-68) 9
64 (59-69) 8
50 (45-55) 13
55 (50-60) 13
57 (52-62) 18
67 (62-72) 24
70 (65-75) 20
75 (70-80) 13
75 (70-80) 15
80 (75-85) 33
Ribacchi R. (1984) Granite
80 (75-85) 35
Plate Loading Test

Page 6.24
0
10
20
30
40
50
60
70
0 10 20 30 40 50 60 70 80 90 100
GSI
E
d

(
G
P
a
)
.
Bieniawski (1978)
Serafim & Pereira (1983)
Santiago (1986)
Georgiadis (1986)
Gifford (1986)
Pavlakis (1980)
Poulton (1986)
Schultz (1995)
Cheng (1993)
Giovanni (1993)
McDonald (1993)
Littlechild (2000)
Bieniawski (1990)
Ribacchi (1984)
UCS = 100MPa
UCS = 10MPa
( ) 40 10
10
10

GSI c
d
E

Figure 6.11. E
d
versus GSI case study data and Hoek et al (1995) equation for
c
100MPa and
c
= 10MPa
The author believes that the fact that the modulus of a schistose rock is over predicted
by an empirical equation does not justify the extension of GSI to these types of rocks.
The manner in which Figure 6.10 has been extended is also questionable. The author
does not believe that there is a continuum between poorly interlocked, heavily broken
rock mass and thinly laminated or foliated rock masses with closely spaced
schistosity. Therefore extrapolating the lines of equal GSI across this boundary can not
be justified.
The author has decided to adopt the Hoek et al (1995) position and omit highly
schistose rock from the rock mass criterion.
0
10
20
30
40
50
60
0 10 20 30 40 50 60
E
d
test (MPa)
E
d

p
r
e
d

Figure 6.12. E
d
test from case studies versus E
d
pred from Hoek et al (1995)
equation
6.3.2 Estimation of Parameters from GSI
6.3.2.1 The Rock Mass Disturbance Factor, D
The rock mass disturbance factor, D, was introduced by Hoek et al (2002) to account for
the degree of disturbance to which the rock mass has been subjected by blast damage
and stress relaxation. It varies from zero for undisturbed insitu rock masses to unity for
very disturbed rock masses. Table 6.3 shows the suggested values for D for slopes.
Table 6.3. Guidelines for estimating disturbance factor D (Hoek et al, 2002)
Description of rock mass Suggested value for D
Small scale blasting in civil engineering slopes results in modest
rock mass damage, particularly if controlled blasting is used.
However, stress relief results in some disturbance.
D = 0.7
Good blasting
D = 1.0
Poor blasting
Very large open pit mine slopes suffer significant disturbance
due to heavy production blasting and also due to stress relief
from overburden removal.
In some softer rocks excavation can be carried out by ripping
and dozing and the degree of damage to the slopes is less.
D = 1.0
Production blasting
D = 0.7
Mechanical excavation

The introduction of D was in response to published experience using the Hoek-Brown
criterion. Sjberg et al (2001) and Pierce et al (2001) stated that the Hoek-Brown
criterion overestimated the strength of rock masses based on experience with mine
slopes. The author (Douglas and Mostyn, 1999) agrees with these findings and
discusses this further on in the chapter. Sjberg et al (2001) claimed that the equations
(6.14 and 6.15) presented by Hoek and Brown (1988) for disturbed rock masses were
more appropriate. Hoek et al (2002) in response modified their formulas for the
parameters m
b
and s, using D such that they could reduce to the Hoek and Brown (1988)
formulas (Table 6.1).

,
_
14
100
exp
RMR
m
m
i
b
(6.14)

,
_
6
100
exp
RMR
s (6.15)
Prior to excavation, the only tools available to the designer are boreholes and possibly
surface exposures. The model that is developed can therefore only be considered as
preliminary (see Section 6.3.1 for discussion). De-stressing and blasting of the rock
mass may reduce the interlocking of the rock mass in the original model however, the
author does not believe a catch all parameter D is appropriate.
Different types of rock masses with different stress histories will be affected differently
when subjected to blasting and destressing affects. Good quality blasting should not
affect a large region of rock mass. This region is likely to be remote to the failure
surface. If it is considered that this is not the case or that de-stressing will affect the rock
mass the GSI can be reduced accordingly. The GSI is considered the appropriate
parameter as it is meant to be a representation of the degree of interlocking of the rock
mass as per Hoek et al (1995).
The reduction of the GSI should follow an observational type approach. Changes to the
rock mass (interlocking and defect surface conditions) due to blasting and de-stressing
can be predicted using past experience in the particular rock mass under consideration.
As excavation proceeds mapping of exposed surfaces should show whether these
predictions are reasonable or whether they need to be modified.
The author has not used the disturbance factor, D, in this thesis. Instead, where analysis
is performed the GSI at the time of interest (i.e. the time of slope failure or pit wall
completion) is used.
6.3.2.2 The Variation of the Hoek-Brown Parameters with GSI
Figure 6.13 shows the variation of the Hoek-Brown parameters m
b
/m
i
, a and s with GSI
based on Hoek et al (2002) and D = 0. It should be noted that if the Hoek (1997) table,
Figure 6.5, is used then the minimum and maximum values of GSI are 5 and 80
respectively (Table 6.4). m
b
/m
i
, which mainly accounts for friction, varies gradually
from unity as could be expected for a rock mass. The value of s (which mainly accounts
for cohesion) dimi nishes rapidly with a reduction in GSI thus, indicating a rapid
reduction in compressive strength and an even more rapid reduction in tensile strength
as the quality of the rock mass decreases. This is as expected. As rock mass defects
become more cohesive it would be expected that s would be non-zero so as to avoid
zero compressive strength. But GSI reduces for increasing cohesion and if s is predicted
from GSI then s approaches zero not a finite value. This may be why the Hoek-Brown
criterion will underpredict the shear strength of clayey bench slopes. It should be
remembered at this point that the initial Hoek-Brown criterion was developed for hard
rocks and has only recently been accepted for use with very poor quality rock masses by
Hoek and Brown (1997). Thus, it could be expected that the experience with using the
criterion for poor quality rock masses (particularly for slopes) would be limited.
0
0.2
0.4
0.6
0.8
1
0 20 40 60 80 100
GSI
m
b
/
m
i
;

s
;

a
mb/mi
s
a
m
b
/m
i
s
a

Figure 6.13. Variation of a, s and m
b
/m
i
with GSI
Table 6.4. Maximum and minimum values of Hoek-Brown parameters using
Figure 6.10
Parameter
Minimum
GSI = 5
Maximum
GSI = 80
m
b
/m
i
0.034 0.49
s 0.000026 0.11
a 0.62 0.50

The value of a remains relatively constant and has a maximum value of 0.62 (Table
6.4). This is not consistent with what is known about compacted rockfill strength (a
material that could represent a lower bound to poor quality rock masses) where an a of
0.95 would be expected (Chapter 4). Chapter 3 on intact rock indicates that a actually
varies from 0.2 to 1.0, with a reasonable estimate of 0.4 to 0.95 depending on m
i
. Thus it
is not correct at two limits presented in this thesis. Where the intact rock approaches that
of a soft rock or hard soil the curvature is much less pronounced than an exponent of
0.65 would suggest (Johnston & Chiu, 1984).
As has been shown for intact rock (Chapter 3), fixing or limiting a has a very large
impact on the estimation of the other parameters (m
b
and s) and therefore a cannot
simply be changed without addressing the other parameters as well.
6.4 VALIDATION OF THE HOEK-BROWN CRITERION
This section describes some analyses performed in order to assess the ability of the
Hoek-Brown criterion to predict the shear strength of rock mass.
6.4.1 Chichester Dam
The first case analysed was Chichester Dam, a 41m high concrete gravity dam
constructed in the 1920s. The dam was upgraded with drainage holes and post
tensioning due to a perceived weakness in the foundation. The foundation consisted of
interbedded tuffaceous sandstone and thin shale layers. The best estimate GSI for the
foundation was 75. Unfortunately the analysis of the foundation (assuming no post
tensioning and drainage) using the distinct element program UDEC found that the
stresses beneath the dam were insufficient to give meaningful bounds on the Hoek-
Brown criterion. Note that sliding of the dam along bedding surfaces was a potential
failure mode. It was decided at this point to direct attempts at validating the criterion
toward higher stressed rock mass such as large failed slopes.
6.4.2 Nattai North Escarpment Failure
The Nattai North escarpment failure is located 80km south west of Sydney. The failure,
with a total volume of 14 million cubic meters and height ranging between 200 and
300m, is one of the largest rock mass failures to have occurred in Australia in modern
times. The failure has been well documented with studies in the area by Kotze & Pells
(1980), McGregor (1980), Cunningham (1986) and Pells et al (1987). Figure 6.2 shows
a photograph and diagram of the failure.
Figure 6.14 shows the escarpment stratigraphy consists of Hawkesbury Sandstone
overlying sandstones and claystones of the Narrabeen Group which in turn overlies the
Illawarra Coal Measures. Of most interest to stability considerations are the weaker
more fractured, and finer grained claystone and siltstone known as the Wombarra
Claystone and the 2.7m thick Bulli Coal Seam.

Figure 6.14. Typical section of the Nattai North failure (Pells et al, 1987)
Persistent vertical defects dipping at approximately 84 out of the slope together with
bedding planes dipping 5 into the slope, resulted in large pillars that were prevented
from sliding or toppling. The pillars were supported on the highly stressed weaker
Wombarra Claystone. These conditions are believed to be sufficient to cause shearing
through the claystone and failure of the escarpment under natural processes. It is
believed that the failure was triggered by mining of coal seams beneath the escarpment.
The mining is suspected to have caused further fracturing to the claystone and stress
concentrations at the toe (Pells et al, 1987). Figure 6.15 gives a general illustration of
the failure mode. Shearing occurs through the claystone at the base of the pillars. The
base of the pillars slide out and the mass collapses (Cunningham, 1986).
The failure of the escarpment at Nattai North is seen as predominately shear failure
along defects (natural and mining induced) and through rock mass. The nature of the
defects in the claystone lend themselves well to the description of rock mass by Hoek
(1980) and hence the Hoek-Brown empirical rock mass failure criteria.
a) Pre-mining: failure may
occur as a natural process if
the vertical joints are
continuous and the base rock
is of poor quality.
b) De-pillaring proceeds toward the
escarpment. Vertical joints open and
extend to form continuous joints.
Fracturing of the Wombara claystone
is occurring.
c) Massive columns are
formed. Yielding and shearing
at the base is initiated.
d) Shearing through of base and
complete failure.
Fracturing of Wombara
Claystone due to high
shear stress. Further
fracturing due to tensile
strain.
Highly stressed
Fractured due to
high shear and
tensile stresses.
Coal seam
shown
exaggerated.

Figure 6.15. Illustration of the failure mechanism at Nattai North (Helgstedt,
1997)
The geometry and rock mass properties of the failure were estimated using data from
Cunningham (1986), McGregor (1980), Kotze and Pells (1980) and Richmond and
Smith (1979). The strength and deformation properties of the intact rock were estimated
from Bhattacharya (1976) and Evans (1978) who had performed testing on similar rocks
in a different area. The author validated as much of the information as possible during
surface field mapping (both on top and at the toe) of the landslide.
The properties used in the analysis are shown in Table 6.5 and Table 6.6. Table 6.7
shows the GSI estimated for the claystone where rock mass failure was deemed to have
occurred.
Helgstedt (1997) carried out numerical modelling of the failure under the supervision of
the author and Garry Mostyn using the distinct element code UDEC.
Two series of runs were made, with and without mining. The slope was modelled as two
units comprising sandstones and claystones. The geometry of the slope and cliff line
prior to failure was reconstructed using the Burragorang 1:25000 topographical map.
All parameters adopted for the model were best estimate average values. Extreme but
possible values were chosen for the sensitivity analysis. During modelling the material
parameters (marked with an asterix in Table 6.6) were progressively reduced until
failure occurred. The stresses at failure were then obtained along the shear surface
through the claystone.
Due to computational constraints, the actual spacing for the joint sets and bedding
planes could not be implemented in UDEC in areas remote to the zone of shearing.
However, the true spacing for most of the slope was not required to model the
mechanism of slope failure. Three different areas needing different spacing were
recognised and are outlined below. Stresses of 3.5+0.055z and 1.6+0.034z MPa were
used as the upper and lower bounds respectively for
1
(horizontal stress).
The base of the slope in the Wombarra Claystone and lower section of the Scarborough
Sandstone (rough zone) required a spacing as close to actual as possible. Due to the
failure mechanism being one of shear failure both along discontinuities and through
intact rock, randomly oriented fictitious joints having intact rock strength were added.
It should be noted that although the ficticious joints are given intact rock strengths, once
they fail UDEC assumes their strength to reduce to zero rather than that expected for a
rock joint in claystone/sandstone.
The section above the failing base required continuous vertical joints to permit the
mechanism of failure to be modelled. The presence of these vertical joints was
confirmed by the author during field mapping. For simplicity bedding plane spacing
was made equal to the vertical joint spacing. Away from the failure the joint spacing
was increased. The purpose of modelling this area was purely to reduce any edge effects
in the model.

Table 6.5. Joint orientation, spacing and persistence for Nattai North
Parameter Applied to
Upper
bound
Adopted
value
Lower
bound
vertical joints 87.9 84 75.6 Apparent Dip
() bedding - -5 -
vertical joints 20,40,150,250 14,30,125,250 8,20,100,250 Apparent spacing
(m)
(finest to roughest)
Bedding
discontinuities
20,40,110,180 14,30,90,180 8,20,70,180
Discontinuity
persistence
all discontinuities - continuous -
Table 6.6. Summary of parameters used for the Nattai North Escarpment Failure
Adopted initial value
Parameter
Sandstone Claystone
Density (kg/m
3
) 2560 2650
Youngs modulus, E (GPa) 14.4 10.5
Shear modulus, G (GPa) 5.54 4.04
Bulk modulus, K (GPa) 12.00 8.75
Poissions ratio 0.3
Friction angle of intact material ()*
40 35
Cohesion of intact material (MPa)* 6 5
Tensile strength of intact material,
ti
(MPa)*
0.75 0.6
Normal stiffness, K
n
(GPa/m) 50
Shear stiffness, K
s
(GPa/m) 25
Defect friction angle ()*
32 28
Defect cohesion (MPa)* 0.025 0.03
In-situ in plane horizontal stress,
1
(MPa)
2.5+0.044z (where z = depth)
In-situ out of plane horizontal stress,
2
(MPa)
1.2+0.026z
In-situ vertical stress,
3
(MPa)
0.026z
* Changed globally for sensitivity analysis.
Table 6.7. Summary of Hoek-Brown parameters for Nattai using RMR and the
Hoek-Brown chart
GSI
source
RMR
89
Hoek-Brown Chart
(Hoek, 1997)
Estimate LB BE UB LB BE UB
GSI 55 61 69 49 45 55
m
i
4 6.5 9 4 6.5 9
LB = Lower bound; BE = Best estimate; UB = Upper bound
The results of the analysis using UDEC correlated well with the assumed failure mode.
Average shear and normal stresses along the failure zone are presented in Table 6.8. The
mode of failure did not change where mining was modelled however, the shear zone
was found to extend deeper into the toe of the slope.
Table 6.8. UDEC output: average shear and normal stresses along the predicted
failure plane
Average stress along failure
zone (MPa)
Case
Strength parameters
used Normal
stress
Shear
stress
Best estimate of all values, no mining 0.8 x best estimate 1.21 1.39
Lower bound in-situ stresses, no mining 0.7 x best estimate 1.16 1.32
Upper bound in-situ stresses, no mining 1.3 x best estimate 1.26 1.63
Best estimate of all values, mining 1.4 x best estimate 1.96 1.90

6.4.3 Katoomba Escarpment Failure
The Katoomba (or Dogface Rock) escarpment failure is located at Katoomba in the
Blue Mountains west of Sydney. The main failure occurred on the 28
th
January 1931.
Smaller failures have been recorded through to July 1977. The total volume of all
failures comprises approximately 75000 to 100000m
3
. The depth to the failure surface
was up to 280m. The failure has been documented by Pells et al (1987). Figure 6.16
shows a photograph from after the failure.

Figure 6.16. Katoomba Escarpment Failure (photo courtesy Mrs Gwen Silvey of
the Blue Mountains Historical Society)
The geology and mode of failure was similar to that of the Nattai Escarpment failure.
The rocks were part of the Narrabeen Group with hard competent sandstones (Banks
Wall Sandstone and Burramoko Head Sandstone) overlying weaker material (including
Hartley Vale Claystone, Victoria Pass Claystone, Beauchamp Falls Shale). Bedding
planes were near horizontal and joints in the sandstones persistent and subvertical.
Figure 6.17 shows an example of the vertical defects in the sandstone.

Figure 6.17. Katoomba Escarpment Failure, column prior to collapse (photo
courtesy Mrs Gwen Silvey of the Blue Mountains Historical Society)
No site specific testing was carried out on the materials and so all parameters were
estimated from testing on similar rocks in other areas (Evans, 1978 and Bhattacharyya,
1976). Table 6.9 shows the material properties used and Table 6.10 shows the
estimation of the GSI.
Table 6.9. Summary of parameters used for the Katoomba Escarpment Failure
Adopted initial value
Parameter
Sandstone Claystone
Density (kg/m
3
) 2560 2650
Youngs modulus, E (GPa) 14.4 10.5
Shear modulus, G (GPa) 5.9 3.86
Bulk modulus, K (GPa) 8.57 12.5
Poissions ratio 0.22 0.36
Friction angle of intact material ()*
38 33
Cohesion of intact material (MPa)* 15 12.9
Tensile strength of intact material,
ti
(MPa)*
1.91 1.61
Defect friction angle ()*
32 28
Defect cohesion (MPa)* 0.025 0.03
In-situ in plane horizontal stress,
1
(MPa)
2.5+0.044z (where z = depth)
In-situ out of plane horizontal stress,
2
(MPa)
1.2+0.026z
In-situ vertical stress,
3
(MPa)
0.026z
Table 6.10. Summary of Hoek-Brown parameters for the Claystone in the
Katoomba Escarpment Failure using RMR and the Hoek-Brown chart
GSI
source
RMR
89
Hoek-Brown Chart
(Hoek, 1997)
Estimate LB BE UB LB BE UB
GSI 28 50 69 48 53 58
m
i
4 6.5 9 4 6.5 9

Tarua (1997) carried out the numerical modelling under the supervision of the author
and Garry Mostyn using the distinct element code UDEC. As the mode of failure and
geology/geometry was simlar, the analysis was carried out in the same manner as that
for the Nattai Escarpment failure.

6.4.4 Aviemore Dam Insitu Shear Tests
Aviemore Dam, constructed between 1963 and 1968, is located on the Waitaki river,
185km south west of Christchurch, New Zealand. The dam is comprised of both a
340m long, 57m high concrete gravity section and a 350m long, 49m high earth
embankment section (Read et al, 1996). During construction eight large-scale in-situ
shear tests were carried out on the foundation in order to determine the strength of the
concrete/rock interface for stresses relevant to those induced by the proposed dam.
The dam is founded on silty to sandy greywacke of Mesozoic age and coal measure
sediments of Tertiary age. However all in-situ tests were carried out on greywacke rock.
The rock mass is closely jointed, veined, and often is crushed and sheared. Read et al
(1996) identified three semi -orthogonal joint sets as follows:
joints sub-parallel to bedding, 150-190/75NE-75SW (strike/dip)
joints striking approximately perpendicular to bedding, 100-120/70NW-90
joints striking approximately perpendicular to bedding, 080-110/30-50SW
The range of the orientation for the joint sets is very wide and in addition several other
randomly orientated joints are present. Read et al (1996) stated that a statistical analysis
gives the impression that the overall jointing pattern at Aviemore was random. The most
common spacing has been found to be between 60 and 200mm (Read et al, 1996). Most
of the joints were found to have a slightly wavy profile, were rough to smooth and
closed with a persistence of a few metres. Moreover sheared and crushed zones occur,
although these are very widely spaced at the location of the in-situ shear tests.
Read et al (1996) mapped approximately 350 defects on available outcrops. The
outcrops were considered similar to the rock mass in the tests based on photos and other
documents. By scrutinising old photos and documents Read et al (1996) concluded that
the tested rock mass was of similar quality to that covered by the joint survey. A
summary of this survey is shown in Table 6.11. The GSI was estimated using the results
of this survey and is shown in Table 6.12.
Figure 6.18 illustrates the setup and dimensions of the tests at Aviemore. One test block
(0.76mx1.8m) and one reaction (anchor) block (0.9mx1.8m) were cast directly onto
clean rock foundations (Foster & Fairless, 1994). The vertical load was applied to the
blocks using stressed bars anchored into the rock foundation. An ungrouted length of
4.9m was used for at least two of the tests. Freyssinet flat jacks were used to force the
blocks apart as constant vertical load was maintained on the test blocks. Six tests were
performed using a horizontally acting thrust and two using an inclined thrust acting at a
downward angle through the centre of the blocks. The latter two test results were not
used.
Table 6.11. Summary of the Joint Properties from the Joint Survey carried out by
Read et al (1996)
Persistence of joints (m) 0-0.75 0.75-1.5 1.5-2.5 2.5-3.5 3.5-4.5 4.5-7 7-(15)
% per metre interval 27 30 19 13 9 5 0.5
Defect types shear/
fault
joint bedding schistosity
/foliation

% 10 84 6 0
Aperture (mm) >200 60-200 20-60 6-20 2-6 <2 tight
% 0 0 0 1 2 97 0
Nature of infill clean surface stain. non-cohesive cemented
% 12 72 1 12
Roughness polished slickensided smooth rough defined
ridges
small
steps

% 0 1 58 35 4 1
Table 6.12. Hoek Parameters for Aviemore Shear Tests using RMR, and the
Hoek-Brown Chart
GSI
Source
RMR
Hoek-Brown Chart
(Hoek, 1997)
Estimate UB BE LB UE BE LB
GSI 52 43 31 40 35 30
m
i
15 12 10 15 12 10
c
(MPa) 50 35 20 50 35 20

Foster & Fairless (1994) stated that all tests failed through the rock mass,
approximately 50-100mm below the rock-concrete contact, involving failure of several
defects. During the test both the test block and the reaction block failed at the same
time. Foster & Fairless (1994) analysed the failures and proposed that both blocks
failed at the same /
N
ratio and that the - curves should have been similar. Since the
- curves were similar it was likely that vertical force had been transferred between
the blocks via the stiff jack.
Foster and Fairless (1994) adjusted the vertical stresses on the test and reaction blocks
assuming that vertical forces were transferred from the reaction block to the test block
due to differential displacement. The values were adjusted to yield similar /
N
ratios
for the reaction and the test blocks.
Helgstedt (1997) carried out the numerical modelling under the supervision of the
author and Mr Garry Mostyn using the distinct element code UDEC. The modelling
was used to investigate whether the assumptions of Foster and Fairless (1994) were
reasonable. The general model behaviour was studied to see if it agreed with
observations from the real tests and to confirm the theory that vertical stress had been
transferred between the blocks through the Freyssinet jack. The failure stresses from the
model were not used.

Figure 6.18. Direct Shear Test Set up (Foster & Fairless, 1994)
The defects in the rock mass were created by Monte Carlo generation. The joints were
assumed to be randomly orientated with a negative exponentially distributed joint trace
length. The distribution of the discontinuities in space was assumed to be random.
These assumptions were verified against the data from the joint survey and data
presented by Read et al (1995 and 1996). Figure 6.19 illustrates the mesh generated for
one of the UDEC models. The actual true joint frequency was only modelled close to
the simulated test and reaction blocks, joints were more widely spaced in the far field.
The vertical load applied by stressed bars, anchored in the rock below the shear tests, at
Aviemore was simulated in the UDEC model by applying vertical loads at the top of the
blocks. This simplification is however not likely to influence the results. The horizontal
load from the Freyssinet jack was simulated by four blocks as can be seen in Figure
6.20 (between the test and reaction block). The left and right blocks represent
pressurised sides of the jack. A pressure was exerted in the vertical slot between these
blocks. The upper and lower blocks represent the casing of the jack. The two horizontal
cracks dividing the upper and lower blocks from the left and right blocks were given
zero friction angle and cohesion, but very high tensile strength. This allows the left and
right blocks to slide freely in a horizontal direction, while simultaneously, the simulated
jack can act as a rigid device capable of transferring moment.
The parameters that were used for the numerical model are given in Table 6.13 and
Table 6.14.
The general displacement (vertical and horizontal) behaviour matched the behaviour of
the field tests. The assumptions of Foster and Fairless (1994) were validated as vertical
stress was shown to be transferred through the jack and failure occurred simultaneously
in the reaction and test blocks. The statistically generated models were run several
times. All with results similar to that discussed above.
Table 6.13. Intact material parameters
Material Tensile
strength
(MPa)
Cohesion
(MPa)
Friction
angle
()
Bulk
Modulus
(MPa)
Shear
Modulus
(MPa)
Density
(t/m
3
)
Concrete blocks 5 15 45 25000 10400 2.4
Jack and stressed bars 200 400 0 133333 80000 7.6
Intact rock 3 8 45 20000 18000 2.2

Table 6.14. Defect and Interface Material Parameters
Material Tensile
strength
(MPa)
Cohesion
(MPa)
Friction
angle
()
Normal
Stiffness
(MPa)
Shear
Stiffness
(MPa)
Conc-steel interface 0 0 45 50000 30000
Conc-rock interface 5 15 50 25000 15000
Jack casing interface 200 0 0 2500000 1500000
Press. void in jack 0 0 0 2500000 1500000
Defects in rock mass 0 0 44 15000 5000

6.4.5 Discussion of the Results of the Analysis
The Hoek-Brown criterion, together with Figure 6.5 to estimate GSI, was used to
estimate the shear strength of the rock mass. Balmers method was used to convert from
principal stresses to shear and normal stresses. Figure 6.21 shows a plot of the ratio of
calculated shear strength,
in-situ
, versus the shear strength estimated from GSI for the
cases. The results show that the Hoek-Brown criterion provides a reasonable estimate of
the strength at the Nattai Escarpment. The Aviemore insitu shear tests show the
predicted strength using the Hoek-Brown criterion to range from 0.8 to 1.2 times the
computed strength from the shear tests. The Hoek-Brown criterion predicts twice the
strength as that calculated using UDEC for the Katoomba escarpment failure. However
this may be due to the lack of site specific strength data at Katoomba. The GSI
estimated from the RMR was noticeably higher than that estimated using Figure 6.5 for
the Aviemore insitu shear tests and Nattai escarpment failure. Thus, the strengths
estimated using RMR would lead to an overprediction of the strength of the rock mass

Figure 6.19. Example of mesh used (Helgstedt, 1997)

Figure 6.20. Close up of the simulated jack (Helgstedt, 1997)


0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
n
(MPa)
i
n
-
s
i
t
u
/
t
a
b
l
e
Katoomba escarpment failure
Aviemore shear tests
Nattai escarpment Failure

Figure 6.21. Back analysis results using Figure 6.5 for GSI

6.5 A NEW ESTIMATION OF ROCK MASS STRENGTH
6.5.1 Development of a Modified Criterion
As discussed earlier the Hoek-Brown criterion is virtually universally accepted. It also
appears to give reasonable results if used in conjunction with Figure 6.5 to estimate
GSI. The current method of calculating the parameters m and a appears to be flawed at
the limits of intact rock (Chapter 3) and rockfill (Chapter 4). It has therefore been
decided to modify this criterion so that it incorporates some of the results presented in
this thesis rather than to develop a completely new approach.
The terminology used for the modified criterion is outlined below.

,
_
+ s
m
ci
ci
3
3 1
(6.15)
For intact rock (GSI=100):
s = s
i

m = m
i

=
i

For rock mass (GSI<100):
s = s
b

m = m
b

=
b
In general,
ci
is the unconfined compressive strength,
c
, of the intact rock; unless the
scale of the discontinuities affects the strength of the blocks in the rock mass under
consideration.
The author proposes to use the form of the Hoek-Brown criterion and to modify the
method of calculating the parameters , s and m. The basic assumption is that the rock
mass parameters will be factored versions of the intact parameters developed in Chapter
3.
6.5.1.1 Exponent
Hoek and Brown (1980) restricted to 0.5. Hoek (1994) modified this for GSI<25 to
allow to vary up to a theoretical maximum of 0.65 (but actually 0.625 for a minimum
GSI of 5) using the equation below.
200 / 65 . 0 GSI (6.16)
This equation resulted in a step at GSI = 25 from 0.525 to 0.5. Hoek et al (2002)
eliminated this step by modifying the equation for all GSI values to:
( )
3 20 15
6
1
2
1

+ e e
GSI
(6.17)
This equation results in a maximum of 0.62 using a minimum GSI of 5.
Chapter 3 has shown that is not fixed at 0.5, and can vary considerably for intact rock.
A rock mass can be considered a transitional material between intact rock and soil. At
the soil limit it is expected that would approach unity (Mohr-Coulomb material). A
statistical analysis of 929 triaxial tests on rockfill given in Chapter 4 showed that
b
for
rockfill is approximately 0.9 and that m
b
2.5.
If the results of Chapter 3 are used for intact rock then a modification to the current
Hoek-Brown equation is required where
b
is not restricted to a range of 0.5 to 0.65,
and is a function of m
b
and GSI. It is assumed that as the rock mass approaches a soil
like material that
b
will approach 0.9 and m
b
will approach 2.5.
6.5.1.2 Parameter m
Hoek and Brown provide estimates of m
i
based on rock type however, as discussed in
Chapter 3, rock type is a poor predictor of m
i
. The parameter m
i
should be estimated
using triaxial tests on intact rock samples or can be approximately estimated as
ci
/
ti
.
This method may also be appropriate for intact soil-like materials (clays). Limited Al-
Hussaini (1981) found that the unconfined compression strength obtained was about 5
to 10 times larger than the tensile strength for samples of high and low plasticity clay
and sandy clay. Fang and Fernandez (1981) found that the ratio of
c
/
t
was between
15-28 for clays at their optimum moisture content and 3-7 for air dried samples. These
ratios from the literature fall within the authors suggested m
i
range.
There is some limited information in the literature that shows that m
i
=
c
/
t
, for intact
soil-like materials (clays), is within the range found for intact rock (between 4 and 40)
(Al-Hussaini, 1981 and Fang and Fernandez, 1981). However, the information in the
literature is insufficient for the author to make a definitive judgement on whether the
authors criterion can be applied to clays. As the criterion is empirical and has been
developed for rock using triaxial tests on rock samples the author does not recommend
the extrapolation and use of the criterion for clays.
The parameter m (in association with the exponent ) predominately affects the friction
angle of the rockmass. Therefore as GSI drops it can be expected that the rockmass will
become less interlocked and the frictional strength of the rockmass will reduce
(predominately through a reduction in dilation). The parameter m
b
should therefore
reduce from m
i
to a limiting value as a function of GSI. The limiting value for a non-
cohesive rockmass could be taken as 2.5 from the analyses of rockfill discussed earlier.
6.5.1.3 Parameter s
The parameter s contributes predominately to the cohesiveness of the rock mass. Thus,
as GSI decreases s should also decrease as per the Hoek-Brown criterion. The limiting
value should be zero where the soil-like rock mass is non-cohesive.
Once a rock mass no longer behaves as intact it can be expected that cohesion and thus
s should decrease rapidly with a decrease in GSI. Hoek and Brown use an exponential
drop in s versus GSI as shown in the following equation.

( )
,
_
9
100
exp
GSI
s (6.18)
Hoek and Brown consider a well-interlocked rock mass to have a GSI of 80. Therefore,
s may be close to unity in the region of GSI equal to 80-100.
The Hoek-Brown criterion was developed for hard rock masses. As such, at the soil
limit (GSI 0) the parameter s in the Hoek-Brown equation approaches zero. This is
reasonable for a non-cohesive or frictional rock mass. However, where the rock mass is
controlled by cohesion (i.e. the rock mass is controlled by the clay in the mass) this
assumption is incorrect. The author believes that where there is sufficient clay material
in the rock mass such that there is no rock to rock contact during shearing the shear
strength of the soil should be used. The Hoek-Brown criterion is inappropriate, as the
properties of the intact rock (e.g.
c
) will have at the most a minor effect on the strength
of the mass. The cohesive soil limit is not considered in this thesis.
6.5.2 Development of the Equations to Estimate the Parameters in the
Hoek-Brown Criterion
The previous section discussed the development of modifications to the parameters of
the Hoek-Brown criterion. These were based on the authors work on intact rock,
rockfill and rock masses. The bounds have been well quantified in this thesis however,
due to the limited information available about failures and the addition of a variable
dependent on m, it meant good quality triaxial tests were needed to provide equations
for intermediate rock masses. The analysis of triaxial testing on rock mases could also
give confidence to the theory developed in the previous section. Habimana et al (2002)
published four sets of triaxial tests on quartzitic sandstone. These were part of an
extensive laboratory testing programme carried out by the rock mechanics laboratory of
the Swiss Federal Institute of Technology Lausanne (EPFL) for the hydroelectric power
plant of Cleuson-Dixence and the reconnaissaince gallery for the Lotschberg Tunnel in
the Swiss Alps. New sampling techniques were developed to ensure as little disturbance
as possible. To the authors knowledge, these are the best set of published triaxial tests
on rock mass available. The quartzitic sandstone had varying degrees of tectonic
crushing. Habimana et al (2002) classified the rocks used into four GSI groupings (GSI
= 15, 25, 50 and 80). Figure 6.22 shows the triaxial test results for each GSI data set.
The samples were tested perpendicular to anisotropic planes (if any) to induce rock
matrix failure. Habimana et al (2002) noted that the degree of anisotropy decreased with
an increase in tectonic crushing. At a low GSI the material was considered isotropic.
Each data set was analysed statistically using the Hoek-Brown equation. Figure 6.23
shows the Habimana et al (2002) data together with the authors statistical fits. During
this analysis the loss function used in the statistical analysis had to be modified to allow
the solution to converge and to avoid local minimums outside the bounds of the Hoek-
Brown parameters i.e.
0 s 1
0 m 40
0 1
The results of the statistical analyses are shown in Table 6.15.
Table 6.15. Results of statistical analysis of Habimana et al (2002) test data
GSI m s
c

(MPa)
Variance explained
(%)
15 2.46 0 0.84 16 93.4
25 3.9 0.016 0.65 16 99.3
50 14 0.10 0.62 16 99.996
80 20 0.7 0.55 16 93.3

Table 6.15 was used together with the discussion and models in the preceding section to
develop equations for the parameters s
b
,
b
and m
b
.
6.5.2.1 A New Equation for m
b

The results for the parameter m
b
showed that a linear equation gave the best fit to the
data. The best-fit equation (variance explained = 95.5%) was:

4
GSI
m
b
(6.19)
If this line was extrapolated to GSI = 100 then m
b
= 25. The equation was therefore
rewritten as:

100
GSI
m m
i b
(6.20)
Figure 6.24 shows this equation together with the data used.
A lower limit of m
b
was set at 2.5 as indicated from the analysis of rockfill. Thus the
equation becomes:

'
5 . 2
100
max
GSI
m
m
i
b
(6.21)
Page 6.52
0
20
40
60
80
100
0 5 10 15 20
3
(MPa)
1
GSI = 15
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 25
0
20
40
60
80
100
0 5 10 15 20
3
(MPa)
1
GSI = 50
0
20
40
60
80
100
0 5 10 15 20
3
(MPa)
1
GSI = 80

Figure 6.22. Test results for tectonised quartzitic sandstone (Habimana et al, 2002)
Page 6.53
0
20
40
60
80
100
0 5 10 15 20
3
(MPa)
1
GSI = 15
0
20
40
60
80
100
0 5 10 15 20
3
(MPa)
1
GSI = 25
0
20
40
60
80
100
0 5 10 15 20
3
(MPa)
1
GSI = 50
0
20
40
60
80
100
0 5 10 15 20
3
(MPa)
1
GSI = 80

Figure 6.23. The authors statistical fits to Habimana et al (2002) data
The linear relationship is different in form to the exponential Hoek-Brown relationship:

( )
,
_
28
100
exp
GSI
m m
i b
(6.22)
The author does not see this as a problem as:
Frictional strength may not decrease as rapidly with GSI as indicated by the Hoek-
Brown equation for m
b
.
is no longer almost constant and as it is directly related to m
b
, the Hoek-Brown
equation will not be appropriate.
An exponential relationship is not supported by the data.
0
5
10
15
20
25
0 20 40 60 80 100
GSI
m
b

Figure 6.24. m
b
versus GSI
6.5.2.2 A New Equation for s
b

The results from the statistical analysis of the Habimana et al (2002) data showed a
rapid decrease of s with a decrease in GSI as predicted in the discussion in Section
6.5.1.3. Therefore an exponential relationship, similar to the Hoek-Brown relationship
for s, was deemed appropriate. One issue raised from the test results was the relatively
high value for s for a GSI of 80. This result was taken to indicate the possibility of a
plateau in the value of s at high GSI. This can be justified by the argument that at high
GSI (>85) the rockmass is very strongly interlocked and thus the cohesive strength is
similar to that of intact rock. This is supported by the GSI table by Hoek (1999) that
includes a new row for intact or massive rock that has a GSI ranging from 80 upwards
for very good defect quality (Figure 6.25).

Figure 6.25. GSI for an intact or massive rock structure (Hoek, 1999)

A statistical analysis of the data in Table 6.15 using an exponential curve together with
a plateau (maximum) of s = 1 for GSI>85 was performed and produced the following
equation (variance explained = 99.97%):

( )
'
,
_
1
15
85
exp
min
GSI
s
b
(6.23)
This relationship is plotted on Figure 6.26 together with the data used.

0
0.2
0.4
0.6
0.8
1
0 20 40 60 80 100
GSI
s
b

Figure 6.26. s
b
versus GSI
6.5.2.3 A New Equation for
b

The exponent,
b
, varies with GSI and m
b
. The limits, as discussed in Section 6.5.1.1,
on
b
are:
GSI = 100,
b
=
i

GSI 0,
b
0.90
A statistical analysis of the data using these limits produced the following equation
(variance explained = 83.3%).
( )
,
_

+
i
b
i i b
m
m 30 75
exp 9 . 0 (6.24)
Figure 6.27 shows the relationship for
b
together with the data. It should be noted that
for this analysis m
i
was taken as that extrapolated from the equation derived earlier for
m
b
. The exponent for intact rock,
i
, was also estimated from the test data.
0
0.2
0.4
0.6
0.8
1
0 20 40 60 80 100
GSI
b

Figure 6.27.
b
versus GSI
Figure 6.28 shows the relationship plotted on a m versus curve (Figure 6.29 shows an
example of the transition from intact rock to rock mass). Plotted on this curve is a
relationship between m
i
and
i
(Equation 6.25). This relationship is a slight
modification to that derived in Chapter 3 (Equation 6.26). This was done to allow the
curve to pass through = 1 when m
i
= 0.

,
_
+
+
7
exp 1
2 . 1
4 . 0
i
i
m
(6.25)

,
_
+
+
455 . 7
exp 1
08585 . 1
4032 . 0
i
i
m
(6.26)
Figure 6.30 shows that the modification does not affect the results significantly.
Figure 6.27 and Figure 6.28 show that the increase in occures rapidly over a short
range of m
b
. This increase starts at approximately m
i
/4 or GSI 20-25. Interestingly this
is very similar to the Hoek et al (1995) relationship for a where a remains constant for
GSI approximately 25-100 and increases below a GSI of 25.
Page 6.58
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 5 10 15 20 25 30 35 40
m
m
i
= 40
m
i
= 20
6
m
i
= 10
4
( )

,
_

+
i
b
i i b
m
m 30 75
exp 9 . 0
,
_
+
+
7
exp 1
2 . 1
4 . 0
i
i
m

Figure 6.28. Graphical representation of the equations for and m
Page 6.59
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 5 10 15 20 25 30 35 40
m
Intact sample
GSI = 100
m
i
,
i
Mass sample
m
b
= function of GSI
i
= function of m
b
(GSI)
Transition curve from
GSI =100 to GSI = 0
Rock mass limit
GSI = 0
Relationship between m
i
and
i
for intact rocks

Figure 6.29. Transition of and mfrom intact rock to rock mass
Page 6.60
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 5 10 15 20 25 30 35 40 45 50
m
i
i
ai
ainew
inew
,
_
+
+
455 . 7
exp 1
08585 . 1
4032 . 0
i
i
m
,
_
+
+
7
exp 1
2 . 1
4 . 0
i
inew
m

Figure 6.30. Original and modified relationships for
i
and m
i

6.5.2.4 The Overall Equation
Figure 6.32 shows the original plots of data together with curves developed using the
preceding equations. These curves provide a good fit to the data. For a GSI of 50 there
is a slight underestimation of the strength and for a GSI of 25 there is a slight
overestimation of the strength. A look at the original test curves presented by Habimana
et al (2002) (Figure 6.31) shows that there is a larger gap between GSI = 25 and 50 than
could be expected for a regular transition from an intact rock to a weak rockmass.
Habimana et al (2002) claim that their estimates of GSI are within t5. It may be that the
gap is due to a slight error in the estimation of GSI.

Figure 6.31. Shear strength curves for tectonised quartzitic sandstone (Habimana
et al, 2002)
A final global statistical analysis of all the data was performed using the general Hoek-
Brown criterion together with the equations developed in this section. This was carried
out to check whether any errors were introduced due to each parameter equation being
derived seperately and then being recombined into the Hoek-Brown equation. Figure
6.33 shows the plots for GSI = 15, 25, 50 and 80 provide good fits to the data (variance
explained 95.6%).
Figure 6.34 and Figure 6.35 show non-dimensional example plots of the shear strength
criterion for an m
i
of 40 and an m
i
of 4 respectively for GSI varying from 10 to 100.
Figure 6.36 and Figure 6.37 compare the authors modified criterion with that of the
current Hoek-Brown criterion for an m
i
of 40 and an m
i
of 4 respectively. The plots
show that the modified criterion gives a higher strength in the compressive region for an
m
i
of 4. For an m
i
of 40, the modified criterion gives a lower strength for a GSI of 100, a
similar strength for a GSI of 80 and a higher strength for a GSI of 10. There is a larger
relative drop in strength between GSI 100 and 80 for the Hoek-Brown criterion
compared with the modified criterion. This is in accordance with the discussion in
Section 6.5.2.2. It should be noted that the two criteria use different methods of
approximating m
i
.
Page 6.63
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 15
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 25
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 50
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 80

Figure 6.32. Results of analysis of Habimana et al (2002) data using the new equation and parameters from equations
Page 6.64
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 15
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 25
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 50
0
20
40
60
80
100
0 5 10 15 20 25
3
(MPa)
1
GSI = 80

Figure 6.33. Results of global analysis of Habimana et al (2002) data using new equations
Page 6.65
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
3
/
c
1
/
c
GSI = 10
GSI = 40
GSI = 60
GSI = 80
GSI = 100
,
_
s
m
ci ci ci
3 3 1

i
= 40 and varying GSI
Page 6.66
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
-0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
3
/
c
1
/
c
GSI = 10
GSI = 40
GSI = 60
GSI = 80
GSI = 100
,
_
s
m
ci ci ci
3 3 1

i
= 4 and varying GSI
Page 6.67
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
-0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
3
/
c
1
/
c
Hoek 2002
Author's criterion
,
_
s
m
ci ci ci
3 3 1

Figure 6.36. Comparison of the authors criterion and the Hoek-Brown criterion (Hoek, 2002) for m
i
= 40
Page 6.68
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
-0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
3
/
c
1
/
c
Hoek 2002
Author's criterion
,
_
s
m
ci ci ci
3 3 1

Figure 6.37. Comparison of the authors criterion and the Hoek-Brown criterion (Hoek, 2002) for m
i
= 4
6.5.3 Summary of Method
The basic form of the shear strength equation remains unchanged from the Hoek-Brown
criterion.

,
_
+ s
m
ci
ci
3
3 1
(6.27)
For intact rock m = m
i
and =
i
. These should preferably be measured from triaxial
tests on intact rock samples. Alternatively an approximation can be made using the
uniaxial compressive strength,
ci
, and tensile strength,
ti
, of the intact rock and the
equations below.

ti
ci
i
m
(6.28)

,
_
+
+
7
exp 1
2 . 1
4 . 0
i
i
m
(6.29)
The estimation of m
b
,
b
and s
b
can be made using the following equations:

'
5 . 2
100
max
GSI
m
m
i
b
(6.30)
( )
,
_

+
i
b
i i b
m
m 30 75
exp 9 . 0 (6.31)

( )
'
,
_
1
15
85
exp
min
GSI
s
b
(6.32)
The equations presented by Hoek et al (2002) (Figure 6.6) can be used to estimate the
cohesion, c, and friction angle, , of the rock mass, as the form of the Hoek-Brown
equation has not been changed.
Conclusions and Recommendations Page 7.1
7. CONCLUSIONS AND RECOMMENDATIONS
7.1 CONCLUSIONS
This thesis is divided into two sections. The first section of this thesis (Chapter 2)
describes the creation and analysis of a database on concrete and masonry dam
incidents known as CONGDATA. The second and main section of this thesis (Chapters
3-6) had its origins in the results of Chapter 2 and the general interests of the author. It
was found that failure through the foundation was common in the list of dams analysed
and that information on how to assess the strength of the foundations of dams on rock
masses was limited. This section applies to all applications of rock mass strength such
as the stability of rock slopes.
7.1.1 The Analysis of Concrete and Masonry Dams
The author has collected a far more extensive database on concrete and masonry dam
incidents (CONGDATA) than has previously been attempted
Generally, unlike some failure modes for embankment dams, e.g. internal erosion and
piping, the stability of concrete and masonry dams is analysable and hence can readily
be checked. The major unknowns for these dams lie in the foundation where sliding and
piping failures can occur. It is for this reason that foundation problems (sliding, leakage
and piping) are the main causes of failure to concrete dams. Overtopping tends to play a
bigger part in the failure of masonry dams, possibly reflecting limited understanding of
floods when they were built. It should be noted that the actual failure mode for
overtopping failures was often unknown.
Foundation shear strength is the main cause of failure for dams with rock or unknown
foundations. Shale (interbedded with other sedimentary units) has a greater tendency to
be involved with sliding failure because of the likely presence of weaknesses in the
bedding such as bedding surface shears. Shale and limestone (often interbedded) have a
high incidence of failure. The limestone has a high proportion of accidents generally
due to excessive leakage through dissolution.
An analysis of the water levels at failure show most dams failed at their highest
recorded water level (regardless of the age of the dam). Several of these were only
slightly higher than that recorded previously. The database showed that, where
information was available, most dam failures had some warning that could have
resulted in the warning and evacuation of residents downstream. Often the warning was
a sudden increase in the amount and rate of leakage. The actual volume of leakage was
not as significant a guide. This indicates that although a dam may have performed
satisfactorily in the past, increases in water level (above historical maxima) should be
treated with caution and the dam sufficiently monitored as the water level rises.
Piping is the main cause of failure for concrete and masonry dams with soil
foundations. The alluvial soils have a tendency to pipe under the high gradients
imposed. No gravity dam has been reported to have failed by sliding on alluvial soils.
Piping tends to occur early in a dams life (<5 years, with one exception).
From the data collected it appears that the failed dams suffered from a lack of good
engineering. Very few dams were found with galleries (1 dam); drainage (1 dam);
grout curtains (4 dams); and shear keys (1 dam). The downstream slopes appeared to be
too steep. Six gravity failures had downstream slopes of 0.6:1 (H:V) or less. Failed
dams, particularly gravity dams, were usually located in relatively wide valleys or were
composite sections with earthfill dams. Three-dimensional effects are unlikely to have
contributed any strength in these cases. h
wf
/W ratios ranged from 0.6 to 2.1 with an
average of 1.35.
The author has used the analysis of CONGDATA and a population database to
develop a method for assessing the first order probability of failure of masonry or
concrete gravity dams. The method accounts for dam age, year commissioned and type;
failure mode; foundation geology; height to width ratio; and monitoring and
surveillance. General probabilities of failure for arch, buttress and multi-arch dams,
based on failure and population statistics, are included.
The author cautions that this approach should only be used as a first order
approximation of the annual probabilities of failure. It is clearly very approximate, and
suffers from being based on small numbers of failures, and limited quality data. Where
significant decisions on dam safety are being made, detailed deterministic and/or
probabilistic methods should be used.
Whilst all care has been taken in compiling the data in CONGDATA, it should be
remembered that the information in CONGDATA has come from numerous sources, not
all of which could be validated. The analysis of dams in CONGDATA does not take
into account such things as: surveillance; quality of construction; and quality of
geological description. It is therefore recommended that this work be used in a
qualitative sense only.
7.1.2 The Shear Strength of Intact Rock
An overview of the strength of intact rock has been presented. It was demonstrated that
the method of fitting the criterion to the test data has a major effect on the estimates
obtained of the material properties. The results of a recent analysis of a large database
of test results demonstrated that there are inadequacies in the Hoek-Brown empirical
failure criterion as currently proposed for intact rock and, by inference, as extended to
rock mass strength. The parameters m
i
and
c
are not material properties if the exponent
is fixed at 0.5. Published values of m
i
can be misleading as m
i
is not related to rock type.
The Hoek-Brown criterion can be generalised by allowing the exponent to vary. As
expected, this change resulted in a better model of the experimental data. The most
accurate method of estimating m
i
and is through using triaxial tests on intact rock.
The recommended method for regression of the data is modified least squares, Equation
7.1, combined with the extended formulation of the generalised Hoek-Brown criterion,
Equation 7.2. The equations are repeated below.

( )
( )

;
>
3 1 3 3
3 1 1 1

i
m
(7.1)

>
,
_
+
i c
i c
c
i
c
m
m
m

3 3 1
3
3
3 1
for
for 1
(7.2)
Analysis of individual data sets indicated that the exponent, , is a function of m
i
which
is, in turn, closely related to the ratio of
c
/
t
. A regression analysis of the entire
database provided a model to allow the triaxial strength of an intact rock to be estimated
from a reliable measurement of its uniaxial tensile and compressive strengths. The
method proposed is the most accurate of those methods that do not require triaxial
testing and is adequate for preliminary analysis. An analysis was presented that showed
applying the Hoek-Brown criterion to most rocks results in systematic errors. Simple
relationships for triaxial strength that are adequate for slope design were presented.
7.1.3 The Shear Strength of Rockfill
A general overview of the shear strength of rockfill is presented. An analysis of a large
database of test results was used to develop two new shear strength equations, one
relating the secant friction angle and normal stress (Equation 7.3) and the other the
principal stresses (Equation 7.4). The parameters can be found using equations 4.21-
4.23 and 4.25-4.30 in Chapter 4 respectively. The equation for principal stresses
provided a much better fit to the data and is recommended. The equations presented
effectively give the mean strengths of the data. Graphs are provided showing the range
of strengths and the affect of various parameters on the shear strength of rockfill. Of the
parameters statistically investigated, the unconfined compressive strength, particle
angularity, fines content, maximum particle size and void ratio were found to have the
most significant effect on the shear strength of rockfill.

c
n
b a + (7.3)

3 1
RFI (7.4)
7.1.4 Empirical Slope Design
A review of current empirical methods of slope design using rock mass characterisation
has also been presented. The findings highlighted the lack of well tested methods.
Current slope design methods were based on limited databases with no failures and
slopes of limited height. Many of the methods were incorrectly advocated for
structurally controlled slopes. The author has presented new slope design curves, based
on slopes that have had rock mass failure components, that can be used for preliminary
slope design.
7.1.5 The Shear Strength of Rock Masses
The Hoek-Brown criterion is the most commonly used strength criterion for rock mass
and has thus been the main subject of this section of the thesis. The author has
examined the appropriateness of the equation for predicting strengths at the two limits
of rock mass (intact rock and rockfill). A study of the strength of intact rock using a
large database of triaxial tests shows that the exponent, a, in the Hoek-Brown criterion
should vary from about 0.2 to 0.9. The study of a database of 988 triaxial tests on
rockfill shows that, if it is assumed a rockfill is representative of a very poor quality
rock mass, the exponent, a, should approach 0.9 to 0.95 (with m
i
approximately 2.4-2.7)
as GSI approaches zero. The current Hoek-Brown criterion assumes for most rock
masses a is 0.5 and limits a to approximately 0.62 for a very poor quality rock mass. A
problem with simply modifying a is that a and m
i
(or m
b
)

are interrelated.
The author has developed a new method of determining the parameters in the Hoek-
Brown criterion to overcome these problems. It is strongly suggested that the intact rock
parameters m
i
& a
i
should be obtained using triaxial testing and statistical methods
discussed in Chapter 3. The author has provided approximate methods of determining
m
i
and a
i
where no triaxial test results are available. Equations have been derived for
rock mass to address the limit (a
b
0.95, m
b
2.5, s
b
=0) of very poor quality rock
masses. The equations developed allow for a reduction in m
b
from m
i
(and associated
increase from a
b
from a
i
) and s
b
from s
i
to this limit. A summary of the method is
presented below.
The basic form of the shear strength equation remains unchanged from the Hoek-Brown
criterion.

,
_
+ s
m
ci
ci
3
3 1
(7.5)
For intact rock m = m
i
and =
i
. These should preferably be measured from triaxial
tests on intact rock samples. Alternatively an approximation can be made using the
uniaxial compressive strength,
ci
, and tensile strength,
ti
, of the intact rock and the
equations below.

ti
ci
i
m
(7.6)

,
_
+
+
7
exp 1
2 . 1
4 . 0
i
i
m
(7.7)
The estimation of m
b
,
b
and s
b
can be made using the following equations:

'
5 . 2
100
min
GSI
m
m
i
b
(7.8)
( )
,
_

+
i
b
i i b
m
m 30 75
exp 9 . 0 (7.9)

( )
'
,
_
1
15
85
exp
min
GSI
s
b
(7.10)
The equations presented by Hoek et al (2002) can be used to estimate the cohesion, c,
and friction angle, , of the rock mass, as the form of the Hoek-Brown equation has not
been changed.

7.2 RECOMMENDATIONS FOR FURTHER RESEARCH
7.2.1 The Analysis of Concrete and Masonry Dams
The analysis of CONGDATA and the method for predicting probabilities of failure of
concrete and masonry dams is based on field data of varying quality. Further detailed
analysis of new incidents would improve the confidence in the conclusions presented in
this thesis. Further detailed information on the geology of the foundations of dams
would allow a better prediction of the likelihood of failure. Research into the
effectiveness of monitoring and warning systems using the outcomes from this thesis
would be of value.
The author believes that it is better to do a probabilistic analysis of stability modelling
uncertainty in the geology, shear strengths of the foundation and foundation pore
pressures (uplift) as modified by grouting and drainage.
Where large defects exist below a dam the shear strength in the foundation will be
governed by these defects. As discussed in Section 7.2.2 the shear strength of field scale
defects is still poorly understood. Further work in this area is required. Studies could
examine large-scale failures, preferably with insitu shear tests and laboratory scale shear
tests for comparison. However, these would be limited in number and quality.
Alternatively studies could use numerical modelling to look at the effect of increasing
the scale of defects. The use of the program PFC which models the movement and
interaction of circular particles by the distinct element method has shown some promise
in this area.
7.2.2 The Shear Strength of Rock Masses
The methods presented for estimating the shear strength of intact rock and rockfill are
based on substantial databases. These provide good bounds on the shear strength of rock
masses. The development of the equations for estimating the strength of the transitional
rock masses is based on a limited amount of field and laboratory data. Further analysis
and reporting of well-documented failures and lab testing of rock masses of varying
quality would improve the confidence in the results presented in this thesis and would
also provide a better understanding of the degree of uncertainty in the results obtained
by the equations presented in this thesis. The publishing of more data on failures would
also assist in improving the slope design curves presented in this thesis.
The equations for rock masses provided in this paper are principally for cohesionless
rock masses. Further work could be carried out to assess cohesive rock masses.
Modifications to the parameter s based on cohesive properties of the rock mass would
allow the Hoek-Brown criterion to better model these types of rock masses at low
confining stresses. This would be of value for predicting strengths for pit slope benches.
The effect of the intermediate principal stress,
2
, could be incorporated into the
equations for intact rock and ultimately rock masses. The use of Lades (1993) work
would be of benefit here.
Appendix Page A.1
APPENDIX A: CONGDATA DATABASE
The full CONGDATA database is contained on the accompanying CD-ROM.
APPENDIX B: DAM LIST - FAILURES
Table B1. Dam list - failures
Dam Name Country Dam Type Year
Commissioned
Year
Failed
Hlf
(m)
Kohodiar India PG/TE 1963 1983 36
Zerbino Italy PG 1925 1935 16
Mohamed V Morocco PG 1966 1963 62
Torrejon-Tajo Spain PG 1967 1965 62
Xuriguera Spain PG 1902 1944 42
Bayless (A) USA PG 1909 1910 17
Bayless (B) USA PG 1909 1911 17
Elwha River USA PG 1912 1912 51
Hauser Lake II USA PG 1911 1969 40
St Francis USA PG 1926 1928 62
Cheurfas Algeria PG(M) 1884 1885 42
Fergoug I Algeria PG(M) 1871 1881 43
Fergoug II Algeria PG(M) 1885 1927 43
Habra (A) Algeria PG(M) 1871 1872 40
Habra (B) Algeria PG(M) 1872 1881 40
Habra (C) Algeria PG(M) 1881 1927 40
Sig Algeria PG(M) 1858 1885 21
Bouzey France PG(M) 1881 1895 26
Chickahole India PG(M) 1966 1972 30
Khadakwasla India PG(M) 1879 1961 33
Kundli India PG(M) 1924 1925 45
Pagara India PG(M) 1927 1943 30
Tigra India PG(M) 1917 1917 28
Santa Catalina Mexico PG(M) 1900 1906 15
Granadillar Spain PG(M) 1930 1933 22
Puentes Spain PG(M) 1791 1802 69
Elmali I Turkey PG(M)/TE 1892 1916 23
Angels USA PG(M) 1895 1895 16
Austin (A) USA PG(M) 1893 1900 21
Lower Idaho Falls USA ER/PG(M) 1914 1976 15
Appendix Page A.2
Dam Name Country Dam Type Year
Commissioned
Year
Failed
Hlf
(m)
Lynx Creek USA PG(M) 1891 1891 15
Komoro Japan CB 1927 1928 16
Selsford Sweden CB/TE 1943 1943 21
Ashley USA CB 1908 1909 18
Overholser USA CB 1920 1923 17
Vega de Tera Spain CB(M) 1956 1959 35
Austin (B) USA CB(M) 1915 1915 20
Stony River USA CB(M) 1913 1914 15
Gleno Italy MV 1923 1923 35
Leguaseca Spain MV 1958 1987 20
Malpasset France VA 1954 1959 66
Moyie River USA VA 1924 1926 16
Vaughn Creek USA VA 1926 1926 20
Meihua China VA(M) 1981 1981 22
Bacino di Rutte Italy VA(M) 1952 1965 15
Gallinas USA VA(M) 1910 1957 32

Appendix Page A.3
APPENDIX C: DAM LIST - POPULATION OF DAMS
Table C1. Dam list - USBR population
Dam Name Type Year
Commissioned
Height
(m)
Altus PG 1945 33.5
American Falls PG 1927 31.5
Angostura PG 1949 58.8
Black Canyon PG 1924 55.8
Brantley PG 1988 33.5
Camp Dyer PG 1929 24.1
Canyon Ferry PG 1954 68.6
Elephant Butte PG 1916 91.7
Folsom PG 1956 103.6
Friant PG 1942 97.2
Grand Coulee PG 1942 167.6
Jackson Lake PG 1911 20
Keswick PG 1950 47.9
Kortes PG 1951 74.4
Marshall Ford PG 1942 84.7
Nimbus PG 1955 26.5
Olympus PG 1949 21.3
Savage Rapids Diversion PG 1921 13.1
Shasta PG 1945 183.5
Upper Stillwater PG 1988 88.4
Yellowtail Afterbay PG 1965 21.9
Bartlett CB/MV 1939 94
Coolidge (BIA) CB/MV 1928 75.9
Minidoka CB 1906 26.2
Pueblo CB 1975 76.2
Red Bluff Diversion CB 1963 15.8
Stony Gorge CB 1928 42.4
Thief Valley CB 1932 22.3
Anchor VA 1960 63.4
Arrowrock VA 1915 106.7
Buffalo Bill VA/PG 1910 106.7
Clear Creek VA 1914 25.6
Crystal VA 1976 98.5
Deadwood VA 1931 50.3
East Canyon VA 1966 79.2
Appendix Page A.4
Dam Name Type Year
Commissioned
Height
(m)
East Park VA 1910 42.4
Flaming Gorge VA 1964 153
Gerber VA 1925 26.8
Gibson VA 1929 60.7
Glen Canyon VA 1964 216.4
Hoover VA 1936 221.4
Horse Mesa VA 1927 93
Hungry Horse VA 1953 171.9
Monticello VA 1957 92.7
Mormon Flat VA 1926 68.3
Morrow Point VA 1968 142.6
Mountain Park VA 1975 40.5
Nambe Falls VA 1976 45.7
Owyhee VA/PG 1932 127.1
Parker VA 1938 97.5
Pathfinder VA(M) 1909 65.2
Santa Cruz VA 1929 46
Seminoe VA 1939 89.9
Stewart Mountain VA 1930 63.1
Swift VA 1967 62.5
Theodore Roosevelt VA(M) 1911 108.5
Warm Springs VA 1919 32.3
Wild Horse VA 1967 33.5
Yellowtail VA 1966 160

Appendix Page A.5
Table C2. Dam list - Australia/New Zealand population
Dam Name Type Year
Commissioned
Height
(m)
Bendora VA 1961 47
Cotter PG 1915 31
Lower Molongolo PG 1994 32
Scrivener PG 1963 33
Wrights PG 1989 16
Avon PG 1927 72
Back Creek VA 1937 15
Borenore Creek VA 1928 18
Bundanoon VA 1960 35
Burrinjuck PG 1928 93
Captains Flat PG 1939 19
Carcoar VA 1970 58
Cataract PG 1907 56
Chichester PG 1923 44
Coeypolly Creek No I VA 1932 19
Cordeaux PG 1926 67
Crookwell PG 1937 16
Danjera CB 1971 36
Deep Creek PG 1961 21.3
Dunn Swamp VA 1930 16
Flat Rock Creek VA 1933 16
Fountaindale VA 1915 15
Glenquarry Cut PG 1974 18
Greaves Creek VA 1942 19
Guthega PG 1955 33.5
Happy Jack PG 1959 76.2
Hume PG 1936
Ingleburn MV 1933 16
Island Bend PG 1965 48
Junction Reefs MV 1897 19
Keepit PG 1960 55
Lake Medlow VA 1907 21
Lake Rowlands CB 1953 25
Lithgow No 2 VA 1907 26
Loyalty Road PG 1995 30
Maldon Weir PG 1968 20
Manly PG 1892 20
Medway VA 1964 25
Middle Cascade (No 1) VA 1915 15
Molong PG 1987 16
Appendix Page A.6
Dam Name Type Year
Commissioned
Height
(m)
Mooney Upper VA 1961 28
Moore Creek VA 1898 19
Murray 2 VA 1968 42.7
Nepean PG 1935 82
Oaky River PG 1956 18
Oberon CB 1949 35
Parramatta VA(M) 1857 15
Porters Creek PG 1968 18
Puddledock Creek VA 1928 19
Redbank Creek VA 1899 15
Rylstone VA 1953 20
Suma Park VA 1962 35
Tallowa PG 1976 43
Tantangara PG 1960 45.1
Timor VA 1961 22
Tumut 2 PG 1961 46.3
Tumut 3 Pipeline PG 1971 34.7
Tumut Pond VA 1959 86.3
Umberumberka PG 1914 41
Upper Cordeaux No 2 VA 1915 22
Warragamba PG 1960 142
Warragamba Weir PG 1940 21
Wellington VA 1933 15
Winburndale PG 1936 22
Woodford Creek VA 1928 16
Woronora PG 1941 74
Wyangla PG 1971 85
Atiamuri PG 1958 46
Aviemore PG 1968 57
Clyde PG 1993 105
Lake Onslow VA 1982 17
Mangahao No. 1 PG 1926 36
Mangahao No. 2 PG 1924 32
Marslin VA 1982 19
Roxburgh PG 1956 70
Waihopai VA 1927 34
Waitaki PG 1934 37
Whakamaru PG 1956
Beardmore PG 1972 17
Boggabilla Weir PG 1991 16
Burdekin Falls PG 1987 55
Burton Gorge PG 1992 34
Appendix Page A.7
Dam Name Type Year
Commissioned
Height
(m)
Cedar Pocket PG 1984 20
Chinaman PG 1993 19
Cooloolabin PG 1979 20
Copperfield PG 1984 40
Dumbleton PG 1992 15
Ibis PG 1906 16.5
Julius MV 1976 38
Koombooloomba PG 1961 52
Kroombit PG 1992 23
Lake Manchester PG 1916 38
Leslie PG 1965 33
Little Nerang PG 1961 47
Moogerah VA 1961 37
North Pine PG 1975 46
Rifle Creek VA 1929 21
Somerset PG 1955 50
Theresa Creek PG 1982 19
Tinaroo Falls PG 1958 47
Greenstone Ck Dam VA 1969 20
Wappa PG 1961 20
Wuruma PG 1969 46
Aroona PG 1955 26.2
Barossa VA 1902 36
Beetaloo PG 1890 31
Clarendon Weir PG(M) 1896 15
Middle River PG 1968 20
Mount Bold VA 1938 58
Myponga VA 1962 52
Sturt VA 1966 41
Ullabidinie PG 1914 22
Ulbana PG 1911 11.1
Warren PG 1916 26
Yeldulknie PG 1913 17
Bowden 1984 18
Catagunya PG 1962 49
Clark VA 1949 67
Cluny PG 1967 30
Craigbourne PG 1986 25
Devils Gate VA 1969 84
Gordon VA 1974 140
Henty PG 1988 23
Lake Margaret PG 1918 17
Appendix Page A.8
Dam Name Type Year
Commissioned
Height
(m)
Liapootah PG 1960 40
Meadowbank CB 1966 43
Mount Paris CB 1936 18
Pine Tier PG 1953 39
Repulse VA 1968 42
Ridgeway VA 1919 59
Trevallyn PG 1954 33
Clover CB 1956 20
Dartmouth PG 1980 25
Evansford PG 1887 17
Glenmaggie PG 1927 37
Goulburn Weir PG 1891 15
Hume Weir 1919
Junction CB 1945 26
Lauriston CB 1941 33
Lower Stoney Creek PG 1875 21
Maroondah PG 1927 46
Mt Cole PG 1903 28
Nicholson River CB 1976 16
Rocklands PG 1953 28
Swingler PG 1977 18
Yallourn Storage CB 1961 21
Canning PG 1940 70
Conjurunup PG 1992
Harvey PG 1916 24
Kununurra Diversion 1963 20
Mundaring PG 1902 71
New Victoria PG 1991 52
Serpentine Pipehead PG 1957 16
Wellington PG 1933 37

Appendix Page A.9
Table C3. Dam list - Portugal population
Dam Name Type Year
Commissioned
Height
(m)
Alto Cavado PG 1964 29
Alem da Fazenda PG 1967 20
Carrapatelo PG 1972 57
Corgas PG 1991 25
Cova do Viriato PG 1962 28
Fratel PG 1973 43
Monte Novo PG 1982 30
Penha Garcia PG 1980 25
Pocinho PG 1982 49
Raiva PG 1981 36
Ranhados PG 1986 41
Regua PG 1973 42
Torrao PG 1988 70
Touvedo PG 1996 43
Valeira PG 1975 48
Gameiro PG/TE 1960 20
Andorinhas PG(M) 1945 25
Burgaes PG(M) 1940 30
Covao do Ferro PG(M) 1956 35
Freigil PG(M) 1955 17
Guilhofrei PG(M) 1938 49
Idanha PG(M) 1949 54
Lagoa Comprida PG(M) 1958 29
Poio PG(M) 1932 18
Povoa PG(M) 1928 32
Vale do Rossim PG(M) 1956 27
Penide PG(M) 1951 15
Caia CB/PG/TE 1967 52
Roxo CB/PG/TE 1968 49
Miranda CB 1961 80
Pracana CB 1951 60
Odivelas MV/TE 1972 55
Aguieira MV 1981 89
Alto Lindoso VA 1993 110
Bravura VA 1958 41
Cabril VA 1954 136
Caldeirao VA 1996 39
Appendix Page A.10
Dam Name Type Year
Commissioned
Height
(m)
Fagilde VA 1984 27
Fronhas VA 1984 62
Funcho VA 1991 49
Picote VA 1958 100
Varosa VA 1976 76
Vilarinho das Furnas VA 1972 94
Alto Rabagao VA/PG 1964 94
Bemposta VA/PG 1964 87
Castelo do Bode VA/PG 1951 115
Covao do Meio VA/PG 1953 25
Venda Nova VA/PG 1951 97
Alto Ceira VA 1949 36
Bouca VA 1955 65
Canicada VA 1955 76
Salamonde VA 1953 75
Santa Luzia VA 1942 76
Appendix Page A.11
APPENDIX D: CAUSES OF INCIDENTS
Table D1. Causes of incidents - all dams
Cause Fail Acc. Major
Repairs
Total Cause Fail Acc. Major
Repairs
Total
1.1.1 1 2 3 3.1.2 1 1 2
1.1.2 1 8 4 13 3.1.3 3 1 2 6
1.1.3 5 5 2 12 3.1.4 2 3 9 14
1.1.4 7 16 7 30 3.1.5 5 1 6
1.1.5 6 13 1 20 3.1.9 2 1 1 4
1.1.5.1 4 2 6 3.1.12 1 2 3
1.1.5.2 1 1 2 3.2 1 1
1.1.6 1 1 3.2.2 3 1 22 26
1.1.8 1 2 3 3.2.3 9 9
1.1.9 1 1 2 3.2.5 3 3
1.1.10 1 1 3.2.6 4 1 5
1.1.11 4 3 7 3.2.7 4 1 5
1.1.12 3 4 7 3.2.8 3 10 13
1.1.14 1 1 3.2.9 3 2 1 6
1.2.1 6 13 19 3.2.10 3 3
1.2.2 1 6 22 29 3.3.2 1 1 2 4
1.2.3 1 6 53 60 3.4.1 5 5
1.2.5 1 1 2 3.4.2 8 6 1 15
1.2.6 1 1 3.4.3 1 1
1.2.7 1 1 3.4.4 4 4
1.2.8 9 22 31 3.4.5 1 1
1.2.9 9 17 26 3.4.6 10 10
1.2.10 3 4 7 3.5.1 2 2 4
1.2.11 7 13 20 3.5.2 5 3 1 9
1.2.13 1 2 3 3.5.3 1 1
1.3.1 1 4 5 3.5.4 1 1 2
1.3.2 4 3 15 22 3.5.5 1 3 4
1.3.3 6 1 7 3.7.2 1 1
1.3.4 4 28 32 4.1.5 2 2
1.3.5 5 16 21 4.1.8 0
1.3.7 3 1 4 4.2.1 1 1
1.3.7.2 1 1 4.2.2 1 1
1.3.7.3 1 1 4.2.3 6 6
1.4.1 1 1 2 4.2.4 1 1
1.4.2 8 8 4.2.5 1 1 2
Appendix Page A.12
Repairs
Repairs
Total
1.4.3 1 1 2 4.2.6 1 1
1.4.4 1 1 2 4.2.7 1 2 3
1.4.6 1 2 3 4.2.8 4 1 5
1.4.7 2 3 5 4.2.9 1 2 3
1.5.1 1 2 2 5 4.2.10 1 1
1.5.2 2 4 6 12 4.2.12 3 13 16
1.5.4 2 3 5 4.2.13 2 12 14
1.5.5 2 2 4.4.2 1 1
1.5.6 1 1 4 6 4.4.3 1 1
1.6.1 2 2 4.4.4 3 1 4
1.7.1 1 1 4.5.1 1 1
1.7.2 8 8 4.5.5 4 4
2.3.9 5 5 4.6 1 10 5 16
4.6.1 1 1 4.9.1 2 2 4
4.6.2 1 1 2 4.9.2 2 2 4
4.6.3 1 1 4.11.1 1 1
4.7.1 6 13 10 29 4.11.6 10 6 16
4.7.2 4 1 5 4.11.7 1 4 5
4.7.3 1 1 4.12.6 1 1
4.7.4 1 1 5.1 5 5
4.7.6 1 1 2 5.3 4 1 5
4.7.7 1 1 5.4 9 2 11
4.7.8 4 9 13 6.1 1 1
4.7.9 1 3 4 6.2 5 2 7
4.8 2 16 6 24 Total 121 283 450 854

Appendix Page A.13
Table D2. Causes of incidents - PG dams
Repairs
Repairs
Total
1.1.1 1 1 4.1.5 2 2
1.1.2 2 2 4 4.1.8 0
1.1.3 4 3 1 8 4.2.1 1 1
1.1.4 4 7 5 16 4.2.2 1 1
1.1.5 1 8 1 10 4.2.3 6 6
1.1.5.1 1 2 3 4.2.4 1 1
1.1.5.2 1 1 4.2.5 1 1 2
1.1.6 1 1 4.2.7 1 2 3
1.1.9 1 1 4.2.8 1 1 2
1.1.10 1 1 4.2.9 1 1
1.1.11 1 3 4 4.2.12 1 13 14
1.1.12 4 4 4.2.13 11 11
1.2.1 8 8 4.4.3 1 1
1.2.2 1 15 16 4.4.4 1 1
1.2.3 1 40 41 4.5.1 1 1
1.2.5 1 1 4.5.5 4 4
1.2.7 1 1 4.6 1 7 4 12
1.2.8 2 15 17 4.6.2 1 1
1.2.9 2 11 13 4.7.1 2 5 10 17
1.2.10 1 3 4 4.7.2 1 1 2
1.2.11 1 7 8 4.7.3 1 1
1.3.1 4 4 4.7.4 1 1
1.3.2 2 2 15 19 4.7.6 1 1
1.3.4 1 6 7 4.7.7 1 1
1.3.5 1 2 3 4.7.8 9 9
1.3.7 1 1 4.7.9 2 2
1.4.1 1 1 4.8 4 3 7
1.4.6 2 2 4.9.1 1 1 2
1.5.1 1 1 2 4 4.9.2 2 2
1.5.2 1 1 3 5 4.11.1 1 1
1.5.4 1 1 4.11.6 5 4 9
1.5.5 2 2 4.11.7 4 4
1.5.6 4 4 4.12.6 1 1
1.6.1 2 2 5.1 2 2
1.7.1 1 1 5.3 2 2
1.7.2 8 8 5.4 2 2 4
3.1.4 1 1 6.1 1 1
3.2.2 1 1 6.2 3 1 4
3.2.8 1 1 Total 19 82 263 364

Appendix Page A.14
Table D3. Causes of incidents - PG(M) dams
Repairs
Repairs
Total
1.1.3 1 1 3.2.9 3 2 1 6
1.1.5 1 1 3.2.10 3 3
1.2.2 1 1 3.3.2 1 2 3
1.2.3 1 1 3.4.1 5 5
1.2.8 1 1 3.4.2 7 6 1 14
1.3.1 1 1 3.4.3 1 1
1.3.3 1 1 3.4.4 4 4
1.3.7 1 1 2 3.4.6 8 8
1.4.7 1 1 3.5.1 2 2 4
1.5.6 1 1 3.5.2 5 3 8
2.3.8 1 1 3.5.3 1 1
2.3.9 5 5 3.5.4 1 1
3.1.12 1 2 3 3.5.5 1 3 4
3.1.2 1 1 2 3.7.2 1 1
3.1.3 3 1 1 5 4.2.8 1 1
3.1.4 1 2 8 11 4.6 1 1
3.1.5 4 1 5 4.6.2 1 1
3.1.9 2 1 1 4 4.7.1 2 1 3
3.2 1 1 4.7.8 1 1
3.2.2 3 1 20 24 4.7.9 1 1
3.2.3 8 8 4.8 1 1
3.2.5 2 2 4.9.1 1 1
3.2.6 4 1 5 4.11.6 1 1
3.2.7 4 1 5 5.3 1 1
3.2.8 2 10 12 5.4 2 2
Total 62 39 80 181

References Page Ref.1
REFERENCES
Abel, J.F. and Jowis, J.E. (1979) Concrete shaft lining design. Concrete Shaft Lining Design,
20th U.S. Symposium on Rock Mechanics, Austin, Texas, pp. 627-640.
Abraham, T.J. (1970) Selection and Design of a Compacted Rock Fill Dam with a Slopin g
Earth Core on Foundation Rock with Weak Horizontal Bedding Planes. Dixieme Congres
des Grands Barrages, Montreal, pp. 339-359.
Abrahams, A.D. and Parsons, A.J. (1987) Identification of strength equilibrium rock slopes:
further statistical considerations. Earth Surface Processes and Landforms, 12 (6), pp. 631-
635.
Adachi, T., Ogawa, T. and Hatashi, M. (1981) Mechanical properties of soft rock and rock
mass. Proceedings of the Tenth International Conference on Soil Mechanics and
Foundation Engineering, Stockholm, Balkema. Vol. 1 pp. 527-530.
Akai, K., Yamamoto, K. and Arioka, M. (1970) Experimental research on the structural
anisotropy of crystalline schists. 2nd International Congress on Rock Mechanics, ISRM,
Belgrade, Vol. 2 pp. 181-186.
Aldrich, M.J. (1969) Pore pressure effects on Berea Sandstone subjected to experimental
deformation. Geological Society of America Bulletin, 80 pp. 1577-1586.
Alheid, H.J. (1982) Friction processes on shear surfaces in granite at high pressure and
temperature. High Pressure Researches in Geoscience, Behaviour and Properties of Earth
Materials at High Pressures and Temperatures, (Schreyer ed.) Stuttgart, E.
Schweizerbart'sche Verlagsbuchhandlung. pp. 95-102.
Al-Hussaini, M. (1981) Tensile properties of compacted soils, Laboratory Shear Strength of
Soils, ASTM STP 740, pp. 207-255.
Al-Hussaini, M. (1983) Effect of particle size and strain conditions on the strength of crushed
basalt. Canadian Geotechnical Journal, 20 pp. 706-717.
Alva-Hurtado, J.E., McMahon, D.R. and Stewart, H.E. (1980) Apparatus and techniques for
static triaxial testing of ballast. Laboratory Shear Strength of Soil, ASTM STP 740,
Chicago, American Society for Testing Materials. pp. 94-113.
Anagnosti, P. and Popovic, M. (1982) Evaluation of shear strength for coarse-grained granular
materials. Fourteenth Congres on Large Dams, Rio de Janeiro, ICOLD. Vol. Q55 pp. 753-
767.
ANCOLD (1991) Guidelines on Design Criteria for Concrete Gravity Dams. Australian
National Committee on Large Dams, November, pp. 34.
Anderson, J.G.C. and Trigg, C.F. (1976) Dams and reservoirs. In Case-histories in Engineering
Geology Elek Science, London, pp. 20-45.
Andreev, G.E. (1995) Brittle Failure of Rock Materials Test Results and Constitutive Models.
Balkema, Brookfield, pp. 446.
Araujo, R.G.S., Sousa, J.L.A.O. and Bloch, M. (1997) Experimental investigation on the
influence of temperature on the mechanical properties of reservoir rocks. International
Journal of Rock Mechanics and Mining Sciences, 34 (3-4), pp. 298.
ASCE (1929) Essential facts concerning the failure of the St. Francis Dam - Report of
committee of board of direction. A.S.C.E. Proceedings, 55 (Oct.), pp. 2147-2163.
Attewell, P.B. and Sandford, M.R. (1974) Intrinsic shear strength of a brittle anisotropic rock -
I: Experimental and mechanical interpretation. International Journal for Rock Mechanics,
Mining Sciences and Geomechanical Abstracts, 11 (11), pp. 423-430.
Aversa, S. and Evangelista, A. (1998) The mechanical behaviour of a pyroclastic rock: Yield
strength and "destructuration" effects. Rock Mechanics Rock Engineering, 31 (1), pp. 25-
42.
Ayers, M.E., Thompson, M.R. and Uzarski, D.R. (1989) Rapid shear strength evaluation of in
situ granular materials. Transportation Research Record, 1227 pp. 134-146.
Babb, A.O. and Mermel, T.W. (1968) Catalog of Dam Disasters, Failures and Accidents.
USBR, Washington D.C.
Baker, E.A.. (1897) Failure of the Bouzey dam (Discussion) Procedures of the Institution of
Civil Engineers pp. 68-79.
Baladi, G.Y. and Wu, T.T.H. (1986) Interpretation of triaxial test results of cohesionless soils: a
new model. Advanced Triaxial Testing of Soil and Rock, ASTM STP 977, Louisville,
Kentucky, American Society for Testing and Materials. pp. 567-581.
Balmer, G. (1952) A general analytical solution for Mohr's envelope. Proceedings American
Society for Testing Materials, 52 pp. 1260-1271.
Bandis, C.S. (1993) Engineering properties and characterization of rock discontinuites. In
Comprehensive Rock Engineering (E. T. Brown ed.), Pergamon Press, pp. 155-183.
Bandis, C.S., Lumsden, A.C. and Barton, N.R. (1981) Experimental studies of scale effects on
the shear behaviour of rock joints. International Journal for Rock Mechanics, Mining
Sciences and Geomechanical Abstracts, 18 pp. 1-21.
Banks, D.C. and Maciver, B.N. (1971) Discussion of: Leps (1970) Review of shearing strength
of rockfill. Journal of the Soil Mechanics and Foundations Division Proceedings of the
American Society of Civil Engineers, 97 (SM5), pp. 791-794.
Barla, G., Sharp, J.C. and Rabagliati, U. (1991) Excavation and support optimisation for a large
underground storage facility in weak jointed chalk. 7th ISRM International Congress on
Rock Mechanics, (Herausgeber ed.) Aachen, ISRM. Vol. 2 pp. 1067-1072.
Bartholomew, C.L. (1989) Failure of concrete dams. In 6th ASDSO Annual Conference,
Albuquerque, New Mexico, pp. 428-445.
Barton, N.R. (1971a) Estimation of in situ shear strength from back analysis of failed rock
slopes. Rock Fracture: Proceedings International Symposium Rock Mechanics, Nancy,
France, Vol. 1 pp. II-27.
Barton, N.R. (1971b) A Model Study of the Behaviour of Steep Excavated Rock Slopes. PhD,
University of London (Imperial College of Science and Technology).
Barton, N.R. (1971c) A relationship between joint roughness and joint shear strength. Rock
Fracture: Proceedings International Symposium Rock Mechanics, Nancy, France.
Barton, N.R. (1973) Review of a new shear strength criterion for rock joints. Engineering
Geology, 7 pp. 287-332.
Barton, N.R. (1976) The shear strength of rock and rock joints. International Journal for Rock
Mechanics, Mining Sciences and Geomechanical Abstracts, 13 pp. 255-279.
Barton, N.R. and Bandis, S. (1982) Effects of block size on the shear behavour of jointed rock.
Proceedings of the 23rd U.S. Symposium on Rock Mechanics, Berkelay, California, USA.
Barton, N.R. and Bandis, S. (1990) Review of predictive capabilities of JRC-JCS model in
engineering practice. Rock Joints, (B. &. Stephansson ed.) Leon, Norway, Balkema,
Rotterdam. pp. 603-610.
Barton, N.R. and Choubey, V. (1977) The shear strength of rock joints in theory and practice.
Rock Mechanics, 10 pp. 1-54.
Barton, N. and Kjaernsli, B. (1981) Shear strength of rock fill. Journal of Geotechnical
Engineering Division, pp. 373.
Barton, N., Lien, R. and Lunde, J. (1974) Engineering classification of rock masses for the
design of tunnel support. Rock Mechanics, Vol. 6 pp. 189-236.
Becker, E., Chan, C.K. and Bolton Seed, H. (1972) Strength & Deformation Characteristics of
Rockfill Materials in Plane Strain and Triaxial Compression Tests. Geotechnical
Engineering TE 72-3, University of California, Berkeley.
Behrestaghi, M.H.N., Seshagiri Rao, K. and Ramamurthy, T. (1996) Engineering geological and
geotechnical responses of schistose rocks from dam project areas in India. Engineering
Geology, 44 pp. 183-201.
Bell, F.G. and Jermy, C.A. (2000) The geotechnical character of some South African dolerites,
especially their strength and durability. Quarterly Journal of Engineering Geology and
Hydrogeology, 33 pp. 59-76.
Bellier, J., Londe, P. and Langbein, J. (1976) The Malpasset Dam. In The Evaluation of Dam
Safety, California, A.S.C.E. pp. 72-137.
Benassini, A. and Barona, F. (1962) Foundation treatment for the Benito Juarez dam. Journal of
the American Concrete Institution, 59 (October), pp. 1439-1488.
Berga, L. (1997) Failures and hydrological incidents of dams in Spain. In Dix-neuvime
Congrs des Grands Barrages, Florence, ICOLD. Vol. Q75 pp. 417-430.
Bernabe, Y. and Brace, W.F. (1990) Deformation and fracture of Berea Sandstone. The Brittle-
Ductile Transition in Rocks, The Heard Volume, American Geophysical Union. pp. 91-
101.
Berry, P., Manfredini, G. and Ribacchi, R. (1977) The mechanical properties of the rocks of the
Laga formation in the Camposto area. Capri '77, International Symposium The
Geotechnics of Structurally Complex Formations, Capri, Vol. 2 pp. 41-50.
Bertacchi, P. and Bellotti, R. (1970) Experimental research on materials for rockfill dams. Tenth
Congres on Large Dams, Montreal, ICOLD. Vol. Q36 pp. 511-529.
Betournay, M.C., Gorski, B., Labrie, D., Jackson, R. and Gyenge, M. (1991) New
considerations in the determination of Hoek and Brown material constants. 7th
International Congress on Rock Mechanics, ISRM, (Wittke ed.) Aachen, Vol. 1 pp. 195-
200.
Bieniawski, Z.T. (1967) Mechanism of Brittle Fracture of Rock. MEG 580, National
Mechanical Engineering Research Institute, Council for Scientific and Industrial Research
(CSIR), Pretoria, South Africa.
Bieniawski, Z.T. (1973) Engineering classification of jointed rock masses. Transactions of the
South African Institution of Civil Engineers, Vol 15, pp. 335-43.
Bieniawski, Z.T. (1974) Estimating the strength of rock materials. Journal of the South African
Institute of Mining and Metallurgy, 74 (8), pp. 312-320.
Bieniawski, Z.T. (1975) Case studies: Prediction of rock mass behaviour by the geomechanics
classification. Proceedings of the 2nd Australia New Zealand Conference on
Geomechanics, Brisbane, Australia, IEAust. pp. 36-41.
Bieniawski, Z.T. (1976) Rock mass classifications in rock engineering. Proceedings of the
Symposium on Exploration for Rock Engineering, Johannesburg pp. 97-107.
Bieniawski, Z.T. (1978) Determining rock mass deformability: Experience from case histories,
International Journal for Rock Mechanics, Mining Sciences and Geomechanical Abstracts,
15, pp. 237-247.
Bieniawski, Z.T. (1979) The geomechanics classification in rock engineering applications.
Proceedings ISRM 4th International Congress on Rock Mechanics, Montreux. Balkema,
Boston, Chapter 5, pp. 55-95.
Bieniawski, Z.T. (1983) The geomechanics classification (RMR system) in design applications
to underground excavations. International Symposium on Engineering Geology and
Underground Construction, Portugal, IAEG. Vol. 2 pp. II33-II47.
Bieniawski, Z.T. (1988) The rock mass rating (RMR) system (geomechanics classification) in
engineering practice. In Rock Classification Systems for Engineering Purposes (Kirkaldie
ed.), ASTM, Philadelphia, pp. 17-34.
Bieniawski, Z.T. (1989) Engineering Rock Mass Classifications. Wiley, New York.
Bieniawski Z.T. and Nicholson G.A. (1990) A Nonlinear Deformation Modulus based on Rock
Mass Classification, International Journal of Mining and Geological Engineering, Vol 8,
pp. 181 202.
Bieniawski, Z.T. and Orr, C.M. (1976) Rapid site appraisal for dam foundations by the
geomechanics classification. Douzieme Congres des Grands Barrages, Mexico, pp. 483-
501.
Biswas, A.K. and Chatterjee, S. (1971) Dam disasters: an assessment. Engineering Journal,
(March 1971), pp. 3-8.
Blanton, A.C. (1915) Caverns under the Austin, Texas, dam. Engineering News, 73 (No. 16;
Apr. 22), pp. 788-789.
Blanton, T.L. (1981) Effect of strain rates from 10^-2 to 10 sec^-1 in triaxial compression tests
on three rocks. International Journal for Rock Mechanics, Mining Sciences and
Geomechanical Abstracts, 18 pp. 47-62.
Blind, H. (1983) The safety of dams. Water Power & Dam Construction, (May.), pp. 17-21.
Bodonyi, J. (1970) Laboratory tests of certain rocks under axially-symmetrical loading
conditions. Proceedings of the 2nd Congress International Society for Rock Mechanics,
Beograd, Vol. 1 pp. 389-397.
Bollo, M.F. (1965) Technical study of the bursting of the Vega de Tera Dam. Review de la
Federation Intl du Batiment et des Travaux Publics (45), pp. 15-22.
Borsetto, M., Frassoni, A., Rossi, P.P., Garbin, C. and Moro, T. (1970) The Anapo pumped-
storage power station: geomechanical investigations and design criteria. Design and
Performance of Underground Excavations ISRM/BGS, Cambridge, pp. 73-80.
Boughton, N.O. (1970) Elastic analysis for behaviour of rockfill. Journal of the Soil Mechanics
and Foundations Division Proceedings of the American Society of Civil Engineers, 96
(SM5), pp. 1715-1733.
Bowers, N.A. (1928) St. Francis Dam catastrophe - A great foundation failure. Engineering
News Record, 100 (12), pp. 466-473.
Brace, W.F. (1964) Brittle fracture of rocks. State of Stress in the Earth's Crust (Judd ed.)
Elsevier. pp. 111-174.
Brace, W.F., Paulding, B.W. and Scholz, C. (1966) Dilatancy in the fracture of crystalline rocks.
Journal of Geophysical Research, 71 (16), pp. 3939-3953.
Brady, B.H.G. and Brown, E.T. (1993) Rock Mechanics for Underground Mining, Chapman &
Hall.
Bray, J.W. (1967) A study of jointed and fractured rock, Rock Mechanics and Engineering
Geology, 5(2-3), pp. 119-136 & 197-216.
Bredthauer, R.O. (1956) Strength characteristics of rock samples under hydrostatic pressure.
American Society of Mechanical Engineers Transactions, 79 pp. 695-708.
Broch, E. (1974) The influence of water on some rock properties. Proceedings of the 3rd
Congress International Society for Rock Mechanics, Denver, Vol. 11-A pp. 33-38.
Brown, E.T. (1970) Strength of models of rock with intermittent joints, Journal of Soil
Mechanics, Foundation Division, ASCE, 96(SM6), pp. 1935-1949.
Brown, E.T. (1971a) Brittle fracture of rock at low confining pressures. First Australia-New
Zealand Conference on Geomechanics, Melbourne, Vol. 1 pp. 31-36.
Brown, E.T. (1971b) The strength of jointed rock masses, Queensland Division Technical
Papers, The Institution of Engineers Australia, 12(9), pp. 1-16.
Brown, E.T. (1981) Suggested method for petrographic description of rocks. In Rock
Characterization, Testing & Monitoring, I.S.R.M. Suggested Methods (Brown ed.), pp. 75-
77.
Brown, E.T. and Hoek, E. (1988) Discussion on Paper No. 20431 by R. Ucar, entitled:
'Determination of shear failure envelope in rock masses'. A.S.C.E., Journal of the
Geotechnical Engineering Division, Vol. 114 (3), pp. 371-373.
Byerlee, J.D. (1975) The fracture strength and frictional strength of Weber sandstone.
International Journal for Rock Mechanics, Mining Sciences and Geomechanical Abstracts,
12 (1-4).
Byerlee, J.D. (1968) Brittle-ductile transition in rocks. Journal of Geophysical Research, 73
(14), pp. 4741-4750.
Carlier, M.A. (1974) Causes of the failure of the Malpasset Dam. In Foundations for dams,
California, A.S.C.E. pp. 5-11.
Carter, B.J., Scott Duncan, E.J. and Lajtai, E.Z. (1991) Fitting strength criteria to intact rock.
Geotechnical and Geological Engineering, 9, pp. 73-81.
Cecconi, M. and Viggiani, G. (1998) Physical and structural properties of a pyroclastic soft
rock. The Geotechnics of Hard Soils - Soft rocks, Proceedings of the Second International
Symposium, (Picarelli ed.) Naples, Italy, Balkema. Vol. 1, pp. 85-91.
Cecconi, M., Viggiani, G. and Rampello, S. (1998) An experimental investigation of the
mechanical behaviour of a pyroclastic soft rock. The Geotechnics of Hard Soils - Soft
rocks, Proceedings of the Second International Symposium, (Picarelli ed.) Naples, Italy,
Balkema. Vol. 1 pp. 473-482.
Chan, S.S.M., Crocker, T.J. and Wardell, G.G. (1972) Engineering properties of rocks and rock
masses in the deep mines of the Coeur d'Alene Mining District, Idaho. Transactions of the
Society of mining Engineers of AIME, 252 (4), pp. 353-361.
Charles, J.A. (1976) The use of one-dimensional compression tests and elastic theory in
predicting deformations of rockfill embankments. Canadian Geotechnical Journal, 13 (3),
pp. 189-200.
Charles, J.A. and Boden, J.B. (1985) The failure of embankment dams in the United Kingdom.
In Proceedings of the Symposium on Failures in Earthworks, London, Thomas Telford. pp.
181-202.
Charles, J.A. and Watts, K.S. (1980) The influence of confining pressure on the shear strength
of compacted rockfill. Geotechnique, 30 pp. 353-367.
Chen, Z. (1995) Recent developments in slope stability analysis. ISRM International Congress
on Rock Mechanics, (Herausgeber ed.) Tokyo, ISRM. Vol. 3 pp. 1041-1048.
Cheng Y. and Liu S.C. (1993) Power Caverns of Mingtan Pumped Storage Project Taiwan,
Comprehensive Rock Engineering Principles Practice & Projects, Vol 5, pp. 112 131.
Chiu, H.K. and Johnston, I.W. (1980) The effects of drainage conditions and confining
pressures on the strength of Melbourne mudstone. 3rd Australia New Zealand Conference
on Geomechanics, Wellington, pp. 1-185 to 1-189.
Clements, T. (1969) St. Francis dam failure of 1928. In Engineering Geology in Southern
California (R. Lung and R. Proctor ed.), AEG, pp. 89-91.
Coates, D.F., McRorie, K.L. and Stubbins, J.B. (1963) Analysis of pit slides in some
incompetent rocks. Transactions of the Society of Mining Engineers, AIME, 226 pp. 94-
101.
Collado, A. and Gili, J.A. (1988) A Geological Study for Highway 420, University Pol de
Catalunya. In Spanish.
Commission Administrative d'Enqute - France (1965a) The experts' report on the collapse of
the Malpasset Dam. Le Gnie Civil, 142 (11), pp. 239-246.
Commission Administrative d'Enqute - France (1965b) The experts' report on the collapse of
the Malpasset Dam. Le Gnie Civil, 142 (12), pp. 277-283.
Commission de Contre-Expertise - France (1966) The report of the re-surveying committee on
the failure of the Malpasset Dam. Le Gnie Civil, 143 (1), pp. 14-20.
Cook, J.M., Sheppard, M.C. and Houwen, O.H. (1991) Effects of strain rate and confining
pressure on the deformation and failure of shale. SPE Drilling Engineering, 6 (2), pp. 100-
104.
Coutinho Rodrigues, J.M. (1987) Failure and deterioration of arch dams. In International
Workshop on Arch Dams, (J. Laginha Serafim and R. W. Clough ed.) Coimbra,
A.A.Balkema. Vol. 1 pp. 495-499.
Courtney, C.F. (1897) Failure of the Bouzey Dam. In Masonry Dams - From inception to
completion Crosby Lockwood and son, London, pp. 91-101.
Cunningham, D.M. (1986) The Nattai North Escarpment Failure. BE Thesis, University of
Wollongong.
Das, M.N. and Sheorey, P.R. (1986) Triaxial strength behaviour of some Indian coals. Journal
of Mines, Metals & Fuels, 34 pp. 118-122.
David Rogers, J. (1995) A man, a dam and a disaster: Mulholland and the St. Francis dam. The
St. Francis dam disaster revisited: Southern California Quarterly (& Ventura County
Historical Society Quarterly 40 Nos. 3-4), LXXVII (1-2).
David Rogers, J. and McMahon, D.J. (1993) Reassessment of the St. Francis Dam failure. In
10th ASDSO Annual Conference, Kansas City, Missouri, pp. 333-339.
Dayre, M. and Giraud, A. (1986) Mechanical properties of granodiorite from laboratory tests.
Engineering Geology, 23 (2), pp. 109-124.
Dlugosz, M., Gustkiewicz, J. and Wysocki, A. (1981) Apparatus for investigation of rocks in a
triaxial state of stress. Part II. Some investigation results concerning certain rocks.
Archiwum Gornictwa, 26 pp. 29-41.
Dobereiner, L. and de Freitas, M.H. (1986) Geotechnical properties of weak sandstones.
Geotechnique, 36 (1), pp. 79-94.
Donaghe, R.T. and Cohen, M.W. (1978) Strength and Deformation Properties of Rock Fill.
Final report Tecknical Report S-78-1, U.S Army Engineers Waterways Experiment Station
Soils and Pavements Laboratory.
Donath, F.A. (1964) Strength variations and deformational behaviour in anisotropic rock. In
State of Stress in the Earth's Crust (Judd ed.), Elsevier, New York, pp. 281-297.
Doruk, P. (1991) Analysis of the Laboratory Strength Data Using the Original and Modified
Hoek-Brown Failure Criteria. M.A.Sc, University of Toronto.
Douglas, K.J. (1998) Case studies in the assessment of rock mass criteria. 3rd Young
Geotechnical Professionals Conference, Melbourne.
Douglas, K.J. and Mostyn, G. (1999) Strength of large rock masses field verification. Rock
Mechanics for Industry, Proceedings of the 37
th
U.S. Rock Mechanics Symposium, Vail,
Colorado, USA. 1:271-276. Balkema, Rotterdam, ISBN 90 5809 099 X0.
Douglas, K., Spannagle, M. and Fell, R. (1998a) Estimating the probability of failure of
concrete and masonry gravity dams. 1998 ANCOLD-NZSOLD Conference on
Dams, Sydney.
Douglas, K., Spannagle, M. and Fell, R. (1998b) Report on Analysis of Concrete and Masonry
Dam Incidents. UNICIV, The School of Civil and Environmental Engineering, The
University of New South Wales.
Douglas, K., Spannagle, M. and Fell, R. (1999a) Analysis of Concrete and Masonry Dam
Incidents. The International Journal on Hydropower & Dams. 6(4):108-115. Aqua~Media,
Surrey, ISSN 1352-2523.
Douglas, K., Spannagle, M. and Fell, R. (1999b) Estimating the probability of failure of
concrete and masonry gravity dams. ANCOLD Bulletin. No. 112:53-63. Australian
National Committee on Large Dams, ISSN 0045-0731.
Duevel, B. and Haimson, B. (1997) Mechanical characterization of pink Lac du Bonnet granite:
evidence of nonlinearity and anisotropy. International Journal of Rock Mechanics and
Mining Sciences, 34 (3-4), pp. 117.
Dunikowski, A., Korman, S. and Kohsling, J. (1969) Laboratory test indices of physico-
mechanical properties of rocks in three-axial state of stress (in Polish). Przeglad Gorniczy,
25 (11), pp. 523-528.
Duran, A. and Douglas, K. (1999) Do slopes designed with empirical rock mass strength
criteria stand up? Proceedings ISRM 9
th
International Congress on Rock Mechanics,
Paris, France, 1, pp. 87-90. Balkema, Rotterdam, ISBN 90 5809 070 1.
Duran, A. & Douglas, K.J. (2000) Experience with empirical rock slope design. GeoEng2000:
An International Conference on Geotechnical & Geological Engineering, 19-24
November, Melbourne, Australia, 2, pp. 41 and CD-Rom paper no. SNES1186, Technomic
Edmond, J.M. and Paterson, M.S. (1972) Volume change during the deformation of rocks at
high pressures. International Journal of Rock Mechanics and Mining Sciences, 9 pp. 161-
182.
Einstein H.H., Veneziano D., Baecher G.B. and O'Reilly K.J. (1983) The effect of discontinuity
persistence on rock slope stability, International Journal of Rock Mechanics, Mining
Sciences and Geomechanical Abstracts, 20(5), pp. 227-236.
Elliot, G.M. and Brown, E.T. (1985) Yield of a soft, high porosity rock. Geotechnique, 35 (4),
pp. 413-423.
Engineering News (1897?) Failure of the dam of the Bouzey Reservoir, France. Engineering
News, XXXIII (21), pp. 332.
Engineering News (1900a) The failure of the Austin dam. Engineering News, 43 (16), pp. 250-
253.
Engineering News (1900b) Failure of the great masonary dam across the Colorado River at
Austin, Tex. Engineering News (Apr 12), pp. 244-247.
Engineering News (1900c) More light on the failure of the Austin dam. Engineering News, 43
(19), pp. 308-309.
Engineering News (1900d) The silt deposit behind the Austin Dam and the relation of silt
deposits to storage reservoirs. Engineering News, XLII (25), pp. 410-411.
Engineering News (1900e) Silting-up of Lake McDonad and the leak at the Austin dam.
Engineering News (Feb 22), pp. 135-136.
Engineering News (1901) Coefficient of friction in dam design, and the failure of the dam at
Austin, Tex. Engineering News (May 30), pp. 392-393.
Engineering News (1902) An estimate for rebuilding the Austin dam. Engineering News, 47
(14), pp. 272-273.
Engineering News (1908a) (Austin dam estimates for rebuilding) Engineering News (Nov. 12),
pp. 530-532.
Engineering News (1908b) A report on rebuilding the Austin dam. Engineering News (Nov.
12), pp. 530.
Engineering News (1909) The undermining of a reinforced-concrete dam at Pittsfield, Mass.
Engineering News, 61 (13; Apr. 1), pp. 345-347.
Engineering News (1910a) Partial failure of a concrete dam at Austin, Pa. Engineering News, 63
(No. 11; Mar. 17), pp. 321-323.
Engineering News (1910b) Designs and contract for rebuilding the Colorado River dam at
Austin, Texas. Engineering News, 63 (No. 15; Apr. 14), pp. 440-441.
Engineering News (1911) The partial failure of a concrete dam at Austin, Pa., on Jan. 23, 1910.
Engineering News, 66 (14), pp. 417-423.
Engineering News (1914a) Break in the Stony River Dam. Engineering News, 71 (4), pp. 211-
214.
Engineering News (1914b) Editorial: The failure of the Stony River Dam. Engineering News,
71 (4), pp. 199-201.
Engineering News (1914c) The failure of the Stony River Dam. Engineering News (Jan 22), pp.
199-201.
Engineering News (1915) Reconstruction of the Austin, Texas, dam. Engineering News, 73 (No.
11; Mar. 18), pp. 528-529.
Engineering News (1916a) Austin dam section criticized. Engineering News (Aug. 3),.
Engineering News (1916b) More trouble at Austin dam. Engineering News, 75 (No. 14; Apr.
6),.
Engineering News Record (1911) Ignorance of the coefficient of sliding friction between rock
substrata a cause of dam failures. Engineering News Record, 36 (18; Nov 1), pp. 453-454.
Engineering News Record (1918) Will cost half million to make Austin dam usable.
Engineering News Record, 80 (No. 8; Feb. 21), pp. 353-354.
Engineering News Record (1924a) Details of the failure of an Italian multiple-arch dam.
Engineering News Record, 92 (5), pp. 182-184.
Engineering News Record (1924b) Failure of Gleno Dam charged to buttress shear. Engineering
News Record, 92 (12), pp. 486-488.
Engineering News Record (1924c) Official report on collapse of Gleno Dam. Engineering News
Record, 93 (6), pp. 213-215.
Engineering News Record (1926a) Abutment of dam fails through undermining. Engineering
News Record, 96 (4; Jan. 28), pp. 172.
Engineering News Record (1926b) Two arch dams fail through undermining of abutments.
Engineering News Record, 97 (16; Oct. 14), pp. 616-618.
Engineering News Record (1928a) Commission finds failure of St. Francis Dam due to
defective foundations. Engineering News Record, 100 (14), pp. 553-555.
Engineering News Record (1928b) Geological formation at the St. Francis Dam site.
Engineering News Record (1928c) News of the week: All St. Francis Dam reports agree as to
reason for failure. Engineering News Record, 100 (Apr. 19), pp. 639.
Engineering News Record (1928d) News of the week: Mead committee is in agreement with
others on St. Francis failure. Engineering News Record, 100 (Apr. 26), pp. 675.
Engineering News Record (1928e) News of the week: St. Francis Dam of Los Angeles water-
supply system fails under full head. Engineering News Record, 100 (11), pp. 456.
Engineering News Record (1928f) Sixth report on St. Francis Dam offers new theories.
Engineering News Record, 100 (23), pp. 895.
Engineering News Record (1928g) St. Francis Dam catastrophe: A review six weeks after.
Engineering News Record (1928h) The St. Francis Dam disaster. Engineering News Record,
100 (Mar. 22), pp. 465.
Engineering News Record (1928i) St. Francis Dam failure - An engineer's study of the site.
Engineering News Record (1928j) St. Francis Dam of Los Angeles water-supply system fails
under full head. Engineering News Record, 100 (11), pp. 456.
Engineering News Record (1929a) Rebuilding Power Plant No. 2 after the St. Francis Dam
break. Engineering News Record, 102 (Mar. 28), pp. 494-495.
Engineering News Record (1929b) Standing section of St. Francis Dam razed with dynamite.
Engineering News Record, 103 (Jul. 11), pp. 51.
Engineering News Record (1935) Overtopping of small dam causes flood in Italy. ENR (Oct.
31), pp. 607-610.
Engineering News Record (1957) Flood waters hit Las Vegas as three small dams fail.
Engineering News-Record (Aug), pp. 28.
Engineering News Record (1959a) French dam collapse: Rock shift was probable cause.
Engineering News Record, 163 pp. 24-25.
Engineering News Record (1959b) Masonry dam crumbles in Spain. Engineering News Record,
162 (Jan. 15), pp. 28.
Engineering News Record (1960a) Malpasset Dam's failure is half solved. Engineering News
Record, 164 (Apr. 7), pp. 25.
Engineering News Record (1960b) Rock did shift: Malpasset report delayed as engineers check
fault. Engineering News Record, 164 (Jan. 21), pp. 26.
Engineering News Record (1960c) Court convicts dam designers. Engineering News Record,
164 (Mar. 17), pp. 26.
Engineering News Record (1962) New probe of dam collapse asked. Engineering News Record,
168 (6), pp. 22.
Engineering News Record (1963) New Malpasset collapse study due. Engineering News
Record, 170 (Mar. 14), pp. 23.
Engineering News Record (1964a) Engineer on trial for dam collapse. Engineering News
Record, 173 (18), pp. 14-15.
Engineering News Record (1964b) Malpasset trial decision awaited. Engineering News Record,
173 (19), pp. 19.
Engineering News Record (1967) Dam designers focus on safety. Engineering News Record
(Sep 14), pp. 15-16.
Engineering Record (1900a) The failure of the Austin dam. Engineering Record, 41 (15), pp.
340-342.
Engineering Record (1900b) Failure of the Austin dam. Engineering Record, 41 (16), pp. 372-
375.
Engineering Record (1911a) Another account of the Austin dam failure. Engineering Record, 64
(No. 16; Oct. 14), pp. 444.
Engineering Record (1911b) Masonry dams and their foundations. Engineering Record, 64 (17),
pp. 492.
Engineering Record (1911c) The probable cause of the Austin dam failure. Engineering Record,
64 (19), pp. 530-532.
Engineering Record (1911d) The destruction of the Austin dam - An account of the conditions
which caused its failure on September 30. Engineering Record, 64 (15), pp. 429-436.
Engineering Record (1911e) The destruction of the Austin dam. Additional facts. Engineering
Record, 64 (16), pp. 442.
Engineering Record (1911f) The expert testimony at the Austin Dam investigation. Engineering
Record, 64 (20), pp. 578.
Engineering Record (1912) Washout of base of Port Angeles dam. Engineering Record, 66 (22;
Nov. 30), pp. 600-601.
Engineering Record (1914) Failure of Stony River Dam. Engineering Record, 69 (4), pp. 88.
Engineering Record (1915a) Hollow reinforced-concrete structure replaces dam at Austin,
Texas, which failed fifteen years ago; Part I. Engineering Record, 71 (No. 22; May 29), pp.
672-673.
Engineering Record (1915b) Reinforced-concrete structure at Austin, Texas, replaces dam
which failed. Engineering Record, 71 (Jun. 5), pp. 707-708.
Evans, B., Fredrich, J.T. and Wong, T. (1990) The brittle-ductile transition in rocks: recent
experimental and theoretical progress. In The Brittle-Ductile Transition in Rocks,
Geophysical Monograph 56, Duba et al. Eds, American Geophysical Union, Washington,
D.C., pp. 1-20.
Everling, G. (1960) Rock mechanical investigations and basis for determination of rock pressure
according to deformation of drill holes (in German). Gluckauf, 96 (7), pp. 390-409.
Fairhurst, C. (1964) On the validity of the Brazilian test for brittle materials. International
Journal for Rock Mechanics and Mining Sciences, 1, pp. 535-546.
Fang, H.Y. and Fernandez, J. (1981) Determination of tensile strength of soils by unconfined
penetration test, Laboratory Shear Strength of Soils, ASTM STP 740, pp. 130-144.
Fayed, L.A. (1968) Shear strength of some argillaceous rocks. International Journal of Rock
Mechanics and Mining Sciences, 5 (1), pp. 79-85.
Feld, J. (1968) Construction Failure. Wiley Series of Practical Construction Guides pp. 91-109.
Fell, R. (1996) Estimating the probability of failure of embankment dams under normal
operating conditions. In Repair and Upgrading of Dams, Stockholm, KTH. pp. 567-576.
Fell, R., McGreggor, P. and Stapledon, D. (1992). Geotechnical Engineering of Embankment
Dams. A.A. Balkema, pp. 675.
FEMA (1995) Water Control Infrastructure - National Inventory of Dams Update 1993-1994,
[CD-ROM] Federal Emergency Management Agency.
Finch, J.K. (1914) Letters to the Editor: The Stony River Dam: Sliding as a danger in hollow
dams. Engineering News, 71 (4), pp. 202.
Flagg, C. Geological causes of dam incidents. General 125146 79A, USBR.
Fleming, R.W., Spencer, G.S. and Banks, D.C. (1970) Empirical Study of Behaviour of Clay
Shale Slopes. NCG Technical Report No.15, U.S. Army Engineer Nuclear Cratering
Group, Livermore, California.
Foster, M., Spannagle, M. and Fell, R. (1998) Report on the Analysis of Embankment Dam
Incidents. UNICIV Report, School of Civil and Environmental Engineering, The
University of New South Wales.
Foster, P.F. and Fairless, G.J. (1994). Waitaki Dam - Review of Aviemore rock strength.
(unpublished) ID: 9W551.BO, Power Engineering Office Works Consultancy Service Ltd
(Wellington, N.Z.).
Frassoni, A., Hegg, U. and Rossi, P.P. (1982) Large-scale laboratory tests for the mechanical
characterization of granular materials for embankment dams. Fourteenth Congres on
Large Dams, Rio de Janeiro, ICOLD. Vol. Q55 pp. 727-751.
Franklin, J.A. (1971) Triaxial strength of rock materials. Rock Mechanics, 3 pp. 86-98.
Franklin, J.A. and Hoek, E. (1970) Developments in triaxial testing technique. Rock Mechanics,
2 pp. 223-228.
Freeman, G.L. and Alsop, R.B. (1941) Underpinning Austin dam. Engineering News Record
(Jan. 30), pp. 52-57.
Fry, J. (1996) Causes et probabilites de rupture des barrages en Europe. In COGECH 1996.
Fujii, Y., Kiyama, T. and Ishijima, Y. (1997) Condition insensitive damage indicator for brittle
rock. International Journal of Rock Mechanics and Mining Sciences, 34 (3-4), pp. 086.
Fumagalli, E., Mosconi, B. and Rossi, P.P. (1970) Laboratory tests on materials and static
models for rockfill dams. Tenth Congres on Large Dams, Montreal, ICOLD. Vol. Q36 pp.
531-551.
Georgiadis M. and Michalopoulos A.P. (1986) Dilatometer Tests for the Design of Grouted
Piles in Rock, USA, Use of In-situ Tests in Geotechnical Engineering, Proceedings of In-
situ 86, pp. 560-568.
Gifford A.B., Green G.E., Buechel G.J. and Feldman A.I. (1986) In-situ Tests and Design of a
Cylinder Pile Wall, USA, Use of In-situ Tests in Geotechnical Engineering, Proceedings of
In-situ 86, pp. 569-587.
Giovanni B. (1993) Case Study of Rock Mechanics in the Masua Mine - Italy, Comprehensive
Rock Engineering Principles Practice & Projects, Vol 5, pp. 291 333.
Glastonbury, J. & Douglas, K.J. (2000) Catastrophic rock slope failures. GeoEng2000: An
International Conference on Geotechnical & Geological Engineering, 19-24 November,
Melbourne, Australia, Vol. 2 pp. 21 and CD-Rom paper no. SNES0507, Technomic
Gomez Laa, G., Alonso Franco, M. and Hernandez, J.L.R. (1979) Reflections on some incidents
in Spanish dams. Thirteenth International Congress on Large Dams, New Dehli, ICOLD.
Vol. Q49 pp. 721-740.
Gonzalez Garcia, A.J. (1985) Rockfill strength with dimensionless parameters. Proceedings of
the Eleventh International Conference on Soil Mechanics and Foundation Engineering,
San Francisco, Balkema. Vol. 2 pp. 479-485.
Good, R.J. (1976) Kangaroo Creek Dam use of a weak schist as rockfill for a concrete faced
rockfill dam. Twelth Congres on Large Dams, Mexico, ICOLD. Vol. Q44 pp. 645-659.
Gosselin, M. (1960) The report on the failure of the Malpasset Dam. Civil Engineering and
Public works review, 55 (648), pp. 921-922.
Goto, S., Suzuki, Y., Nishio, S. and Oh-oka, H. (1992) Mechanical properties of undisturbed
tone-river gravel obtained by in-situ freezing method. Soils and Foundations, 32 (3), pp.
15-25.
Gowd, T.N. and Rummel, F. (1980) Effect of confining pressure on the fracture behaviour of a
porous rock. International Journal for Rock Mechanics, Mining Sciences and
Geomechanical Abstracts, 17 (4), pp. 225-229.
Graham, L.P. and Bartsch, M. (1995) Risk analysis safety assessment for use at Swedish dams.
In Research and Development in the Field of Dams, Crans-Montana, Switzerland, Swiss
National Committee on Large Dams. pp. 627-638.
Gray, K.E. and Holder, J.T. (1991) Towards characterizing shale stability. Proceedings of the
44th Canadian Geotechnical Conference, Calgary, Alberta, Canada, Vol. 1 pp. 37/1 to
37/7.
Gray, P.A. (1988) The problem of estimating the shear strength of unstable rock slopes. Fifth
Australia-New Zealand Conference on Geomechanics, Sydney, pp. 375-380.
Griffith, A.A. (1924) The theory of rupture. Proceedings of the 1
st
Congress on Applied
Mechanics, Delft, pp. 55-63.
Gruner, E. (1963) Dam disasters. Proceedings of the Institute of Civil Engineering (London), 24
pp. 47-60.
Gruner, E. (1967) The safety of reservoirs. World Dams Today, The Japan Dam Association pp.
104-109.
Grunsky, C.E. and Grunsky, E.L. (1928) Letters to the Editor: The Grunsky Report on the
failure of the St. Francis Dam. Engineering News Record, 101 (Jul. 26), pp. 144.
Guerreiro, M., Fernandez Cuevas, R., Guerreiro, M.J. and Gomez Laa, G. (1991) Causes of
failure of the Fonsagrada Dam. In Dix-septieme Congres des Grands Barrages, Vienne,
ICOLD. Vol. Q65 pp. 49-63.
Gulan, I.J. (1995) A Study on the Failure of Concrete Gravity Dams Through the Concrete
Mass. Unpublished BE Thesis, The University of New South Wales.
Gupta, K.K. and Ramamurthy, T. (1978) Prediction of the behaviour of rockfill materials.
Proceedings of GEOCON India Conference on Geotechnical Engineering, New Delhi,
Vol. 1 pp. 25-31.
Gupta, K.K. and Soni, M.L. (1994) Prediction of rockfill material responses. Proceedings of the
Thirteenth International Conference on Soil Mechanics and Foundation Engineering, New
Delhi, Balkema. Vol. 3 pp. 963-966.
Gurtu, S.K. (1925) Masonary dams and weirs. Indian Engineering (Aug 1), pp. 67-69.
Gustkiewicz, J. (1985) Transition of rocks from brittle to ductile state; strain at failure as a
function of confining pressure. Acta Geophysica Polonica, 33 (2), pp. 169-181.
Habimana, J., Labiouse, V. and Descoeudres, F. (2002) Geomechanical characterisation of
cataclastic rocks: experience from the Cleuson-Dixence project, International Journal of
Rock Mechanics and Mining Sciences, 39, pp.677-693.
Hadizadeh, B. and Rutter, E.H. (1983) The low temperature brittle-ductile transition in a
quartzite and the occurence of cataclastic flow in nature. Geologische Rundschau, 72 (2),
pp. 493-509.
Haied, A. and Kondo, D. (1997) Strain localization in Fontainebleau sandstone: macroscopic
and microscopic investigations. International Journal of Rock Mechanics and Mining
Sciences, 34 (3-4), pp. 161.
Haines, A. (1993) Rock slope classification for the optimal design of monitoring networks.
Geotechnical Instrumentation and Monitoring in Open Pit and Underground Mining,
(Szwedzicki ed.) Kalgoorlie, Balkema. pp. 201-209.
Haines, A. and Terbrugge, P.J. (1991) Preliminary estimation of rock slope stability using rock
mass classification systems. Proceedings 7th International Society Rock Mechanics,
(Herausgeber ed.) Aachen, Vol. 2 pp. 887-892.
Hall, E.B. and Gordon, B.B. (1963) Triaxial testing with large-scale high pressure equipment.
Laboratory Shear Testing of Soils, ASTM STP 361, Ottawa, Canada, American Society for
Testing and Materials. pp. 315-328.
Handin, J. Strength at high confining pressure and temperature of serpentinite from Mayaguez,
Puerto Rico. .
Handin, J.H. and Hager, R.V. (1957) Experimental deformation of sedimentary rocks under
confining pressure: Tests at room temperature on dry samples. Bulletin of the American
Association of Petroleum Geologists, 41 (1), pp. 1-50.
Handin, J.H., Hager, R.V., Friedman, M. and Feather, J.N. (1963) Experimental deformation of
sedimentary rocks under confining pressure: Pore pressure tests. Bulletin of the American
Association of Petroleum Geologists, 47 (5), pp. 717-755.
Hanna, K., Conover, D. and Haramy, K. (1991) Coal mine entry intersection behaviour study.
Report of Investigations 9337, United States Department of the Interior, Bureau of Mines.
Hareland, G., Polston, C.E. and White, W.E. (1993) Normalised rock failure envelope as a
function of grain size. International Journal for Rock Mechanics, Mining Sciences and
Geomechanical Abstracts, 30 (7), pp. 715-717.
Hatem, G.A. (1985) Development of a Data Base on Dam Failures in the United States
Preliminary Results. BE Thesis, Stanford University.
Hatton, C. (1912) The Austin dam and its failure. Engineering News (Oct 3), pp. 605-606.
Helgstedt, M.D. (1997) An Assessment of the In-Situ Shear Strength of Rock Masses and
Discontinuities. M.E., University of Lulea.
Helgstedt, M.D., Douglas, K.J. and Mostyn, G. (1997) A re-evaluation of in-situ direct shear
tests at Aviemore Dam, New Zealand. Australian Geomechanics, 37 (June), pp. 56-65.
Hencher, S.R. (1995) Interpretation of direct shear tests on rock joints, Proceedings of the 25
th

US Symposium on Rock Mechanics (Daeman and Schultz eds) Reno, pp. 99-106.
Hill, W.R. (1902) A list of failures of American dams. Engineering Record, 46 (13; Sep 27), pp.
290-293.
Hinds, J. (1953) Continuous development of dams since 1850. A.S.C.E. Transactions pp. 489-
520.
Hirth, G. and Tullis, J. (1989) The effects of pressure and porosity on the micromechanics of the
brittle-ductile transition in quartzite. Journal of Geophysical Research, 94 (B12), pp.
17,825-17,838.
Hobbs, D.W. (1964) The strength and the stress-strain characteristics of coal in triaxial
compression. Journal of Geology, 72 pp. 214-231.
Hobbs, D.W. (1966) A study of the behaviour of broken rock under triaxial compression, and its
application to mine roadways, International Journal on Rock Mechanics and Mining
Sciences, 3, pp. 11-14.
Hobbs, D.W. (1970) The behaviour of broken rock under triaxial compression. International
Journal for Rock Mechanics, Mining Sciences and Geomechanical Abstracts, 7 pp. 125-
148.
Hoek, E. (1968) Brittle failure of rock. In Rock Mechanics in Engineering Practice, John Wiley
& Sons, London, pp. 99-124.
Hoek, E. (1970) Estimating the stability of excavated slopes in open cut mines. Trans.
Institution of Mining and Metallurgy, 79 pp. A109-A132.
Hoek, E. (1983) Strength of jointed rock masses. Geotechnique, 33 (3), pp. 187-223.
Hoek, E. (1997) Reliability of Hoek-Brown estimates of rock mass properties and their impact
on design. Technical Note. International Journal of Rock Mechanics and Mining
Sciences.
Hoek, E. (1997) Practical estimates of properties of very weak to strong rock masses. Draft
paper (Melbourne City Link Tunnel - review of geotechnical aspects).
Hoek, E. (1998) Reliability of Hoek-Brown estimates of rock mass properties and their impact
on design, International Journal of Rock Mechanics and Mining Sciences, 35(1), pp. 63-
68.
Hoek, E. (1999) Putting numbers to geology an engineers viewpoint, Quarterly Journal of
Engineering Geology, 32, pp. 1-19.
Hoek, E. (2001) Personal communication, 1
st
October.
Hoek, E. (2002) A brief history of the development of the Hoek-Brown failure criterion.
Accessed through the program RocLab, pp. 4.
Hoek, E. and Brown, E.T. (1980a) Empirical strength criterion for rock masses. A.S.C.E.,
Journal of the Geotechnical Engineering Division, 106(GT9), pp. 1013-1035.
Hoek, E. and Brown, E.T. (1980b) Underground Excavations in Rock, The Institution of Mining
and Metallurgy, London.
Hoek, E. and Brown, E.T. (1988) The Hoek-Brown failure criterion - a 1988 update.
Proceedings of the 15th Canadian Rock Mechanics Symposium, Toronto.
Hoek, E. and Brown, E.T. (1997) Practical estimates of rock mass strength. International
Journal of Rock Mechanics and Mining Sciences, 34(8), pp. 1165-1186.
Hoek, E., Carranza-Torres, C. and Corkum, B. (2002) Hoek-Brown failure criterion - 2002
edition. Proceedings of the North American Rock Mechanics Society Meeting in Toronto in
July 2002.
Hoek, E., Kaiser, P.K. and Bawden, W.F. (1995) Support of Underground Excavations in Hard
Rock. Balkema.
Hoek, E. and Karzulovic, A. (2001) Rock mass properties for surface mines. In Slope Stability
in Surface Mining (Hustrulid, McCarter and Van Zyl ed.), Society for Mining, Metallurgy,
and Exploration, Inc. (SME), pp. 59-69.
Hoek, E. Marinos, P. and Benissi, M. (1998) Applicability of the geological strength index
(GSI) classification for very weak and sheared rock masses. The case of the Athens Schist
Formation. Bulletin of Engineering Geology and the Environment, 57, pp. 151160.
Hoek, E., Wood, D. and Shah, S. (1992) A modified Hoek-Brown failure criterion for jointed
rock masses. Eurock '92 pp. 209-213.
Holtz, W.G. and Gibbs, H.J. (1956) Triaxial shear tests on pervious gravelly soils. Journal Soil
Mechanics and Foundation Division Proccedings of the American Society of Civil
Engineers, 82 (SM1), pp. 1-22.
Horii, H. and Nemat-Nasser, S. (1986) Brittle failure in compression: splitting, faulting and
brit tle-ductile transition. Philosophical Transactions of the Royal Society of London, Series
A: Mathematical and Physical Sciences, 319 (1549), pp. 337-374.
Horino, F.G. and Ellickson, M.L. (1970) A method for estimating strength of rock containing
planes of weakness. Report of Investigations 7449, United States Department of the
Interior, Bureau of Mines.
Hornaday, W.D. (1899) Municipal ownership at Austin, Texas. Electrical World and Engineer,
XXIV (16), pp. 575-576.
Hoshino, K., Koide, H., Inami, K., Iwamura, S. and Mitsui, S. (1972) Mechanical properties of
Japanese tertiary sedimentary rocks under high confining pressures. Geological Survey
Report 244, Japanese Geological Survey.
Hoskins, E.R. (1969) The failure of thick-walled hollow cylinders of isotropic rock.
International Journal of Rock Mechanics and Mining Sciences, 6 (1), pp. 99-125.
Hossaini, S.M.F. (1993) Some Aspects of the Strength Characteristics of Intact & Jointed
Rocks. PhD, UNSW.
Hossaini, S.M.F. and Vutukuri, V.S. (1993) On the accuracy of multifailure triaxial test for the
determination of peak and residual strength of rocks. Australian Conference on
Geotechnical Instrumentation and Monitoring in Open Pit and Underground Mining,
(Szwedzicki ed.) Kalgoorlie, pp. 223-228.
Hudson, J.A. and Harrison, J.P. (1997) Engineering Rock Mechanics: An Introduction to the
Principles, Pergamon.
Hunter, J.K. (1964) Failure of the Panshet and Khadakwasla dams, Bombay. Water Power, 16
(Jun), pp. 251-252.
Ibanez, W.D. and Kronenberg, A.K. (1993) Experimental deformation of shale: mechanical
properties and microstructural indicators of mechanisms. International Journal for Rock
Mechanics, Mining Sciences and Geomechanical Abstracts, 30 (7), pp. 723-734.
ICOLD (1969) Progress Report on ICOLD Committee on Failures and Accidents to Large
Dams other than in Connection with the Foundations. ICOLD.
ICOLD (1973) World Register of Dams. International Commission on Large Dams, Paris.
ICOLD (1974) Lessons From Dam Incidents. International Commission on Large Dams, Paris.
ICOLD (1976) World Register of Dams 1st Updating. International Commission on Large
Dams, Paris.
ICOLD (1978) Technical Dictionary on Dams. International Commission on Large Dams,
Paris.
ICOLD (1979) World Register of Dams 2nd Updating. International Commission on Large
Dams, Paris.
ICOLD (1983) Deterioration of Dams and Reservoirs. International Commission on Large
Dams, Paris.
ICOLD (1984) World Register of Dams 3rd Updating. International Commission on Large
Dams, Paris.
ICOLD (1988) World Register of Dams 4th Updating. International Commission on Large
Dams, Paris.
ICOLD (1995) Bulletin 99 - Dam Failures Statistical Analysis. International Commission on
Large Dams, Paris.
INCOLD (1967) Dam construction in India from 1963-1966. Neuvieme Congres des Grands
Barrages, Istamboul, ICOLD. Vol. 4 pp. 931-962.
Indraratna, B. (1994) Implications of non-linear strength criteria in the stability assessment of
rockfill dams. Proceedings of the Thirteenth International Conference on Soil Mechanics
and Foundation Engineering, New Delhi, Balkema. Vol. 3 pp. 935-938.
Indraratna, B., Haque, A. and Aziz, N. (1998a) Comparison of shear behaviour under direct
shear and constant normal stiffness control. Geomechanics/Ground Control in Mining and
Underground Construction, (Aziz & Indraratna ed.) Wollongong, Australia, Vol. 1 pp. 85-
98.
Indraratna, B., Ionescu, D. and Christie, H.D. (1998b) Shear behaviour of railway ballast based
on large-scale triaxial tests. Journal of Geotechnical and Geoenvironmental Engineering,
124 (5), pp. 439-449.
Indraratna, B., Wijewardena, L.S.S. and Balasubramaniam, A.S. (1993) Large-scale triaxial
testing of greywacke rockfill. Geotechnique, 43 (1), pp. 37-51.
Ingles, O.G. (1984) A short study of dam failures in Australia, 1857-1983. Civil Engineering
Systems, 1 (June), pp. 190-194.
Inglis, C. (1962) The Poona flood disaster of 12 July, 1961. Proceedings of the Institution of
Civil Engineers, London, 22(May), pp. N10-N11.
Institute of Civil Engineering (1897?) Failure of the Bouzey dam. Procedures of the Institution
of Civil Engineers pp. 461-467.
Ismail, I.A.H. and Murrell, S.A.F. (1990) The effect of confining pressure on stress-drop in
compressive rock fracture. Tectonophysics, 175 (1-3), pp. 237-248.
I.S.R.M. (1981) Commission on classification of rocks and rock masses. International Journal
for Rock Mechanics, Mining Sciences and Geomechanical Abstracts, 18 pp. 85-110.
ITCOLD (1967) General Paper No 6. Neuvieme Congres des Grands Barrages, Istamboul,
ICOLD. Vol. 4 pp. 687-734.
Jaeger, J.C. (1969) Behavour of closely jointed rock. Proceedings of the 11th U.S. Symposium
on Rock Mechanics, Berkeley, California, pp. 57-68.
Jaeger, J.C. and Cook, N.G.W. (1997) Fundamentals of Rock Mechanics, Chapman & Hall.
Jamison, J.R. and Teufel, L.W. (1979) Pore volume changes associated with failure and
frictional sliding of a porous sandstone. Concrete Shaft Lining Design, 20th U.S.
Symposium on Rock Mechanics, Austin, Texas, pp. 163-170.
Jang, M.-H. and Yang, H.-S. (1996) Thermomechanical failure criteria of some granites for
muclear waste repository. Proceedings of the Korea-Japan Joint Symposium on Rock
Engineering, (Li & Moon ed.) Seoul, Korea, The Korean Institute of Mineral & Energy
Resources Engineers. pp. 95-99.
Jansen, R.B. (1980) Dams and Public Safety. U.S. Government Printing Office, Denver.
Jansen, R.B. (1988) Advanced Dam Engineering for Design, Construction and Rehabilitation.
Van Nostrand Reinhold,, New York.
Jeremic, M.L. (1994) Elasto-plastic deformation and failure. In Rock Mechanics in Salt Mining,
Balkema, Rotterdam, pp. 167-197.
John, K.W. (1962) An approach to rock mechanics, Journal of the Soil Mechanics and
Foundations Division, ASCE, 88(SM4), Proc. Paper 3223, pp. 1-30.
John, K.W. (1969) Engineering methods to determine strength and deformability of regularly
jointed rock, Preprint, 11
th
Symposium on Rock Mechanics, Berkeley, California.
Johnson, B., Friedman, M., Hopkins, T.N. and Bauer, S.J. (1987) Strength and microfracturing
of Westerly granite extended wet and dry temperatures to 800degC and pressures to 200
MPa. 28th US Symposium on Rock Mechanics, Tucson, pp. 399-412.
Johnston, I.W. (1985a) Comparison of two strength criteria for intact rock. A.S.C.E., Journal of
the Geotechnical Engineering Division, 111 (12), pp. 1449-1454.
Johnston, I.W. (1985b) Strength of intact geomechanical materials. A.S.C.E., Journal of the
Geotechnical Engineering Division, 111 (6), pp. 730-749.
Johnston, I.W. (1988) Geotechnics of weak and jointed rock - General report. Fifth Australia-
New Zealand Conference on Geomechanics, Sydney, pp. 236-242.
Johnston, I.W. (1989) Discussion leaders report: Material properties of soft rocks. Proceedings
of the Twelfth International Conference on Soil Mechanics and Foundation Engineering,
Rio de Janeiro, Balkema. Vol. 5 pp. 2831-2848.
Johnston, I.W. (1991) Geomechanics and the emergence of soft rock technology, Australian
Geomechanics, Dec.
Johnston, I.W. (1995) Rational determination of the engineering properties of soft rocks.
Procedures of the Institution of Civil Engineers; Geotechnical Engineering, 113 (Apr), pp.
86-92.
Johnston, I.W. and Chiu, H.K. (1984) Strength of intact weathered Melbourne sandstone.
Journal of Geotechnical Engineering, 110 (7), pp. 875-898.
Johnston, I.W. and Novello, E.A. (1985) Cracking and critical state concepts for soft rocks.
Proceedings of the Eleventh International Conference on Soil Mechanics and Foundation
Engineering, San Francisco, Balkema. Vol. 2 pp. 515-518.
Johnston, I.W., Williams, A.F. and Chiu, H.K. (1980) Properties of soft rock relevant to
socketed pile design. International Conference on Structural Foundations on Rock,
Sydney, pp. 55-64.
Jorda, L., Seron, J.B. and Romana, M. (1999) Application of the geomechanical classification
SMR to slopes in limestones and marls. 9th ISRM International Congress on Rock
Mechanics, Paris, Balkema. Vol. 1 pp. 111-114.
Jorgensen, L. (1920) Record of 100 dam failures. Journal of Electricity (San Francisco) (Apr.),
pp. 320-321.
Jorgensen, L.R. (1928) Letters to the Editor: The structural safety factor in the St. Francis Dam.
Kaloustian, E.S. (1984) Statistical analysis of distribution of concrete dam rock foundations
failures. International Conference on Safety of Dams, (Serafim ed.) Coimbra, A.A.
Balkema. pp. 311-319.
Kany, M. and Becker, H. (1967) Determination of the strength of a coarse-grained rock fill by
means of a large-scale test. Proceedings of the Geotechnical Conference Oslo 1967, Oslo,
NGI. Vol. 1 pp. 209-214.
Kate, J.M. and Gokhale, C.S. (1998) Influence of moisture on triaxial compression behaviour of
soft sandstone. The Geotechnics of Hard Soils - Soft Rocks, (Evangelista & Picarelli ed.)
pp. 245-251.
Kimura, T., Esaki, T., Kameda, N. and Nishida, T. (1987) Experimental and theoretical studies
on strain softening behaviour of rocks. 28th U.S. Symposium on Rock Mechanics, Tucson,
pp. 197-202.
King, J.R. and Huber, B. (1993) The history of Austin Dam. In 10th ASDSO Annual
Conference, Kansas City, Missouri, pp. 217-223.
Knoll, P. (1979) On the fracture mechanism of pillars under dynamic loading (in German). Rock
Mechanics, Suppl. 8 pp. 209-226.
Kotze, G.P. and Pells, P.J.N. (1980) Report on Stability of the Escarpment above Nattai North
Colliery, NSW, McMahon Burgess & Yeates.
Kovari, K. and Tisa, A. (1975) Multiple failure state and strain controlled triaxial tests. Rock
Mechanics, 7 pp. 17-33.
Kovari, K., Tisa, A. and Attinger, R.O. (1983) The concept of "continuous failure state" triaxial
tests. Rock Mechanics and Rock Engineering, 16 pp. 117-131.
Kulkarni, D.N. and Kulkarni, M.K. (1994) Safety improvement of some large dams in
Maharashtra, India. Eighteenth Congress on Large Dams, Durban, ICOLD. Vol. Q68 pp.
143-155.
Kung, W.H.R. (2001) Estimation of the in-situ deformation modulus using rock mass
classification systems, BE(Hon) Thesis, unpublished, The University of New South Wales.
Kwasniewski, M. (1983) Deformational and strength properties of the three structural varieties
of carboniferous sandstones. 5th International Congress on Rock Mechanics, ISRM,
Balkema. Vol. 1 pp. A105-A115.
Kwasniewski, M. (1987) A new linear criterion of brittle failure for rocks. Proceedings of the
6th Congress International Society for Rock Mechanics, Montreal, Canada, Vol. 2 pp.
1031-1038.
Kwasniewski, M. (1990) Effect of moisture on the mode of deformation and failure of coal
measure mudstone under triaxial compression. Proceedings of the 11th Plenary Scientific
Session of the International Bureau of Strata Mechanics/World Mining Congress,
Novosibirsk, Balkema. pp. 169-186.
Kwasniewski, M. and Mogi, K. (1990) Effect of the intermediate principal stress on the failure
of a foliated anisotropic rock. Mechanics of Jointed and Faulted Rock, (Rossmanith ed.)
Technical University of Vienna, Balkema, Rotterdam. pp. 407-416.
Lade, P.V. (1993) Rock strength criteria: the theories and the evidence. In Comprehensive Rock
Engineering (E. T. Brown ed.), Pergamon, Oxford, New York, pp. 255-284.
Lane, K.S. (1961) Field slope charts for stability studies. 5th International Conference on Soil
Mechanics & Foundation Engineering, Paris, ISSMFE. Vol. 2 pp. 651-655.
Laubscher, D.H. (1977) Geomechanics classification of jointed rock masses - mining
applications. Transactions of the Institution of Mining and Metallurgy, Section A, Mining
Industry, 86 (Jan), pp. A1-A8.
Laubscher, D.H. (1990) A geomechanics classification system for the rating of rock mass in
mine design. Journal of the South African Institute of Mining and Metallurgy, 90 (10), pp.
257-273.
Lajtai, E.Z. (2001) Personal Communication from the University of Manitoba, 5
th
February.
Leger, P., Tinawi, R., Bhattacharjee, S.S. and Leclerc, M. (1997) Failure mechanisms of gravity
dams subjected to hydrostatic overload: Influence of weak lift joints. Dix-neuvime
Congrs des Grands Barrages, Florence, ICOLD. Vol. Q75 pp. 11-37.
Lemprire, F. (1993) Dams that have failed by flooding: an analysis of 70 failures. Water
Power & Dam Construction (Sep./Oct.), pp. 19-24.
Lemprire, F., Fry, J.J., Reverchon, B. and Royet, P. (1997) Analysis of failures of dams less
than 30m high practical lessons. Dix-neuvime Congrs des Grands Barrages, Florence,
ICOLD. Vol. Q75 pp. 673-710.
Leps, T.M. (1970) Review of shearing strength of rockfill. A.S.C.E., Journal of the Soil
Mechanics and Foundations Division, 96 (SM4), pp. 1159-1170.
Lewis, M. (1988) Two Hundred Years of Concrete in Australia. Concrete Institute of Australia,
pp. 137.
Li, H.B., Zhao, J. and Li, T.J. (1999) Triaxial compression tests on a granite at different strain
rates and confining pressures. International Journal of Rock Mechanics and Mining
Sciences, 36 (8), pp. 1057-1063.
Li, T. (1985) The strength of intact rocks and rock masses. Earth Science - Journal of Wuhan
College of Geology, 10 (1), pp. 179-189.
Liao, Q.H. and Hencher, S.R. (1997) The effect of discontinuity orientation and spacing on
failure mechanisms in rock slopes results from systematic numerical modelling. CIM
Vancouver 97, Vancouver, Canadian Institute of Mining.
Lin, Y. (1998) An introduction of the Chinese Standard for engineering classification of rock
mass (GB50218-94). In Advances in Rock Mechanics (Lin ed.), World Scientific
Publishing Co., Singapore, pp. 317-327.
Littlechild B.D., Hill S.J., Statham I., Plumbridge G.D. and Lee S.C. (2000) Determination of
Rock Mass Modulus for Foundation Design, Geotechnical Specially Conference
Innovations and Applications in Geotechnical Site Characterization, pp. 213 228.
LNEC (1992) Large Dams in Portugal. [Online] National Laboratory for Civil Engineering.
Available: http://www.lnec.pt/bin/barragens/bin/aceder.pl. Accessed on: 19 Dec. 1996.
Londe, P. (1977) Panel Discussion in: Session 2 - Description and classification of the types of
complexities and relative geotechnical models. In The Geotechnics of Structurally
Complex Formations, Capri, Associazione Geotecnica Italiana. Vol. 2 pp. 205-211.
Lubarda, V.A., Mastilovic, S. and Knap, J. (1996) Brittle-ductile transition in porous rocks by
cap model. Journal of Engineering Mechanics, 122 (7), pp. 633-642.
Lutton, R.J. (1970) Rock slope chart from empirical data. Transaction of the Society of Mining
Engineers, AIME, 247 pp. 160-162.
MacDonald, A., Kerr, J.W. and Coats, D.J. (1994) Remedial works to Upper Glendevon Dam.
Dix-huitime Congres des Grands Barrages, Durban, ICOLD. Vol. Q68 pp. 1193-1207.
MacGregor, P. (1980) Report on the Nature of the Defects in the Narrabeen Group Rocks in the
Southern Coalfields Water Catchment Area. Coffey Partners International Pty Ltd.
Maksimovic, M. (1989) Nonlinear failure envelope for coarse-grained soils. Proceedings of the
Twelfth International Conference on Soil Mechanics and Foundation Engineering, Rio de
Janeiro, Balkema. Vol. 1 pp. 731-734.
Marachi, N.D., ASCE, A.M., Chan, C.K. and Seed, H.B. (1972) Evaluation of properties of
rockfill materials. Soil Mechanics and Foundation Divisions pp. 95-115.
Marachi, N.D., Chan, C.K., Seed, H.B. and Duncan, J.M. (1969) Strength and Deformation
Characteristics of Rockfill Materials. TE-69-5, Department of Civil Engineering,
University of California.
Marinos, P. and Hoek, E. (2001) Estimating the geotechnical properties of heterogeneous rock
masses such as flysch, Bulletin of Engineering Geology and the Environment, 60, pp. 85-
92.
Marsal, R.J. (1967) Large scale testing of rockfill materials. Journal of the Soil Mechanics and
Foundations Division Proceedings of the American Society of Civil Engineers, 93 (SM2),
pp. 27-43.
Marsal, R.J. (1971) Discussion of: Leps (1970) Review of shearing strength of rockfill. Journal
of the Soil Mechanics and Foundations Division Proceedings of the American Society of
Civil Engineers, 97 (SM3), pp. 590-592.
Marsal, R.J. (1973) Mechanical properties of rockfill. In Embankment Dam Engineering J.
Willey & Sons, N.Y.,, pp. 109-201.
Marsal, R.J. (1976) Mechanical properties of rockfill soil mixtures. Douzieme Congres des
Grands Barrages, Mexico, pp. 179-209.
Mary, M. (1968) Arch Dams - Historic Accidents and Occurrences. In Construction and
Surveillance of Dams Dunod, Paris, pp. 159.
Matsumoto, N., Yasuda, N., Ohkubo, M. and Yoshioka, R. (1991) Monotonic and cyclic
loading tests of rockfill materials. Wind and Seismic Effects; Proceedings of the 23rd Joint
Meeting of the US-Japan Cooperative Program in Natural Resources Panel on Wind and
Seismic Effects. NIST Special Publication No. 820. pp. 149-158.
McDonald, P. and Williams, A. (1993) Large Piles in Weak Rock West Gate Freeway Project,
Comprehensive Rock Engineering Principles Practice & Projects, Vol 5, pp. 728 758.
McDonough, C. (1940) Historic Austin Dam rebuilt. Engineering News Record (Jun 20), pp.
52-55.
McLamore, R. and Gray, K.E. (1967) The mechanical behaviour of anisotropic sedimentary
rocks. Transactions of the American Society of Mechanical Engineers, Series B, Journal of
Engineering for Industry, 89 (1), pp. 62-76.
McMahon, B.K. (1976) Estimatiion of Upper Bounds to Rock Slopes by Analysis of Existing
Slope Data. CANMET Report 76-14, Canada Centre for Mineral and Energy Technology.
McMahon, Burgess and Yeates (1981). Chichester Dam, Remedial Works, Report on
Investigation Into Zone 5.
McMahon, B.K., Douglas, D.J. and Burgess, P.J. (1980) Chichester Dam Remedial Works (1).
Report on geology, Volume 1, McMahon Burgess & Yeates, Consulting Engineers &
Geologists.
McWilliam, T. (2001) Stability Assessment of Rockfill Dams, Unpublished, BE(Hon) Thesis,
The University of New South Wales.
Medhurst, T.P. (1996) Estimation of the in situ strength and deformability of coal for
engineering design. PhD thesis, University of Queensland.
Medhurst, T.P. and Brown, E.T. (1996) Large scale laboratory testing of coal. Seventh
Australia-New Zealand Conference on Geomechanics, Adelaide, pp. 203-208.
Medhurst, T.P. and Brown, E.T. (1998) A study of the mechanical behaviour of coal for pillar
design. International Journal of Rock Mechanics and Mining Sciences, 35 (8), pp. 1087-
1105.
Michelis, P. (1987) True triaxial yielding and hardening of rock. ASCE Journal of Geotechnical
Engineering, 113 (6), pp. 616-635.
Millar, P.J. and Murray, D.R. (1986) Triaxial testing of weak rocks including the use of triaxial
extension tests. Advanced Triaxial Testing of Soil and Rock, ASTM STP 977, Louisville,
Kentucky, American Society for Testing and Materials. pp. 376-386.
Milovanovitch, D. (1958) Recherches sur le Barrage Exprimental de L'Idbar au Moyen
D'Analyse lastostatique, D'Essais sur Modles Rduits et D'Auscultation du Barrage. In
Sixime Congrs des Grands Barrages, New York, ICOLD. Vol. C34 pp. 571-587.
Misra, B. (1972) Correlation of Rock Properties with Machine Performance. PhD, University
of Leeds.
Mogi, K. (1965) Deformation and fracture of rocks under confining pressure, 2: Elasticity and
plasticity of some rocks. Bulletin of the Earthquake Research Institute, Tokyo University,
43 pp. 349-379.
Mogi, K. (1966) Pressure dependence of rock strength and transition from brittle fracture to
ductile flow. Bulletin of the Earthquake Research Institute, University of Tokyo, 44 (1),
pp. 215-232.
Mogi, K. (1973) Rock fracture. Annual Review of Earth and Planetary Sciences, 1, pp. 63-85.
Moomivand, H. and Vutukuri, V.S. (1996) Effect of diameter-to-length ratio on the strength of
cylindrical specimens in triaxial tests. 7th Australia New Zealand Conference on
Geomechanics, Adelaide, South Australia, pp. 134-139.
Moon, B.P. and Selby, M.J. (1983) Rock mass strength and scarp forms in Southern Africa.
Geografiska annaler, Series A, 65A (1-2), pp. 135-145.
Moon, V., Russell, G. and Stewart, M. (2001) The value of rock mass classification systems for
weak rock masses: a case study example from Huntly, New Zealand. Engineering
Geology, 61 pp. 53-67.
Moroto, N. and Ishii, T. (1990) Shear strength of uni-sized gravels under triaxial compression.
Soils and Foundations, 30 (2), pp. 23-32.
Mostyn, G.R. (1988) General report - stability of slopes. Fifth Australia-New Zealand
Conference on Geomechanics, Sydney, pp. 435-445.
Mostyn, G. (1993) Discussion of: Indraratna et al (1993) Large-scale triaxial testing of
greywacke rockfill. Geotechnique, 44 (3), pp. 539-543.
Mostyn, G. & Douglas, K.J. (2000) Issues Lecture: The shear strength of intact rock and rock
masses. GeoEng2000: An International Conference on Geotechnical & Geological
Engineering, 19-24 November, Melbourne, Australia, Vol. 1, pp. 1389-1421, Technomic
Mostyn, G. and Ferguson, A.J. (1984) Analysis of strength test results - Pynang formation rocks
- Papua New Guinea. Fourth Australia-New Zealand Conference on Geomechanics, Perth,
pp. 112-117.
Mostyn, G., M.D. Helgstedt and K.J. Douglas (1997) Towards field bounds on rock mass
failure criteria. International Journal of Rock Mechanics and Mining Sciences Vol. 34 (3-
4): Paper No. 208.
Murrell, S.A.F. (1965) The effect of triaxial stress systems on the strength of rocks at
atmospheric temperatures. Journal of Geophysics, Astronomy Society, 10 pp. 231-281.
Murthy, Y.K., Mane, P.M. and Pant, B. (1979) Tensile failures in some stone masonry gravity
dams in India. Treizieme Congres des Grands Barrages, New Delhi, pp. 461-477.
Murthy, Y.K., Sreerami Reddi, M. and Rama Rao, M.S. (1976) Leakage investigations at the
Bhandardara dam. Water Power & Dam Construction, (Sep.), pp. 43-46.
Murti, N.G.K. (1967) Khadakwasla, the oldest masonry dam in India. In Ninth Congress on
Large Dams, Istanbul, ICOLD. Vol. Q34 pp. 895-915.
Nakayama, K., Itoga, F. and Inque, Y. (1982) Selection and quality control of materials for
rockfill dam of pumped storage project in phyllocrystalline schistose area. Fourteenth
Congres on Large Dams, Rio de Janeiro, ICOLD. Vol. Q55 pp. 23-45.
Nasseri, M.H., Rao, K.S. and Ramamurthy, T. (1997) Failure mechanism in schistose rocks.
International Journal of Rock Mechanics and Mining Sciences, 34 (3-4), pp. 219.
Naylor, D.J., Maranha Das Neves, E., Mattar JR, D. and Pinto, A.A.V. (1986) Prediction of
construction performance of Beliche Dam. Geotechnique, 36 pp. 359-376.
Neves, J.B. (1970) Control, mechanical characteristics and observation of Mira Dam. Tenth
Congres on Large Dams, Montreal, ICOLD. Vol. Q36 pp. 321-333.
Niandou, H., Shao, J.F., Henry, J.P. and Fourmaintraux, D. (1997) Laboratory investigation of
the mechanical behaviour of Tornemire shale. International Journal of Rock Mechanics
and Mining Sciences, 34 (1), pp. 3-16.
Nicholson, P.G. and Seed, R.B. (1992) Laboratory correction method for undrained testing of
coarse gravelly soils. Proceedings of the 45th Canadian Geotechnical Conference,
Toronto, Ontario, Canada, pp. 3/1 to 3/10.
Nishio, S. and Tamaoki, K. (1990) Stress dependency of shear wave velocities in diluvial gravel
samples during triaxial compression tests. Soils and Foundations, 30 (4), pp. 42-52.
Novello, E. and Johnston, I.W. (1995) Geotechnical materials and the critical state.
Geotechnique, 45 (2), pp. 223-235.
Ofoegbu, G.I. and Curran, J.H. (1991) Yielding and Damage of Intact Rock. Canadian
Geotechnical Journal, 28 pp. 503-516.
Ohnaka, M. (1995) A shear failure strength law of rock in the brittle-plastic transition regime.
Geophysical Research Letters, 22 (1), pp. 25-28.
Ohnaka, M., Akatsu, M., Mochizuki, H., Odedra, A., Tagashir a, F. and Yamamoto, Y. (1997) A
constitutive law for the shear failure of rock under lithospheric conditions. Tectonophysics,
277 (1), pp. 1-27.
Okubo, S., Fukui, K. and Nishimatsu, Y. (1996) A recent development of servo-controlled
testing method in rock mechanics. Proceedings of the Korea-Japan Joint Symposium on
Rock Engineering, (Li & Moon ed.) Seoul, Korea, The Korean Institute of Mineral &
Energy Resources Engineers. pp. 67-72.
Olwage, H.J.W. and Oosthuizen, C. (1984) Lessons learnt from South African dam incidents. In
Safety of Dams, Coimbra, A.A.Balkema, Rotterdam, pp. 33-39.
Orr, C.M. (1992) Assessment of rock slope stability using the rock mass rating (RMR) system.
The AusIMM Proceedings, 297 (2), pp. 25-29.
Orr, C.M. (1996) Use of the Rock Mass Rating (RMR) system in assessing the stability of rock
slopes. In Milestones in Rock Engineering the Bieniawski Jubilee Collection,
A.A.Balkema, Rotterdam, pp. 159-171.
Outland, C.F. (1963) Man-Made Disaster the story of St. Francis Dam. The Arthur H. Clark
Co., California.
Ouyang, Z. and Elsworth, D. (1991) A phenomenological failure criterion for brittle rock. Rock
Mechanics and Rock Engineering, 24 (3), pp. 133-153.
Ozbay, M.U., Dede, T. and Napier, J.A.L. (1996) Physical and numerical modelling of rock
fracture. The Journal of The South African Institute of Mining and Metallurgy, 96 (7), pp.
317-323.
Parkin, A.K. and Adikari, G.S.N. (1980) A theoretical investigation of the constructed
behaviour of a rockfill dam. 3rd Australia New Zealand Conference on Geomechanics,
Wellington, pp. 1-7 to 1-12.
Parkin, A.K. and Adikari, G.S.N. (1981) Rockfill deformation from large-scale tests.
Proceedings of the Tenth International Conference on Soil Mechanics and Foundation
Engineering, Stockholm, Balkema. Vol. 1 pp. 727-731.
Paolina, R., Appendino, M., Baldovin, E., Berra, M., Bianchini, A., Carabelli, E., Posta, U. and
Vielmo, I. (1991) Deterioration problems for concrete and masonry dams in Italy. Dix-
septieme Congres des Grands Barrages, Vienne, ICOLD. Vol. Q65 pp. 785-815.
Parker, R.D. (1900) Concerning the Austin Dam. The Engineering Record, 41 (26), pp. 610-
611.
Parry, R.H.G. (1995) Mohr circles, stress paths, and geotechnics, London, New York, Spon.
Patterson, H.B. (1900) The Austin Dam - Report to the Austin Water and Light Commission.
The Engineering Record (Jul. 28), pp. 79.
Patton, F.D. (1966) Multiple Modes of Shear Failure in Rock and Related Materials. PhD,
University of Illinois.
Pavlakis M. (1980) Pressuremeter Testing of Weathered Karroo Siltstone, Seventh Regional
Conference for Africa on Soil Mechanics and Foundation Engineering, South Africa.
Pells, P.J.N. (1975) Predicted displacements of the rock foundations of a major arch bridge.
Proceedings of the 6th Conference for Africa on Soil Mechanics, Durban, SAICE. pp. 207-
212.
Pells, P.J.N., Braybrooke, J.C., Mong, J. and Kotze, G.P. (1987). Cliff collapse associated with
mining activities. Proceedings of an Extension Course on Slope Instability and
Stabilisation, Sydney, Balkema. 359-385.
Pender, M.J. (1980) Friction and cohesion parameters for highly and completely weathered
Wellington greywacke. 3rd Australia New Zealand Conference on Geomechanics,
Wellington, pp. 1-171 to 1-175.
Penman, A.D.M., Charles, J.A. and Humphreys, J.D. (1982) Sandstone rockfill in two dams.
Fourteenth Congres on Large Dams, Rio de Janeiro, ICOLD. Vol. Q55 pp. 279-291.
Pierce, M., Brandshaug, T., and Ward, M. (2001) Slope stability assessment at the Main
Cresson Mine, Slope stability in surface mining, (Hustrulid et al eds), Littleton, Society
for Mining, Metallurgy and Exploration, Inc., pp. 239-250.
Priest, S.D. (1993). Discontinuity analysis for rock engineering, Chapman & Hall.
Queiroz, L. (1964) Geotechnical properties of weathered rock and behaviour of Furnas Rockfill
Dam. Huitime Congrs International des Grands Barrages, Q28 pp. 877-890.
Ramamurthy, T. (1985) Stability of rock mass. Indian Geotechnical Journal, pp. 1-74.
Ramamurthy, T. and Arora, V.K. (1994) Strength predictions for jointed rocks in confined and
unconfined states. International Journal for Rock Mechanics, Mining Sciences and
Geomechanical Abstracts, Vol. 31 (1), pp. 9-22.
Ramamurthy, T. and Gokhale, V.L. (1982) Stress-strain response of some intact sedimentary
rocks. Proceedings 4th International Congress IAEG, India, IAEG. Vol. 2 pp. 103-114.
Ramamurthy, T. and Gupta, K.K. (1980) Prediction of the behaviour of rockfill materials. Third
Australia-New Zealand Conference on Geomechanics, pp. 1-13 to 1-18.
Ramamurthy, T. and Kumar, N. (1978) Strength of rock-like materials. Proceedings of
GEOCON India Conference on Geotechnical Engineering, New Delhi, Vol. 1 pp. 340-346.
Ramamurthy, T., Seshagiri Rao, K. and Gupta, K.K. (1988a) Prediction of strength of intact
rocks. Evaluation and Applications in Geotechnical Engineering, Indian Geotechnical
Conference, Allahabad, Vol. 1 pp. 149-153.
Ramamurthy, T., Venkatappa Rao, G. and Sheshagiri Rao, K. (1988b) A non-linear strength
criterion for rocks. Fifth Australia-New Zealand Conference on Geomechanics, Sydney,
pp. 247-252.
Ramez, M.R.H. (1967) Fractures and strength of a sandstone under triaxial compression.
International Journal of Rock Mechanics and Mining Sciences, 4 (3), pp. 257-268.
Ransome, F.L. (1928) Geology of the St. Francis Dam-site. Economic Geology, 23 (5), pp. 553-
563.
Rao, K.L. (1960) Failures in weirs and dams. Bhagirath, VII (1), pp. 11-17.
Rao, K.S., Venkatappa Rao, G. and Ramamurthy, T. (1988) Strength and deformation behaviour
of sandstones. Journal of the Institution of Engineers (India), 68 (CI4), pp. 181-186.
Rao, N.S.G. (1967) Failure of the Khadakwasla and Panshet Dams. Journal of the Institution of
Engineers (India), 47 (11), pp. 1123-1144.
Read, S.A.L., Dellow, G.D. and Barrel, D.J.A. (1995). Waitaki Power Station. Review of
geological and foundation data. Unpublished client report 353911.01., Institute of
Geological & Nuclear Science Ltd.
Read, S.A.L., Dellow, G.D. and Perrin, N.D. (1996). Aviemore Power Station. Review of
geological and foundation data. Unpublished client report 35504B.10, Institute of
Geological & Nuclear Science Ltd.
Reineking, V.H. (1914) Reconstruction of the Elwha River dam. Engineering Record, 69 (13),
pp. 372-375.
Renner, J. and Rummel, F. (1996) The effect of experimental and microstructural parameters on
the transition from brittle failure to cataclastic flow of carbonate rocks. Tectonophysics,
258 pp. 151-169.
Ribacchi R. (1984) Rock Mass Deformability: In-situ Tests, their interpretation and Typical
Results in Italy, 2
nd
International Symposium on Field Measurements in Geomechanics,
(Sakurai ed.), pp171-192.
Ribacchi, R. (2000) Mechanical tests on pervasively jointed rock material: Insight into rock
mass behaviour. Rock Mechanics and Rock Engineering, 33 (4), pp. 243-266.
Rich, M. (1995) A Study of the Failure of Concrete Gravity Dams Through Their Foundations.
Unpublished BE Thesis, The University of New South Wales.
Richmond and Smith (1979) Prediction of Mining Conditions from Bore Core Information. PR
79-14. Australia. Coal Ind. Research Laboratories Ltd.
Robertson, E.C. (1955) Experimental study of the strength of rocks. Bulletin of the Geological
Society of America, 68 pp. 1275-1314.
Robertson, A.M. (1988) Estimating weak rock strength. SME Annual Meeting, Phoenix,
Arizona, Society of Mining Engineers, Preprint No. 88-145. pp. 1-5.
Robertson, A.M., Olsen, R.S. and Pierce, G.L. (1987) Assessment of weak altered rock masses
at the Island Copper Mine. SME Annual Meeting, Denver, Colorado, Society of Mining
Engineers.
Romana, M. (1985) New adjustment ratings for application of Bieniawski classification to
slopes. International Symposium on the Role of Rock Mechanics, Zacatecas, ISRM. pp. 49-
53.
Romana, M. (1991) SMR classification. 7th International Congress on Rock Mechanics,
(Herausgeber ed.) Aachen, ISRM. Vol. 2 pp. 955-960.
Romana, M. (1996) The SMR geomechanical classification for slopes: a critical ten-years
review. Proceedings of the Eighth International Conference and Field Trip on Landslides,
Granada, Spain, A.A.Balkema. pp. 255-267.
Romana, M. (1997) The geomechnical classification SMR for slope correction. Proceedings of
the International Conference and Seminar on Tunnelling Under Difficult Conditions and
Rock Mass Classification, Basel, Switzerland, Independent Technical Conferences Ltd. pp.
1-16.
Rosenberg, W. (1900) The failure of the Austin Dam. The Engineering Record, 41 (20), pp.
468-469.
Rouve, G., Strack, B., Idel, K., Beckman, J., Hager, M., Boucek, B., Sowers, G. and Gilg, B.
(1977) Lectures of the seminar failures of large dams - Reasons and remedial measures. In
Institute for Foundation Engineering, Soil Mechanics, Rock Mechanics and water Ways
Construction, Aachen, Vol. 4 pp. 1-242.
Rummel, F. and Fairhurst, C. (1970) Determination of the post-failure behavour of brittle rock
using a servo-controlled testing machine. Rock Mechanics, 2 pp. 189-204.
Rutter, E.H. (1991) On the influence of porosity on the low-temperature brittle-ductile transition
in siliclastic rocks. Journal of Structural Geology, 13 (5), pp. 609-614.
Santiago J.L. and Garcia F.A. (1986) Design Parameters for Grouted Piles in Rock from In-situ
Load Test, USA, Use of In-situ Tests in Geotechnical Engineering, Proceedings of In-situ
86, pp. 968-984.
Sarac, Dz. and Popovic, M. (1985) Shear strength of rockfill and slope stability, Proceedings of
the Eleventh International Conference on Soil Mechanics and Foundation Engineering,
San Francisco, pp. 641-645.
Sawyer, W.H. (1911) The probable cause of the Austin Dam failure. Engineering Record, 64
(19), pp. 530-532.
Scheibner, B.J. (1980) Geology of the single-entry project at Sunnyside Coal Mines 1 and 2,
Sunnyside, Utah. Report of Investigations 8402, United States Department of the Interior,
Bureau of Mines.
Schmidt, K.M. and Montgomery, D.R. (1996) Rock mass strength assessment for bedrock
landsliding. Environmental & Engineering Geoscience, II (3), pp. 325-338.
Schnitter, N.J. (1994) A History of Dams - The Useful Pyramids. A.A.Balkema, Rotterdam, pp.
266.
Schnitter, N.J. (1993) Dam failures due to overtopping. In International Workshop on Dam
Safety Evaluation, (Kreuzer & Taylor ed.) Grindelwald, Switzerland, Dam Engineering.
Vol. 1 pp. 13-19.
Schober, W. and Rostek, R. (1979) The shear resistance in the interface of the fill material or
the rock overburden and the bedrock. Treizieme Congres des Grands Barrages, New
Delhi, Vol. Q48 pp. 989-1003.
Schock, R.N., Heard, H.C. and Stephens, D.R. (1973) Stress-strain behaviour of a granodiorite
and two greywackes on compression to 20 kilobars. Journal of Geophysical Research, 78
(26), pp. 5922-5941.
Schultz, R.A. (1993) Brittle strength of basaltic rock masses with applications to Venus. Journal
of Geophysical Research - E - Planets, 98 (E6), pp. 10883-10895.
Schultz R.A. (1995) Limits on Strength and Deformation Properties of Jointed Basaltic Rock
Masses, Rock Mechanics and Rock Engineering, pp. 1 15.
Schuyler, J.D. (1906) The failure of the Santa Catalina Dam near Durango, Mexico.
Engineering News, 57 (17).
Schuyler, J.D. (1908). Reservoirs for Irrigation, Water-Power and Domestic Water-Supply.
John Wiley & Sons, New York,pp.573.
Schwab, H. and Pircher, W. (1979) Roughening of smooth and steeply sloping rock surfaces in
the foundation of Finistertal rockfill dam. Treizieme Congres des Grands Barrages, New
Delhi, Vol. Q48 pp. 387-403.
Schwartz, A.E. (1964) Failure of rock in the triaxial shear test. Proceedings of the Sixth
Symposium on Rock Mechanics, The University of Missouri at Rolla, pp. 109-151.
Scott, G.A. and Von Thun, J.L. (1993) Draft - Interim Guidelines Geotechnical Studies for
Concrete Dams. Bureau of Reclamation, Denver.
Scott, T.E. and Nielsen, K.C. (1991a) The effects of porosity on fault reactivation in sandstones.
Journal of Geophysical Research, 96 (B2), pp. 2353-2362.
Scott, T.E. and Nielsen, K.C. (1991b) The effects of porosity on the brittle-ductile transition in
sandstones. Journal of Geophysical Research, 96 (B1), pp. 405-414.
Selby, M.J. (1980) A rock mass strength classification for geomorphic purposes: with tests from
Antarctica and New Zealand. Zeitschrift fr Geomorphologie, N.F., 24 (1), pp. 31-51.
Selby, M.J. (1982a) Controls on the stability and inclinations of hillslopes formed on hard rock.
Earth Surface Processes and Landforms, 7 (5), pp. 449-467.
Selby, M.J. (1982b) Rock mass strength and the form of some inselbergs in the central Namib
Desert. Earth Surface Processes and Landforms, 7 (5), pp. 489-497.
Selby, M.J. (1987) Rock slopes. In Slope Stability, Geotechnical Engineering and
Geomorphology, John Wiley & Sons Ltd, Chichester, pp. 475-504.
Selby, M.J. (1993) Hillslope Materials and Processes. Oxford University Press, Oxford, pp.
451.
Serafim, J.L. (1981a) Elements for a thorough statistical analysis of dam failures. Revista da
Universidade de Coimbra.
Serafim, J.L. (1981b) Safety of dams judged from failures. Water Power & Dam Construction.
(Dec.), pp. 32-35.
Serafim, J.L. and Pereira, J.P. (1983) Considerations on the geomechanical classification of
Bieniawski, International Symposium on Engineering Geology and Underground
Construction, Portugal, IAEG, Vol. 1, pp. II33-II42.
Seto, M., Nag, D.K., Vutukuri, V.S. and Katsuyama, K. (1997) Effect of chemical additives on
the strength of sandstone. International Journal of Rock Mechanics and Mining Sciences,
34 (3-4), pp. 280.
Seto, M., Vutukuri, V.S., Nag, D.K. and Katsuyama, K. (1996) Chemically induced change in
strength of sandstone. Proceedings of the Korea-Japan Joint Symposium on Rock
Engineering, (Li & Moon ed.) Seoul, Korea, The Korean Institute of Mineral & Energy
Resources Engineers. pp. 151-156.
Shah, S. (1992) A Study of the Behaviour of Jointed Rock Masses. PhD, University of Toronto.
Shah, S. and Hoek, E. (1992) Simplex reflection analysis of laboratory strength data. Canadian
Geotechnical Journal, 29 pp. 278-287.
Shand, N. and Pells, P.J.N. (1970) Experience in the design and construction of reinforced
rockfill dams. Dixieme Congres des Grands Barrages, Montreal,
Sharma, V.M., Venkatachalam, K. and Roy, A. (1994) Strength and deformation characteristics
of rockfill materials. Proceedings of the Thirteenth International Conference on Soil
Mechanics and Foundation Engineering, New Delhi, Balkema. Vol. 3 pp. 959-962.
Shea-Albin, V.R., Hansen, D.R. and Gerlick, R.E. (1991) Elastic wave velocity and attenuation
as used to define phases of loading and failure in coal. Report of Investigations 9355,
United States Department of the Interior, Bureau of Mines.
Sheorey, P.R. (1997) Empirical Rock Failure Criteria. A.A.Balkema, pp. 200.
Sheorey, P.R., Biswas, A.K. and Choubey, V.D. (1989) An empirical failure criterium for rocks
and jointed rock masses. Engineering Geology, 26 (2), pp. 141-159.
Sherman, C.W. (1910) Arched Masonry dam at Las Vegas, N.M. Engineering News, 64 (17;
Oct. 27), pp. 446-447.
Shimada, M. (1991) Comment on "The effects of pressure and porosity on the micromechanics
of the brittle-ductile transition in quartzite". Journal of Geophysical Research, 96 (B7), pp.
11,877-11,879.
Shimada, M. (1993) Lithosphere strength inferred from fracture strength of rocks at high
confining pressures and temperatures. Tectonophysics, 217 pp. 55-64.
Shimada, M. (2000) Mechanical Behavior of Rocks under High Pressure Conditions.
Geomechanics Research Series 2 (Kwasniewski ed.), Balkema, Rotterdam, pp. 178.
Shimada, M. and Cho, A. (1990) Two types of brittle fracture of silicate rocks under confining
pressure and their implications in the earth's crust. Tectonophysics, 175 (1-3), pp. 221-235.
Shimada, M., Cho, A. and Yukutake, H. (1983) Fracture strength of dry silicate rocks at high
confining pressures and activity of acoustic emission. Tectonophysics, 96 (1-2), pp. 159-
172.
Shuk, T. (1965) Discussion of paper by Langejan. 6th International Conference on Soil
Mechanics & Foundation Engineering, Canada, ISSMFE. Vol. 3 pp. 576-577.
Silveira, A. (1990) Some considerations on the durability of dams. Water Power & Dam
Construction. (Feb.), pp. 19-28.
Silveira, A.F. (1984) Statistical analysis of deteriorations and failures of dams. In Safety of
Dams, (Serafim ed.) Coimbra, A.A. Balkema. pp. 55-60.
Sizer, F.L. (1908) The break in the Hauser Lake dam, Montana. Engineering News (Apr. 30),
pp. 491-492.
Sjberg, J., Sharp, J.C., and Malorey, D.J. (2001) Slope stability at Aznalcllar, Slope stability
in surface mining, (Hustrulid et al eds), Littleton, Society for Mining, Metallurgy and
Exploration, Inc., pp. 183-202.
Smith, J.L., DeVries, K.L., Bushnell, D.J. and Brown, W.S. (1969) Fracture data and stress-
strain behaviour of rocks in triaxial compression. Proceedings of the Society for
Experimental Stress Analysis, XXVI (2), pp. 348-355.
Smith, N. (1972) A History of Dams. The Citadel Press, Secaucus.
Smith, N.A.F. (1995) Unhappy anniversary - the disaster at Bouzey in 1895. International
Water Power & Dam Construction (Apr.), pp. 40.
Smoltczyk, U. and Gartung, E. (1979) Geotechnical properties of a soft Keuper sandstone.
Proceedings of the 4th Congress International Society for Rock Mechanics, Montreux,
Vol. 2 pp. 639-644.
Sowers, G.F. (1971) Discussion of: Leps (1970) Review of shearing strength of rockfill. Journal
of the Soil Mechanics and Foundations Division Proceedings of the American Society of
Civil Engineers, 97 (SM5), pp. 794-801.
Stapleton, D.H. (1976) Geological hazards and water storage. Bulletin of the International
Association of Engineering Geology (14), pp. 249-262.
Steffen, O.K.H. (1976) Research and development needs in data collection for rock engineering.
Proceedings of the Symposium on Exploration for Rock Engineering, (Bieniawski ed.)
Johannesburg, Balkema. Vol. 2 pp. 93-104.
Steger, W. and Unterberger, W. (1990) Rock slopes and dam abutments in jointed rock: FE-
Analyses using Gap-Friction Elements and contact procedures. In Mechanics of Jointed
and Faulted Rock, Rossmanith (ed.), (Rossmanith ed.) Technical University of Vienna,
Balkema, Rotterdam. pp. 495-501.
Stephansson, O., Swan, G. and Leijon, B. (1978) Mechanical properties of Stripa granite.
Proceedings of the Seminar on In Situ Heating Experiments in Geological Formations,
Stripa, pp. 191-204.
Stoney, S.M. and Dhir, R.K. (1978) Measurement of anisotropy in rocks. Proceedings of
GEOCON India Conference on Geotechnical Engineering, New Delhi, Vol. 1 pp. 351-359.
Tarua, A. (1997) Field validation of the Hoek-Brown failure criterion at Dog Face Rock,
Katoomba. B.E., Unpublished, The University of New South Wales.
Tavares, L.V. and Serafim, J.L. (1983) Probabilistic study on failure of large dams. A.S.C.E.,
Journal of the Geotechnical Engineering Division, 109 (11), pp. 1483-1486.
Taylor, F.S. (1915) Reconstruction of Austin, Tex., masonry dam. Engineering News, 73 (No.
22; Jun 3), pp. 1089-1093.
Taylor, T.U. (1900) The latest soundings at the break in the Austin Dam. Engineering News,
XLIV (23), pp. 390-391.
Terzaghi, K. (1962) Does foundation technology really lag? Engineering News Record, 168 (7),
pp. 58-59.
The Engineer. (1896) Failure of the great dam at Bouzey. The Engineer (Jan 17), pp. 50-51.
The Engineer (1928a) The collapse of the St. Francis dam. The Engineer (Apr 20), pp. 438.
The Engineer (1928b) The St. Francis dam disaster. The Engineer (May 25), pp. 568-570.
The Engineer. (1942) Historic accidents and disasters. The Engineer (Dec. 4), pp. 452-455.
Thiers, G.R. and Donovan, T.D. (1980) Field density, gradation, and triaxial testing of large-
size rockfill for Little Blue Run Dam. Laboratory Shear Strength of Soil. ASTM STP 740,
Chicago, American Society for Testing and Materials. pp. 315-325.
Tosun, H., Mirata, T., Mollamahmutoglu, M. and Colakoglu, N.S. (1999) Shear strength of
gravel and rockfill measured in triaxial and prismatic wedge shear tests. Electronic Journal
of Geotechnical Engineering pp. Paper No. 9903.
Tsiambaos, G. and Telli, D. (1991) Application of rock mass classification systems on stability
of limestone slopes. Landslides. Proceedings of the Sixth International Symposium, (Bell
ed.) Christchurch, Balkema. Vol. 2 pp. 1065-1069.
Ulusay, R., Arikan, F., Yoleri, M.F. and Caglan, D. (1995) Engineering geological
characterization of coal mine waste material and an evaluation in the context of back-
analysis of spoil pile instabilities in a strip mine, SW Turkey. Engineering Geology, 40 pp.
77-101.
Unal, E. (1996) Modified rock mass classification: M-RMR System. Milestones in Rock
Engineering, The Bieniawski Jubilee Collection, Balkema. pp. 203-223.
USBR (1961) Laboratory tests of rock cores from the foundation of Shihmen Dam Taiwan,
Formosa. Concrete Laboratory Report No. C-1157, United States Department of the
Interior Bureau of Reclamation.
USBR (1965) Laboratory tests of foundation rock cores Morrow Point damsite Colorado River
storage project. C-1157, United States Department of the Interior Bureau of Reclamation.
USBR (1966) Summary of large triaxial shear tests for silty gravels earth research studies. EM-
731, United States Department of the Interior Bureau of Reclamation - Soils Engineerin g
Branch, Division of Research.
USBR (1996) Bureau of Reclamation Concrete Dams. [Online] US Bureau of Reclamation.
Available: http://www.usbr.gov/cdams/. Accessed on: 24 Mar. 1997.
USCOLD (1975) Lessons From Dam Incidents USA. ASCE, New York.
USCOLD (1988) Lessons From Dam Incidents. USA-II. ASCE, New York.
USCOLD Committee on Dam Safety (1996) Anthology of Dam Modification Case Histories.
USCOLD, Denver, pp. 138.
Vallerga, B.A., Seed, H.B., Monismith, C.L. and Cooper, R.S. (1956) Effect of shape, size, and
surface roughness of aggregate particles on the strength of granular materials. Road and
Paving Materials Presented at the Second Pacific Area National Meeting, ASTM STP
212, Los Angeles, American Society for Testing Materials. pp. 63-76.
Varshney, R.S. and Raheem, M.A. (1971) Concrete dam disasters and remedies. Indian Journal
of Power & River Valley Development, (July), pp. 248-256.
Veesaert, C. (198?) Lessons from dam incidents.
Veltrop, J.A. (1988) Concrete arch dams. In Development of Dam Engineering in the United
States, Pergamon Press, New York, pp. 219-532.
Vernik, L., Lockner, D. and Zoback, M.D. (1992) Anisotropic strength of some typical
metamorphic rocks from the KTB pilot hole, Germany. Scientific Drilling, 3 pp. 153-160.
Vogel, A. (1984) Failures of masonry and concrete dams in Europe. In International Conference
on the Safety of Dams, (J. L. Serafim ed.) Coimbra, A.A.Balkema. pp. 45-54.
Vogel, A. (1994) Bibliography of the History of Dam Failures. Data Station for Dam Failures,
Vienna.
Von Thun, J.L. (1985) Application of statistical data from dam failures and accidents to risk-
based decision analysis on existing dams. Bureau of Reclamation Engineering and
Research Centre.
Vutukuri, V.S. (1998) Strength of rock and rock mass, International Conference on
Geomechanics/Ground Control in Mining and Underground Construction, (Aziz &
Indraratna eds), 14-17 July, Wollongong, Australia, pp. 49-58.
Vutukuri, V.S. and Farough Hossaini, S.M. (1993) Correlation between the effect of confining
pressure on compressive strength in triaxial test and the effect of dia/height ratio on
compressive strength in unconfined compression test. 12th Conference on Ground Control
in Mining, (Peng ed.) Morgantown, pp. 316-321.
Wang, R. and Kemeny, J.M. (1995) A new empirical criterion for rock under polyaxial
compressive stresses. 35th US Symposium on Rock Mechanics, Reno, Nevada, Balkema.
pp. 453-458.
Waversik, W.R. and Fairhurst, C. (1970) A study of brittle rock fracture in laboratory
compression experiments. International Journal of Rock Mechanics and Mining Sciences,
7 pp. 561-575.
Wawersik, W.R. and Brace, W.F. (1970) Post-failure behavour of a granite and diabase. Rock
Mechanics, 3 pp. 61-85.
Wegmann, E. (1889) The Design and Construction of Masonry Dams. John Wiley & Sons,
New York.
Williams, D.J. and Walker, L.K. (1984) Laboratory and field strength of mine waste rock.
Fourth Australia-New Zealand Conference on Geomechanics, Perth, pp. 118-123.
Won, G.W. (1985) Engineering properties of Wianamatta Group rocks from laboratory and in-
situ tests. In Engineering Geology of the Sydney Region (Pells ed.), Balkema, Sydney, pp.
143-155.
Wright, A.J. and Hosking, A.D. (1966) Shear strength of rockfill. Construction Materials Report
No. SM.1402, Snowy Mountains Hydro_Electric Authority.
Wu, B., Marsden, J.R. and Hudson, J.A. (1990) Undrained mechanical behaviour of mudstone.
The Engineering Geology of Weak Rock, University of Leeds, pp. 105-114.
Yildiz, D. and zcek, E. (1994) Prediction of scour depth from free falling flip bucket jets.
International Water Power & Dam Construction (Nov.), pp. 50-56.
Yoshida, N., Morgenstern, N.R. and Chan, D.H. (1990) A failure criterion for stiff soils and
rocks exhibiting softening. Canadian Geotechnical Journal, 27(2), pp. 195-202.
Yoshinaka, R., Michiie, T., Norose, S. and Koshizuka, K. (1985) Fracture development in large
underground quarry of soft rock and some consideration by numerical analysis. The Role
of Rock Mechanics in Excavations for Mining and Civil Works - ISRM Symposium,
Zacatecas, Mexico, pp. 257-264.
Yudhbir, Lemanza, W. and Prinzl, F. (1983) An empirical failure criterion for rock masses.
Proceedings 5
th
Congress for I.S.R.M., Melbourne.
Yumlu, M. and Ozbay, M.U. (1995) Technical note: A study of the behaviour of brittle rocks
under plane strain and triaxial loading conditions. International Journal for Rock
Mechanics, Mining Sciences and Geomechanical Abstracts, 32 (7), pp. 725-733.
Zaretsky, Y.K., Chernilov, A.G., Vorontsov, E.I. and Garitselov, M.Y. (1978) Triaxial studies
of mechanical properties of rockfills. Proceedings of GEOCON India Conference on
Geotechnical Engineering, New Delhi, Vol. 1 pp. 145-150.
Intact Rock Database References from Sheorey (1997)
Barat, D. (1995) Personal communication from C.M.R.I., Dhanbad.
Bieniawski, Z.T. and Denkhaus, H.G. (1967) Personal communication from M.A. Kwasniewski,
Gliwice, 1995.
Borecki, M., Kwasniewski, M., Pacha, J., Olesky, S., Berszakiewicz, Z. and Guzik, J. (1982)
Triaxial compressive strength of two mineralogic/diagenetic varieties of coal-measure,
fine-medium grained Pniowek and Anna sandstones tested under confining pressures up to
60 MPa. Prace Instytutu PBKiOP Politechniki Slaskiej, 119 (2),.
Bowen, R. (1984) Geology in Engineering. Elsevier, New York, pp. 416.
Chirkov, S.E. (1973) Test results in properties of rock in the conditions of triaxial nonuniform
compression. In Studies of Strength and Deformability of Rocks (Barton ed.), Nauka,
Moscow, pp. 113-134.
Glushko, V.T. and Kirnichanskiy, G.T. (1974) Engineering - Geological Prognosticating of
Stability of the Openings in Deep Coal Mines. Nedra, Moscow,
Gnirk, P.F. and Cheatham, J.B. (1963) Indentation experiments on dry rocks under pressure.
Transactions of the Society of Petroleum Engineers, AIME, 228 pp. 1031-1039.
Gnirk, P.F. and Cheatham, J.B. (1965) An experimental study of single bit tooth penetration
into dry rock at confining pressures of 0-5000 psi. Journal of the Society of Petroleum
Engineers, 5 pp. 117-130.
Gustkiewicz, J. (1984) Personal communication from M.A. Kwasniewski, Giliwice.
Harza Engineering Co. (1976) Comprehensive ground control study of a mechanised longwall
operation, final report. Vol 2. USBM OFR 5(2)-775(2)-77, United States Bureau of Mines.
Heard, H.C., Abey, A.E., Bonner, B.P. and Schock, R.N. (1974) Mechanical behaviour of dry
Westerly granite at high confining pressure. Lawrence Livermore Laboratory Report
UCRL 51642.
Hoshino, K., Koide, H., Inami, K., Iwamura, S. and Mitsui, S. (1972) Mechanical properties of
Japanese tertiary sedimentary rocks under high confining pressures. Geological Survey
Report 244, Japanese Geological Survey.
Hoshino, K. and Mitsui, S. (1975) Mechanical properties of palaeogene and cretaceous rocks in
Shikoku under high pressure. Journal Japanese Association of Petroleum Technologists,
40 pp. 166-173.
Illnitskaya, E.I., Teder, R.I., Vatolin, E.S. and Kuntysh, M.F. (1969) Properties of rocks and
methods of their determination. Nedra, Moscow,
Kuntysh, M.F. (1964) Investigation of methods of determining the basic physico-mechanical
characteristics of rocks, used while solving the problem of rock pressure. Cand. Technical
Sci. Thesis, Moscow.
Kwasniewski (1995) Personal communication from Gliwice, Poland.
Ramamurthy, T. (1989) Personal communication from I.I.T., Delhi.
Rao, K.S., Rao, G.V. and Ramamurthy, T. (1983) Strength anisotropy of a Vindhyan sandstone.
Indian Geotechnical Conference, Madras, Vol. 1 pp. VI41-VI48.
Singh, S.K. (1995) Personal communication from C.M.R.I., Dhanbad.
Wilhelmi, B. and Somerton, W.H. (1967) Simultaneous measurement of pore and elastic
properties of rocks under triaxial stress conditions. Journal of the Society of Petroleum
Engineers, 7 pp. 283-294.

Shear Strength of Rock

Transféré par

Informations du document

Description originale:

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Shear Strength of Rock

Transféré par

Droits d'auteur :

Formats disponibles

THE SHEAR STRENGTH OF ROCK MASSES

direct shear - quartz schist - e=0.28

+ + + condition defect spacing defect strength intact Ratings RQD GSI

Vous aimerez peut-être aussi