Purdue University

Purdue e-Pubs
JTRP Technical Reports Joint Transportation Research Program

1977

Computerized Slope Stability Analysis for Indiana Highways, Vol. I

Eva Boutrup

Recommended Citation Boutrup, E. Computerized Slope Stability Analysis for Indiana Highways, Vol. I. Publication FHWA/ IN/JHRP-77/25. Joint Highway Research Project, Indiana Department of Transportation and Purdue University, West Lafayette, Indiana, 1977. doi: 10.5703/1288284313961.
This document has been made available through Purdue e-Pubs, a service of the Purdue University Libraries. Please contact [email protected] for additional information.

IS

HfilllEEfiltiG JJBiMii

BRARIES

LIBRARIES

S3iavaan

SBiavaai

N* &

*0j

%
ES

LIBRARIES

LIBRARIES

saiavaan

saiavau

"

ft

&>*
'&*

CO

&
fi
LIBRARIES

ES

LIBRARIES

saiavaan

saiavaan

saiavaa

RARIES

LIBRARIES

LIBRARIES

saiavaan

saia

1133

^
0*
RARIES s*
LIBRARIES
f

V
0*
s*

LIBRARIES

Sdld Vdtii

saiavaan

saii
**-*

*0y N

SfSVV

-1/- 'c

&"r>~
LIBRARIES
LIBRARIES

^
RAklES

ft

saiavaan

saiavaan

S 31

I*

i~m

"0

Digitized by the Internet Archive

in

2011 with funding from

Indiana Department of Transportation

LYRASIS members and Sloan Foundation;

http://www.archive.org/details/computerizedslopOObout

SCHOOL OF
30.57

CIVIL

ENGINEERING
!

UNIVERSITY LIBRARIES

JUN7

1978

JOIN^WGmVAY
RESEARCH PROJECT
JHRP-77-25

COMPUTERIZED SLOPE STABILITY ANALYSIS FOR INDIANA HIGHWAYS

Vol.

J 7!

X~~

(6

UNIVERSITY

HIGHWAY COMMISSION

Final Report

COMPUTERIZED SLOPE STABILITY ANALYSIS

FOR INDIANA HIGHWAYS

TO:

F. McLaughlin, Director Joint Highway Research Project J.

December, 1977
Project:

C-36-36L

FROM:

Michael, Associate Director Joint Highway Research Project

H.

L.

File:

6-14-12

Attached is the Final Report "Computerized Slope Stability Analysis for Indiana Highways", submitted for acceptance in fulfillment of the objectives The research and of the approved JHRP Research Study of the same title. report were performed by Ms. Eva Boutrup, Graduate Instructor on our staff, under the direction of C. W. Lovell and W. D. Kovacs of our staff.
The report presents results of continuous work on the computer program, This includes STABL, developed under a previous JHRP Study (JHRP- 7 5-8 ) results of a parametric study conducted on the STABL program, comparisons with other methods of slope stability analysis, and corrections and modifications of the original program. The corrected and improved STABL program, named STABL2, is a valuable tool for geotechnical engineers in the evaluation of the stability of highway slopes.
.

This Report is in two volumes with Vol. II consisting primarily of the STABL2 Program and having such a subtitle.

Copies of the The Report is submitted for approval and acceptance. report will also be submitted to ISHC for their review, comment and similar acceptance.

Respectfully submitted,

,^/

7k^^^

Harold L. Michael Associate Director

HLM:ms
cc:

Altschaeffl Bevilacqua W. L. Dolch R. L. Eskew G. D. Gibson W. H. Goetz M. J. Gutzwiller

A.
G. 0.

G.
D.
K.

K.

M.

Hoover Marsh R. D. Miles P. L. Owens G. T. Satterly

R.

E. R. F.

Hal lock Hancher

M.

Scholer Scott K. C. Sinha C. A. Venable L. E. Wood E. J. Yoder S. R. Yoder

C. F.
B.

Final Report

COMPUTERIZED SLOPE STABILITY ANALYSIS

FOR INDIANA HIGHWAYS
Vol.
I

by

Eva Boutrup Graduate Instructor in Research

Joint Highway Research Project

Project No.
File No.

C-36-36L
6-14-12

Prepared as Part of an Investigation

Conducted by
Joint Highway Research Project Engineering Experiment Station Purdue University
in cooperation with the

Indiana State Highway Commission

Purdue University West Lafayette, Indiana December, 1977

iii

ACKNOWLEDGMENTS

want to express

ray

thanks to my major Professor, C. W. Lovell,

,

and coadvisor, Professor W. D. Kovacs

support during this research.

Purdue University, for continuing

Thanks also goes to people from the Indiana State Highway Commission
for their participation and cooperation in the improvements of the STABL

program.

Mr. Sisiliano and Mr. Reeves from the Division of Materials and

Tests provided material for the comparisons reported in Chapter VI, and

made several suggestions for improvements of STABL.

Mr. Bellinger and

Mr. McKey from the ISHC Computer Center, Indianapolis, were very helpful

in the process of getting STABL on the IBM computer.

This included a

significant amount of error tracing, for which they introduced a special

subroutine (TURKEY) that provided a regulated error trace key search.

Further
questions
,

want to thank other users of the STABL program for their

,

comments

and suggestions that helped to point out some

uncertainties with respect to the use of STABL.

(and hopefully clarified) in Chapter IV.

These are discussed

Thanks, finally goes to Ms. Edith Vanderwerp and Ms. Jan Bollinger
for typing draft and parts of the final report, and to Ms. Peggy McFarren who drafted the figures.

iv

TABLE OF CONTENTS
Page

LIST OF TABLES

LIST OF FIGURES
LIST OF ABBREVIATIONS AND SYMBOLS

xiv
xxv

HIGHLIGHT SUMMARY
CHAPTER
I

xxxiii

INTRODUCTION
1.1

Purpose of the Present Study

1
2
h h

1.2
II

Organization of Report

SLOPE STABILITY ANALYSIS

2.1

General Description

2.2

Different Methods of Slope Stability Analysis

2.2.1

6
6 7

Early Development
The
cf>

2.2.2

Method

2.2.3
2.2.1*

The Logarithmic Spiral Procedure

The Friction Circle Method

11
11

2.2.5

Method of Slices
2.2.5.1

Ordinary Method of Slices

Bishop's Method

13
17

2.2.5-2 2.2.5.3
2.2.5.H
2.2.6

Modified Bishop Method

18
18

Other Procedures

Methods other than Limiting Equilibrium

20
21 22

2.2.7 2.2.8
2.3

Probabilistic Approach
Comparison of Methods

STABL - a Computer Program for General Slope

Stability Analysis

2U

Page
III

SENSITIVITY ANALYSIS
3.1

28
28

Introduction
Procedure for Analysis
Cases Studied
3.3.1

3.2 3.3

28 29

Sensitivity Analysis of Homogeneous Slopes

3.3.1.1

...

29
31

Analysis by STABL
Results of Stability Analysis of

3.3.1.2

Homogeneous Slopes
3.3.1.3

33

Sensitivity of the Factor of Safety

to Input Variables
1*1

3.3.1.1*

Conclusion and Recommendations

60

3.3.2

Sensitivity Analysis of an Embankment Founded

on a Soil with a Weak Seam
63

3.3.2.1

Factor of Safety for Failure Determined by a Weak Soil Seam

66
68

3.3.2.2

Sensitivity Analysis
Results of Stability Analysis for

3.3.2.3

Embankments
3.3.2.1*

69

Results Applied to Sensitivity

Analysis of Embankments
3.3.2.5
IV

89 95

Summary and Conclusions

.

MODIFICATIONS OF THE STABL PROGRAM

k.l

96 96

The Modified Sliding Block Routine (BL0CK2)

I*.

1.1

Generation of Active and Passive Portions of

the Sliding Block According to Rankine's Theory
.

98 99 99

U.1.2
1*.2

Application of the New Sliding Block Routine

Option of Several Piezometric Surfaces

1+.2.1

Program Modifications
Change in Data Cards

101

U.2.2
1+.2.3

102
.

Application of more than one Piezometric Level

102

k.3

Modified Bishop Factor of Safety for Circular

Failure Surfaces
1*.3.1

lOo
106 108

Program Modifications
CIHCL2 and SURBIS

U.3.2

vi

Page U.U
U.5

Example Problem for Revised STABL Program

Some Advise in the Use of STABL
H.5.1

108

110 110

Preliminary Analysis
Use of STABL
J

Stability Charts

U.5.2

123

.5.2.1

How to Define Limitations for

Search Routines

123

H.5-2.2

Clockwise and Counterclockwise

Direction Limits
1+.5-3
H.5.I*

12**

Strength along Soil Boundaries

126 126

Analysis Involving Earthquake Loadings

k.5.k.l
Effect of Vertical Earthquake Loading
on the Stability of a Slope

126
129 131

U.5.U.2
U.5.1+.3
k. 5 .h.k
i*.5-5

Cavitation Pressure
Illustrative Example
Dynamic Earthquake Analysis

13

1*

Misleading Factors of Safety

13^ 137

COMPARATIVE DESIGN STUDY

5.1

Description of Problem
Types of Analysis

137

5.2
5.3

137
1^2

Methods of Analysis
5.3.1

The Simplified Bishop Method

1^2
ll+2

5.3.2

Morgenstern-Price Analysis

5.3.3
5.k

Method Used in the STABL Program

. .

1U3
lUU

Soil Strength Parameters Used in Stability Analyses

5.1+.1

Corrected Field Vane Shear Strength

SHANSEP Anisotropic Shear Strength

lM
ikk

5.U.2
5.5

Results and Comparisons

5.5.1

1^5
1^5

Preliminary Stability Analyses

5.5.1.1

Frequency Distribution Curves for

the Factor of Safety
153
156
. . .

5.5.2

Final Design Stability Analyses

5.5.2.1

Partially Drained Stability Analyses

157

5.5.2.2 .Long Term Stability Analyses

5.6

157

Summary and Conclusion

166

vii

Page

VI

COMPARISON OF STABL WITH EXISTING PROCEDURES FOR SLOPE

STABILITY ANALYSIS USED BY INDIANA STATE HIGHWAY

COMMISSION (ISHC)
6.1

170
171
171

Problem 1 Problem 2

6.2 6.3
6.k 6.5
VII

Problem Problem

183
188

Conclusion of Comparative Analysis

200
20U
20^+

SUMMARY AND RECOMMENDATIONS FOR FUTURE WORK

7.1

Work and Results

7.1.1

Parametric Study

20U 20U

7.1.2
7.1.3
7.2

Modifications of STABL

Comparative Study

205

Suggestions for Future Work

206
207

LIST OF REFERENCES

APPENDICES Appendix A, Tables and Figures Related to Chapter III

Appendix B, Tables and Figures Related to Chapter V Appendix C, Tables, Figures and Analysis Related to
Chapter VI
C.I
320

....

211 297

The Influence of the Line of Thrust on the

Factor of Safety in Slope Stability Analysis,

Method of Slices
C.l.l

320 320 321

Introduction
Position of Line of Thrust

C.1.2

C.1.3
C.II

Interslice Stress Distribution

322

Influence of Moment Center on the Factor of

Safety for Circular Failure Surfaces

C.2.1

32U
32^

Introduction

C.2.2

Derivation of Equations for F

as a Function of the Problem

and

3F
c

3y

Geometry and Moment Center (y) for the

Case of
<j>

326

viii

Page

Appendix D, Corrections, Adjustments and Smaller Changes in STABL

D.l

373 373 373

. . .

Adjustments for the IBM Computer

D.2

Debugging of STABL
D.2.1
D.2. 2

Errors in the Anisotropic Soil Option

373

Infinite Loop in the Circular Surface

Generator
D.2. 3
D.2.1*

37^

Unit Weight of Water The TURKEY Trace Key

376
377 378

D.3

Other Adjustments D.3.1

Modifications in the Irregular Surface

Generator (RANDOM)
D.3.2
D.3.
3

378
. . .

Modifications of Errors RC11 and BK10

Format Adjustments

380
380 381

D.3. 3.1 D.3.

D.3.
3. 2

Format Statements

Misleading Factors of Safety

Changes of the Calcomp Plot

....

381 382 382

3. 3
3.1*

D.3.

Updating of Comment Cards

Flow Chart of STABL

D.3.

3. 5

383

Appendix E, Listing of Revised STABL Program

Main Program STABL2 Subroutine Subroutine
READER
QUIT

393
39!*

399
1+03
l+ol*

Subroutine Subroutine
Subroutine

PROFIL
ANISO

l+il+

WATER
LOADS

I+18 1*22

Subroutine

Subroutine Subroutine
Subroutine
Subroutine

EQUAKE LIMITS
INTSCT
SURFAC

1+26
1*27

1*30 1+32
1*35
1*51

Subroutine Subroutine

RANDOM
RANSUF
BLKSUF

Subroutine

1*63

ix

Page

Appendix E (cont.
Subroutine Subroutine Subroutine Subroutine Subroutine Subroutine Subroutine
Subroutine
BL0CK2
SORT
1+gS

w^

EXECUT
SLICES

^
l^r
ItSo

WEIGHT
SOILWT
FACTR

^66
1^88

SCALER PLOTIN PLTN

^93
1^95

Subroutine
Subroutine

501

Subroutine

POSTN

506
508

Appendix F, STABL on the IBM Computer

F.l

Introduction

cq8 508 509 509 510

F.2
F.3

The Random Function Ranf(x)

Precision

F.3.1
F.3. 2

Machine Rounding Errors

Use of TOL

LIST OF TABLES

Table
2.1

Page

Equations and Unknowns Associated with the Method of Slices

15

2.2

Factor of Safety for Different Methods of Slope Stability Analysis

23

2.3

Comparison of STABL with Modified Bishop Method after Wright (1969), Using Failure Surfaces Obtained by the Modified Bishop Analysis
Factors of Safety for Homogeneous Slopes

27
*0

3.1
3.2

Variation in Sensitivity of the Factor of Safety to Input Parameters

52

3.3

Typical Values of Partial Coefficients Used in Denmark for Geotechnical Designs

Range in Variables Studiet for an Embankment on a Foundation with a Weak Layer

62

3.h

67

3-5

Geometry Parameters k and a for Embankment Stability Analysis

Statistical Analysis of Geometry Parameters k and a
.

7"

3.6
i+.l

92

Data Setup for Example Problem for Revised STABL Program

Output of Example Problem

10 111

k.2 U.3

Stability Analysis of Slopes Subjected to Earthquake Forces

132

5.1

Calculation of In Situ Undrained Shear Strength (After Ladd, 1975)

s

15

5.2

in Zones A, B, C and D for 95$ Consolidation in Zones A, B, C and D for 100$ Consolidation

160
16U

5.3


xi

Table
5.J+

Page

Results of Stability Analyses from MIT-Report (After Ladd, 1975) Results of Stability Analyses performed with STABL

167

5.5

....

168

6.1

Comparative Study, Problem 1. Different Methods of Analysis

Factor of Safety for

175

6.2

Comparative Study, Problem 2. Different Methods of Analysis

Comparative Study, Problem 3. Different Methods of Analysis
Comparative Study, Problem k. Different Methods of Analysis

Factor of Safety for

180

6.3

Factor of Safety for

187

6.h

Factor of Safety for

201

Appendix Table
A.l
AF
-r-j-7

for Homogeneous Slopes

212

A. 2

T7

AF
AF

for Homogeneous Slopes

213

A. 3

.( C*

for Homogeneous Slopes

21 k

A.h

AF

p-

for Homogeneous Slopes

215

A. 5

Sensitivity Analysis, Homogeneous Soil, Vertical Slope Actual Change in the Factor of Safety F, (6 = 90). due to Changes in the Independent Variables
Sensitivity Analysis, Homogeneous Soil, Slope 1/2 Actual Change in the Factor of Safety F, (3 = 63.h). due to Changes in the Independent Variables Sensitivity Analysis, Homogeneous Soil, Slope 1/1 Actual Change in the Factor of Safety F, (6 = U5 ). due to Changes in the Independent Variables

216

A. 6

217

A. 7

218

A. 8

Sensitivity Analysis, Homogeneous Soil, Slope 2/1 Actual Change in the Factor of Safety F, (8 = 26.6). due to Changes in the Independent Variables
Sensitivity Analysis, Homogeneous Soil, Slope 3/1 Actual Change in the Factor of Safety F, (6 = l8.U). due to Changes in the Independent Variables

219

A. 9

220

xii

Appendix Table

Page

A. 10

y T

for Homogeneous
,

dopes

{%)

221

AF
A. 11
c

for Homogeneous Slopes {%)

6

222

AF
A. 12

Y~?
AF
h F

for Homogeneous Slopes {%) AF

=
y

223

A. 13

.J

for Homogeneous Slopes {%)

22*4

A.lU

Factors of Safety for Embankment on Foundations with a Weak Layer

225

B.l

Example of Input Data for Preliminary Design Analysis, Circular Search Example of Input Data for Preliminary Design Analysis, Sliding Block Search

298

B.2

299

B.3

Example of Input Data for Final Design Analysis, Sliding Block Search
Cases Investigated by STABL for the Comparative Design Study

300

B.k

302

C.l

Subroutine FACTR, Modified to Calculate Factor of Safety by the Modified Bishop Method (Used only for Circular Arc Failure Surfaces)
Subroutine FACTR Substituted in Program STABL to Calculate the Factor of Safety by Spencer's Method

333

C.2

....

33

1*

C.3

Subroutine FACTR, Modified to Calculate Interslice Sideforces E, and Effective Normal Force and Stress at Bottom of each Slice for Normal STABL Procedure (Janbu), Specified Surface
Examples of Input Data for Problem 1, Comparative Study Examples of Input Data for Problem 2, Comparative Study
Examples of Input Data for Problem 3, Comparative Study Examples of Input Data for Problem h, Comparative Study
. .

338

C.U

339

C.5 C.6 C.T

3^0
3^1

3^2

xiii

Appendix Table
C.8

P age

Interslice Forces, Effective Normal Stress at Bottom of Slices, and Line of Thrust for Spencer Analysis, Problem 2, Circle 5
Line of Effective Thrust and Interslice Shear Factors of Safety, Spencer Analysis, Problem 2, Circle 5

3^3

C.9

3hk

CIO

Interslice Forces and Normal Stress at Bottom of Slices for STABL Stability Analysis (Janbu Method), Problem 2, Circle 5

3U8

C.ll

Factor of Safety, Interslice Forces, and Line of Thrust for Spencer Analysis, Problem 3, Surface BL0CK2-1+
Interslice Shear Factors of Safety for Sliding Surface BL0CK2-H, Problem 3, Spencer's Method

3^9

C.12

350

C.13

Factor of Safety and Interslice Forces for STABL Stability Analysis (Janbu Method), Problem 3, Surface BL0CK2-U
Variables Used in Calculation of Line of Effective Thrust, Interslice Stresses, and Interslice Shear Strength for Spencer's Method

351

C.lU

352
35 1*

C.15
D.l

Variables Used in Appendix C.II

Chronologic Listing of Corrections and Changes Made in the STABL Program between August 1975 and June 1977
Illustration of the Use of Subroutine TURKEY for Error Trace of the STABL Program
Changes Necessary to Transfer STABL from the CDC Computer to an IBM Computer

....

38U

D.2

389

->12

xiv

LIST OF FIGURES

Figure
2.1

Page

Principle of the

<f

Method

2.2

Principle of Log Spiral Method

10

2.3
2.h

Principle of Friction Circle Method

12
lit

Division of Sliding Mass into Slices

Forces Acting on a Slice

2.5
3.1 3.2

16
30

Flow Chart Showing Steps in Sensitivity Analysis

Range in Variables for Sensitivity Analysis of Homogeneous Slopes
Critical Failure Surfaces for Vertical Slopes in Homogeneous Soil
Critical Failure Surfaces for Slopes 1/2 in Homogeneous Soil

32

3.3

3^

3.k

35

3.5

Critical Failure Surfaces for Slopes 1/1 in Homogeneous Soil

Critical Failure Surfaces for Slopes 2/1 in Homogeneous Soil
Critical Failure Surfaces for Slopes 3/1 in Homogeneous Soil

36

3.6

37

3.7

38

3.8

Relationship between Slope Parameters, Critical Failure Surface and Factor of Safety for Homogeneous Slopes ....
Isometric Plot of Factor of Safety vs. Slope 1/2
((J)',

^2

3.9

-r-)
I

for
**3

3.10

Factor of Safety vs. Strength Angle Slope 1/2

(J)'

for Various

T
c'
.

,
. . .

kk
**5

3.11

Factor of Safety vs. ^r

for Various

<J>'

Slope 1/2

....

3.12

Equal Factor of Safety Contours for Slope 1/2, Homogeneous Soil

^6

'

XV

Figure
AF
;

Page

3.13

vs.

Slope Angle

f?

for Various

<}>'

1*7

3.1 1*

AF

vs

vs.

4)

for Different Slope Angles 8

U8

A(%)
3.15

ttt
AF
-r-77-

AF

Slope Angle B for Various

<f>'

1*9

3.16

vs.

(J)

for Various Slopes

p

50

3.17

AF

AcQt s

vs. 3 for Various

to 0.3

51

AF Xi 3.18

Relative Sensitivity

for Vertical Slope

55

AF X
3.19

Relative Sensitivity r=r

AF X ,

for Slope 1/2

56

3.20

Relative Sensitivity g-^r

AF X
.

for Slope l/l

57

3.21

Relative Sensitivity ^-^r

AF X
.

for Slope 2/1

58

3.22

Relative Sensitivity , \fi

for Slope 3/1

59

3.23

Variables for an Embankment Built on a Foundation Soil with a Weak Soil Layer
A Sliding Block Type of Failure Surface is most

61*

3.2l*

Critical
3.25

TO

Comparison between Failure to Left and Failure to Right Side of Embankment, D = 10 ft a = 5-7,
,
<t>

72

3.26

Comparison between Failure to Left and Failure to Right = Side of Embankment, D = 10 ft a = 11. 3
, ,
<J>

73

3.27

Change in Critical Failure Surface when Strength of Weak Soil Seam Increases

7^

3.28

Normalized Factor of Safety F

vs. Normalized Strength c

for Embankment 20 feet High, Slopes 1.5/1, Depth to Weak

Seam

feet, Inclined at a = 11.3 to Horizontal

76

xvi

Figure
3.29

PaEe

Example of F

vs. c

for Embankment 20 feet High, '

5

Slopes 1.5/1, Depth to Weak Seam

at a = 11.3 with Horizontal 3.30

feet, Inclined
77

Variation of k and "a" Coefficients with a for Constant Depth D = 0.25h

Variation of k and "a" Coefficients with a for Constant Depth D = 0.5h

79

3.31

80

3.32

Variation of k Coefficients with Depth D for Constant a = 0 Variation of k Coefficients with Depth D for Constant a = 5-7
Variation of k Coefficients with Depth D for Constant a = 11.3

8l

3.33

82

3.3^

83

3.35

Variation of "a" Coefficients with Depth D for Constant a = 0 (and a = 5-7)

Variation of "a" Coefficients with Depth D for ^ Constant a = 11.3

81+

3.36

85

3.37

Variation of F vs.
Ratios of

c
5

F
C

for Various a, and Different

ft

D =

87

3.38

Example of Variation of the Factor of Safety with Geometry of Weak Soil Layer Sliding Block Failure Surface with Rankine Active and Passive Zones

88

!*.l

97

k.2

BL0CK2 Sliding Block Analysis Applied to Anisotropic Soil

Slope Stability Analysis for Problem with a Perched Water Table

100

I+.3

1Q 3

k.k U.5 U.6

Rapid Drawdown Stability Analysis for Embankment

Calcomp Plot of Result of STABL Analysis

105

H8
11 9

Stability Chart for Slopes in Homogeneous Soil (after Taylor, 19U8)

xvii

Figure
k.J

Page

Stability Chart for Slopes in Homogeneous Soil, (after Taylor, 1937)

<J>

120

1+.8

Stability Chart of the Taylor Type for Homogeneous Slopes based on Results from STABL Program
,

121

1*.9

Comparison between STABL and Taylor Stability Chart for Homogeneous Slopes
Illustration of Clockwise and Counterclockwise Direction Limitations for Initiation of Failure Surfaces

122

It.

10

125 128

U.ll

Transmission of Farthquake Forces to the Pore Water

Change in Pore Water Pressure due to a Horizontal Earthquake Force

....

U.12

130

U.13

Critical Failure Surfaces for Slopes Analysed for Influence of Earthquake Loadings
Soil Profile, Geometry and Design Criteria for Embankment (after Ladd, 1975)

133

5.1

138

5.2

Relationship between Total and Effective Strength for a Factor of Safety Larger than One
Stress vs. Strain from CK U Tests with Different
o

1^0

5.3

Modes of Failure (after Ladd, 1975)

5.1+

1^6

Input Data and Results of Preliminary, Undrained, Total Stress Analysis, Circular Failure Surfaces

lU8

5-5

Drained Effective Stress Analysis of Embankment, Preliminary Design

1^9

5.6

Anisotropic Undrained Strength Parameters for Use in Total Stress Analyses (after Ladd, 1975)
Undrained Total Stress Analysis of Embankment with SHANSEP Anisotropic Strengths, Preliminary Design
Different Types of Critical Failure Surfaces Generated by STABL

151

5-7

152

5.8

15

1*

5.9

Frequency Distribution Curves for the Factor of Safety, Cases 1A, 2B, 3B, HB Estimated Soil Profile and Cross Section for Embankment on Varved Clay after Construction

155

5.10

158

xviii

Figure
5.11

P age

SHANSEP Strength Data for Total Stress Analyses at 95$ Consolidation (after Ladd 1975)
,

159

5.12

Total Stress Analyses of Plmbankment with Berms Partially Drained

Drained Effective Stress Analyses of Final Embankment vith Berms SHANSEP Strength Data for Total Stress Analyses at 100$ Consolidation (after Ladd, 1975)
Total Stress Analyses of Final Embankment with Berms, 100$ Consolidated
Soil Profile and Strength Data for Problem 1

l6l

5.13

162

5.1k

163

5.15

165 172

6.1

6.2

Critical Failure Surfaces from Analysis by NYSTAB Circular Arc, and Hand Calculated Sliding Wedges Critical Failure Surfaces from Analysis by STABL, CIRCLE, RANDOM and BL0CK2
-

173

6.3

17

1*

6.14

Soil Profile and Strength Data for Problem 2

177

6.5 6.6 6.7

Stability Analysis by ATEC and STABL, Circular Arc

Sliding Block Analyses, STABL and STABL w/ Spencer

.... ....

178
l8l

Stability Analysis by STABL, Spencer Method, Line of Thrust

Soil Profile and Strength Data for Problem
3

l8 2
l81

6.8 6.9

Stability Analysis by ATEC and STABL, Circular Arc

Stability Analysis by STABL, Sliding Block Failures
Soil Profile and Strength Data for Problem h, = 0) Total Stress Analysis
(<f>

....

185

6.10
6.11

....

186

l8 9

6.12 6.13

Critical Circles Analysed

190
G
,

Factor of Safety F vs. Moment Center y for Various C . r = 2.5h, c = c E p

Factor of Safety F vs. Moment Center y for Various C = 2c r = 2.5h, c
E

192
Q

6.1U

193

'

xix

Figure
6.15

Pa ^ e

Factor of Safety F
r = 5h, c

vs. Moment Center y for Various 9

= 2c

F
vs. Moment Center y for Various 9
,

19U

6.16

Factor of Safety F
r = 2.5h,
c

= 0.5c

p
c

195

F yh

6.17

Stability Number

vs. 9 for Various Radii,

,
<J>

Circular Failure Surfaces, c^ = c_

I96

Fjh
6.18

Stability Number
c

vs. 0 for Various Radii,

'
<$>

Circular Failure Surfaces, c^ = 2c,

E

197

F yh

6.19

Stability Number

Circular Failure Surfaces, c = 0.5c,

6.20

cO

vs. 8 for Various Radii,

<()

198

Correction Factors for the Simplified Janbu Procedure (after Janbu et al., 1956)

203

Appendix Figure
A.l

Factor of Safety vs. Strength Angle

Vertical Slope

<t>

'

for Various

c'

233

A. 2
A. 3

Factor of Safety vs.

for Various
<\>

<f>',

Vertical Slope

23

1*

Factor of Safety vs. Strength Angle

Slope 1/2

'

for Various

c
'

235
-r-

A.k
A. 5

Factor of Safety vs.

for Various
<j>

(f)',

Slope 1/2

....

236

Factor of Safety vs. Strength Angle

Slope 1/1

'

for Various

c'

237

A. 6

Factor of Safety vs.

~
V Yh

for Various $'

<J>'

Slope 1/1

....

238

A. 7

Factor of Safety vs. Strength Angle

Slope 2/1

for Various

c'

239

A. 8
A. 9

Factor of Safety vs.

for Various
<f>

<J>',

Slope 2/1

....

2U0

Factor of Safety vs. Strength Angle

Slope 3/1

'

for Various

c'

2kl

'

Appendix Figure
A. 10 A. 11

Page
c'

Factor of Safety vs.

for Various

<f>'

Slope 3/1

....

2^2

Equal Factor of Safety Contours for Vertical Slope, Homogeneous Soil

2^3

A. 12

Equal Factor of Safety Contours for Slope 1/2, Homogeneous Soil

Equal Factor of Safety Contours for Slope 1/1, Homogeneous Soil Equal Factor of Safety Contours for Slope 2/1, Homogeneous Soil

2kk

A. 13

2^5

A.lU

2U6

A. 15

Equal Factor of Safety Contours for Slope 3/1, Homogeneous Soil

2^7

A.l6

Factor of Safety vs. cot

Theoretical Solution

ft

for Various

<J>',

Yn

~r~ = 0,

2^8
ft

A. IT

Factor of Safetv vs. cot

Factor of Safety vs. cot Factor of Safety vs. cot
Factor of Safety vs. ^7

for Various

<J>

'

,
Yn
c

c'

= 0.1 = 0.2 = 0.3

....
. .

2^9 250
251 252

A.18
A. 19

for Various for Various

<{>

<J>'

ft

<}>'

c'

A. 20 A. 21
A. 22

for Constant

Vertical Slope
Slope 1/2 Slope 1/1 Slope 2/1

Factor of Safety vs. ^7

for Constant $
for Constant
<j>

... ... ...

253
25
1*

Factor of Safety vs. ~V

Factor of Safety vs. ^7 Factor of Safety vs. ^7

A. 23

for Constant
for Constant

<J>

255

A.2H
A. 25

<f>

Slope 3/1

...

256

Equal Factor of Safety Contours vs. Geometry Variables,

<J>'

= 0

257

A. 26

<J>

Equal Factor of Safety Contours vs. Geometry Variables, = 10

258

A. 27

Equal Factor of Safety Contours vs. Geometry Variables,

* = 20 2 59

A. 28

Equal Factor of Safety Contours vs. Geometry Variables,

<f>'

= 30

260

'

'

'

xxi

Appendix Figure
A. 29

Page

Equal Factor of Safety Contours vs. Geometry Variables,

r
A. 30

= ^0

261
<S>'

AF
,

vs. Slope Angle 8 for Various

262

A. 31

AF -7
%>

vs.

<J>'

for Different Slone Angles 8

263

A. 32

AF -7AF , At 5 !-)

vs.

c
1

for Various

<t>'

Vertical Slope and

261*

Yh Slope 1/2
A. 33

vs.

Yh
c

for Various

(J)',

Slope l/l and 2/1

265

Yh

A.

3!+

A(S-) Yh AF
*~7T

vs. rr

Yh

for Various

<j>,

Slope 3/1

266

A. 35

vs. Slope Angle 6 for Various

1
(J)

= 0,

Theoretical Solution
A. 36 A. 37

267
<J>'

vrr ttt
AF

AF

vs. Slope Angle 6 for Various vs.

<j)'

268

for Various Slopes,

c'
:~r

= 0, Theoretical

Solution
A. 38
A. 39

269
<J>'

AF
-rj-7-

vs. vs.

for Various Slopes

for Various Slopes, and
rf>
'

270
e ;V- =

AF

ttt
Af
Atp

<t>'

0.1 to 0.3

271

A.Uo

-rrr

for Various vs. Yn Slopes 1/2 and 2/1

vs.
c_^

p -r!

Vertical Slope,
272

A.

1+1

Txy
AF .^ Atan Ac
Ac
.

^r Yh

f r Various

(f>

'

Slopes 1/1 and 3/1

273
27

A.U2
A. 1+3

7
<J)

, '

vs. cot 6 for Various * T

'

1*

vs. 3 for Various $',

g

A.Ult

A-^5

^ ~ Acot

vs.

for Various

<f>',

^ ~=
r-

= 0.1

275
276

0.2

g B

vs. 6 for Various

<f>',

Yh

" 0.3

27T

'

' '

xxii

Appendix Figure
A kG
'

Page

"

Acot B

Vs

"

6 for

Various

*'.^=0to0.3

278

A.U7

AcQt g

vs.

<t>'

for Various

A.U8
i

Acot g
AF Acot B
AF AcQt g AF AcQt g AF AcQt
o

vs. 4" for Various

~
c

Vertical Slope
Slope 1/2 Slope 1/1 Slope 2/1 and 3/1

279

280
281

Vs

'

*'

for Various ^"

A. 50

vs.

<t>'

for Various
for Various for Various for Various for Various

c
'

282

A. 51

vs.

A. 52

vs.

A. 53

AF AcQt

vs.

c
c
'

cf>

Vertical Slope
Slope 1/2 Slope l/l
Slopes 2/1 and 3/1

283
28U 285

<$>'

(J)',

A.5

Acot g

vs.

<J>'

....

286

A. 55

CIRCLE Search with STABL, Embankment Founded on Soil with a Weak Seam RANDOM Search with STABL, Embankment Founded on Soil with a Weak Seam
BLOCK Search with STABL, Embankment Founded on Soil with a Weak Seam

287

A. 56

2 88

A. 57

289

A. 58

BL0CK2 Search with STABL, Embankment Founded on Soil with a Weak Seam

290
.

A. 59

CIRCLE Search with STABL, Embankment in Homogeneous Soil

RANDOM Search with STABL, Embankment in Homogeneous Soil
Critical Failure Surfaces for Embankment with = 0 Underlying Weak Seam, c_ = c E
,
<J>

291
292

A. 60
A. 6l

293

A. 62

Critical Failure Surfaces for Embankment with = 0 Underlying Weak Seam, c = 2c,
<t>

29*

A. 63

Critical Surfaces for Embankment with Underlying = 0 = 10 Weak Seam, c = 0.5c E

p

<j>

<f>

A.6U

Critical Failure Surfaces for Embankment with = Underlying Weak Seam, c = c E F E

p
,
<f> <J>

296

xxiii

Appendix Figure
B.l

p age

Circular Search by STABL, Preliminary Design Analysis, Undrained Shear

303

B.2

Random Search by STABL, Preliminary Design Analysis, Undrained Shear

Sliding Block Search by STABL, Preliminary Design Analysis, Undrained Shear
Sliding Block Search, Preliminary Design Analysis, Effect of Varying Horizontal Elevation on Factor of Safety
Sliding Block2 Search by STABL, Preliminary Design Analysis, Undrained Shear

30U

B.3

305

B.U

306

B.5

307

B.6

Sliding Block2 Search, Preliminary Design Analysis, Effect of Varying Horizontal Elevation on Factor of Safety
.

308

B.T

Circular Search by STABL, Preliminary Design Analysis, Drained Shear

Sliding Block Search by STABL, Preliminary Design Analysis, Drained Shear

309

B.8

310

B.9

Circular Search by STABL, Final Design Analysis, Drained Shear

Circular Search by STABL, Final Design Analysis, Undrained Shear

311

B.10

312

B.ll

Random Search by STABL, Final Design Analysis, Undrained Shear

Sliding Block Search by STABL, Final Design Analysis, Effect of Varying Horizontal Elevation on Factor of Safety
Sliding Block2 Search by STABL, Final Design Analysis, Effect of Varying Horizontal Elevation on Factor of Safety

313

B.12

31

B.13

315

B.lU

Circular Search by STABL, Final Design Analysis, 100% Consolidated, Assumed Undrained Shear

316

B.15

Random Search by STABL, Final Design Analysis, 100% Consolidated, Assumed Undrained Shear

31

'

xxiv

Appendix Figure
B.16

Page

Sliding Block Search by STABL, Final Design Analysis, 100? Consolidated, Assumed Undrained Shear

318

B.17

Sliding Block2 Search by STABL, Final Design Analysis, 100? Consolidated, Assumed Undrained Shear
Circular Search by STABL, Problem
1

319
357

C.l

C.2
C.3

Deep Circular Search by STABL, Problem 1

358
359

Deep Random Search by STABL, Problem 1

C.k
C.5

Circular Search by STABL, Problem 2

Sliding Block2 Search by STABL, Problem 2

360
36l

C.6
C.T

Circular Search by STABL, Problem 3

Sliding Block2 Search by STABL, Problem
3

362

363

C.8

Circular Search by STABL, Problem h (Simplified Janbu Procedure)

36h

C.9

CIPCL2 Search by STABL, Problem (Modified Bishop Method)

It

365

CIO
C.ll

Sliding Block2 Search by STABL, Problem U

366

Calculation of Effective Line of Thrust, Interslice Stresses and Interslice Shear Strength for Spencer's Method
Line of Effective Thrust for Spencer Analysis, and Effective Normal Stress along Shear Surface for Spencer and for Normal STABL ( Janbu) Analysis

367

C.12

370

C.13

Factor of Safety and Line of Thrust for Surface BL0CK2-U, Problem 3, Spencer Analysis

371

C.lU D.l

Geometry Variables Used in Expression for F (y)

Attempt to Generate a Circular Failure Surface when Conflict Occurs between Overturning Slip Surface and Minimum Elevation

372

390

D.2

Angle Limits for Successive Line Segments of Irregular Failure Surface

391

D.3

Flow Chart of STABL

392

LIST OF ABBREVIATIONS AND .SYMBOLS

Note:

Symbols used exclusively in Appendix C are listed in Tables C.lU and C.15 of Appendix C.

Abbreviations

ASCE
ATFC
CDC

American Society of Civil Engineers

American Testing and Engineering Corporation.

Control Data Corporation (Computer System at Purdue Univ.
K
o

CK U
o

Consolidated

Undrained Shear Test.

DSS
IBM

Direct Simple Shear.

International Business Machines Corporation (Computer).

ICES-LEASE

Integrated Civil Engineering System - Lease Electronic

Accounting System.
ISHC

Indiana State Highway Commission.

Joint Highway Research Project.

JHRP
MP

Morgenstern-Price method of slices for slope stability

analysis.

NYSTAB

New York State, Department of Transportation, Slope

Stability Analysis.
OCR
PI

Over Consolidation Ratio.

Plasticity Index.
Plane Strain Active.
Plane Strain Passive.

PSA PSP

SHANSEP

Stress History and Normalized Soil Engineering Properties.

xxvi

Special Symbols

A prime ('"") indicates that the variable is in terms of effective stresses

A L

(Delta) indicates a finite change or variation in a variable.

(Sigma) used as summation symbol.

Symbol for Center Line.

Symbols
a
-

Moment arm for body forces in simple slope stability

analysis.

a.

to a

Geometry dependent coefficients in standardized equation

for the factor of safety.

A1

,A ,A

Coefficients introduced in the general equation for the

factor of safety.

A,B,C,D.E

Zones of constant stress.

Different types of analysis (see Table B.M.

Base of a slope.
Crown width of an embankment.

b B
c,
c'

Strength intercept in terms of total (c) and effective (c')

stresses, respectively.

c,

to e

n
-

Strength intercepts of soils 1 to n.

c
c c

n
-

Normalized strength of weak seam

w (=
I

r
-

Required strength intercept to make slope in equilibrium,

Undrained strength intercept.

Stress independent strength (=c dl).
c

C, C

r
-

dl

A(-er)

The change in

due

to a change

Ac'

in c'

(= ^jj"~)'

xxvii

A(-^-)

The change in
(c
'

due
C
yh

to a change Ly in y
V
'

(Y + ^y) h

Yh^h

"

The chan Se in
(v "
c'

~ due
C'
v

to a change Ah

in h

Y <h + Ah)

"

Yh

Maximum distance between a shear surface and the chord

of the shear surface, measured normal to the chord.

Depth to a weak layer

Interslice normal forces

As subscript, indicates embankment.

Used as subscript, indicates at failure or on the failure

plane.

ff

Used as subscript, indicates on failure plane at failure.

Partial coefficient for skin friction on piles
Partial coefficient with respect to strength intercept
c.

f
f f g f P f
f,

Partial coefficient for dead load and water pressure,

Partial coefficient for live loads.

Partial coefficient for wind loads.

Partial coefficient with respect to strength angle
$.

f(x)

A function relating the horizontal and vertical interslice

forces along a shear surface as proposed by Morgenstern

and Price (1965)

F
-

Factor of safety.

Used as subscript, indicates foundation.

Modified Bishop factor of safety for circular failure
surface

Janbu's factor of safety (Figure 6.20).

xxviii

F^

The value of F
(=

when moment is taken at infinity

l^m F

= Janbu's factor of safety.

c,

F
F
n

Factor of safety with respect to strength intercept

Normalized factor of safety
(=

-^C

).

E
<j).

Factor of safety with respect to strength angle

AF(Ax

The change in the factor of safety due to a change in the

independent variable
x,
,

other parameters kept constant

AF
Ax.
1

AF(Ax
i

Ax.
1

The tangents or secants of the F vs.

x.
i

curves,

other variables held constant.

AF
,

= = = = = = -

c'

AF(Ac')
AF(Ah)

AFU

AFg

AF(A6)
AF(Ay)
AF(A4>')

AF^
AF,,

EAF
h, H
H
i

EAF(Ax.),
1

1, n.
J

Height of a slope or embankment.

Critical height of a slope.

Subscript indicating an array of variables.

Inclination of ground surface.
Geometry dependent coefficients in standardized equation
for the factor of safety.

k,

In
to k
o

k,

Horizontal earthquake coefficient.

Vertical earthquake coefficient.

Coefficient of earth pressure at rest.

K
1

Length of a failure surface.

dl

Infinitesimal increment of length along the failure surface.

xxix
L -

Length of the chord of a shear surface. Moment around the center

failure surface.

of a circular or log. spiral

Number of slices in a method of slices slope stability

analysis

Depth factor (Taylor's stability chart).

N N

'

Total and effective normal forces acting on a shear surface, Center of a circular or log. spiral failure surface.

q.

Maximum shear stress at failure (=

;r

(a, - a) 1 3 f

).

Q r
r
r

Resultant of uniform surcharge, acting on top of a alice.

Radius of a circular or log. spiral failure surface.

Q -

Initial radius of log. spiral failure surface.

Maximum radius of log. spiral failure surface,

Pore pressure parameter (= ^)
.

r R

Resultant of normal stress (force) and normal stress dependent limiting shear stress (force) on a shear surface.

Ranf(0.).

Ranf(x)

Standard random function, taking any value between and l(for x =

0.
).

R
R.
s
s

__ -

Resultant of strength intercept resistance forces.

te
i

number in a random sequence of Ranf(0.).

Shear strength based on effective stresses.

Undrained shear strength, based on total stresses.

Shear resistance required along a potential shear surface
for equilibrium of slope.

S, S

Shear resistance available along an assumed shear surface.

XXX

u u
U U
c

Pore water pressure.

Pore pressure constant,

Degree of consolidation.
Water force acting on base of a slice.

Water force acting on top of a slice.

W
W, W'

Used as subscript, indicates weak soil layer.

Total and effective weight of a sliding mass, or weight of
an individual slice.

x.

Independent parameters for stability analysis.

Ax
X y

Width of a slice. Interslice shear force.

nun

Minimum elevation to which a failure surface may extend,

Greek Alphabet

Inclination of shear surface with respect to the horizontal. Inclination of weak soil layer. Inclination of shear surface at intersection with top of
s

ct-

lope

6
Bj.,

Slope inclination.

Inclination of left and right side of an embankment.

Unit weight of a soil.

Y
6

Inclination of uniform surcharge, acting on top of a slice,

measured positive counterclockwise from vertical.

9

One half the center angle of a circular failure surface.

Angle between r

and r for a log. spiral surface.

Initial inclination of a failure surface at bottom of slope.

Inclination of a line segment of a trial failure surface.

xxxi

Q-r

Lower angle limitation for line segment of irregular

failure surface
(=

clockwise direction limit).

Upper angle limitation for line segment of irregular

failure surface
(=

counterclockwise direction limit).

9q

Central angle of circular failure surface, measured from

horizontal radius above top of slope to radius through

intersection of failure surface with ground surface at
the top of the slope.
A9
-

See Figure C.lU.

Angle deflection between consecutive line segments of a

circular failure surface.

A0

Maximum angle deflection between consecutive line segments

of a circular failure surface.

AG

mm
.

Minimum angle deflection between consecutive line segments & '

of a circular failure surface.

Stability parameter used by Janbu (195^+b),

(=

t~

Maximum and minimum principal stresses

Total and effective normal stress on failure plane. Horizontal normal stress,

0,0'
o,

h v
'

o a

Vertical normal stress,

Initial vertical effective normal stress,

vo
'

vc

Vertical consolidation stress,

a
T T

'

vm

Maximum past vertical stress,

Shear stress on a potential failure plane.
Shear stress on a failure plane at failure.

Shear stress on a horizontal plane (direct-simple shear

test)

xxxii

$> *'

strength angle in terms of total

stresses, respectively.

(<f>)

and effective

(<J>')

*1 t0 ^n
<J>

~
_

"

tren &tn angles of soils 1 to n.

r
-

Strength angle required for equilibrium of a slope.

<J>

Undrained strength angle.

xxxiii

HIGHLIGHT SUMMARY

This study is a continuation of a previous one by Siegel (1975)

and deals with computerized slope stability analysis.

All investigations are related to a computer program, STABL, which

performs analyses of general slope stability problems.
A parametric study was conducted on the STABL program. The study
<J>

showed that variations in the soil strength parameters

largest influence on the stability of a slope.

and

have the

The STABL program was modified to provide additional options in

terms of a new sliding block surface generator, several piezometric

surfaces, and the modified Bishop factor of safety for circular failure
surfaces.

Results of analysis performed with STABL were compared to results

obtained with other methods of slope stability analysis.

Results by

STABL are in general conservative, but the difference is usually less

than

1055.

However, for deep circular failure surfaces in soil with

strength defined mainly in terms of a strength intercept c, the simplified Janbu method of slices normally applied in the STABL solution may

give erroneous and nonconservative results.

In such cases, the modified

Bishop method should be used.

XXXIV

The modified STABL program, STABL2, is written in FORTRAN IV source

language, developed for a CDC 6500 computer, and
i3

also available in a Both ver-

form compatible vith the IBM 360 (or 370 ) computer system.
sions are verified.

INTRODUCTION

This report presents investigations conducted with the computer

program STABL, which performs analyses of general slope stabilityproblems

.

STABL was developed by R. A. Siegel (1975a, 1975b) under a previous

research program conducted for and in cooperation with the Indiana Gtate

Highway Commission (ISHC).

This current research program is a continu-

ation and extension of the work done by Siegel to develop computerized

slope stability solutions to meet the needs of the ISHC for efficient,

versatile and accurate computational methods.

1.1

PURPOSE OF THE PRESENT STUDY

The purpose of the present study was originally stated as follows.
(1)

To perform a parametric study (sensitivity analysis) of the

existing general program developed by Siegel, or variations
thereof.

This is needed to assess the precision with which

input data, particularly soil data, need to be defined.

High sensitivity to an input parameter requires more

attention to be given to that item.

(2)

To develop a hierarchy of special and simplified programs

to treat specific and common slope stability problems.
It

was felt that the general program, being extremely versatile,

but accordingly rather bulky and time consuming, would be

uneconomic to use for simple problems.

(3)

To compare the results from all programs developed with

existing ISHC solutions.

This will increase the

,

confidence in the new programs

and indicate whether they

are in general more or less conservative than present

solutions

During the course of this research, objective (2) was somewhat

altered.

Instead of developing special and simplified programs for

common slope stability problems, it was found more desirable to retain

and use the general STABL program.
In addition to above objectives, the author was assigned as

consultant on computer conversions of the STABL program from the CDC 6500

computer at Purdue University, where the program is developed, to the

IBM 360 computer at the ISHC computer center, Indianapolis.

1.2

ORGANIZATION OF REPORT
The report is arranged as follows.
Chapter II gives a short introduction to slope stability analysis

in general, and presents some of the methods that are being used.

It

also contains a brief description of the STABL program.

The sensitivity study is reported in Chapter III with additional

illustrations in Appendix A.

Modifications of the STABL program are described in Chapter IV,

including an example of data setup and output for the modified program,
along with some general advise for users.

Chapter V presents a case study for the design of an embankment

on varved clay, comparing STABL solutions with others.

Related computer

plots are contained in Appendix B.

Comparisons of STABL with existing ISHC procedures for slope

stability analysis are reported in Chapter VI, with related additional
analysis and computer plots in Appendix C.

Chapter VII summarizes goals, results and conclusions of the

research, and gives some suggestions for future studies in this area.

Appendix D describes smaller changes and errors that have been

corrected in the original STABL program.
A total listing of the revised

(corrected and modified) program is presented in Appendix E.

Finally, Appendix F reports some of the experiences in converting

the STABL program from the CDC computer to the IBM computer.

Included

in Appendix F is a listing of the changes necessary to transfer the

STABL program from the CDC 6500 computer to an IBM 360 (or 370)
computer system.

Because of the large number of tables and figures included in the

appendices, each of these are arranged with the text (if any) first,

then the tables in consecutive order, followed by the figures in consecutive order.
This was done to avoid interrupting the text, and

because most of the text readily can be interpreted apart from the
tables and figures.

II

SLOPE STABILITY ANALYSIS

2.1

GENERAL DESCRIPTION
Earthen masses located beneath a sloping ground surface, whether

natural or man-made , have a tendency to move downward and outward

under the influence of gravity.

Unless this tendency is suitably

,

counteracted by the shearing resistances within the mass

occurs.

a landslide

Other forces such as seepage forces and surcharge loads may

add to the instability.

Avoiding such instabilities is a major concern

of the geotechnical engineer. The investigation of the stability of a slope has three main
steps
(1) (2)

Identification of the forces which cause and resist failure.

Determination of the shear strengths of the soils beneath the

slope

(3)

Application of a model to investigate the stability based on

(1) and (2).

The forces acting on the slope depend on the profile and density of the soils; the position of the phreatic surface, and associated seepage
forces
;

and external applied loads

The shear strength of a particular soil depends upon its type and
origin, water content, prior stress history, method and type of

compaction, and the like.

The shear strength of a soil was initially expressed by Coulomb

(1776) as
T = c + o

tan

<J>

(2.1)

where

is the strength intercept,

<j>

is the strength angle,

and a

is

the stress normal to the shear plane. The equation commonly used today is modified to express T
in

terms of effective stresses rather than total stresses and has the

form
T f

= c'

(a

- u)

tan

<J>'

or

t
c'

= c"

+ a*

tan

<f>'

(2.2)

where

and

<J)'

are the strength intercept and strength angle with

respect to effective stresses, also named effective strength parameters,

and a*

= (o

- u)

is the effective stress normal to the shear plane.

c' ~ 0,

In effective stress analysis of sands and gravels,

and

equation (2.2) is simplified accordingly.

In total stress analysis, = 0,

and with particular reference to un drained shear in clayey soils,

<J>

and equation (2.1) is reduced to

= c.

Instability of a slope in sands and gravels occurs if the slope

inclination exceeds a certain value, known as the angle of repose.
This

angle is constant for a particular material at a particular void ratio

and is independent of the height of the slope, provided the slope is not
subjected to seepage forces.
In contrast, a slope in a clayey soil has a

critical height, which depends on the slope inclination and the soil

properties

The shearing behaviour of a particular material

ir,

determined

empirically, i.e., by field tests or laboratory tests.

Most common are

the laboratory triaxial and direct shear tests and the field vane test.

The results are then applied to the slope stability analysis by

assuming certain mechanisms and modes of failure.

A common approach to slope stability analysis

is to

investigate

failure along an assumed slip surface, and express the stability in terms
of a safety factor F, defined as the ratio of the available shearing
forces along the slip surface, S
,

to the shearing forces required for

static equilibrium, S

F = <T r

(2.3)

An F value equal to or below unity indicates that the slope is

unstable and failure can be expected.

To identify the more critical slip surfaces (lower F values) it

is generally necessary to conduct the analysis for a number of surfaces

of different shape and location.

2.2

DIFFERENT METHODS OF SLOPE STABILITY ANALYSIS

2.2.1

Early Development

Until Coulomb (1776) [Skempton (19^6)

soils were considered as

strictly frictional, i.e., a slope was stable if the slope angle was
less than the angle of repose of the material.

Coulomb analysed the

stability of a vertical bank of cohesive soil with respect to a plane

slip surface, and obtained the equation

=
c

k y

[-

cos .^ 1 - sin

J
<f>

(2.10

where H

is the critical height, c is the strength intercept,

<J>

is the

strength angle, and y is the unit weight of the soil.

In 1820 Francais gave a solution to the more general problem of a

sloping bank.

He also assumed a plane slip surface and found

ti

H
c

_ if "

r [

cos 4> sin B 1 - cos (8 - <fr)

< ]

(o k\ (2 5)
'

where

is the slope inclination.

Collin (181+6) introduced, based on his observations of deep curved

slip surfaces in clay slopes, an analysis assuming a cycloidal slip

surface and a

<J>

value of zero.

However, not until the twentieth century did greatly improved

measurements of shear strength and refined procedures of analyses

appear.

Wright (I969) and Siegel (1975a) have described in detail the

different methods of slope stability analysis available today.
As this

study is a continuation of that begun by Siegel, only a resume of the

most common methods will be given here

2.2.2

The

<ft

Method

Based on field observations in Sweden, a circular slip surface was

introduced by Petterson (1916) [Skempton (l9*+6)], which combined with

the assumption that
<j>

equals zero, gave what we today know as the

<J>

method.

Figure 2.1 shows the principle.

c

When $ equals zero, the shear

strength T_ equals

(Equation 2.1) and is independent of the actual

Equilib-

value of the normal stress at any point of the failure surface.

rium of tte potential sliding mass is satisfied by moment equilibrium

o o I
I-

o
H

I h o
id _j cl

Id

o z
a:

CM Id

L.

u.

h
i

around the center of the circle.

The forces acting normal to the

failure surface pass through the center of the circle and do not con-

tribute to the moment, hence the shear stress required for equilibrium
(c
)

is determined directly by the moment equation, and the factor of

safety F can be found from Equation 2.3.

Different centers and radii are

investigated to determine F
gives the stability number depth factor n.

min

Taylor (1937) made a chart that directly

as

c
r yli

a function of the slope angle 3 and

Knowing c, y and H, F can be found.

Gee Figure ^.7-

The

<}>

method is commonly used for short term stability analysis

of saturated clay slopes, where a failure would occur undrained.

2.2.3

The Logarithmic Spiral Procedure

When

4>

'(<!>)

is not taken as zero, t

is not a constant, but depends on

the normal stress O'(a) along the potential failure surface, and a dis-

tribution of 0'{a) has to be assumed in order to determine F.

To over-

come this problem the logarithmic spiral was introduced as a proposed

failure surface.

Such a surface has the characteristic that the resultants

of the normal stresses 0' and frictional components of shear strength

a'

tan

4>'

pass through the center of the spiral.

Consequently moment

equilibrium around the center will only involve the body forces and
strength intercept force (C) of the soil as in the Figure 2.2.
The equation of a logarithmic spiral can be expressed in polar
<j)

method.

See

coordinates as

r = r

etanO'/FJ
<j>

(2 6)
.

10

11

where r is the radius from the center, r

the angle between initial radius r

is the initial radius,

1

is

and radius r,

(J)

is the effective

strength angle and F

angle.
If F
,

is the factor of safety with respect to the strength

is taken as unity [Taylor (1937)] the procedure directly

gives F

the factor of safety with respect to the strength intercept, c'.

In most cases, however, one will require an F larger than unity for both
<J>'

and c'.

Assuming F
o

= F
c

several spirals with the same center and

To find F

initial radius r

are investigated to find the value of F.

min

the center 0, and initial radius r

have to be varied.

2.2.U

The Friction Circle Method

For a circular shear surface the resultants of the normal stresses

and the frictional components of shear strength will lie on lines tangent
to a circle of radius r sin
<}>',

called the friction circle.

Hence the

moment about the center of the circle will involve the normal stress dis-

tribution which is unknown.

The assumption commonly made is that the

resultant of all normal stresses and frictional shear stresses also lies

tangent to the friction circle.

This assumption is equivalent to assuming

that the normal stresses are concentrated at a single point [Wright (1969)]

and the solution obtained is a lower bound solution (conservative).

The

principle is shown in Figure 2.3.

Taylor (1937) also presented charts

See

for the stability J number =-rr based on the friction circle method. FyH

Figure U.6.

2.2.5

Method of Slices
The methods described thus far are particularly suited to simple

slope stability problems with homogeneous soil conditions and no forces

12

13

other than those due to weight and shearing resistance.

For problems

involving changing soil conditions and additional forces such as seepage

and earthquake forces, several limiting equilibrium methods have been developed, all based on the method of slices. The soil mass above an assumed slip surface is divided into slices
in such a way that each slice has a minimum change in variables (Figure
2 ,U)
.

The system of equations and unknowns associated with complete

equilibrium of the entire soil mass in terms of n slices is summarized

in Table 2.1 (see also Figure 2.5).

For any value of n, other than unity,

there are (2n - 2) more unknowns than equations, and the problem is

indeterminate.

The differences

in the various methods of stability

analyses are contained in the different assumptions made with respect

to these unknowns in order to render the problem determinate.

2.2.5.1

Ordinary Method of Slices

The assumptions made

This method assumes a circular slip surface.

by Fellenius (1926) to eliminate the unknowns are

(1)

The normal force N on the base of each slice acts at the

center of the base.

(2)

The interslice forces are zero.

3)

These two assumptions eliminate (^n -

unknowns, which is more than

is necessary to make the problem statically determinate.

Remaining are

(n + l) unknowns, namely F and the n normal forces at the base of each

slice.

The equations used are:

(a)

force equilibrium normal to the

base of each slice, which directly gives the normal forces N, and (b)
overall moment equilibrium around the center of the circle forming
the slip surface, which directly gives the shear strength required for

in

-i

in

S
u.

o > 2
a

15

Table 2.1

Equations and Unknowns Associated with the Method of Slices

Equations
n

Moment equilibrium for each slice.

Force equilibrium in 2 directions for each slice.

2n
n

Mohr-Coulomb relationship between shear strength and

normal effective stress.

ln

Total number of equations.

Unknowns
1

Factor of Safety F. Normal force at base of each slice N.

n
n n

Position of N.
Shear force at base of each slice S. Horizontal interslice forces E.

n - 1
n - 1

Vertical interslice forces X.

Location of interslice forces (line of thrust)

n - 1
6n - 2

Total number of unknowns.

n is the number of slices.

16

A X

line of

thrust

lassumed shear surface

FIGURE 2.5

FORCES ACTING ON A SLICE

17

equilibrium.

The factor of safety F will be expressed in terms of an

average value (summing over the n slices) of the shear strength available divided by the shear strength required.

Because the problem is overdetermined the values obtained for the

normal forces will usually not satisfy equilibrium in directions other
than those normal and parallel to the base of each slice.

Hence the

neglect of interslice forces may lead to unrealistic results.

2.2.5.2

Bishop's Method

Bishop (1955) introduced a method satisfying complete equilibrium.

His primary equations are
(1) (2)

no. of eq.
1

Overall moment equilibrium

Vertical force equilibrium for each slice

Force equilibrium parallel to base of each slice

n n n
3n + 1

(3)
(h)

Shear force as function of normal force and F Total of

In addition he has two equations with respect to overall equilibrium

of interslice forces X and E.

In the primary analysis Bishop does not make any assumptions

about the line of thrust (see Figure 2.5)-

This line may be determined

by adding the remaining (n - l) moment equations for the individual

slices, but the primary equations can be solved independent of the

location of the line of thrust.

The normal force, as well as the

resultant weight of each slice, is assumed to act through the center

of the base of the slice.

The remaining unknowns are

18

no. of eq.
(1)
(2) (3)
(1+)

Factor of safety F

Normal force at base of each slice

Shear force at base of each slice

V.

n
n n - 1

Vertical interslice forces X

(5)

Horizontal interslice forces E

Total of

n - 1
Un - 1

This total of unknowns is compared to

an indeterminacy of (n - 2).

3n + l) equations, which gives

The procedure is to assume values of X

that satisfy the two overall interslice force equations (E can be expressed
in terms of X from the force equation parallel to the base of a slice)
.

To judge whether a particular solution is reasonable, one may calculate

the corresponding line of thrust using the remaining (n - l) moment

equations for the individual slices.

2.2.5.3

Modified Bishop Method

The simplest solution to Bishop's method is to assume X equal to

zero which directly satisfies equilibrium with respect to the vertical

interslice forces, but generally not equilibrium with respect to the

horizontal interslice forces.

simplified Bishop method.

The solution is called the modified or

It can be directly arrived at by neglecting

the interslice forces as for the ordinary method of slices, but using

vertical force equilibrium for each slice instead of force equilibrium

normal to the base of each slice.

2.2.5.^

Other Procedures

Many other methods have been developed by different investigators

making various assumptions with respect to the interslice forces.

19

Spencer (1967) assumed parallel interslice forces and developed a

statically determinate solution with 3n equations and 3n unknowns.

Morgenstern and Price (1965) assumed a relationship between the

horizontal and vertical interslice forcer, dependent on a function f(x)
along the shear surface.
surfaces of general shape.

They extended their solution to failure

Janbu (195^+a) formulated an expression for F in terms of vertical

force equilibrium for each slice combined with overall horizontal force

equilibrium.

As with the Bishop method, this procedure requires

X.

assumptions of the vertical interslice forces

Janbu made his

assumptions so that they satisfy interslice force equilibrium as well

as moment

equilibrium for each slice, assuming a position of the line

A simplified procedure results if X is assumed equal to

of thrust.
zero.

This procedure is similar to the modified Bishop method, the

difference being that overall horizontal force equilibrium is used

versus the use of overall moment equilibrium
in the modified Bishop

procedure.

Carter (1971) reached the same expression as Janbu's simplified

procedure, but arrived at it in a different manner by considering the

modified Bishop procedure applied to slip surfaces of general shape.

In the case where the slip surface is not a circular arc, no specific

point exists that is obvious to use as moment center.

Carter found

that the value of the factor of safety was independent of the x-coordinate

of the moment center, but varied with the y-coordinate, and reached a

20
#

minimum value for y =

lie

concluded that y =

was the only consis-

tent moment center to use for an irregular as well as a circular slip

surface.

2.2.6

Methods other than Limiting Equilibrium

is to

An alternative method of slope stability analysis

investigate

the shear stresses imposed under the slope by means of the theory of

elasticity [Perloff and Baron (1976), Romani (1970), Romani, Lovell and

Harr (1972)].

The criterion is that the shear stress at any point does The factor of safety is defined as the

not exceed the shear strength.

shear strength divided by the shear stress at the point where this ratio
is the least, hence it gives the safety at the most critical point, whereas

F defined by a limiting equilibrium method gives the average value of

safety over the entire slip surface.

The method may be useful when dealing with soils where progressive

failure is likely to occur, because it will discover the areas of stress

concentration from which a failure is likely to initiate.

On the other

hand, it does not take into account the redistribution of stress which
occurs, when the stress level at a point approaches the strength.
In a case with concentrated surface loads
,

methods based on an

elastic model can take into account the distributional effect of the
slope material.

This cannot be done with the method of slices in the

limiting equilibrium models, and an error may be introduced by applying

*Further investigations by the author show that this is not always true, see Chapter VI and Appendix C.II.

21

such a model in this case.

For important and complex problems it

may-

be advisable to perform both types of analysis.

2.2.7

Probabilistic Approach

All the methods presented thus far have been discussed from a
deterministic viewpoint, as if the parameters involved in the slope
stability analysis were single fixed values.
This is a gross simplifi-

cation of the real situation, where uncertainties exist with respect to soil properties as well as the soil profile.

Several people, e.g., Wu et al (1970), Athanasiou-Grivas (1976),

and Catalan and Cornell (1976), have worked on probabilistic models for

slope stability analysis, which make use of slip surface and limiting

equilibrium models, but which operate in terms of probability of failure

rather than the traditional factor of safety.
In this approach the soil

properties are defined, not as exact values, but as distributions with

a mean value and standard deviation.

The relationship between the factor of safety as defined by the

traditional methods, and the probability of failure was investigated in

a case study by Athanasiou-Grivas (1976)
,

where a slope having a conThe probability of failure

ventional factor of safety of 1.2 failed.

was found to be approximately 0.3% with zero pore pressure and 1% for a

pore pressure parameter (r

= u/yh) equal to 0.6.

*The significance of varying soil properties may be investigated within the deterministic model by a sensitivity study (see Chapter III)

22

2.2.8

Comparison of Methods

Wright (1969) has done a comprehensive study of the many methods

available for slope stability analysis.
He found, in agreement with

others [Bishop (1955), Lowe and Karafiath (1967)], that for homogeneous

soil the modified Bishop procedure is as accurate as any of the more

rigorous methods (Spencer, Janbu, Morgenstern and Price), whereas the

ordinary method of slices, as well as Janbu' s simplified procedure, gave

factors of safety as much as 10$ and 5% lower than the lower bound value

given by the friction circle method.

Some of the results are shown in

Table 2.2.

The difference is larger if pore pressures are present, and

for cases with non-homogeneous soil it may be even larger.

In any case where the slip surface can be approximated by a circular

arc, Bishop's modified procedure is recommended, as it combines reason-

able accuracy with simplicity and small computer time requirements.

If the slip surface cannot be approximated by a circular arc, Bishop's

method cannot be used without modifications.

The method that is most

convenient for irregular slip surfaces is undoubtedly Janbu' s simplified

procedure.

Unfortunately it is not so accurate and may lead to overly

If this is a problem,

conservative (and hence uneconomic) design.

Spencer's procedure seems to be the more rigorous method that is the

least time consuming.

However, it requires analysis of the position of

the line of thrust, together with soil properties, to judge whether the

result is conservative or not (see example in Appendix C.l).

A study comparing the modified Bishop method with a linear elastic

finite element method is reported by Wright et al (1973) with the

following results:

23

Table 2.2
Factor of Safety for Different Methods of Slope Stability Analysis
cot 6

Method
1 2
3 h

1.5/1 1.171
1.125

2.5/1
1.1*85

3.5/1

1.768
1.659
1.71
1*

1.102

l.lVf
1.103
5

1.M3
1.351

1.595
3.095

1
2
3
h

1.831

2.U80
2.33
1*

1.738
1.785

2.91 1*
3.008

2.1*09

1.703
8

2.277
3.399

2.850
1.333

1
2 3
k

2.U30 2.301 2.368

2.262

3.210
3.309
3.15
1*

U.105

k.226
't.0U5

1 2
3
U

20

2.309 2.197
2.257 2.176

3MU
3.268
3.3'+7

U.503
U
1

.332
1

*.

425

3.2U2

U.307
1+.035

1
2
3
h

50

1.93U

2.988

1.863
1.902 1.855

2.901

3.936
3.991
3.928

2.9^8
2.893

X ~

c/yh

'

P =

slope ansle
F
>

'

The stability number N =

is a function of X and B only,

and is constant for constant X and 3 although c, $, y and h may

vary.

The methods compared are (l) Simplified (or Modified) Bishop

Method, (2) Ordinary Method of Slices, (3) Friction Circle and

(h) Janbu's simplified procedure.

2k

(1)

The distribution of normal stress along the failure surface was approximately identical for the two methods.

(2)

Although the factor of safety by the finite element method

varies along the failure surface, the average value is nearly
the same as for the modified Bishop method.

(3)

Because the finite element method involves a completely

different set of assumptions than those employed in the

modified Bishop method, the close agreement between the two

methods indicates that they both are reliable.
Because the

limiting equilibrium model is much simpler than a finite

element analysis, the former will usually be preferred.

2.3

STABL - A COMPUTER PROGRAM FOR GENERAL SLOPE STABILITY ANALYSIS

Many programs exist, based on one of the previous described methods

of slope stability analysis.

When STABL was developed, the major purpose

was to develop a program with the least number of restrictions imposed on

the input data, and the maximum utilization of the computer in the search
for the most critical failure surfaces, combined with ease of use.

STABL can handle almost any problem with respect to boundary geometry and loading, soil conditions including anisotropic soils, equilibrium
or excess of equilibrium pore pressures and pseudo-static earthquake
forces.
It has routines that search for the more critical failure sur-

faces by generating surfaces of circular, sliding block or irregular

shape and compare the corresponding safety factors.

It can as well

handle specified failure surfaces.

It is limited to 2-dimensional analy-

sis, which is the case with almost all methods available today.

*For an evaluation of 3-dimensional effects on slope stability analysis, see Baligh and Azzouz (1975)-

25

The method chosen for the calculation of the factor of safety is

Janbu's simplified procedure satisfying vertical force equilibrium for

each slice and overall horizontal force equilibrium, assuming horizontal

interslice forces.

For the simple case of dry soil and no external

loading, the equations are as follows:

Vertical force equilibrium for a slice

N cos a + S sin a - W =

(2.7)

Overall horizontal force equilibrium

(U

sin a - S cos a) =

(2.8)

Mohr-Coulomb relationship

C
From (2.7) and (2.9)
(N = N

+ N'

tan

d>'

F
1

(2.9)

N = W/cos a - S tan a _

(2.10)
, x

+ W tan (ji'/cos a F + tan a tan (J)'

K-u.i

Substitute (2.10) into (2.8)

(W tan a - S/cos a)

=0

(2.12)

Substitute (2.11) into (2.12)

_ Z

rif

[W tan a -

C* + W tan r=, -

7 tttJ cos a (F + tan a tan <p')

<j)'/cos

-i

_ n

Rearranging gives
W cos g (F tan a - tan 4>') cos a (F + tan a tan <t>')

j,

C'/cos a + W (tan F + tan a tan

<j>'

- F tan a)
1

2 13

(b

26

This expression is the same as that derived by Carter [Siegel (1975a)]

A similar expression including surface loads, water table and earthquake forces is presented by Siegel (1975a); see also Chapter IV,

Equation

(*+.l).

This method was chosen for its simplicity and because it can be

applied to failure surfaces of any shape.

It is usually conservative
A com-

and may in some cases give factors of safety which are too low.

parison was made with the modified Bishop method from Wright (1969).
STABL gave values of F from 2.3% to 8.8% lower than those obtained by
the modified Bishop method, the largest difference occurring for low

values of X (see Table 2.3).

Siegel compared STABL with Spencer's pro-

cedure and found a maximum difference of 15% for steep slopes and high
values of A, with a pore pressure parameter r
= 0.5, STABL giving the

lower value.
In most slope stability problems, however, there is an uncertainty
in the determination of the input data such as the soil profile and

particularly the soil properties in terms of density and strength.

The

factor of safety may be more sensitive to variations in these properties

than to different methods of analysis [Singh (1970)].
This point is

further investigated in Chapter III.

A detailed discussion of how the program performs its functions

given by Siegel (1975a).

is

Later modifications of the program are reported

in Chapter IV of this report.

^Important exceptions, see Chapter VI, Problems

and k, and Appendix C.II.

27

Table 2.3 Comparison of STABL with Modified Bishop Method after Wright (1969),
Using Failure Surfaces Obtained by the Modified Bishop Analysis
cot 6 2.5/1
1.1*85

Factor of Safety
Mod. Bishop

1.5/1

3.5/1

1.171 1.106
5-9

1.768

STABL
% difference

1.367
8.6
2.1*80

1.625
8.8

Mod. Bishop

1.831

3.095

STABL
% difference

1.720
6.5
8 2.1*30

2.318
7.0

2.890
7.1
1+.333 1+.099

Mod. Bishop

3-399 3.209
5-9
3 -UlU

STABL
% difference

2.290
6.1 20

5.7
1+.503

Mod. Bishop

2.309
2. 201*
1+.8

STABL
% difference

3.275

U.3U6
3-6
1*.035 3.91+5

U.2

Mod. Bishop

50

1.931+

2.988

STABL
% difference

1.873
3.3

2.912
2.6

2.3

X =

*?*

4*
,

6 = slope angle

(see also Table 2.2).

20

III

SENSITIVITY ANALYSIS

3.1

INTRODUCTION
A sensitivity study was performed on the STABL program to assess

the precision with which input data need to be defined, particularly

soil data.

Although the particular results apply to this program, the

trend is believed to be applicable to slope stability analysis in

general.

Little research has been published on sensitivity analysis.

An

article by Singh (1970) concludes that the factor of safety for slope

stability analysis is much more sensitive to the choice of strength

parameters as interpreted from soil tests than to the choice of the

computational method of analysis.
This conclusion is of particular

interest with respect to the STABL program, because the procedure of

analysis used in this program usually is somewhat conservative compared

to other methods of analysis (see Chapters II, V and VI).

3.2

PROCEDURE FOR ANALYSIS

The procedure initially chosen for the sensitivity analysis is as

follows

(l)

For a given slope (geometry, soil profile, water table, seepage, load conditions), a STABL searching routine is used to

generate a failure surface, which will be close to the critical one, and the safety factor is calculated.

29

(2)

Using this specified surface, the factor of safety is calculated for a small change in an input parameter equal to the

expected variation in the field or scatter in test results.

(3)

The influence of the change in the factor of safety due to a

change in one of the input parameters, relative to the total

change in the factor of safety due to changes in all parameters, is investigated.

(k)

A conclusion may be reached with respect to the required pre-

cision in definition of the different variables to limit

uncertainties in the factor of safety to tolerable levels.

The steps are illustrated on the flow chart of Figure 3.1.

3.3

CASES STUDIED
The sensitivity analysis was performed for the idealized case of

a slope in homogeneous soil with no pore water pressure, and for an

embankment founded on a soil, which has a weak layer.

3.3.1

Sensitivity Analysis of homogeneous Slopes

It was decided to perform the sensitivity study for a relatively

wide range in parameters for slopes in homogeneous soils, to get a good

feel for how the factor of safety is likely to change, when different

parameters are changed.

For homogeneous soils the factor of safety depends on

pendent (to some extent) parameters, which are:
Slope geometry defined by

to 6 inde-

Height

h
6

Slope angle

30

STEP

Slope geometry
Soil

characteristics

Locate

STABL
Water conditions

critical

failure

surface

Loads

STEP

Variation

in

'STABL
critical

w
surface

Variation

in

problem variables

Factor of Safety

STEP 3

AF(Axi)

vs.

AF(Axj)

STEP 4
Restrictions

on

Ax

to

limit

AF(Ax:)

FIGURE

3.1

FLOW CHART SHOWING STEPS

SENSITIVITY

IN

ANALYSIS

'

'

31

Soil properties defined

by-

Unit weight Strength intercept Strength angle

Y
c'

(J>

Pore water pressure, if present, is important, as well.

was neglected in this study.

This parameter

Figure 3.2 shows the range in these parameters, covered by the

sensitivity analysis.
Five different slopes were examined.
Height (h)

and unit weight (y) were chosen arbitrary as 100 feet and 100 pcf, respectively.
However, the results can be applied to any values of h and

Y by introducing the dimensionless number

Taylor ( 1937)

The slopes

were analysed for all combinations of (c',

<f>')

shown in Figure 3.2, with

c
*

the exception of those where a combination of low strength intercept low strength angle
<f>' ,

and a steep slope, gave values of the factor of

safety much below unity.

3.3.1.1

Analysis by STABL

For each problem case (slope geometry, soil properties) limitations

for initiation and termination of failure surfaces were set according
to the particular problem, e.g., all failure surfaces for slopes with a

slope angle 8 greater than 53 passed through the toe; for

53 and
<j>

less than

= 0, always a base failure.

A hundred surfaces were generated

for each problem.

If the critical surfaces were located on the boundary

of the search limitations, these were changed accordingly, and a new

search was carried out.

An exception to this procedure was the

<J>

case, where the critical failure surface theoretically extends to an

infinite depth.

Only the CIRCLE search was used because of the homoge-

neity of the soil (see also p. 69).

32

Vertical

Slope

Slope 1/2

Slope

l/l

Slope

2/1

Slope 3/1

Height of slopes
Soil

100

feet

characteristics;

unit

weight

y* 100
c'-

pcf

strength intercept

1000-3000
0-50

psf

strength angle

<

FIGURE 3.2

RANGE

IN VARIABLES FOR SENSITIVITY ANALYSIS OF HOMOGENEOUS SLOPES

33

The most critical failure surfaces generated by STABL for each par-

ticular problem were plotted for comparison with the others, as shown
on Figures 3.3 to 3.7.

Where the location of a failure surface deviated

,

from the logical position with respect to the other failure surfaces

more reasonable trial surface was introduced by the author and analysed
as a specified surface. If this surface was significantly more critical

(lower factor of safety) than the most critical surface generated by STABL, the former surface was used.

These surfaces are drawn with

The reason why this can happen

dashed lines on Figures 3.3 through 3.7.

is explained in the word "random".

STABL generates failure surfaces

randomly within the defined limits, and selects the 10 most critical of
the surfaces generated.

However, it is possible that there are some

more critical failure surfaces than those found by STABL.

3.3.1.2

Results of Stability Analysis of Homogeneous Slopes

The results of the stability analysis in terms of factors of safety,

F, are given in Table 3.1.

Following the procedure outlined in Section

3.2, critical failure surfaces (Figures 3.3 to 3.7) were used as speci-

fied surfaces
c
'

and the factor of safety calculated for small changes in

and

<f

'

Another approach to determine AF(Ax

)*

is to plot F vs.

x^

keeping

other parameters constant. This method has the advantage that it also
reflects changes in F due to a slight change in the critical failure

surface when the soil parameters change.

It was possible to apply the

method in the case of homogeneous slopes, because a relatively wide

The change in the factor of safety due to a change in x other parameters constant.

keeping

3^4

J*--

0.2

yfc'--

FIGURE 3.3

CRITICAL FAILURE SURFACES FOR

VERTICAL SLOPES
SOIL

IN

HOMOGENEOUS

35

7% -0.1

7%-02

7^ = 0.3

FIGURE 3.4

CRITICAL FAILURE SURFACES FOR SLOPES

1/2
IN

HOMOGENEOUS

SOIL

36

j-o.

3^-0.2

^-0.3

FIGURE 3.5

CRITICAL FAILURE SURFACES FOR SLOPES

l/l

IN

HOMOGENEOUS SOIL

37

rh -o

rh-o.2

^K-o.3

<o

FIGURE 3.6

CRITICAL FAILURE SURFACES FOR SLOPES

2/1
IN

HOMOGENEOUS

SOIL

38

rh

=20'- 50

rh

-o.i

7R-0.2

FIGURE 3.7

CRITICAL FAILURE SURFACES FOR SLOPES

3/1
IN

HOMOGENEOUS

SOIL

39

rfr-o.3

4>'-o
E Xh 0'

NolS

Dashed

circle

was forced deep,

but result is the

same as
For
yfi

the shallower circle.

failure

=0.2 and 0.3 the

depth.)

surface

is

at limit

of boundaries. (Theoretical
infinite

surfaces extend to

FIGURE 3.7

(CONT.)

1+0

Table 3.1

Factors of Safety for Homogeneous Slopes

1

Nscot

Yh

0.5

1.0

2.0

3.0

0.757

10

.787
.677
.792 .915

.936

1.152

1.388 2.087

0.1

20
30

.987

1.210
1.531+

1.618
2.181+

1.196
1.1+31

2.871
3.806

40
50

1.883

2.808
3.61+2

1.057
.801

1.719

2.331
1.106
1.575

5.011
1.11+3

1.088
1.31+7

1.129

10 0.2
20 30

.967

1.817

2.039
2.819
3.620
1+.582

1.120
1.261+

1.577
1.818

1.891 2.230

2.366

2.928
3.587
1+.519

40
50

1.1+08

2.087
2.1+26

2.622
3.115

1.598
1.202

5.879
1.711+

1.633
1.886
2.11+1+

I.658 2.232
2.51+1

1.707
2.1+37
3.021+

10

1.368 1.521 1.683 1.860

2.075

2.67I+ 3.1+90
1+.351+

0.3

20
30

2.1+06

2.888
3.303 3.901

3.631
1+.327

40
50

2.701
3.066

5.331

5-232

6.627

Note

The underlined results are obtained from specified failure surfaces, shown

with dashed lines on Figures 3.3 through 3.6,

see also top of page 33.

Ui

range in the independent variables was studied.

The dimensionless

number

is
c'

used as an independent variable, reducing the number of

c' -r*
,

independent variables to three:

is the factor of safety, F.

<p'

and 8.

The dependent variable

The location of the more critical failure

surface may be considered another dependent variable, see Figure 3.8.

From the graphical plots, the sensitivity of F to changes in the

independent variables can be found.

The graphs may be plotted as func-

tions in 3-dimensional space, giving F as a function of two independent

variables, while other independent variables (one in this case) are held
constant, see Figure 3-9.

For any practical purposes, the two-dimension-

al projections of the three dimensional plots are the most useful

(Figures 3.10 to 3.12).

The variables are combined in different ways and graphical results

obtained, which are convenient to use for a quick estimate of the in-

fluence of change in an independent variable x

on the safety factor F.

These graphs are presented in Appendix A, Figures A.l through A. 29.

3.3.1.3

Sensitivity of the Factor of Safety to Input Variables

)

To determine AF(Ax.

for a particular problem,

AF ^

was determined

over the total range of parameters investigated.

This was done by deter-

mining the tangents or secants to the F vs.

points.
AF of v
i

x.

curves at particular

Since these curves are based on empirical values, the values are subjected to some uncertainty.
AF Examples of variations of -^
i

with x
J

are illustrated in Figures 3.13 through 3.17.

Tables and more

figures are included in Appendix A.

3.2, page
52.

The results are summarized in Table

1*2

Q)

O
cfl

<H
>H

3
CO

3
rH
<H

-p
<L)

<m

O
fc -P
a;

a)
<H

H
o
H
a)

i *

"*-

o
(h

E O
a>

P
u o

p o
a
fc

C5

p
ft

o ^

a; d)

H
C

ai

p
H
.c

<u

H be
3

p
be

p
X! b0

aJ

ft
a>

W p
1
0)

d)

& o H CO p
Jh d)

p
CO

c 0) u

x p be
c

H
0)

>
+3 H

p
CO

I) Jh

w
0)

&
ft

Pi

H
H

CO

'O

6-

V)

>

W Ll < <n
Ll
en

\-

o u 2
\-

LU

LiJ

cn

ii
0>

J AidJDS

JO

JOPDJ

kh

4.0

FIGURE 3.10

FACTOR OF SAFETY VS., STRENGTH ANGLER' FOR VARIOUS y^ SLOPE 1/2

>5

2.0-

rh

FIGURE

3.11

FACTOR OF SAFETY VS. VARIOUS 4>' SLOPE 1/2

t

yfi

FOR

1*6

same

critical

surface lines

^0.3

20
<p',

30
degrees

FIGURE 3

12

EQUAL FACTOR OF SAFETY CONTOURS FOR SLOPE 1/2, HOMOGENEOUS SOIL

17

Valid

in

the range

0.08

J^ 0.35

(range

? 0.35 not investigated)

AF

*(*)

theoretical
infinite

solution for

depth
only

For /390 valid

for

y^O.2

10

20

30

40

Slope Angle

70 50 60 degrees @,

80

90

FIGURE

3.13

AF

vs.

SLOPE ANGLE

/3

FOR VARIOUS

Jlfi

Valid

in

the range

0.08^^^0.35
not investigated)

(range

^ > 0.35

FIGURE 3.14

^1

AF

vs.

</>'

FOR DIFFERENT

SLOPE ANGLES

U9

UK
Average over range

0.1*^0.3
0.10

0.08

\\

\
\jOt\ *

Yfc.

0.06

&.

0.04

0.02

1,1
40 50

10

20

30

60

70

80

90

Slope Angle ^9, degrees

FIGURE 3.15

4^

vs.

SLOPE ANGLE

FOR

VARIOUS $'

50

0.12

Average over range

o.i^y^ ^o.3

0.08

0.02

Vertical Slope (/^90[L

10

20
4>

30
,degrees

40

50

FIGURE 3.16

jM^- vs.

FOR VARIOUS SLOPES

51

2.0

AF
Acot/3

1.0 *

10

20

30

40

50
/B,

60
degrees

70

80

90

Slope Angle

AF
FIGURE 3.17

Acot
y
=

VS.

FOR VARIOUS

<,

TO 0.3

52

r\ rH
r. <D

p
tS

c
n
bC
r-i

13
aj

,^
CM
-3-

c
r-l

P
10

tn
bJ CJ

u o

-H H H
td to

O E

J
faP
II

<
(U

-P U"\

^
bfi

aa H
fe

CJ

c o u >> O iH <w T3
H

P. <M

o
CO
rH

O
<M

r-

3-^
Cd

r,

O
r.

b^ >, ft

H
p

aj r<

m
0)
f-.

H
H
rH

X!
b H rH
10

P M C O p H H p
CJ
0)

ft

rH
.

>
jr bO H ,C

00
0) Jh

c C

%H

in 0)
Ih CJ

03

.p

<D

3
b

>

o
tO

a1
ro

o
<fH

IH

-H

O
O

bO V- T3 -H

h O

P
M

a.'

rn

O
Vc

o nT> m < o\
.

CO

tn

0)

U O
Ml

ft

H
to

r. cd 0)

p
t-

r-i-a-

S
X, bD H
10
<U

bd V bd c c H H W e- (D
a!
<1)

tr

-e-

II

H
to a)

P
03

<
K
Ih to (D

rH Vh
Ih

ft -P
c8

a!

r<
<fH

(l)

10

h'H

r< CJ 0)

o u

Q
=8

o c mI

U O

H -P

OJ

ai ih

U c

o
<h

ft p*
0)

3
far

-H
r-

CM

00

P
to

V
~

o|
CM

c o

()

r> rH

H a
to

O E

ir\

<
o p H
LT\

-e-

m<
M

u o
tM

H O 3o
II

h
a! 0)

C >> H rH A to >d Olr- Bl H 0) -eT) ft

-H

CM

C
"3 ot 0)

bC

H H P P
m
E
CtH
ctf

V ft c O
to

<
-P
ir.

rH

3
Hp
to

a>

Ih

ai r-

u c
HH

c
a]

rl

u CJ c
>-<

&

ht

u c o o Uh O

H
U-

fafl

H 43
(h
td

>

'

53

A general observation is that

(x
c

<j>

'

or cot 8) is
c
'

relatively insensitive to variations in

than 0.1.
1*5

Yh

for values of

Yh

larger

An exception is the t

AF

75-

ratio for slopes steeper than

and
-r

For

a
'

values between
AF
;

AF and 0.1, -rry

At*
1

increases with

c'

Yh

whereas

Yh Yh Figures 3.13 to 3.16 show that the factor of safety F, is most

A( cl }

c* is constant or decreases with -r-

sensitive to variations in the

flat slopes.

Yh
c'

parameter as well as

$'

for relatively

For

angles up to

1+5

w^
Wh
;
,

AF

increasing 8; for 8 angles larger than U5

larger.

decreases slightlv with AF is the decrease in a/c\

Also ttt
AF _
,

AF

decreases with increasing value of 8; however, conthe decrease is large for 8 angles up to U5
to 50,

trarily to a/C\
^Yh

and less for 8 angles larger than 50

Figure 3.17 shows the sensitivity of F to variations in cot 8,

being the slope angle.

Cot 8 was chosen as independent variable instead

8

of 8, because the relationship between F and cot that between F and

8.

was more simple than

For values of
,

angles 8 less than U5

Acot p

5AF

Yh
c'

higher than 0.1 and slope

is almost independent of the value of

as well as R, but increases with increasing ip.

For steeper slopes,

AF Acot 8

increases with

yh
c
'

as well as with p.

, ,

.,

To investigate actual changes in factors of safety due to variation

in the variables, specific variances were assigned to the variables as

follows
A<(>'

(constant)

=T~ c

0.20

20

'

5U

^
n

0.05

5%

0.05

5%

Assuming 5# variance in height h and base of slope b, we find

Acot 6 -

rrf
0.10
=

I
10*

- - 1

- - 1 cot

or

^2t6=
cot
8

~-

20$

gives

A(^-)

=
c

205$

At ,

gives

M ^) h

<>,

5*

The total change

A(
Yh

due to c', y, and h will be

<>"

Li.05

1 - 0.2 _ 2 "

- 0.271.

^
5

27?,

which is less than the sum of the three (20 +

procedure will overestimate E AF a little.

5 +

= 30$).

Hence the

For positive values of

it is opposite; however, all variations are investigated with respect

to a decrease in F, because this is the critical situation.

In prac-

tice, the coincidence of all variances being in the direction of less

stability is unlikely, but in the deterministic approach, we design for

the combination that gives the most critical situation.

The specific variances are applied to the five slopes investigated,

in the range of factors of safety from 0.8 to 3.

The calculations are

tabulated in Appendix A, Tables A. 5 to A. 13.

Figures 3.18 through 3.22

show the relative sensitivity of F to the variables.

55

Vertical Slope

100%

ZAF
F
=

24

to

33%

av.

28%

0.8 to 3

w
>
'5

<3

c
a*

50%
> O
a>

*'

/ondh

0%

jL
01

0.2
_c_

0.3

0.4

FIGURE 3.18

RELATIVE SENSITIVITY VERTICAL SLOPE

ARXj SAF

FOR

56

Slope 1/2

-=^^ =26

to

37%

av.

31%

100%

0.8

to

>
c
</>

50%

c^__

> o
a*

cr

^^\^
-^_*^
0%
0.1

///i///v
0.3
c'

w,
0.4

0.2

FIGURE

3.19

RELATIVE SENSITIVITY

AFXi EaF FOR

SLOPE

1/2

>7

Slope

l/l

SAF
100%

29

to

39%

av.

31%

0.8

to

>

$ 50%
0>

>

D
CC

0%

RGURE 3.20

RELATIVE SENSITIVITY

AR*L ZAF FOR

SLOPE

l/l

Slope 2/1

ZAF
100%

28

to

41%

av.

33%

F =0.8

to

7U-

<

50%

0%
0.4

FIGURE 3.21

RELATIVE SENSITIVITY

*FXi EaF FOR

SLOPE

2/1

59

Slope 3/1

-^E =30

to

43%

av.

33%

100%
F
=

0.8 to 3

to

0%

FIGURE 3.22

RELATIVE SENSITIVITY

EAF FOR

SLOPE

3/1

60

It is seen that for slopes with 6 >

1*5

and

> 0.1,

c*

is the

most important parameter vith respect to variation in the factor of

safety F, contributing 30% to 60% of the total change in F.

For flatter slones and not too high values of

<$>'

Yh
c
'

/C
,

( yh

'

< 0.1

cot 8),

is the more significant, contributing 35%> to

60% of the total change

in F.

The total change in F, if variances in all parameters are assumed

to decrease F, amounts to 2h% to
is no relationship between F and
1*3$

of F, which is quite large.

.

There

AF

If different values of the variances are assumed, different results

may be obtained.

The values chosen here are thought to be reasonable.

Very little information was found in the literature about actual variances in soil parameters as obtained from field and/or laboratory tests. Singh (1970) uses maximum variances of 50$ for c' and 10 for
<f>',

which

shifts the results in the direction of c' being most significant with

respect to variations in F for a larger range of slopes

3.3.1.1*

Conclusion and Recommendations

The results of this sensitivity study show that for homogeneous

slopes the factor of safety is most sensitive to variations in the

strength parameters c' and

<J>*.

These two parameters may contribute 65$

to 80$ of the total change in F due to variances in the independent

variables.

For steep slopes and

Yh

c
'

> 0.1,
<j>*.

variation in c' has a larger influFor flatter slopes with

ence on F than has variation in

0.1 cot 8, variation in
<f>*

c'

<

has the larger influence on F.

61

These two parameters are of course the most difficult to determine

with high accuracy. This is already reflected in the assumption of

= 20, whereas

^ Y

and

~
h

5*.

The fact that it is impractical to obtain precise values of c' and

4>'

is already accounted for, to some extent, in the requirement that

the value of the safety factor F should lie in the range of 1.2 to 1.5
for slope stability.

These values, although they are based on experiThis depends on how reliable

ence, may not always assure a safe design.

the soil parameters are.

In some countries, difficulties in the determination of soil

strength parameters have led to the introduction of "partial coefficients".

These coefficients are factors that are applied to the soil

parameters, as well as structural and load parameters, if applicable.

The factored variables are used in the stability calculations and the
safe design requires P i 1.

This method has the advantage over the

traditional method of stability analysis, in that the factors can vary

from one variable to another, and hence account for the fact that some

variables are more difficult to determine than others

and that varia-

tion in some parameters has more effect on the result than variation in
others.

Table 3.3 shows some of the partial coefficients used in

For stability and
)

Denmark [from Lundgren and Brinch Hansen (1965)].

earth pressure the partial coefficients for c'
(f

and

<j>'

(f.

are 1.5

and 1.2, respectively, indicating that c' is the one of the two parameters believed to have the most uncertainty and the highest influence on
the result.
c'

The value of c' used in stability analysis would be

'

lab
c

62

Table 3.3

Typical Values of Partial Coefficients used in Denmark for

Geotechnical Designs*

Symbol

Partial Coefficient for

formal

Extraordinary
1.0
1.5

g
f

Dead

load, water pressure

1.0
1.5 1.5

P
f

Live loads

V
*

Wind loads
Friction (stability & earth pressures)

1.0
1.1

1.2

Friction (bearing capacity, foundations & piles)

f
c

1.25

1.15

Cohesion (stability & earth pressures)

Cohesion (bearing capacity, foundations) Cohesion (bearing capacity, piles)

1.5

l.i

1.75 2.0

1.6

1.8

Skin friction, piles

2.0
<

1.8

From Lundgren and Brinch Hansen (1965)

63

For practical purposes it is recommended that Figures A. 11 through

A. 15 he used for determination of AF due to variances in soil data.

For a particular problem, the variances Ac' and A<K should be estimated
from the different possibilities of interpreting the test data (different ways of fitting the Mohr-Coulomb envelope to the test data).

The

same curves may also be used to determine the maximum variation in the
c' and
<J>

'

parameters that can be tolerated if a certain minimum factor

h

of safety is required (Step

in Figure 3.1).

Figures A. 25 through A. 29 may be used in a similar way to determine

the change in F due to changes in the geometry parameters h and 3, when
the soil parameters are considered fixed.

3.3.2

Sensitivity Analysis of an Embankment Founded on a Soil with a

Weak Seam

Moving from the simple problem of a slope in homogeneous soil to

problems encountering inhomogeneous soils
,

the number of independent

variables increases significantly.

mined.
(a) (b)

The following items must be deter-

Independent variables x.

Range of each of these variables.

The range of some

variables may influence the range of others.

(c)

Range of variation in each of the variables, Ax

(Constant, percentage of mean, etc.)

The 16 variables

which apply to an embankment built on a foundaFor some of the

tion with a weak soil layer, are shown in Figure 3.23.

variables, e.g., the soil parameters, certain limits in the values can
easily be established.

The range in geometric variables may impose

6k

<>;<:,<**

Variables
Height and width of embankment

"ymbols
h, B D

Depth to weak layer from base of embankment measured at center line of embankment

Inclination of ground surface

Inclination of weak soil layer

a
6

Inclination of left and right sloDe of embankment

Soil parameters of embankment fill

L>eR

(Y,c,(f>)
]

Soil parameters of foundation soil Soil parameters of weak soil layer

(Y,c,*)
]

(Y,c,(f>)

FIGURE 3.23

VARIABLES FOR AN EMBANKMENT BUILT ON A FOUNDATION SOIL WITH A WEAK

SOIL LAYER

65

further limits on soil strength parameters, when we are only interested

in problems having factors of safety less than 3.
in the parameters must be determined by trial.

Many of the limits

For example, at what

depth will a weak seam in a foundation soil have no effect on the sta-

bility of an embankment?
Because of the excessive number of possible combinations of the

variables (if each of the 16 variables were given

of combinations would be
3

values, the number

= l+3,0U6,72l)

it was necessary to cover

only changes in those variables which are considered to be most important.

The variables that are held constant are mainly those with respect
to geometry, particularly those which are easy to get information about
in a real situation, such as the surface profile (i) and the geometry

of the embankment proper.

The ground surface was taken as horizontal;

the embankment geometry was assumed as: 20 feet in height, ho feet of

crown width, and slopes 1.5/1 (33.7).

The number of soil parameters

is reduced by using an average unit weight y for &

<(>

soils, and taking

as equal to zero.

After this reduction, the following parameters

remain as variables:
D

Depth to weak seam

Inclination of weak seam

y
(c,
<{>)

Average unit weight of soil

Strength intercept and strength angle of embankment soil Strength intercept and strength angle of foundation soil Strength of weak soil seam

(c,

<)>)_,

66

The relationship between the factor of safety F and the unit weight
Y is relatively simple, which means that y can he held constant, leaving

seven variables
In this study, c

w
p

has a maximum of five values.

Ten combinations

of (c,

<J>)

and (c,

<J>)

are investigated, for up to nine different depths

See Table
3.1*.

D, and three different angles a.

3.3.2.1

Factor of Safety for Failure Determined by a Weak Soil Seam

The factor of safety for an embankment founded on a soil with a

weak layer should be less than

(a)

the factor of safety for the embankment itself

(toe failure with 9
>

0),

(b)

the factor of safety for an embankment founded on

a similar soil with no weak layer.

If the factor of safety becomes larger than either (a) or (b), the weak

seam does not influence the stability of the embankment.

Where the geologic conditions, such as Joints in a rock slope or a

weak soil seam beneath an earth slope, determine the shape and location
of the critical failure surfaces, the factor of safety F can be

expressed as

F = F
c

+ P.
<p

(3.1)

where

F
c

^[^tkj
a, 1

^
tan
<f>

...

* n

ej
tan $

(3.2)

and

F.
<p

tan

<J>,

+ a

+ + a

(3.3)

67

Table 3.k Range in Variables Studied for an Embankment on a Foundation with a Weak Layer

Variable
Height of embankment

Symbol
h B D

Range
20 feet
HO feet

Width of

embankment

Depth to weak layer from base of embankment

(measured at centerline of embankment)

to 60 feet

Inclination of ground surface

Inclination of weak soil layer Inclination of left and right slope of

embankment

to 11.3
33.7

V*R
Y

Average unit weight of soil

130pcf
(100 to lUOpcf)

Strength of embankment soil

<J>

lOOOpsf, 0 2000psf, C 500psf, 10 lOOOpsf, 2000psf, 0 500psf, 10^

,

Strength of foundation soil

F' ^F

30:
C

200psf, 30
,

32

Strength of weak soil

200 to lOOOpsf

68

The variables are

n

Number of soil types that the failure surface passes through

Mean value of unit weights of soils above the failure surface
Height of slope or some other characteristic length
to c
to
1
<|>

Strength intercepts of soil No. 1 to soil No. n

(J>

Strength angles of soil No. 1 to soil No. n

,

k.

to a

Geometry dependent variables constant for a certain slope geometry and soil profile.
,

3.3.2.2

Sensitivity Analysis

Using the above expression for the safety factor F, it may be seen
that F, is independent of the unit weight y as well as h, the height of
the embankment.

Any change in y or h will only be reflected in F

It is easy to show that the following relationships hold:

AF
-r

F
=
-

for other parameters held constant

(3.h)

77-

AF Ah

F
=
:

for other parameters held constant

(3.5)

From equations (3.2) and (3.3) the sensitivity of F with respect

to the soil strength parameters
c.

and

<f>.

can be defined as

#
Ac
.

ST*. i yh
a.
i

(3.6)

AF A(tan

=
<f>.

(3.7)

69

Sensitivity of F with respect to the relative geometry can be

expressed as

AF

= E

~
3F
3k.
i

3F,

Ak.
i

+ I

s-*
3a.
1

Aa,
i

E Yh

c,
l

Ak. + I tan
l

<J>,

Aa.
l

(3.8)

3.3.2.3

Results of Stability Analysis for Embankments

Different search routines in the STABL program were applied to the

problem to locate the shape and position of the most critical failure
surfaces.
It was found, as expected, that the sliding block mode of

failure was the most critical, with the base of the block located along
the weak soil seam. See Figure
3. 2U

and Figures A. 55 through A. 53.

However, this was also the case for homogeneous soil, see Table A. 1^4
and Figures A. 59 and A.60.

When comparing these results with results

obtained in the comparative analyses, Chapters V and VI, it may be concluded that the method of slices used in the STABL program (Janbu's

simplified procedure) is likely to give more conservative results for

irregular failure surfaces than for circular failure surfaces.
It was also found that a sliding block type of failure surface with

active and passive portions of the surface inclined at angles corre-

sponding to the Rankine theory,

**5

1+5

<j>'/2

for the active portion and

- <J>'/2 for

the passive portion, was more critical than the sliding

block with randomly generated active and passive portions of the failure
surface (Figures A. 57 and A. 58).

For this reason, the BL0CK2 sliding

block search was introduced in addition to the BLOCK search, see

Chapter IV.

TO

weak seam

FIGURE 3.24

A SLIDING BLOCK TYPE OF FAILURE

SURFACE
IS

MOST CRITICAL

71

When the weak soil layer is inclined with the horizontal, the sta-

bility should be investigated for both sides of the embankment.

This

was done in a few cases, but in most cases the stability was investi-

gated only to the side of the embankment where the weak layer is dipping
down, since this is most difficult to change.

On the side where the

weak soil layer is inclined upwards, it may more easily be replaced

with stronger soil, if the stability is threatened.
The results for the few cases where stability was investigated to

both sides show, that for foundation soil of the same strength as the
embankment soil, failure is more critical towards the down-dipping weak
seam.
In the case of stronger foundation soil, it may be more critical
,

towards the side where the weak seam is inclined upwards

3.25 and 3.26.

see Figures

In most cases the foundation soil will be weaker than

the embankment soil, so that failure towards the side where the weak
layer dips down is the most likely.

The failure surfaces remain the same, or very close to the same,
for different foundation soils within a relatively large range in soil

parameters (y, c,

<J>),

so long as the soil geometry and relative strength

. .

of the weak soil remain unchanged, see Figures A.6l through A 6U

But

the safety factor varies.

As the strength of the weak seam increases, the active and passive

portions of the failure surface move closer to each other. involved in the failure decreases, as shown in Figure 3.27-

The volume

Despite
(3.2),

this change in the failure surface, the initial equations [(3.1),

(3.3)] can still be applied to some extent, because the change in shape

and location of the failure surface is relatively small.

Exceptions to

Embankment 20

feet high, Slopes 1.5/1

max

Failure
Failure

to left

to

right

200

400

600

800
cw

1000

psf

Strength of

Weak Seam

FIGURE

3.25

COMPARISON BETWEEN FAILURE TO LEFT AND FAILURE TO RIGHT SIDE OF EMBANKMENT, D= 10 FT, a = 5.7, <*>0

73

Embankment 20
3

feet high, Slopes 1.5/1

pmox

>

t u

'

Failure to left
r

see Figure 3.26

Failure to right

200

400
Strength of

600

800
c

1000

psf

Weak Seam

FIGURE 3.26

COMPARISON BETWEEN FAILURE TO LEFT AND FAILURE TO RIGHT SIDE OF EMBANK= MENT, D=IOFT, a-H.3?
<*>

7^

Type
weak seam

surface

Type 2 surface

FIGURE 3.27

CHANGE IN CRITICAL FAILURE SURFACE WHEN STRENGTH OF WEAK SOIL SEAM INCREASES

75

this are the situations where the active part of the failure surfaces

intersects the ground surface beyond the crown of the embankment, see

Figure 3.27 (surface type 2).

Tn these cases the geometry of the

problem changes so much that new values of "a" and "k" parameters apply
in equations (3.2) and (3.3).

Type 1 surfaces (Figure 3.27) occur for

small relative depths of the weak soil layer and small inclinations of
a.

Type 2 surfaces occur for larger relative depths of the weak soil

layer and for larger inclinations of a.

From the results (Table A.ll+), curves are plotted showing variations of the normalized factor of safetv F
n
c

vs. the normalized strength

=
n

of the weak seam c

where F

= n

Fyh
'

and

w
c

These curves

show a very obvious change when the failure surface changes from type 1
to type 2, as a break appears in the otherwise linear relationship

between F

and

see Figures 3.28 and 3.29.

From these curves the

"k" and "a" values can be found.

Table 3.5 summarizes the results.

Curves of "k" and "a" values are plotted vs. a for constant depth D
(Figures 3.30 and 3.31) and vs. depth D for constant a (Figures 3.32

through 3.36), where a is the inclination of the weak soil seam.

The a-graphs show that the strength angle
<(>_,

for the foundation

soil has a larger influence on the stability than has the strength

angle

<$>

of the embankment soil.

The difference increases with

increasing depth D, as a_ increases and &- decreases with increasing

depth.

This seems reasonable, as the frictional strength increases

with the overburden, and the overburden is larger for the part of the
failure surface that passes through the foundation soil.

76

FXh

cw

5
N O
E

0.2

0.4

0.6

0.8

1.0

Normalized Strength of

Weak Seam

cn

cn

0.2

0.4

0.6

0.8

1.0

CF

CE

Fn
Fn Cn
^n

3.73
5.35
0.1

4.29
5.93
0.2

4.69
6.47
0.3

4.92
6.93
0.4

5.13
7.23
0.5

1000
I

3.39

3.72

4.02

4.29

4.54

2000

FIGURE 3.28

NORMALIZED FACTOR OF SAFETY Fn VS. NORMALIZED STRENGTH c FOR EMBANKMENT 20 FEET HIGH SLOPES 1.5/1 DEPTH TO WEAK SEAM 5 FEET, INCLINED AT a= 11.3 TO HORIZONTAL
, ,

77

FXh

Fn * 5.85
Fn

+ 1.14 c n

"2.75+ 2.93 c n

'2 (type of failure surface)

N o i o

0.2

0.4

0.6

0.8

1.0

Normalized Strength of Weak Seam c n

cn

0.2

0.4
3.91

0.6
4.31

0.8

1.0

cE

-C7

*E
0

*r
10

Fn
cn

3.34
0.4

4.54
1.6

4.75
2.0

0.5

08
6.76

1.2

Fn

5.67

728

7.73

8.12

10

10

FIGURE 3.29

EXAMPLE OF Fn VS. c n FOR EMBANKMENT SLOPES 1.5/1 DEPTH 20 FEET HIGH TO WEAK SEAM 5 FEET, INCLINED AT
,

a -11.3 WITH HORIZONTAL

78

Table 3.5

Geometry Parameters k and a for Embankment Stability Analysis

Surface type*

D
ft

tan a

(2)
1 1
5 5

0.20 0.20

2.1+1
2.1+1+

1.01+ 1.1+7

2.80
1.57

0.71 0.65

0.86 0.72

2.71+

1.06 I.69

1.87 1.33 1.11

2.72 1.50 1.20
2.00
1.1+6

1
1

0.10

2.31+

0.83
0.1+7
0.51+

1.01
1.51+

5
5

0.20
0.20

1.90 1.50
1.99

2.11 1.66 1.72 2.19 2.25 2.50

2.1+6

(2)
1

0.92 1.20 1.60 1.71

10
10
10

0.82 0.59
0.1+5

1 2

0.10 0.20 0.20

1.69

1.15
1.05
1.51+

2
1

15 20

0.88
1.07
0.61+

0.1+7

2.25
3.01

2
2

20
30
i+o

0.20 0.20

0.83

2.82
3.28
3.
31+

0.33

0.58
O.56

2 2
2

0.5!+
0.1+1+

0.12

1+.73

i+o

0.20

0.1+1

3.60
3.9*+

0.08

(5.1+3)

50

0.20

0.27

0.3!+

Surface types 1 and 2 are illustrated in Figure 3.27.

79

D C
o

s
u
'
0)

E o

Type

surfaces

Type 2 surfaces
(for </>soils

mainly

0.10

0.20

0.10

0.20

Inclination of

Weak Seam, tana

FIGURE 3.30

VARIATIONS OF

WITH a FOR CONSTANT DEPTH D0.25h

AND "a" COEFFICIENTS

Ro

15

o c
D

o
8.

Type

surfaces

Type 2 surfaces

0.10

0.20

0.10

0.20

Inclination of

Weak Seam

tan

FIGURE

3.31

VARIATIONS OF k AND "a" COEFFICIENTS WITH a FOR CONSTANT DEPTH D-0.5h

81

D k.

0>

Type

surfaces

0.25

0.50
Depth Ratio
--

|.o

to

Weak Seam

FIGURE 3.32

VARIATION OF k COEFFICIENTS WITH DEPTH D FOR CONSTANT a =0

82

tan a

0.10 (a -5.7)

b.
tt

<D

E o c o

Z
E

Type

surfaces

0.25

0.50
"pf ro

1.0

Depth Ratio

Weak Seam

FIGURE 3.33

VARIATION OF k COEFFICIENTS WITH DEPTH D FOR CONSTANT 0=5.7

tana

0.20

(a-ll.3)

4>

O
0.25

Type

surfaces

Type 2 surfaces

0.50
Depth Ratio
-^

to

Weak Seam

FIGURE 3.34

VARIATION OF

COEFFICIENTS WITH
a -11.3

DEPTH D FOR CONSTANT

8U

tan a
(

A-tana

-0.10)

Type

surfaces

Type 2 surfaces

V.

5!
01

E 2
S.

I
o

01

0.5

1.0

2.0
"h~

Depth Ratio

to

Weak Seam

FIGURE 3.35

VARIATION OF

"a"

DEPTH D FOR CONSTANT

(AND a- 5.7)

COEFFICIENTS WITH a * 0

85

tana

0.20

o
</>

Type

surfaces

Type 2 surfaces

S E 2 o

a*

Depth Ratio -^

to

Weak Seam

FIGURE 3.36

VARIATION OF "a" COEFFICIENTS WITH DEPTH D FOR CONSTANT a = 11.3

86

With respect to the strength intercepts c, the k-graphs show that

for small angles of a and relatively small depths (D = 0.3h to 0.5h),
the strength intercept c of the embankment soil has a larger influence

on the stability than the strength intercepts c and c

r

of the founda-

tion and the weak soil, respectively.

larger influence.
angles of a (= 0
)

For greather depths, c^ has the

r
,

Comparing k^ and k

it is seen that for very small

c

and depths D = 0.5h, the strength intercept

of

the weak seam has a larger influence on the stability than has the

strength intercept c of the foundation soil.

r

For larger depths and/or

larger angles a,

c,, r

has the larger influence.

Over the range of variables investigated, the factor of safety F

decreases with increasing angle a or depth D of the weak seam, to a

certain depth, then F increases with increasing a or D, other parameters

being held constant.

The depth at which the change occurs depends upon

the strength of the foundation soil relative to the embankment soil, and to some extent on the relative strength of the weak soil seam.

For

foundation soil of equal or lesser strength than the embankment soil,

the shift for F vs. a occurs at a depth D approximately equal to twice
the embankment height, whereas for stronger foundation soil, the shift

occurs at much smaller depths.

Figure 3.37 shows F vs.

The shift for F vs
.

for

F E

= 1

and 2, and a = 0, 5.7 and 11.3.

D occurs

at

depths D between 0.25h and 2h for a foundation soil of the same strength
as the embankment soil.

For a stronger foundation soil, F increases

with depth D over the total range of D.

Figure 3.38 shows the variation

in F with respect to a and D for foundation soil of the same strength

as the embankment soil

87

a = 0

U.

Embankment 20

feet high
1.5/1

Slopes

200

400

600

800
cw

1000

psf

Strength of

Weak Seam

FIGURE 3.37

VARIATION OF F VS. c* FOR VARIOUS a .AND DIFFERENT RATIOS OF e., D5

FT

88

Embankment 20
c E = cF =

feet high, Slopes 1.5/1

1000

psf

<f>

2.0

c
>

w -600 psf

o
in

cw

200

psf

11.3

inclination

of

weak seam

10

20
Depth of Weak

30
Soil

40
Seam
D,
ft

50

60

FIGURE

3.38

EXAMPLE OF VARIATION OF THE FACTOR OF SAFETY WITH GEOMETRY OF WEAK SOIL LAYER

89

3 2
.

. '*

Results Applied to Sensitivity A nalysis of Embankments

Sensitivity of F with respect to y

From Equations (3.l), (3.2) and (3.3) the following expression

was derived

Example
:

-r

<**>

An embankment has a factor of safety with respect to strength

intercept F

=1.2, and
=
.

a unit weight Y = 120 pcf. This gives

& Ay

L
120

_ o U,U1 oi

If 120 ucf is the average value of several measurements of y,

and say the standard deviation of these measurements is + 2.5,

this gives a range in F of

F = 1.2 + AF = 1.2 + (-0.0l)-2.5 = 1.2 + 0.025

or

f-

= ~

f-

lf|o

2 1%
'

>

or F = 1 115 to 1>225,
-

Sensitivity of F with respect to height h

Ah
Example
:

1 -r h

(3 5)
-

Again we assume an embankment with F

of 20 feet.

= 1.2, and a height h

AZ
Ah

1^2
20

0>o6

Considering whether we can add say

feet to the height

-

we

find AF = - 0.06*5 = - 0.3, which gives F = 1.2

so the answer is no.

0.3 = 0.9,

90

Considering an uncertainty in the measurement of the height

during construction, which leads to a height that is 21 feet

instead of 20 feet.
AF =
-

This decreases the safety with

0.06-1 =

0.06, hence F = l.lU.

Sensiti vity of F with respect to strength intercept c

f~ Ac
.

ik
Yh

(3.6)

Example

An embankment 20 feet in height with slopes 1.5/1 and a ^0

feet wide crest is built on a foundation soil which has a

weak seam located 10 feet below the original horizontal

ground surface at the center of the embankment
5.7 with the horizontal.
,

and inclined
3

This problem has a maximum of

types of soil, and the k parameters found were k = 1.69, 2.19, k k jr

=1.20.

If the variation in strength intercepts

is Ac = + 200 psf , and Y = 125 pcf

AF Ac

AF_
Ac_
E

AF_
Ac F
=

AF.
Ac

J.
yh
E

w,

5.08

125'20

0.002

and AF = O.Hl, which is quite high.

If the embankment is
c

designed with a safety of F = 1.5, this variation in

result in an F value of only 1.1.

may

Notice that k^ is almost twice as large as kw> which means

that greater effort should be used to determine c p than
in this particular case.
c

However, complete neglect of a

weak seam could be catastrophic.

91

Sensitivity of F with respect to strength angle

<j>

AF

AltalT^T
E xample
:

a
i

(3 T)
-

For the same embankment as described above, it was found that

a
E
= 0.59 and a^ = 1.60.
~ 30 to 35,

If A*. = 5, we find for

<J>.

A(tan
and for
-

<f>

= 0.123

AF = 0.27

<j.

10 to 15,

A(tan

(J).)

= 0.092

AF = 0.20,

which is also a significant variation in the factor of safety,

although the strength intercept
c

in this particular case has

a larger influence on F than has the strength angle

<J>.

From Tables 3.5 and 3.6 it can be seen that although k, k and k
h
r

vary with location of the weak soil layer (as represented by a and D),
the sums of the k parameters vary only a little. Hence the result for

AF(Ac) is almost independent of the problem geometry, and depends mainly on the values of Ac.
a-j,

This is not the case with respect to

AF(A<}>)

as

increases almost linearly with depth D to the weak seam, and a

de-

creases only slightly with D.

AF(A(J>)

So for greater depths D of the weak seam,

may be larger than AF(Ac).

Sensitivity of F with respect to relative geometry

AF

-r Yh

E c.
1

Ak.
1

+ I tan

<J>.

Aa.
1

(3.8)

92

Table 3.6
statistical Analysis of Geometry Parameters k and a

Dominating Failure Surface

(2) 1

D ft

tan a

*? + ku

+ k

\l

*F

0.20 0.20
5
5 5 5

3.H5 3.91 3.80

3.81+
3.01+

6.25
5.1t8

1.57 1.37
1.81+

1 1 1
(2)

0.10 0.20 0.20

1+.03
It.

01 3.16 3.71 3.88

3.
'40

2.93 3.02 3.22 U.38

3.22 3.39 U. 25

5.67 5.36 5.12 5.88

2.01
1.1+6

1 1 2
2 1 2 2

10 10 10
15

0.10 0.20 0.20

5.21 5.08
5.1+0

2.02 2.19 2.16

-

3.51*

3.96
3.
31+

5.00
1+.88

20 20
30
1*0 1+0

i+.oo

2.72
3.31+

0.20

3.65

3.89
3.92

U .72
1+.50
1+.1+1+
It. 1+5

0.20

3.86 3.90

1+.85

2 2
2

3.88
It.Olt

0.20
0.20

1+.01

(5.51)
-

50

U.21
3.78

it.

28

U.55

Mean
Variance

3.66
0.21+

5.12
0.29

2.32
1.03

0.08

Standard
Dev.

0.28

0.1+9

0.53

1.01

Variance
Coef.

iM

13.3$

10.lt*

1+3.75?

93

Example

Consider the same embankment as in the previous example,

with a safety factor F = F

+ F, corresponding to c^ = 1000 E
(J)

psf, c = 500 psf, $_ = 10, and c

r

= 200 psf, which gives

=
=

J:
1.U9.

(1.69-1000 + 2.19-500 + 1.2-200)

+1.6 tan

10

Let us assume that there is an uncertainty in the determina-

tion of the position of the weak soil seam, reflected as

+ 2 feet in the depth D and

+0.05

in the inclination (tan a).

We now have to determine which sign, plus (+) or minus (-),

will decrease the factor of safety.

c and c
r

Since c v is larger than

r

the value of k is compared to k_ and k

Jii

We are

first investigating the influence of a change in the depth D.

Figure 3.33 shows that for a = 5-7, k decreases with inr,

creasing depth D, whereas k increases approximately the same

amount.

Since c = 2c_, the result is a decrease in F.

L
r

Further, k

also decreases with increasing depth.

On the

other hand, a increases very much with increasing depth (see

Figure 3.35), so it is not obvious whether F = F

decrease of increase with increasing depth.
situations have to be investigated.
AD Ak
= + 2 feet

+ F

will

Therefore both

= - 0.25, Akp = 0.20, Ak

= - 0.05, Aa^, = 0.2U

AF

lg c. go

(- 0.25-1000 + 0.20-500 - 0.05-200)

+ 0.2^ tan 10

- 0.06 + 0.0H = - 0.02,

which shows that increased depth decreases the stability,

91*

although the change is negligible.

For AD =

- 2

feet, the

changes in the "a" and "k" parameters are of opposite sign, but of the same magnitude, so AF = + 0.02.

Applying the same procedure to a change in the angle a,

Figure 3.31 shows that k decreases for increasing a;
k^,

in-

creases, but is almost constant for a between 5.7 and 11.3

at this -particular depth (D = 10 feet = 0.5h), whereas k

increases for increasing as well as decreasing a, because the

failure surface for increasing a changes from type 1 to type

However, a increases with increasing a, so like the previous
case, it is not obvious whether a decrease of an increase in

2.

a will result in a decrease in F.

Atan a = + 0.05

Ak

= - 0.25, Akj, = 0.08, Ak

= 0.16, Aa^ = 0.09

AF

jpc.gQ (-0.25-1000 + 0.08-500 + 0.16-200)

+0.09 tan
Atan q = - 0.05
Ak

10

0.07 + 0.02 = - 0.05

= 0.19, Akp = - 0.19, Ak

= 0.08, Aa^, = - 0.16

AF

2 c. g0

(0.19-1000
=

0.19-500 + 0.03'200)
_

- 0.16

tan 10

0.01+

0.03 = 0.01,

so F decreases for increasing q.

Comparing the changes in the factor of safety F due to variations

in the independent variables, we find that the only significant change
in F is due to variation in the soil strength parameters c and
(J).

95

The above results are based upon a rather limited study, particularly with respect to the range in the a angles investigated.
For very

steep angles of a, the influence of the weak seam may be negligible.

However, this was not investigated.
3.3.2.5

Summary and Conclusions

The task of performing a sensitivity analysis on a slope stability

problem becomes almost impossible when moving from the simple case of
slopes in homogeneous soils to more practical, but also more complica-

ted problems, involving different types of soils and changing soil

profiles.

The number of variables increases drastically, and the posIn this

sible combinations of these variables increases even more.

study, the number of independent variables was reduced from sixteen to

seven, compared to the three independent variables that apply to the

homogeneous slone stability problem.

The results, in terms of safety

factors, occupy ten tables, compared to one for homogeneous slopes, and
even so they cover only a small range in the possible combinations of

the independent variables.

Despite the limited investigation, the analysis did show that

almost the same conclusion may be reached with respect to problems en-

countering inhomogeneous soils, as was reached for homogeneous slope

stability problems
,

namely that the dominating influence on the stabil-

ity and changes therein are determined by the soil strength parameters
c

and

(j)

However, the neglect of a weak soil seam may have a large in-

fluence on the result of a stability analysis, and it may lead to

failure of a slope that otherwise would have been stable with a safety
factor of F = 1.2 to 1.5.
correct c parameter (or
cf>

But this is in fact caused by using an in-

parameter).

96

IV

MODIFICATIONS OF THE STABL PROGRAM

This Chapter describes the major modifications of the GTABL

program that were performed during the course of this research.

adjustments and corrections are reported in Appendix D.

Other

Also included in this Chapter is some general advice to users,

specifically regarding problems that frequently have been raised by
users of the STABL program.

k.l

THE MODIFIED SLIDING BLOCK ROUTINE (BL0CK2) During the sensitivity analysis of an embankment founded on a soil

with a weak soil layer, as described in Section 3-3.2, it was observed

that a sliding block type of failure surface with active and passive

portions of the sliding block surface inclined at

(1+5

<j>'/2)

and

U5

- <j>'/2)

with the horizontal, respectively, gave the lowest factor

This type of failure surface corresponds to Rankine' s earth
See Figure k.l.

of safety.

pressure theory [Rankine (l857)]>

As the original STABL sliding block search (BLOCK) generates the

active and passive portion of the sliding block more randomly, it was
felt that a sliding block using the Rankine theory for active and

passive portions of the failure surfaces would be useful in the analysis of relatively simple slope stability problems.

The original sliding block, on the other hand, was still felt to
be a better solution for cases of soil profiles that are complex.

97

weak

soil

seam

FIGURE

4.1

SLIDING BLOCK FAILURE SURFACE

WITH RANKINE ACTIVE AND PASSIVE

ZONES

98

The new sliding block procedure (BL0CK2) is modified from the

original, with much of the original procedure retained.
This applies

to the random technique used to pick control points on the sliding

block failure surface.

Only the technique for generation of active

The data cards

and passive portions of the sliding surface is changed.

are unchanged except for the command code.

1 1
.

Generation of Active and Passive Portions of the Sliding Block

According to Rankine's Theory

The procedure for generation of the active and passive portions of
the new sliding block is as follows.

The passive portion of the failure surface starts from the point
of the failure surface chosen randomly within the first box defined,
see Siegel (1975a, p. 96).

The number of the nearest soil boundary lo-

cated above this point is determined by checking for intersection with

a vertical line through the point.

When this soil boundary is deter-

mined, the soil type below is known, and the direction of the first
line segment of the passive wedge is determined as (U5
<f>'

/2

with

respect to the horizontal, where

4>'

is the strength angle of the soil.

Each time a line segment is generated, intersection with soil

boundaries is investigated.

If an intersection occurs, the previously

<{>'

described procedure is repeated to determine the new

angle to use.

The active portion of the failure surface is generated similarly

and from the point chosen within the last box specified.
from the horizontal is
(1*5

The angle

+ <J>'/2), instead of (U5

- 4>72).

The

<!>'

angle used is that defined as the isotropic strength angle.

If the soil is specified with anisotropic strength parameters, the

99

active and passive zones of the sliding block BL0CK2 will still be

generated according to the

<J>'

angle defined as the isotropic

<J>'

<f>'

angle.

This means that one should specify the isotropic

parameter with the

value of the anisotropic

<j>'

parameter.

In the active wedge range

between approximately + ^5 and + 65, and in the passive wedge range between approximately - 15 and 1*5

See Figure k.2.

If

(J)'

differs

considerably from the active to the passive regions, one may have to
divide the anisotropic soil into two soil types, one in the active
region, and another in the passive region.

Although the failure sur<j>'

face will be generated according to the isotropic

value, the aniso-

tropic

<(>'

(if different) will be used in calculation of the available

shear strength along the failure surface.

Further, limitation boundaries are ignored in this failure surface

searching routine.

The program prints out a warning if they are speci-

fied, but generates failure surfaces as if they did not exist.

U.1.2

Application of the New Sliding Block Routine

The new sliding block routine, BL0CK2, is particularly suited for

relatively simple soil profiles, where a sliding block type of failure

is obviously the most likely.

This is, for example, the case with the

problem analysed in Chapter III, an embankment built on a foundation

with a weak soil layer. If the soil profile is complex, it is recom-

mended that the original routine, BLOCK, be used.

k.2

OPTION OF SEVERAL PIEZOMETRIC SURFACES The original STABL program had three possibilities for defining the

pore water pressure within a slope.

These were: the groundwater table;

100

Anisotropic

direction ranges
I

for soil types

and 2

90

Soil type 2 Define isotropic

soil

type 3 weak seam

FIGURE 4.2

BLOCK2 SLIDING BLOCK ANALYSIS APPLIED TO ANISOTROPIC SOIL

101

the pore water pressure parameter r

which may be used to present

excess pore water pressures due to shear; and finally the pore pressure

constant u

that defines a constant pore pressure for a defined zone.

This would seem to be sufficient to model any type of pore water

pressures, equilibrium and excess of equilibrium, that would be encoun-

tered in a slope stability analysis.

However, a modification which

allows different piezometric levels to be assigned to different layers of soil, was proposed by the Division of Materials and Tests, ISHC.

This approach is primarily applied to end-of-construction stability

analysis of embankments built upon clayey foundations, where the con-

struction of the embankment creates excess pore pressures in soil layers

with poor drainage characteristics.

Using the r

parameter in such cases would give pore pressures that

were proportional to the total overburden from the soil layer considered to the gound surf ace
,

whereas the excess pore pressures imposed by the

embankment is believed to be proportional only to the increase in over-

burden which the embankment represents.

The pore pressure constant u

might be applied, but this would re-

quire an increase in the number of soil types to be defined in the area

where the embankment slopes, and a corresponding increase in soil boundaries.

It was felt that it would be much simpler to introduce a

second

piezometric surface to represent the excess pore pressures built up in

the less permeable soil layers due to construction of an embankment.

1+.2.1

Program Modifications
The option of defining more than one phreatic surface required

significant changes in the subroutines WATER and WEIGHT in STABL.

WATER

102

is the subroutine that reads, checks and stores the data for the

piezometric surfaces, if any.

acting on each slice.

WEIGHT calculates the resultant forces

These include the total weight of the slice and

To be sure

water forces acting on the top and bottom of the slice.

that the correct unit weight, moist or submerged, is used for each soil

type within a slice, it is necessary to calculate and store the level

of each piezometric surface, where it intersects the center line of the

slice.

Only the piezometric surfaces that apply to the top soil and

bottom soil of a slice are used in the calculation of the water forces
U
a

and U
8

where U

is the water force acting on the bottom of a slice,

and U

is the water force acting on the top of a slice.

p

H.2.2

Change in Data Cards

To define more than one piezometric level, an integer is added to

the soil parameters, indicating which level applies to the particular

soil.

If no piezometric surfaces are defined, any number may be used.

Also the data cards for WATER are changed to allow up to 10 piezometric
surfaces to be defined.

These changes are all included in the revised

STABL User Manual.

U.2.3

Application of more than one Piezometric Level

The option of defining more than one piezometric level can be

applied to problems where either artesian water pressures or a perched

water table exist, see Figure k.3.
The program will treat the perched

water table problem as follows.

For a slip surface located within the

clayey soil, the water pressure on the bottom as well as on the top of a
slice will be calculated as the hydrostatic pressure corresponding to the

103

I H 2 m o
K Q.

UJ -j

or
ij-

o
cn en >_J

< _i z m < ?
>1_i
UJ

in

b < m 1 hw Q UJ O t o _I UJ
ft!

en

0.

ro <f UJ cc

3
u.

e>

10U

perched water table.

If a slip surface extends down into the sand layer,

the water pressure at the bottom of a slice, if located in the sand,

will be calculated as the hydrostatic pressure corresponding to the

ground water table. The top of the slice will still have a water pressure

corresponding to the perched water table.

Another stability problem, where the option of defining two water

tables was applied, is that of a reservoir embankment analysed for rapid

drawdown, see Figure U.k.

A slight modification of the method proposed

by Taylor (1937) is applied.

Taylor assumes that during the rapid draw-

down no drainage takes place in the embankment.

While the water pressure

on the surface of the slope is removed, the water pressure within the em-

bankment remains the same as it was under the submerged condition.

This

is considered a very conservative assumption [Dvinoff and Harr (1971)].

The modified assumption used in the above is that within the clayey soil

core no change in water pressure occurs during the rapid drawdown, whereas
in the outer shell consisting of granular soil of very high permeability,

the water pressure changes according to the drawdown water table.

The

stability analysis is performed defining the drawdown water table for the

outer shell of granular material, and the original high reservoir water
table for the impermeable core material.

The strength of the core material has to be defined in terms of

effective strength parameters

<t>'

and c'.

If the strength is defined

<J>

solely in terms of a total strength c, with

= 0, the presence or ab-

sence of one or several water tables does not influence the result

unless the water table that applies to the top soil extends above the

ground surface

acting as a surcharge

105

Reservoir High water table

_S7

shell of

granular
material

(a)

Embankment With

Full

Reservoir

(b)

Water

Table After Drawdown

(approximation)

FIGURE 4.4

RAPID DRAWDOWN STABILITY ANALYSIS FOR EMBANKMENT

106

This may also be seen directly from the expression for the factor

of safety (Section U.3, Equation k.l), where the water force at the

bottom of a slice, U

will disappear from the equation if

<f>'

= 0,

whereas U, the water force at the top of a slice, if any, still is

contained in the expression for A

it.

MODIFIED BISHOP FACTOR OF SAFETY FOR CIRCULAR FAILURE SURFACES During the comparative analysis described in Chapter VI, it was

discovered that the simplified Janbu method of slices used in the STABL
program does not always provide conservative results compared to other
methods.
In fact it may give erroneous and unconservative results

where a failure surface intersects the ground surface at the top of the
slope at very steep angles
,

and the strength of the soil is defined

c

mainly in terms of a strength intercept

(or c').

Surfaces with such

steep angles are seldom the most likely failure surfaces, so in most

situations we will not meet this problem.

But it may become important

where deep circular failure surfaces are analysed, e.g., for an embankment on a soft foundation. This problem is investigated in Section 6.k.

As a result of the findings, it was decided to add the option of

using the modified Bishop factor of safety in addition to the simplified

Janbu factor of safety for circular failure surfaces.

1+.3.1

Program Modifications
The modification of the program was relatively simple. The Janbu

factor of safety by the method of slices is calculated by iteration of

the following equation A

-FA

'

107

where
A.,

cos a U

tan

<J>*

(W (1 - k

- k,

tan a]

LL (cos B + tan a sin B)

+ Q (cos 6 + tan a sin 6)}

A
d.

W {(1 - k

tan a +

k,

+ U (tan a cos

f>

- sin B)

Q (tan a cos

sin 6)

tan a tan

(J)'

[see

Siegel (1975a)
This equation is derived by taking the moment of the forces acting

on the slices around infinity (overall horizontal force equilibrium).

If the moment is taken around an arbitrary point, the equation becomes

A,

-PA

where y is the vertical distance from the center of the bottom of a

slice to the moment center. If we, for a circular shaped failure surface, take the moment

around the center of the circle, y = r cos a (see, e.g., Figure 0.1*0,
and Equation k.2 becomes

A
E

r cos a

- F A - - - 2

108

or since r = constant,

A
Z cos a
x

- F A

- -

(U.3)

which is the equation used for the modified Bishop factor of safety-

calculation

A more detailed description and discussion of the equations and

assumptions applied in the Bishop method as well as the Janbu method

may be found in Chapter II of this report, by Siegel (1975a), and

Wright (1969).

U.3.2

CIRCLE2 and SURBIS

To use the modified Bishop factor of safety option for a circular

failure surface search, one should use the command word CIRCL2 in place
of CIRCLE.
It may also be applied to a specified surface by using the

command word SURBIS in place of SURFAC.

To avoid erroneous use, a

warning is printed out

that this method should be applied only to

failure surfaces that approximate a circular shape.

k.k

EXAMPLE PROBLEM FOR REVISED STABL PROGRAM

The problem analysed in Chapter VI, Section 6.2, is used to illus-

trate the data setup and output for the revised STABL program.

Figure

6.k shows the problem with soil parameters and positions of piezometric

surfaces.

Table U.l shows the data setup for a sliding BL0CK2 analysis.

Note

that the unit weight of water (the second variable defined under the com-

mand word WATER) is specified as 0., which means that the default value
of 62. k pcf is used.

109

Table h.l

Data Setup for Example Problem for Revised GTABL Program

PRDFIL EXAMPLE PROBLEM

23 11
46. 10. 46. 4 10. 46. 17. 42. 4 17. 42. 26. 42. 4 26. 42. 34. 46. 4 34. 46. 56. 45. 4 56. 45. 90. 56. 1 90. 56. 145. 75. 2 145. 75. 156. 76. 2 156. 76. 240. 105. 1 240. 105. 259. 108, 259. 10S. 320. 103. 156. 76. 320. 78. 2 90. 56. 320. 58. 1 56. 45. 216. 38. 4 216. 38. 320. 43. 4 0. 36. 20. 36. 5 20. 36. 136. 29. 5 136. 29. 216. 32. 5 216. 32. 320. 28. 5 136. 22. 4 0. 22. 0. 17. 31. 17. 1 31. 17. 136. 22. 1 136. 22. 320. 22. 1
0.

1
1

SDIL
5 125. 125. 125. 122. 122.
125. 0. 35. 0. 0. 125. 0. 33. 0. 0. 125. 0. 35. 0. 0. 122. 320. 29. 0. 122. 740. 16. 0.
1

1
1

0. Q.

2
2

UATER
2 5
0. 0.

45. 56. 45. 126. 51. 216. 47.

320. 47.
-?

45. 56. 45. 126. 51. 136. 60. 150. 64. 250. 112. 320. 110.
0.

BLQCK2 50 2 50.
50. 29. 90. 26. 8. 140. 26. 280. 26. 6.

110

Table k.2 is a listing of the corresponding output.

plot is shown in Figure U.5.

The computer

This problem illustrates the use of the BL0CK2 sliding block search,
and of the WATER command.
in Chapter VI, Section 6 .h
,

An example of the CIRCL2 search can be found

with data setup in Appendix C, Table C.7.

*t.

SOME ADVICE IN THE USE OF STABL

This section is intended to give general advice in the use of the

STABL program, and to discuss some of the questions and problems that

frequently have been raised by users of the program. First it should be emphasized that a computer program gives results
that are no better than the input data.
But let us assume that we have

determined our problem parameters with reasonable accuracy, or at least

with a known variability.

How do we get the best results from the com-

puter with the least amount of effort?

Preliminary Analysis - Stability Charts

Whether the problem is relatively simple or more complicated, it is

usually a good time investment to try to determine an approximate solution by means of stability charts for homogeneous slopes as, for example,

presented by Taylor (1937, 19^8)

see Figures k.6 and

1+.7.

A chart similar to Figure h.6 was constructed based on STABL results

from the analysis of homogeneous slopes presented in Chapter III.
It was

found, however, that charts constructed from equal contour lines of factors
of safety F = 2 and F =
3

differed somewhat from the chart based on F =

contours.

See Figure U.8.

The STABL stability chart was compared to Taylor's chart (Figure

k.6)
.

Both charts are superposed on Figure h.9.

It is seen that for

Ill

Table k.2
Output of Example Problem

SLOPE STABILITY ANALYSISSIMPLIFIED JANBU METHOD OF SLICES IRREGULAR FAILURE SURFACES

PROBLEM DESCRIPTION

EXAMPLE PROBLEM

BOUNDARY COORDINATES

BOUNDARIES 11 TDP 23 TOTAL BOUNDARIES

BOUNDARY
MO.
1

X-LEFT
<FT>

Y-LEFT
<FT)

X-RIGHT
<FTJ

Y-RIGHT
<FT>

SOIL TYPE BELOW BND

4 4
4 4

2 3
4

5 6 7

8 9 10
11

10.00 17.00 6.00 34.00 56.00 90.00 145.00 156.00 240.00

12 13 14 15 16 17 13 19

259.00 156.00 90.00 56.00 216.00

20.00 136.00 216.00
31.00 136.00

46.00 46.00 42.00 42.00 46.00 45.00 56.00 75.00 76.00 105.00 108.00
76. 00

10.00 17.00 6.00 34.00 56.00 90.00 145.00 156.00 40.00

259. 00 320. 00

56.00 45.00
38. 00

320.00 320.00 216.00

30. 00

36.00
36. 00

20
21

22 3

9.00 32.00 22.00 17.00 17.00 .00

0.00 136.00 216.00 320.00 136.00 31.00 136.00 320.00

46.00 42.00 42.00 46.00 45.00 56.00 75.00 76.00 105.00 108.00 108.00 78.00 53.00 38.00 43.00 36.00 29.00 32.00 8.00 2.00 17.00 22.00 22.00

4
1

2
2
1 1 1

2
1

4
4

5 5 5 4
1
1 1

ISOTROPIC SOIL PARAMETERS

5 TYPE<S> DF SOIL

SOIL TYPE
NO.

TOTAL SATURATED UNIT UT. UNIT WT.

<PCF> <PCF)

COHESION INTERCEPT <PSF>

FRICTION ANGLE
<DEG)

PORE PRESSURE PARAMETER

PRESSURE CONSTANT
<PSF>

PIEZQMETRIC SURFACE
NO.
1 1

125.0 125.0 125.0 1.0 1.0

125.0 125.0 125.0 182.0 122.0

320 740

35.0 33.0 35.0 9.0 16.0

112

Table k.2 (cont.

PIEZOMETRIC SURFACE<S> HAVE BEEN SPECIFIED

UNITUEIGHT OF WATER =

62.40

PIFZOMETRIC SURFACE NO.

POINT
NO.
1

SPECIFIED BY

5 COORDINATE POINTS

X-UATER
<FT>

Y-UATER
<FT>

45.00
56.00 126.00 216.00 320.00
45. 00 51. 00

2 3
4 5

47.00 47.00

PIEZOMETRIC SURFACE NO.

POINT
NO.
1

2 SPECIFIED BY

7 COORDINATE POINTS

X-WATER
<FT)

Y-UATER
<FT>

45.00
56.00 126.00 136.00 150.00 250.00 320.00
45. 00

2
3 4 5 6 7

51.00 60.00 64.00 112.00 110.00

USING A RANDOM TECHNIQUE FOR GENERATING SLIDING BLOCK SURFACES? HAS BEEN SPECIFIED.
A CRITICAL FAILURE SURFACE SEARCHING METHOD*

THE ACTIVE AND PASSIVE PORTIONS DF THE SLIDING SURFACES ARE GENERATED ACCORDING TO THE RANKINE THEORY.

50 TRIAL SURFACES HAVE BEEN GENERATED.

2 BOXES SPECIFIED FDR GENERATION DF CENTRAL BLOCK BASE

LENGTH OF LINE SEGMENTS FOR ACTIVE AND PASSIVE PORTIONS OF SLIDING BLOCK IS 50.
BOX
NO.

X-LEFT
<FT>

Y-LEFT
<FT>

X-RIGHT
<FT>

Y-RIGHT
<FT>

WIDTH
<FT)

50.00 140.00

29.00 26.00

90.00 280.00

26.00 26.00

8.00 6.00

113

Table k.2 (cont.

FOLLOWING ARE DISPLAYED THE TEN MOST CRITICAL OF THE TRIAL FAILURE SURFACES EXAMINED. THEY ARE ORDERED - MOST CRITICAL
FIRST.

FAILURE SURFACE SPECIFIED BY

POINT
NO.
1

9 COORDINATE POINTS

X-SURF
<FT)

Y-SURF
<FT)

3
4 5 6 7

8 9

52.43 73.46 86.96 11.84 215.84 219.47 229.39 240.15 256.03

1.658 ***

45.16 32.77 2.61 26.69 31.99 38.17

57.21 77.03 107.53

FAILURE SURFACE SPECIFIED BY

>OINT NO.
1

9 COORDINATE POINTS

X-SURF
<FT>

Y-SURF
<FT>

3 4 5 6 7

8
9

39.19 59.85 69.02 212.99 218.17 221.92 231.78 242.55 258.62 1.673 ***

45.76 33.60 26.68 25.05 31.92 38.28 57.23 77.06 107.94

FAILURE SURFACE SPECIFIED BY

PDINT
NO.
1

9 COORDINATE POINTS

X-SURF
<FT>

Y-SURF
<FT)

2 3 4 5

6 7
8

9.76 45.78 55.09 218.26 224.74 228.83 238.55 249.33 265.39

1.691 ***

43.83 34.44 27.43 23.08 31.66 38.62 57.29 77.14 103.00

Ill*

Table

cont

FAILURE SURFACE SPECIFIED BY

PDINT
NO.
1

9 COORDINATE POINTS

X-SURF
(FT)

Y-SURF
(FT)

2 3 4 5 6
7

32.78 51.99 59.31

220. 04

45.39
34. 07 3. 56

8 9

226.04 230.20 239.89 250.67 266.73

1.696 ***

3.65 31.61 33.63 57.30 77.15 103.00

FAILURE SURFACE SPECIFIED BY 10 COORDINATE POINTS

POINT
NO.
1

X-SURF
(FT)

Y-SURF
(FT)

2 3
4

5 6 7 8 9 10

58.03 59.59 80.83 86.95 225.52 30.13 34.51 44.11 54.90 70.93
1.697
*'**

45.66 44.34 32. 33 7.72 5.34 31.46 33.39 57.34

77. 1

103.00

FAILURE SURFACE SFECIFIED BY 10 COORDINATE POINTS

POINT
NO.
1

X-SUPF
<FT)

Y-SURF
(FT)

3
4 5 6
7

8 9 10

53.60 60.60 81.88 89.76 30.83 34.18 38.77 48.29

59. 08

45.84 44.80 32.27 6.3 6.86 31.30

39. 09

57.38 77.26
103. CO

75.09
1.701 ***

115

Table k.2 (cont.)

FAILURE SURFACE SPECIFIED BY

POINT
NO.
1

9 COORDINATE POINTS

X-SURF
<FT)

Y-SURF
<FT>

39.15
59.81 67.11 8.59 30.78 35.19 44.73 55.57 71.59 1.710 ksk

3 4

5
6 7 8 9
*

45.77 33.60 8.09 8.53 31.43 38. 9 57. 35

77.31
103. 00

FAILURE SURFACE SPECIFIED BY

POINT

9 COORDINATE POINTS

m.
J

X-SURF
<FT>
8. SO

Y-SURF
<FT>

3 4 5 6 7 8 9

43.80
5.7<*

.8 6.96 31.17 40.84 51. 6 67.67

1.714 ***

43.40 34.56 7.83 5.37 31.58

38. 73 57.31 77.17 108.00

116

Table k.2 (cont.

FAILURE SURFACE SPECIFIED BY 10 COORDINATE POINTS

POINT
NO.
1

X-SURF
<FT)

Y-SURF
<FT>

2 3 4 5 6 7 8 9 10

56.32 56.56
77.71 84.40 238.81 241.41 246.38
255. ?4

266.55 282.51
1.725 ***

45.10 44.98 32.52 27.48 27.56 31.02 39.46 57.44 77.35 108.00

*xx

FAILURE SURFACE SPECIFIED BY

PDiriT NO.

9 COORDINATE PDINTS

X-SURF
<ft;>

Y-SURF
<FT>

35.03 55.57 59.55 212.46 218.07 221.81 31.68 242.44 258.51

1.728 ***

45.95
33. 85

30.85 24.48 31.92 38.28 57.23 77.05 107.92

117

Table h .2

cont

fl*

M*

kk

+ 00 "OSS

6
S

0C083

e MI
5 6 2
i

6 9 66

+ 00"0v3

S9 e eas i 3 e M * *13
t>

+ 00 "003

+ 00'09l

rt

+ 00 "031

ST e
I" 3'

,#

+ oo "03

S 3fr * WIS" e 3
*t>

+ 00"01>

mm
+
+ +
+

~ ~

*M+*
00

a-*

+ 00

"

00 "003

00 '091

00"03l
s

00*08

'Ofr

00"

118

a s

'

'

119

0.35
-

-r1

11

IMIT

Tl~i

T
-f-

rn

-<< ?gje_
__

nH

::

_/7-:
Typical Cross Sec-.
tion Failure Arc In Zone A. Critical Circle Passes through Toe and

-+"1

>s^

^-*ag_ 3
_ '

Typical Cross Section Showing Various Cases

0.30

Considered
Cose
I:

in

Zone

8.

['

Z " -

The Most Dangerous of The Circles Passin through The Toe; Represented by Full Lines in Cha rt. Where Full Lines do not Appear, This Case Is not
Appreciably Different from Case 2.

.. --j

Stability

"
-

Number
In

Represented
Lines

Chart by Full

0.25-

Case 2: Surface of Ledge or a Strong Stratum c -r The Elevation of The Toe (n = I), Represented by Q3 * Short Dashed Lines in Chart i> \ V
1

\_*
~?
yt-

cf

Represented by Long Cashed Lines

in

Chart.

I
0.20

JIH
'il

Where Long Dashed Lines do not Appear, The Critical Circle Passes Through The Toe.
1

-L* o ~z

y" ^2 / -, /___Z

II II
,

II

jT

<

n co

""

rT
=i"
1

r~^

ri

T.> X^Y
<r
1

Ipf

t?' X^"
^Z. 7

A- ^ y\ sOX

/ H7 /
7
>^

?
/

u
i

v>

.< D
4i
J5
.

*/

7
^

"Z.

0.15
i

o
if)
/

~4 ' I
i

7
/

*
!

^/y

^ 5^
>

^
/
^"
-

^p"

j?
t

\ /
L*

_^
,

V
1
,

tz T> xr T'y
/I

/ t
*l * /

^
1

xt ' ^T

XI

~^^ ' Yl ' /

i

0.10
*R
t
'

p,/ff
*-

y ^r u
i i

-r

yk:

/
.

/i

>/
i,
, '

/
!/V
t/l

Hi/'
jx
,4y Y

7+i

i
! i

tC-

'^

V
1
'

f
"
i

/ f
,'/

y
/

A
1

'

H
1

'y^
'1

if

0.05
1

<

V
<\

~7\

j/f
'

h
la -?

'

/
,

XI XI
1
'

|
,

*
'

'
1

1
1

1
"~*
f

o
/
-

y
;

//

l/i*

!/T

0.0

yi
1 1

J_-

_J_

10

20

30
Slope

40
Angle

50
i

60
(Degrees)

70

80

90

FIGURE 4.6

STABILITY CHART FOR SLOPES IN HOMOGENEOUS SOIL (AFTER TAYLOR, 1948)

120

mM FTT

orJ DTPTtf

Lm\TAT\oN
,

bh

-ioM - ^TA&'iLH.Y Number-

-p-

W|H

LOTS

33

FIGURE 4.7

STABILITY CHART FOR SLOPES

IN

HOMOGENEOUS
(AFTER

SOIL, * TAYLOR, 1937)

121

0.30
cr
4> r

required strength intercept

required strength angle

0.25

10

20

30

40
/3
,

50

60

70

80

90

Slope Angle

degrees

FIGURE

4.8

STABILITY CHART OF THE TAYLOR TYPE FOR HOMOGENEOUS SLOPES, BASED ON

RESULTS FROM STABL PROGRAM

122

0.30
Cr
required strength intercept
required

%
0.25

strength angle

0.20
cr

Xh
0.15

0.10

0.05

0.0

10

20

30

40

50
(B

60

70

80

90

Slope Angle

(degrees)

FIGURE 4.9

COMPARISON BETWEEN STABL AND TAYLOR STABILITY CHART FOR HOMOGENEOUS SLOPES

123
slopes with slope angles 3 < 50, the GTABL solution is conservative

compared to Taylor's solution, particularly for low values of

whereas Taylor is more conservative for steeper slopes.

'

Taylor's chart

is based on a friction circle method with assumed sinusoidal stress

distribution along the shear surface.

This stress distribution should

give a more accurate result than the conventional friction circle method

described in Chapter II, Section 2.2.H.

These charts may also be used to estimate the variability in F due

to variation in problem parameters (see also Chapter III).

Where water pressures are present, the stability coefficients pre-

sented by Bishop and Morgenstern (i960) may be used.

The preliminary estimation of the factor of safety for a problem
by simple means, provides a good base from which to start the
analysis.
''TART,

It gives a check on the computer output, and helps avoid

some troubles that frequently occur when problems having very low safety

factors (< 1) are analysed by GTABL.

it.

5. 2

Use of STABL
The next step is to set up the data for the GTABL analysis. The

basic steps are described in the GTABL User Manual [Giegel (1975b)],
but some additional information may be useful.

H.5.2.1

How to Define Limitations for Gearch Routines

The simple stability charts can provide us with preliminary estimates.

Again we look at Figure h.6.

The full lines on the chart represent cases

where the most critical failure surfaces pass through the toe of the slope.
If our problem is located here
,

we need only investigate such types of

failures, hence the left as well as the right limitation point for

12U

initiation of the failure surfaces should he specified as the xcoordinate to the toe of the slope.
The dashed line on the chart divides the problem into zone
zone R.
A

and

Zone A represents those problems where the critical failure

>

surface does not extend below the toe of the slope: a -

for the total

failure surface, where a is the inclination of the failure surface with

the horizontal.
If our problem lies in zone A, we can initially define our direction

limitations accordingly.

Since the default values used in the program

are - ^5 for the clockwise direction limit, and B - 5 for the counter-

clockwise direction limit, the use of these values for problems located
in zone A of Taylor's stability chart does not provide the most efficient

search limitations.

Instead we should specify, for example, the clock- 5.

wise direction limit as 0, and the counterclockwise as B

k
.

2 2
.

C lockwise

and Counterclockwise Direction Limits

At this point it may be appropriate to discuss some of the confusion

that has arisen with respect to the use of the term "clockwise" and

"counterclockwise direction limits".

The terms are used relative to each other, not
to-

the X-axis, whereas

the angles are defined with respect to the X-axis.

An angle is defined positive counterclockwise from the positive

X-axis, and negative clockwise from the positive X-axis, see Figure
k.

10(a)

A counterclockwise direction limit does not have to be a counterclock-

wise angle and vise versa.

See Figure

it.

10 (b)

Both angles are clock-

wise with respect to the positive X-axis, so they are both negative.

1P5

positive angle

* X-axis
negative angle

(a)

Definition of positive

ana negative angles

Clockwise
direction limitation

-60
-

->

X-axis
limitation

Counter clockwise airection

-40

<

soft clay

(b)

FIGURE 4.10

ILLUSTRATION OF CLOCKWISE AND COUNTER CLOCKWISE DIRECTION LIMITATIONS FOR INITIATION OF FAILURE SURFACES

126

1+.5.3

Strength along Goil Boundaries

If we do not know whether one soil is stronger than another, it

may be tempting to specify a surface or a sliding block search on the

boundary between two soil layers, assuming that the program will analyse the stability for both types of soils and give the most critical

factor of safety.
It must be clearly understood that this is not the case.

If a

failure surface is specified or generated exactly on the boundary be-

tween two types of soil, the strength parameters of the soil above the
failure surface will be used in the stability analyses, independent of

which soil is the weaker.

One should always specify a failure surface

or a sliding block search totally above or totally below a soil boundary.

If one does not know which case will be the most critical, both situa-

tions should be analysed.

k.5. U

Analysis Involving Earthquake Loadings

A slope may be analysed for the influence of earthquake forces by

pseudo-static earthquake coefficients.

with respect to this option.

Some uncertainty has existed

For example, why is the vertical earthWhat value for the cavi-

quake coefficient defined positive upwards?

tation pressure should be used, and how does it affect the stability?
k 5 k .1

Effect of Vertical Earthquake Loading on the Stability of a

Slope

Let us first consider the case of dry soil.

Specifying a vertical

earthquake coefficient k

has, for dry soil, exactly the same effect as

changing the unit weight of the soil mass from Y to

yd

- k

).

'

127

For granular soil, where the strength

stress (a* tan
<f>',

in

solely dependent on the

c'

= 0), the factor of safety is independent of y,

so a vertical earthquake loading alone will neither decrease, nor in-

crease the safety of the slope.

Where

c'

? 0, the factor of safety F is proportional to

see

Figure h.Q; so a change in Y causes a change in F.

F decreases with

increasing Yi so the critical direction of the vertical earthquake

loading is downward, hence k

should be specified negative

If part of the sliding surface extends below a water table, the

result changes for the cases where

<ji

4 0.

The earthquake loading is a

dynamic force, momentarily applied.

If therefore the sliding mass is

submerged, the component of earthquake forces normal to the base of a

slice will be carried by an increase or decrease in the pore water

pressure.
and
4>'
,

This means that a soil with strength defined in terms of

<)>

c'

will act as a soil with

with respect to the earthquake

This

forces, for the part of the soil that is under the water table.

again means that a granular soil (c' = 0) that is submerged will react

exactly as a cohesive soil

so k

(<t>

= 0) with respect to the earthquake forces,

should be defined negative also in this case,

At this point it should be mentioned that the way the earthquake

forces are transmitted to the pore water in the case of a partially or

totally submerged slice is approximate.

principle.

Figure

l+.ll

illustrates the

Since the Janbu procedure as well as the modified Bishop

procedure satisfies vertical force equilibrium for individual slices,

equation (b) in Figure U.ll is the most correct.

However, it has AS^,

the change in required shear strength, as an unknown, and cannot be

120

(a)

Force equilibrium normal

to

base of

slice:

AU

K H sina

K v cosa

(b)

Vertical

force equilibrium

AUcosa

AS r sina

Kv

FIGURE

4.

TRANSMISSION OF EARTHQUAKE FORCES TO THE PORE WATER

129

solved directly with respect to the change in the pore pressure force
All.

Hence equation (a), the projection of forces normal to the slice

base, is used to give an approximate value for AU.

1+.

5.^.2

Cavitation Pressure

The pressure in the pore water between soil particles that are

submerged can be negative; but it cannot go below the cavitation

pressure, where water starts to evaporate.

This pressure is found to

be about -

atmosphere for granular soil, which is equivalent to

2115 psf [Ceed and Lee (1967)].

Cavitation is unlikely to occur.

Only a large upward directed

vertical earthquake force would be able to reduce the pore pressure

to a cavitation pressure below - 1 atmosphere, and the critical direc-

tion of the vertical earthquake force was found to be downward.

Horizontal earthquake forces, on the other hand, would result in

a decrease in pore water pressure for slices with positive a's, when

defined positive (outward directed), which is the critical direction,

see Figure '+.12.

However, this decrease in the pore water pressure is

>

small (except for slices with a

U5

compared to the component of the

force parallel to the base of the slice, which acts as a driving force.
It is most likely that this driving force will decrease the stability

of the slope until the point of failure, long before the component normal to the base of the slice will create a pore water pressure that
reaches the cavitation pressure.
If, however, the pore pressure does reach the cavitation pressure,

the soil particles will have to "carry" the remaining unloading. This

130

a(+)
Ct(-)

AU -K M sina>0

Outward directed earthquake force

+)

Wk h

W-

weight of slice

FIGURE

4.12

CHANGE IN PORE WATER PRESSURE DUE TO A HORIZONTAL EARTHQUAKE FORCE

131

will result in a decrease in the effective normal stresses, and a decrease in the normal stress dependent strength
h 5 k 3
. .

(<b

strength).

Illustrative Example

A slope kO feet high, inclined 26.6 with the horizontal (slope 2/1),

was analysed for the effect of vertical as well as horizontal earthquake

loadings.

Three different situations were considered.

In case A the strength of the soil is defined in terms of strength

intercept

only

((f)

= 0).

In case B the strength of the soil is defined

<J)'

in terms of strength angle

alone (c' = 0); the soil is assumed dry.

Case C is similar to case B, hut a water tahle is introduced.

The slopes are first analysed for normal conditions with no earth-

quake forces, and then for five different combinations of earthquake

loadings, see Table U.3.

Figure H.13 shows the three slopes with the

The critical failure surface for each case

critical failure surfaces.

remained the same or very close to the same for all analyses.
The results agree with the previous considerations.
C the

For cases A and

factor of safety increases with a positive (upward directed) verti-

cal earthquake loading and decreases for a negative (downward directed)

vertical earthquake loading, whereas no change results for case B

granular soil).
A horizontal outward directed (positive k
)

(dry-

earthquake loading re,

duces the stability drastically for all three cases

(the
(J)

but most for case A

case)

A combination of horizontal and vertical earthquake forces does not

give the same result as a direct superposition of the two forces individ-

ually acting, but the tendency is the same.

An upward directed (positive)

132

Table

It.

Stability Analysis of Slopes Subjected to Earthquake Forces*

Factors of safety

Case A

Case B
c*

Case C

Earthquake loadings

0=0

c =

lOOOpsf

= 0, *' = 35

dry soil

water table

(1)

k k

= k

= h

1.2U
=

1.1+2

1.21

(2)
(3) (h)
(5)
(6)

= 0.1,

^
h

1.38

1.U2
1.1*2

1.33

k
k
k

= - 0.1, k = y h

1.13
0.86 0.92

1.11

= o, k

= o.i

1.12
1.09
l.llt

0.98
1.05 0.92
-

v
y

=k h=0 .l
= -o.l, k

= 0.1

0.80
-

A(3) in % of (1)
A(l)

9 - 21
- 23
-

in % of (1)

- 31 - 26 - 35

- 19 - 13 ~ 2k

A(5) in % of (1)
A(6) in % of (1)
1

20

The slopes are Uo feet high, inclined 2/1 (B = 26.6 Y = 125pcf.

133

Case A

Case B

Case C

c'0
</>'

35

FIGURE

4.13

CRITICAL FAILURE SURFACES FOR

SLOPES ANALYSED FOR INFLUENCE OF EARTHQUAKE LOADINGS

13U

vertical force increases the safety compared to the case where only a

horizontal force is acting, and a downward directed (negative) vertical

force decreases the safety.

There is one exception, namely case B, the

In this case a combination of upward and

slope in granular dry soil.

outward directed earthquake loadings was a little more critical (lower

factor of safety) than the outward directed (horizontal) alone, and the

outward and downward directed earthquake forces were slightly less

critical than the outward directed alone.
This does not appear logical,

and may be due to some of the assumptions made in the analytical model,
e.g.
,

interslice shear forces are ignored.

Dynamic Earthquake Analysis

k.5.h.k

The method of earthquake analysis provided in the STABL program in

terms of a pseudo-static loading, is the conventional approach.

However, in recent years dynamic earthquake analyses have been introduced, which more accurately similates the forces that are involved

[Newmark (I965); Seed (1966)].

The stability in a dynamic earthquake

analysis is assessed in terms of embankment (or slope) deformations,

rather than a single value of the factor of safety.

For more important

structures, where the consequence of failure is great, this type of

analysis should be used [Seed et al (1975)].

Dynamic analysis cannot

be applied in the STABL program.

^5-5

Misleading Factors of Safety

Most users of the STABL program will sooner or later experience an

output with the comment that the factor of safety for a failure surface
is misleading.

135

There are two reasons why a factor of safety calculated by STABL

can be misleading.
A problem occurs when, for some reason, the iteration process for

the factor of safety does not converge.

This is usually associated

with a factor of safety which is either very high or very low (below
unity).

However, in most cases STABL will handle factors of safety as

high as 10 and as low as 0.6.

In one case, where a user had troubles with no convergence for a

low factor of safety, the program was modified to define the initial

value of the factor of safety for the iteration process to 1.2 instead
of 1.5 (arbitrarily defined in the program), and this change made it

work [Maunsell

8c

Partners Pty. Ltd., Australia (1977)].

A misleading factor of safety may also occur if the denominator of

the expression for the factor of safety (F + tan a tan
negative.
4>'
)

is zero or

This may happen for deep failure surfaces, where the angle a

(see Figure U.12) has a relatively large negative value for some of the

slices.

The normal force on the base of a slice will thus become very

large or negative.

This implies an unreasonable state of stress, hence

the result is not reliable. The STABL program therefore always checks for the sign of
(F +

tan a tan

<j>'

for each slice, and if the expression turns negative,

the message of a misleading factor of safety is printed, together with

the coordinates of the failure surface in question, and the factor of

safety.
low.

This factor of safety will often be either very high or very

Frequently an asterisk will be printed, indicating that the value

is either larger or smaller than the computer is programmed to handle.

136

The negative value of (F + tan a tan

<f>')

will often occur for probHence an output

lems having an actual factor of safety below unity.

with misleading factors of safety should be a warning that the slope

may be unsafe.
[cos a + (tan
<j>'

Whitman and Bailey (1967) suggests the requirement

sin a/F)
<

0.2] be used as a basis for disregarding a

result.
In no case should a factor of safety that is labeled as "misleading
1

be used as a result.

It could be close to an actual factor of safety,

but it can also be very far from the true value.

Even if some of the factors of safety are misleading, STABL may

give a correct result for the rest of the surfaces examined.

These

will be listed under the ten most critical surfaces.

surfaces give a factor of safety of

If some of these

'500", this indicates that they are

surfaces giving a "misleading" factor of safety.

See also Appendix

D,

Section D.3.3.2.
If all results by STABL are misleading, another method of analysis has to be used.

137

COMPARATIVE DESIGN STUDY

A comparison was made among results obtained by STABL and those obtained by other methods of analysis for an example design problem posed
by Ladd
(1975).

All results in terms of factors of safety are sum5-**

marized in Tables

and 5.5 (pp. 167-168).

5.1

DESCRIPTION OF PROBLEM
The problem is stated in Figure 5.1.
A 30 feet high embankment

for a four lane highway is to be constructed over 80 feet of medium to soft varved clay.

The embankment will extend between bridge abutments,

and a two-dimensional analysis is appropriate.

The design process includes a consolidation-settlement-preload analysis, influenced by drains to satisfy requirements of maximum

allowable settlements and available construction time.

the slope stability analyses are considered here.

However only

5.2

TYPES OF ANALYSIS

Analyses are performed for the following situations

(1)

Undrained shear for end of construction stability.

This case assumes that no consolidation takes place during

construction.
(2)

Drained shear for long term stability.

This case assumes that all excess pore pressure has disappeared,

consolidation is completed.

138

rr
ui

O t 2

rj

0)

<P

>

<
rr LU

D
cm ^r
Al

c 3 O

O 0J
~

>

CD
v|

c
0)

e c

o z
UJ

or

t
in

-o
.fa

ii

I u
IT)

Sl*i
E

_p *

E
UJ

o ^
o|G

CVJ

C C o c c .2 2 2 '^ U

W|

MP
:

M
uj .3

< Q

I
Q
<=*J

3<S'
ro

o
o
UJ

<

(M

o 2
o
o
Q_

u c

a>

ui h_j u. UI

v>

a>

o GO

</>

Z < _j a) 2 O uj

rr iu

O
in

e8
<t ro

m
UJ

8 CD

o to o

o n
IJ

o g

o CO

o
CD

'UOIjOAdG

139
(3)

Partially drained conditions for end of construction stability.

This case assumes that some consolidation takes

place during construction, increasing the strength of the

soil.

Undrained shear is often examined by total stress analysis while

drained shear is examined by effective stress analysis.
However, both

total and effective stresses may in general be applied, regardless of drainage

,

although the results expressed in terms of factors of safety

will be different for the two types of analysis unless the factor of
safety equals one.
This point is illustrated in Figure 5.2:

Total undrained shear strength is defined by

= c

a. tan
f

d>

(5.1)

For saturated clay

d>

=0,sos u =c;s u u
1
- a
3 f
)

is normally defined as

= q

= l/2(a

(5.2)

where l/2(a

- a

) is the
s

maximum shear stress at failure.

is defined as

In this

design problem, however,

u
<J>'

= q _ cos *' f

(5.3)

where q. cos
failure.

is the maximum shear stress on the failure plane at

Effective shear strength is defined by

s

_, = c* + a' el i f

tan

<J>*

(5.M

where a' is the effective normal stress on the potential failure plane,
and c' and
<f>'

are the strength intercept and strength angle, respect-

ively, in terms of effective stress.

For normally consolidated clay

iUo

</>

> UJ Z OO z u. U. <
Lil

Id

UJ

I-

Z*
-

z
III

u.

<

en

mo a. iO

Iu. UJ

-I oc UJ a: g:

j.

ssaJis Joans
cm id
in

tu a:

u.

c'

= 0, hence s

ell

__ = a'
I

tan

<f>'

In the case where F = 1, a' = a'

s

and
= q cos f
1

gff

+ a^

tan

<t>'

= T

ff

(J)

= s

(5-5)

where a'

is the effective normal stress on the failure plane at

failure, and T-_ is the shear stress on the failure plane at failure.
In design situations, however, we require F > 1.

This does not

change the value of

a'

but the effective strength will change since

is not the same as a'

(corresponding to undrained failure) unless

=
s

the pore pressure parameter A

sin
<j>'),

l/2(l
.

sin

(J)').

For A

> 1/2

(l -

will be less than

Circle 1 (Figure 5-2) represents

the effective stress conditions for a point on the potential failure

plane.

Assuming that a sample of the soil can be brought to the same

state of effective stress and then sheared undrained, the sample will
fail when the effective stress conditions given by circle 2 is reached.
This gives a value of
s

= q cos f
s

(J)'.

However, an effective stress

(see Figure 5.2) which is usually

analysis would use the strength

larger than
T
,

The actual shear stress on the potential failure plane,

is independent of the type of analysis, hence total stress analysis

will in most cases yield a lower factor of safety than effective stress
analysis. The above considerations apply to the ideal case, where un<j>'.

drained and drained shear tests result in the same value of

quently this is not the case.

Fre-

In this analysis, the drained and undrained

<'

direct simple shear tests gave values of

equal to 21 and 9 .^

respectively.

This explains the very large difference in results by

effective and total stress analysis (see Section 5.5.2.2).

lU2

5.3

METHODS OF ANALYSIS
In the report by Ladd (1975) two methods of analysis were applied:

the simplified Bishop method with corrected field vane strengths (see

Section 5-^.1)

and the Morgenstern-Price analysis using SHANSEP* an-

isotropic strengths (see Section 5.^.2).

5-3.1

The Simplified Bishop Method

The simplified Bishop method assumes a circular failure surface

and applies the method of slices with the simplifying assumption that
side forces between slices are zero. The method satisfies overall

moment equilibrium, but not moment equilibrium for individual slices.

It satisfies vertical force equilibrium, but not horizontal.

In Ladd's report the analyses are performed by a computer program

ICES LEASE**.

A grid system combined with a search routine is used to

locate the center of the most critical circle.

5.3.2

Morgenstern-Price Analysis
The Morgenstern-Price (MP) method of slices analyses failure sur-

faces of any shape composed of straight line segments

Various assump-

tions can be made with respect to interslice forces, and the analysis

satisfies complete equilibrium.

Different failure surfaces have to be

The line of

specified until the more critical surfaces are found.

thrust and the magnitude of interslice forces are examined to judge the

reasonableness of the result (see also Appendix C.l). If either one is

deemed unreasonable, a new set of side forces must be assumed until a

satisfactory solution is found.

* Stress

History And Normalized Soil Engineering Properties.

** Integrated Civil Engineering System - Lease Electronic Accounting

System.

1U3

Such a procedure is rather time consuming, hence is only recom-

mended for problems where the simplified Bishop circular arc does not
provide reasonable results.
This may be the case when dealing with

non-homogeneous or anisotropic soil conditions.

5.3.3

Method Used in the STABL Program

The STABL program uses a modification of the simplified Bishop

method of slices in which overall horizontal force equilibrium is satisfied instead of overall moment equilibrium.*
This modification

is introduced in order to be able to analyse irregular as well as cir-

cular failure surfaces.

In case of a circular failure surface,

STABL will in most cases

give a factor of safety somewhat lower than the simplified Bishop

method, provided the same (or approximately the same) most critical failure surface is found.**
The advantage of the STABL program is most significant when dealing with cases that require failure surfaces of irregular shape.
In

this example design problem, the RANDOM and BLOCK search routines were

used in connection with the SHANSEP anisotropic soil strength parameters where the Morgenstern-Price analyses were used in Ladd's report.

BLOCK generates sliding block type failure surfaces.

surfaces of more irregular shape.

RANDOM generates

The advantage of the STABL program is

that it searches for the more critical failure surfaces, whereas the

engineer has to conduct this search himself when using the MP-analysis.

**

This method is also known as Janbu's simplified procedure (Wright, 1969) Exceptions, see Chapter VI, Problem 3 and h, and Appendix C.II.

ikk

From the results obtained in this comparative study, it is seen that the
STABL program gives results very close to those obtained by the MP-

procedure

5.U

SOIL STRENGTH PARAMETERS USED IN STABILITY ANALYSES

5.^.1 Corrected Field Vane Shear Strength

Bjerrum (1972) introduced the corrected field vane shear strength

to evaluate the undrained stability of embankments constructed on soft
clay foundations by circular arc analysis
.

The correction factor u is

based upon case studies of embankment failures and y was plotted versus
Plasticity Index (Pi).
However, many other case studies show a wide de-

viation from Bjerrum' s recommended curve.

the clay is layered and anisotropic:

Special problems arise when

Which PI should be used?

How far

is the circular arc from actual failure surfaces that

would develop in

these types of soils?

Instead of using Bjerrum'

curve, the correction factor for the

varved clay was evaluated based on case studies of embankment failures

on varved clays A further check of the stability analysis performed by the simpli-

fied Bishop method and corrected field vane shear strengths was made by

analysing the undrained situation with the Morgenstern-Price method and

SHANSEP anisotropic shear strengths.

5.U.2

SHANSEP Anisotropic Shear Strength

The SHANSEP

method is described

by Ladd and Foott (197U).

The

method provides an evaluation of the undrained (and partially drained)

H*5

stress-strain-strength properties of clays and consists of three basic

steps
(1)

Determine the stress history of the clay foundation, i.e., variation with depth of the effective overburden stress a' vo
,

the maxi1

mum past pressure

(2)

a'

vm

and the overconsolidation ratio OCR = a

vm

/a

vo

Perform K

consolidated-undrained tests on undisturbed samples to

s

find the relationship between the undrained strength ratio

/a*

vc

and OCR for representative modes of failures (e.g., as shown on

Figures 5.3 and
(3)
5.*+).
s

Compute the variation in

with depth for the different modes of

failure from the stress history and normalized strength data ob-

tained in steps (l) and (2).

The method is applied to initial in situ undrained shear strengths and is also used to predict increase in strength due to consolidation.
The anisotropy of the varved clay is taken into account by the modes of

failure investigated in the testing program (Figure 5.3) and applied in

the stability analysis.

5.5

RESULTS AND COMPARISONS

5.5.1

Preliminary Stability Analyses

Preliminary analyses were performed on the initial design crosssection (Figure 5.1) for undrained end-of-construction stability and

drained long term stability using the simplified Bishop circular arc
analysis with corrected field vane strengths.

For the undrained analyses,

various embankment heights were investigated with 2:1 and k:l slopes,
in order to determine the maximum height that could be constructed

IhC

03

V
L.

PSA PSP DSS o

Plane Strain Active Plane Strain Passive Direct Simple Shear

to

o
0;

i_

JZ Ul

0.1

o
a*

Varved

clay

"5

E -0.2

(gy-^h)

PSP

0.3
)

4
Shear Strain /,

10

Data from Lacasse

Test cr^c
=

et. al.
In

(1972)
<7
'

4.1

TSF,

situ

m =3.5

TSF

FIGURE

5.3

STRESS VS. STRAIN FROM CKoU TESTS WITH DIFFERENT MODES OF FAILURE (After Ladd, 1975)

1U7

rapidly without consolidation (i.e., assuming no increase in strength

due to consolidation).

For comparison, the CIRCLE search routine in STABL was used.

How-

ever, only the design cross-section corresponding to a 30 feet embanhment

and 2:1 slopes was analysed

(Figures 5.1 and 5-'0.

Data and results

are summarized in Figures 5'+ and 5.5.

Results show STABL to give a

slightly lower factor of safety than the simplified Bishop method in the

undrained analysis and about 5% lower values in the drained effective

stress analyses. The Morgenstern-Price method with SHANSEP anisotropic strength para-

meters was also applied to the undrained, end-of-construction stage, and

showed very good agreement with the result obtained by the simplified
Bishop method with corrected field vane strengths.
The anisotropic

strengths are given in Table 5-1 and correspond to the relationship

shown in Figure 5.6.
The Morgenstern-Price method was modeled by the RANDOM and BLOCK

procedures in the STABL program, where the RANDOM routine randomly generates irregular failure surfaces and analyses their stability, and the

BLOCK routines randomly, but more restrictedly

generate sliding block (or

(For a detailed

wedge) modes of failure and analyses their stability.

description of the surface generation routines, see Siegel, 1975a).

The results of Morgenstern-Price (MP) and STABL, RANDOM and BLOCK
are given in Figure 5.7.

The factors of safety for the STABL procedures

are equal to or slightly lower than the one obtained by the Morgenstern-

Price procedure. width.

All failure surfaces lie within a band of 20 feet

ll*fl

m
(VI

jo o>

If)

if)

ifimmmifimmm
in UJ rr

i^'uo!it>Aai3

O
Ll

'

1^9

r^m^
_l

Z
UJ
CO ro in CD

CD

< K
CO
Ll.

HCM 2C

^^
a.

O
" v

ji

2 < GO
-

o i CO m
u.

~o

UJ

CD

CJ r-

u "
< o f_i CO

c
*

<D

si

<
/
/
v
.

~ w

ro

ftl

CM

"> oj

o
UJ CO

(/) (/)

UJ CO

< o

CJ

w o

$ S

y \ /// \ ^i/ y
/

Siu> to
/

E
UJ

<

<

>J

< z <
</)

c\
'

<~

a c
cr
a.

fV// \ \y7 v y y ^^
\

o
E o

a>

en UJ rr h-

is

'/

O
ii

Si
Guj
UJ Ll U. UJ

u
\

UJ CO

ro CM

S
y

c*J

"

Q
UJ

< o

uj "9CO

<

QQ.
1 1

in CM

If)

m
If)

CD

if)

If)

'uoiiDAaig

UJ f

150

IA

O O

rH

rH

On

OO

ON J

ON J-

0\ J-

U) IA

VO

oj w

o o o
p - >
J

o o o o

It)

VO

On

0\
OJ

VO
OJ

On t

LTN

VD
On

VO

VO

CO

ON

3
3
I

O O O C
EOJ

>

w
A,'

OO

CO

o o On o o O O o o O o o o H rH
OJ
LTN

VD

t-

-3-

on t

On

VO

H o
0O

a:

rH

W\

ON

"

O
Ox

oo

vo

LTN itn

D-

vo

-*

oo

00

LTN

LTN

OO
OJ

a,

r
OJ

vo
OJ

NO
OJ

OJ

PC

OJ
LTN

OO
t oo

-3LTN

-3-

CO
rH

OJ
Ll^

Ol
-3-

VC
OO

rH

ON
OJ

00
rH

VO

OJ

rH

rH

H
O
-3" l/N

e >

*-<

in

a;

o O O o o O o On VO VO o
l/N

LTN

J-J-

O
On
-3-

VO

-s-

on

00

00

-3-

b
1 j

> m
as

<w

o
OJ

o
ON
rH
rH

LPv

ON

l/n ir\

LT\

LTN

LfN

On

h00
0J

OO
OJ

UA r
OJ

ITN

00 C 00

UA On
rH
-3"

o
c
o

H H
CO

00

H
O
CO

IS\

VO

ON

o rH Ol H H H

00
rH

c o
H
2 ed

u\
>

LTN

+j

OJ

oo

H W

<U

rH

o O O o o o o CM H H 3 c On CO r VO
t
CM
1

151

DSS

<

OCR
FIGURE
5.6

o~y

/<rvc

ANISOTROPIC UNDRAINED STRENGTH PARAMETERS FOR USE IN TOTAL STRESS ANALYSES (After Ladd 1975 )
,

15?

8 ro

X
WIT
ESIGN

ENT

D
VIBANKM

INARY

<^-

O O
CM

ixl

5
_i
[7j

jp

U-

F c o

o
<o -

&
o

C Li x < K z e> < z UJ

"> fc uj en f 1(/) Q_

erf

c
S c
k.

o o

0>

E o L.
*TOTAL
ISOTRO

0*

u o

Q Z < rr Q z 3
in

z <
</>

z < x

C/>

ID
CVi

ID CM CM

m O

mm s
00

in

m
CJ

IT)

m
00

m
CO

UJ rr

Jj'U0UDAd|3

g
u.

153

Figure 5.8 shows the most critical failure surfaces and correspond ing factors of safety found by
of surfaces.

STABL for the four different types

5.5.1.1 Frequency Distribution Curves for the Factor of Safety

The STABL program normally prints out only the factors of safety
for the ten most critical surfaces.
In this study, however, a modifi-

cation provided output of the factors of safety for all surfaces investigated.
A histogram was made with intervals of AF equal to 0.1 (F =

1.0 includes values of F from 0.950 to I.0H9, etc.) for each of the four

procedures used in the undrained, total stress analyses, i.e., CIRCLE

(Case 1A), RANDOM (Case 2B)
,

BL0CK1 (Case 3B)

and BL0CK2 (Case Ub).

From the histograms, frequency distribution curves were drawn as shown

in Figure 5-9.
It is seen that each of the curves has a clear peak

varying from F = 0.9 to 1.3 for the different cases.

From a statistical point of view, the peak value is the value that
is most likely to occur, whereas the variance gives an indication of

the scattering in the results.

It is seen that Cases 1A and UB have

less variance than cases 2B and 3B.

This may indicate that 1A and UB

For a truer compar-

surfaces are more likely than 2B and 3B surfaces.

ison, only B cases should be compared with each other, as soil parameters
are changed from Case A to Case B.

However, the curves are too close to

each other to justify specific conclusions with respect to individual

behavior.
The frequency distribution curve can also be interpreted in terms
of likelihood of failure.

For the CIRCLE case, 9 out of 100 surfaces

have a factor of safety less than 1.0, which means that the probability

>

15U

UJ
z: *:

CO 2 CO
>_i ex

z
en

CD

CL

LU
to

>X) t

Q UJ
2
ex ce CD

P
ft
wtro

CO

z
en

co UJ

Q
>
>-

X)

JOP
uj
UJ
i

5
CI _J
c_>

-low

I ex

cr ex
Q_ s:
e_)

uj

a
>

ex o a: >

"si
312 5m a

1J/U0IJDA9I3

155

Factor of Safety

FIGURE

5.9

FREQUENCY DISTRIBUTION CURVES FOR THE FACTOR OF SAFETY, CASES IA, 2B,3B,4B

156

of failure is

out of a 100 or 9%.

When dealing with probability,

however, many more things have to be taken into consideration.

soil parameters should not be handled as exact values
,

The

but distributed

around mean values with a certain variance that can be estimated from

laboratory tests.

Gee for example the work by Athanasiou-Grivas (1976).

This aspect is not considered here.

5-

5-2

Final Design Stability Analyses

From preliminary stability analyses it was found necessary to improve the stability of the design cross-section by placing berms 15
feet high and approximately 60 feet wide on both sides of the embank-

ment.

Settlements were then calculated, and it was found that to com-

pensate for settlements, the height of the fill should be increased by

8 feet.

To accelerate the rate of settlements, sand drains should be

installed under the embankment and additional load applied by a temporary surcharge.
It was found,

however, that for a minimum F of 1.2,

This height would still lead

the maximum safe height is only k2 feet.

to excessive post construction settlements if the required time schedule

were followed.

It was

suggested that more fill could be placed if the

construction was carefully monitored with a fairly extensive program of

field measurements. For this example, however, the stability analyses

of the final design is based unon the \2 feet of fill, with stability berms as mentioned above.
final design:

Two situations are investigated for the

the partially drained condition and the long term,

totally drained condition.

157

5.5-2.1

Partially Drained Stability Analyses

1*2

The

feet high embankment with stability berms was investigated

for a partially drained condition corresponding to 95% consolidation.

At this stage almost 8 feet of settlement

has taken place, see Figure

5.10.

The stability analysis was carried out as a total stress analysis

with Morgenstern-Price wedges and SHMSEP anisotropic strength values

increased by consolidation.
See plots of strengths in Figure 5-11 and

Table 5.2.

STABL was applied with all four types of analyses and aniso-

tropic soil parameters.

Results of MP and STABL analyses are given in Figure 5.12.
STABL
'

four procedures (Cases 1E.1 through i*E.l) give results from + 6% to - 5$

of that obtained by the MP analysis.

5.5.2.2

Long Term Stability Analyses

For the long term stability the embankment is analysed for a final
design soil profile corresponding to
feet of settlement and a final
h

height of 30 feet above the ground, which means that

is removed compared to the partially drained stage.

feet of material

Conventional drained effective stress analyses were performed with

Bishop and STABL circular arc failure surfaces.
given in Figure 5-13.

Data and results are

Results obtained by STABL are approximately 13$

less than results by the Bishop method.

Total stress analyses were also applied with SHAH SEP anisotropic
strengths corresponding to stresses under the embankment at 100% consolidation, see Figure 5-1^ and Table 5-3.

Again MP and STABL 's four types

The factors

of analysis were applied.

Figure 5-15 shows the results.

of safety obtained with the four procedures of STABL (Cases IE. 2 through

158

u
<^J
c

D
(

E n
1

-O
-

q:

O O ho
uj
</>

b_

$
(m)
-i

01

T>

>

9
0)

r-

1
3
1

UJ

o "

'

c.
c
a>

5 o

it
_

E c o

\
(S

St <
O
>-

\|

UJ
<*-

>)

a>

c c

83 <V
UJ -1 HI

t/> </>

V)
i

01

"
1

a>

E
a*
1-

o c o

i
<

c o
0;
"

o
01

*'
E

in

c o

b o ? o
i i

u c
rvi

Si ^
V) \-

2
ANKME

5 5)

<

cL

u
MATED

/
i

/-

N V t

a
o o

O
in
CVJ

FS? 5
(T)

UJ UJ

o
-

10

' i i

in

m
<D

c^

f ^ n o u 52
in 4J

^
<^

ID

It!

'U0!iDA9|3

159

4j'UO!4DAd|3

D
Ll

It!

i6o

CO

o
OJ

O
CVI

On

CO CO

C\J

LTN

NO
LTN

OJ

o ON
NO

NO

NO f

CO CJ CO

ITS

CO

CM

CO

NO

CO NO

co

o o

NO
co

OJ OJ

On

CO o\

o u
%1
i/n

O < CO

o o
OJ

LfN

CO
OJ

r 00

ON
OJ

u o
CO
PQ CO

CO

a
Eh

Q
l

CO
-a-

CO
on

On
rH

o H

<-\

<
PQ CO Cm

o H H
CV

H H H

LTN

O
-a-

rH

H O

O o H
OJ

NO
r^

LT\

O
On rn

OJ
P~>

l/N ITS

\s\

OJ

rH

H o H

-3

UN
rH

O
c
(0
ctf

T3

CO < <
p*

LT\

_3
OJ

O
.H rH

On

PO
OJ rH

tCO
OJ

^t

m m
rH

-3

o O H

<D

l/N

ITN

-P
01

NO -3

rH
LTN

tLTN

T3

rH

H rH

O
to

C o o
>*i

to

a
CO

o
IA

CO
rH

-P
Ul

H H
o

NO

CO

On

O H

H H

OJ rH

m rH

C a H
o

c o
H

IA

l/N

s
CtJ

+J

m H

OJ

r
OJ

o
C\J

rH

o H
l-\

O O H

ON

CO

O
fr-

NO

H a

>

161

V)
in
(VJ

q: LU CD

o o

H-

4>
fc

J X)

c o

X K Z UJ S *
1-

LU

O
0

c
c

m S LU
U_

k_ a* a;

O
<n LU
_i

o o

E o
-

o c o

2? LU

<2
DRA

8
STRESS

PARTIALLY

TOTAL

CVj

io

i^'U0jiDAd|3

LU or

Ll.

1&

o o
fO

2
UJ

o
\D OJ

< m 5
UJ

_ >*-

_i

o o
f\l

c
JL 1

< z
u_ u.

E C E

o
en UJ 01

UJ <-

o c
c

If)

h <
V) UJ q:

o
8

a;

E o
u

c o
l/t

UJ

>
1EFFEC

u 8

BERMS

ED

o
DRAIN

WITH

if)

UJ or
4J/ U0!iDA3|3
u_

163

in

lCh

a, Pw Oh

CO

o o o vo vo o
(J\

CO CO

VO

o CO o UN o
UN

UN

VO

UN On
t~-

t CO

t-

VO

CO

CO

ON

On

4
o o c
cq CO CO

c o

CJ

< CO

H O
OJ
CNJ

UN

UN

CM

CO CO

CO

O H

VO CO CO

VO
on

c\ t

vo
CNJ

ON

O O O

CO

00
rH

< pq co
a,

1A
CM CM

VO W H H

UN

UN

-3-

O o UN H H
UN
CM

CUN

1A
CM CM

UN
t
CM

UN _*
on

O O
-3-

UN

VO J-

O
CM
rH

UN

UN VO UN

on

VO

rH

UN

VO

CO

ON

p
cS (U

-P

rH

165

UJ

<
CD UJ

<
!5J

U.

O
UJ
(J)

LJ

2 _l O
if) -z.

<o
UJ a: \(j)

o o

O
I-

UJ

io

m ro

in
CVJ

m 5

m N
4J

in

in
c\j

m O

m
00

m
CO

in UJ a:

'U0.UDA9I3
u.

166

UE.2) varied from + k% to - J% of that obtained by the MP analysis.

Total stress analysis is not normally used to investigate long

term stability, where it is assumed that the embankment foundation has

consolidated

100/*.

reason for using total stress analysis in this

case, is the large difference between the factor of safety by the total

stress analysis and the effective stress analysis (1.3

vs.

2.25).

Also, it is believed that if a failure occurred, it would occur rapidly

and develop excess pore pressures. almost
100/u

Thus even though the embankment was

consolidated, an undrained shear would occur, and the un-

drained total stress analysis is considered most appropriate.

5.6

SUMMARY Aim CONCLUSION Results of stability analyses obtained by STARL for a design

example problem (Figure 5.1) were compared with results obtained by programs based on the simpli fied Bishop method and Morgenstern-Price procedures. The results are summarized in Tables $.h and
5- 5-

Most of

the cases investigated by GTABL are documented with graphical computer

plots in Appendix

3.

For the same type of failure surfaces, GTABL gave lower values of
the factor of safety than the other programs.

The difference was in

most cases less than 10$, with a maximum of l6% (0.83 vs. 1.05 for the

preliminary design, undrained total stress analysis).

Considering the

uncertainty in the input data, this difference is tolerable.

For problems with complicated soil conditions, GTABL is recommended

compared to the other programs, because STABL utilizes the computer

more in the search for the more critical failure surfaces.
STABL

usually gives a factor of safety on the conservative side (see Section

167

H >;
CJ

--Pl,

o o

a ^
I

a,

-H

C
CD

H
y,

w oo p < o s
o
co

P,

0\

CO

O ft h O a x
t On

u o
p<

Vh

H O

O E
C\J

K
u
CO
a;

a)

p
rH

p
CO

H H
ni

<v

<) V,

,0

<P
CO

rH

p o H

ed

P
O
C-l

cd

T3
<d
i

C
H

<P

O
cd Jh

o h

c o

168

p
u o
ft
o>

aS

C
H tCD

oo 00

r J-

h
1

Eh

-P

CO

(X H ft co

CJ

W S

O
03

w'

CO

,0
B)
="*

O h
<M
C3

o CO c O o o

-p

H ft

J-

H -h

W o K
i-q

^
a?
OJ

P
a>

(1)

3'
T3

a)

o
u
aS

fr

o m
u

rH

O
>~

ft

ft
en
en

X P
<v

c
H
CD CO as CD

W)

U
P
CO o>

c
0)
J-.

P
aS CO

0) (h

-p
CO ai

<

H
P H
a)

>
H
-p
CJ cd

>
H -P
CJ (D

a c
H
>-.

W
T)
a.
1

<M Vh

<w <M
CJ

H H
o
CO CJ

<iH

CD

T3
K)

O
i-l

c
H
a)

T3
01

c
H

-P
s)

O o

H o c 3

O U -P o
CO

ft

H ft

e c
CD CD

o u
p

,o
CO as

al tn

o
CO

CO 0>
co

S
CVJ

5
Ch

P H

0)

2
gj oj J3 -P H

5 O
cj CJ aS

<tH

bD
CO CD

-h p O 3 S Q
CO CO

a>
Ch

>

*
*

P B

169

2.3).

If for any reason it is desirable to determine a "more precise"

value of the factor of safety, other methods of slices may be incorporated in the CTABL program (see Chapter VI; the modified (or simplified]

Bishop method was later introduced as a permanent option in the CTABL

program, applicable to circular failure surfaces, see Chapter IV).

Differences between factors of safety for the different kinds of

failure surfaces generated by CTABL is less significant in this example

problem than in the example with a weak layer of soil investigated in

Chapter III.
The largest difference (12$) is between CIRCLE and BL0CK2

for the total stress analysis.

170

VI

COMPARISON OF STABL WITH EXISTING PROCEDURES FOR SLOPE STABILITY

ANALYSIS USED BY INDIANA STATE HIGHWAY COMMISSION (ISHC)

Four problems from the Division of Materials and Tests, ISHC, have
been analysed for stability by existing ISHC procedures and by STABL.
The first problem involves analyses by
a)

the computer program NYSTAB* which is based on circular arc

failure surfaces and the modified Bishop method,

b)
c)

hand calculated sliding wedge analysis, and

STABL.

The second and third problem involves analyses by

a)

a computer program used by ATEC** in which circular arc

failure surfaces are analysed by the modified Bishop

method, and
b)

STABL and modifications of STABL.

The last problem is solved by

a)

NYSTAB,

b)
c)

STABL and modifications of STA3L, and

hand calculation.

NYSTAB - New York State, Department of Transportation, Slope Stability Analysis ** ATEC = American Testing and Engineering Corporation, Indianapolis, Indiana.
*

171

6.1

PROBLEM

The problem is presented in Figure 6.1. An embankment of granular

fill is resting on a foundation of 28 feet of granular material under-

lain by clay of medium to low strength.

The factor of safety for the embankment slope alone is given by
_ =

tan
tli

<t>'

'

where *

is the effective strength angle, and 8 is the

<J>'

angle of the slope. In this case

= 35, tan 8 = 32/56, hence

F = 1.23. This result was also found by STA3L, circular search with no

restrictions on initial angle, the critical failure surface being a

shallow circle.

Stability with respect to the underlying clay soil was analysed

with forced deep failure surfaces, both circle and sliding block shapes.
The result from NYSTAB is F

mm =1.32
. .

for a deep circle. Hand calculated

for the deepest surface analysed J
.

sliding block analvsis gave F

mm =1.26

by this method (medium depth), and 3TABL gave F

= l.l6

for a deep

sliding block. All sliding blocks have approximately horizontal bases

and Rankine active and passive failure zones.

Figures 6.2 and 6.3 show the failure surfaces investigated and

corresponding factors of safety. All results are summarized in Table

6.1.

3TABL, sliding block search, gives a factor of safety 6% lower

than the one found by hand calculation for the same depth (l.l8 vs.
1.26), whereas the deeper sliding block found by STABL is more critical
{Q% lower, 1.16 vs. 1.26).

6.2

PROBLEM 2
The soil profile and strength data for problem 2 are shown in

Figure 6.U. An embankment of granular material is constructed on a


n?

'I

O
ii

a O O g o

S!
ii

o
CO

3
O o
CM

8
ii

o o

173

17*1

175

^ pq < Eh
to

.C
t)

-*
cm rH

on
-a-

VO
JrH

CO
rH rH

V>

^
a) tu to

on
CM rH

Vt
CM rH
1

rH

rH

t3

to

J a
ST

tu

-H .h
spec

* a> o
<t-t

* *

VD VD

CM
l/\

rH m rH

H
to

sur

o
pq

rH

<m

<H
fi

H ,0
a<

<L)

o m Ti o
Xi -p
0)

5
5
<H

< H
rn

>fe

o
a;

O U
rH

c o
H
t3

:>,

rQ
t3

*r

P
aj

VO

* >, T3

P C
<v
rH

rH
-d

M3
CM
r-I

E-t

&.
-31

a 3
CJ
i-H

o
<M
-=c
(1)

rH

H
fl
n)

0)

Q)

a)

-P CO

<H Vn

c o
1>

Eh

>
H

Q
Cm

P
h

& -p
rC Eh CO >H

H rO

o
rH

Ph

a)
a)

ft

-P

^
ai a)

LA
-d

CM

>J^

on

H
>
rH
cd

+J

o
!/)

o w
Vh

rH

3
tu <U

O
rH

w
bD

-P

o
-p

C
(1)

o o

T1 H

u
cd

W
H

rH
to

E>H

o
(Tj

<D

S
0)

o
aj
<H

0) 0) 0)
(1)

ft

<H
rH 10

tu to

rH
CJ
rH

o
CD n)

M 3
CO <H
tu

H
gt

o o
crj

H a

Cl>

bD

6 O o
<(h

3 c
H

-p

o
01

M
(U

tU

t3
<u
r*

be

rH

T3
(U

& w
T3

-H

o
tu

H
H

M 3 3

H rH

o
ft

> o
rH

ft

m
tu

s p
H
T3

>
P0> a>

a)

O C
rt

h Eh H CO >h

a
a

T3
ci>

-p

3
to tu rH
to

H
S-H

O
3

H
C)

a o
rH
cd

H lS

>> 6?

Ix,

.c CO

,c CO

tu

ft

176

foundation of somewhat clayey material, introducing an excess pore

water pressure in this material corresponding to piezometric surface

# 2. # 1.

The free groundwater table is represented by piezometric surface The distance between piezometric surface #
1

and # 2 represents

the excess pore water pressure induced by the embankment in the clayey

foundation.

The ATEC program with modified Bishop circular arc analysis gave
for the three circles (l, 2 and 3) values of F from 1.86 to 2.20, see

Figure 6.5-

STABL was modified to handle more than one water surface (see

Chapter

IV"),

and factors of safety were obtained for the three circles

from the ATEC analyses as well as for STABL circular search and sliding

block searches.
The circular arc failure surface proved to be reasonable for this

problem (a sliding block search gave a factor of safety only 2% lower

than the circle search). To extend the comparison between different

methods of analysis, the STABL program was modified to calculate the

factor of safety by
1)
2)

the modified Bishop method, and Spencer's method.

The Bishop method required only a minor change in the subroutine

FACTR of the program, whereas the Spencer method required a complete
change of this subroutine. The source for the Spencer method was Wright
(1969). See Appendix C for a listing of subroutine FACTR by both methods,

Later, the modified Bishop method was introduced as a permanent option

in the STABL program, applicable to circular failure surfaces (see

177

178

O -I U) CD K < < hV)

Br- - -

-'

$S! CM

CM

CM

o|-

10

"I f
<?>

^JcMCVi

1
o o E

^
in

cm ro

179

Chapter IV). The FACTR subroutine used in the present STABL program is
included in the listing of the program, Appendix
E.

The results of the different methods of analysis are summarized

in Table 6.2. STABL with the normal procedure of factor of safety cal-

culation (Janbu's simplified procedure) gives the most conservative

values of F, with a minimum for a circular arc 11$ lower than the ATEC

minimum value (1.66 vs. 1.86). STABL with the modified Bishop procedure
gives the same value of F

min

as the ATEC method, but for a different

circle (see Figure 6.5). STABL with Spencer's procedure gives a value

slightly * below the ATEC F

nun

value, also for a different circle. The

maximum difference in results occurs between STABL, simplified Janbu

procedure, sliding block search, and the ATEC solution for a circular
arc, namely 12$ (1.63 vs. 1.86). It should be noted, however, that the

result for the sliding block analysis with Spencer's procedure is 16$

larger than the result for the normal STABL procedure (1.95 vs. 1.63). This may indicate that the factor of safety by STABL is too low for
the sliding block analysis. The sliding block failure surfaces are

shown in Figure 6.6.

For the Spencer result, circle 5, the lines of total and effective
thrust are shown in Figure 6.7- They look reasonable except for the
right termination. Also, interslice factors of safety with respect to

shear have been calculated. The minimum value was 2.12, compared to an

overall factor of safety F = 1.83 for the sliding surface, indicating

that the result is reasonable. See Appendix C for calculations of

effective line of thrust and interslice factors of safety, and for more

information about the influence of the line of thrust on the factor of

safety.

180

Table 6.2

Comparative Study, Problem 2

Factor of Safety for Different Methods of Analysis

Surface
1

Method of Analysis
2
3 U

ATEC mod. Bishop Circ. arc

Circle 1
2
3 h
5

STABL

STABL mod. Bishop Circ. arc

1.95

STABL Spencer

1.86

1.T5

1.95

2.13
2.20

1.80
1.T5

2.16

2.15

2.01 1.88
1.86

1.98
1.85

1.66 1.68

1.83

BL0CK1-1
"

2.01
1.63
2.09
1.61+

1-2

1.95

BL0CK2-1
"

2-2

2.01

Max. Diff.

0%

-12%

0%

- 2%

Circles 1, 2 and 3 from ATEC analysis,

Circle
Circle

It

from STABL, circular search, case 2,

from STABL, circular search, cases

3 &

*+.

BL0CK1-1 and 2-1 from STABL, sliding block search in

soil layer no. U, case 2,

BL0CK1-2 and 2-2 from STABL, sliding block search in soil layer no. 5, case 2.

Between minimum values, compared to method 1 (ATEC-solution)

181

CM

1
9
\_

Q.

o
_l

GQ

fi (A

9 CO _
-i-

O
CM

CO

_
CM
(VJ

CM

_
CM

i 1 CD QD

P
CD

CD

182

if)

>

a
E o

tj

183

6.3

PROBLEM

Figure 6.8 shows the problem.

An embankment supported on one side

by a retaining structure is investigated for a total failure.

The same procedures as for problem 2 were applied to this problem,

however, the Spencer method did not converge for most of the failure
surfaces investigated.
Even with convergence, a further check showed
See Appendix C, Tables C.ll and C.12, and

the result to be incorrect.

Figure C.13.
The results for circular arc analysis by ATEC and STABL are shown
in Figure 6.9.

For the ATEC failure surface, STABL gives a factor of

safety 1% lower than the ATEC value (2.15 vs. 2.32), whereas a circular search by STABL results in a failure surface having a STABL factor of
safety 16% lower than the ATEC value.
If the safety factor is calcu-

lated by the modified Bishop method, the same two surfaces result in
factors of safety 1% to 3% higher than the ATEC value.

Hence the values Even lower values

found by the normal STABL procedure are somewhat low.

are found by a STABL sliding block search, namely as low as F = 1.5k,

which is 3k% lower than the ATEC minimum (see Figure 6.10).
In applying the Spencer procedure to the same failure surfaces
,

all

cases except one did not converge in the iteration process for the fac-

tor of safety calculation.

The one case that did converge gave a safety

The

factor of 2.30, where a value of 1.65 was obtained by STABL.

Spencer result was checked by calculation of interslice factors of

safety with respect to shear.

These were found to be less than or

equal to l.k (see Appendix C, Table C.12), hence the value of 2.80 is
not correct.

The results of the analyses are summarized in Table 6.3.

18U

1 l

to

IT)
<\i

/
/(
1

3 8 E

CVJ

< f

// / / / /
%
1

"-

'8

/
/
to n

'A 8
/ M

b m

ii

/*
/
a

8
o v, o
M

1
d

f
*-i

U
i
i
X

Ml

00

Q s
to

^^

Kfe* 1*
\

i"

I
ID CM 8.

$ m

soil

to

yj

z
lies

pplies

H
\
1 I
1 1

in

o o &

s
"=*

O
> xi
n

"

\.

S 8
surf

t
su
ric

H
l 1

o u

c
(VI

ft

35

or

@ @
.1
1

8
U

O o

o m

185

fO

fO

If
-if

II

m Jg <tu

00 ro
CM

<t ro CM

r- t>

!
^%

o
IO
<\i

v>

d
V) LL

m
0)

U
UJ

CM fO

LL
U o <u
(7?
(L>

CM

81
4
IT)

A
0>

CM
a
jl>

o O
8
**

u k,

i_

o o

O s

U-

186

187

Table 6.3

Comparative Study, Problem 3

Factor of Safety for Different Methods of Analysis

Surface
1

Method of Analysis
2
3 k

ATEC mod. Bishop Circ. arc

STABL

STABL mod. Bishop Circ. arc

STABL Spencer

Circle 1
2

2.32

2.15
1.95

2.38
2.3<+

BL0CK2-1
"

1.72 1.63

_2

"
"

-3
-1+

1.80 1.65 1.5U

0%

2.80**
-

"

_2*

Max. Diff.

-3k%

1%

Result for search with 50 surfaces instead of only 25.

** Calculation of interslice factor of safety gave so 2.80 is not correct.

Fi

l.k,

108

6.k

PROBLEM k
The problem with soil profile and strength data is shown in Figure

6.11.

An embankment of cohesive material, 9 feet high, is built on a

relatively soft foundation.

The strength of the foundation decreases

with depth until a firm strata is reached at approximately 13 feet

below the Ground surface.

A search with the NYSTAB program gave a factor of safety F = 1.1*1

for a deep circular failure surface (circle 1, Figure 6.12).

This sur-

face was analysed by the STABL program, resulting in a factor of safety

F = 2.21, 51% higher than the result by NYSTAB.

A circular search with

STABL gave F = 1.7U, while specifying the critical circle according to

Taylor (from Figure U.7, page 120), for a similar slope geometry, but
homogeneous soil with
<J>

(circle 2, Figure 6.12), resulted in a STABL

factor of safety F = 1.70.

This value is still 21$ larger than the

result obtained with the NYSTAB program.

Because of the large differ-

ence in results, and the fact that the STABL program gave an unconser-

vative safety factor, the case was further investigated.

Because a total strength analysis is applied in this problem, it

could easily be hand calculated for a specific circular failure surface

by the
circle

=
<f>

method.

This method gave a factor of safety F = 1.1+5 for

(Figure 6.12) confirming the NYSTAB result.

Finally, STABL

with the modified Bishop method was applied to the same failure surface,

resulting in a factor of safety F = 1.55, again much closer to the

NYSTAB result than STABL with the normal (simplified Janbu) procedure.
The fact that the simplified Janbu procedure, normally used in
STABL, in this case gives a higher factor of safety than the modified

189

o
uJ

O
fO

u
JO

o CM

(0
LU cr
to

O
if)

o
to

o
u

1
190

4;

3& 0Q.C
<t him
"-

^
u.

lO

<fr

<
V) id _l

^S c
-J

CD o <~ 3

K_52

CM
(VJ

p N

n-S

<
-_ "

m
ho) >-

<

t u

u_

<*-

u o

o>

(VJ

a>

k.

u k>

u k.

hi

u o o
to

o
CM

191

Bishop procedure, contradicts Carter's conclusion (1971) that the factor of safety reaches a minimum value when the moment center (y) goes

towards infinity.

Taking Carter's equation for the factor of safety as a function

of the y-coordinate of the moment center, and applying
a constant, the factor of safety F
<j>

and

c =

can be expressed explicitly as a

function of strength intercept c, unit weight y, geometry of slope and

failure surface, and moment center (y).

Replacing the summation over

slices with integration over a circular failure surface, a mathematical

3F

expression is obtained for F

and =

3y

can be found.

Derivation of
3f

these expressions are included in Appendix C.II.

goes to y
-*

It is found that
or_ a

-r

(zero) for y

-*

hence F
9F
~

reaches a minimum
C
.

maximum for

depending on the sign of

3F

F
c

and

-r^
<3y

were calculated for different slopes and different

The results are illustrated in Figures 6.13
9F
=

circular failure surfaces.

through 6.19.
and F 00 = F maX
c

Defining F

lim F

it is found that

is positive,

for circular failure surfaces that intersect the ground

> 60 to 80 surface with a steep angle at the top of the slope (a_ U

3F

see Figure 6.12).

_ min , and F OO = F^
Q

For small angles of intersection,

-r

is negative

Since F

the factor of safety for moment taken around the center

) ,

of the circular failure surface (corresponding to y = y

of
<j>

in the case

= 0, represents a mathematical correct solution to the problem,

.

comparison was made between the values of F ot and F

F
c
,

Figure 6.17 shows

and F

for homogeneous soil conditions (strength of embankment soil,

c
).

equals strength of foundation soil,

It is seen

that whereas

192

Fc Xh

FIGURE

6.13

FACTOR OF SAFETY Fc CENTER y FOR VARIOUS

c E -c F

vs.

MOMENT

0.,r=2.5h,

193

10

Fc /h

FIGURE 6.14

FACTOR OF SAFETY Fc vs. MOMENT CENTER y FOR VARIOUS O r=2.5h,

,

c E =2c F

c
19
1*

Fc Xh

FIGURE

6.15

FACTOR OF SAFETY Fc vs. MOMENT CENTER y FOR VARIOUS 0.,r-5h, e2c.

195

Fc Xh

FIGURE 6.16

FACTOR OF SAFETY Fc VS. MOMENT CENTER y FOR VARIOUS 0. r-2.5h,

,

Cj-O.Scp


196

F.

moment around

center of circle

(modified Bishop)

20
Eo

horizontal

force equ librium

(simplified

Janbu)

R-Xh
_

10

F"
l

r-2.5h

F-

F.

1
r

F.

r-IOh

F.

E.

10
O
,

20
degrees

30

FIGURE

6.I7

STABILITY NUMBER

F c Xh

VS. 9.

FOR VARIOUS

RADII,

CIRCULAR

FAILURE SURFACES. Cj-c,,

*-0

197

Fc

~ ~

moment around center

(modified Bishop)

of circle

20

F^

horizontal
(simplified

force equilibrium

Janbu)

r=IOh

10
O,

20
degrees

30

FIGURE

6.18

STABILITY NUMBER

yh

VS.0 O FOR

VARIOUS RADII, CIRCULAR FAILURE

SURFACES. c e 2c*0

198

F.

rsj

moment around center

(modified Bishop)

of circle

20
Fee

horizontal force equilibrium

(simplified

Janbu)

Fc Xh

rIOh

LL
20
30
0.

r2.5h

10

9a ,degrees

FIGURE 6.19

STABILITY

NUMBER Fc^ h VS

FOR

VARIOUS RADII, CIRCULAR FAILURE SURFACES.c E =0.5c F ,*-0

199

Fco = F

is only slightly less than the corresponding F

c
,

oo

= F
c

may be considerably larger than the corresponding F

and actually goes

towards infinity for a

Fco = F

approaching 90

(0
or,

= 0).
> 80

It is seen that (a. = 90 - 9

max
c

for 8

<

10 corresponding to
,

If the soil is inhomogeneous

with the slope material being stron-

ger than the foundation material, e.g., an embankment on a soft foundation, as in this problem, the result by the simplified Janbu method
(F
co
)

deviates even more from the mathematical correct solution (F

).

For the embankment soil being twice as strong as the foundation soil
3F
(o
r>

= 2c-,),
r

t-5
dy

is positive and F m = F
O

maX
C

for a_ > 60
U

and the dif-

ference between Fo and F

6.18.

increases significantly as shown in Figure

If, on the other hand, the foundation soil is stronger than the

embankment soil, F co and F

are almost identically, except for very small

values of 8. (< 5 for the foundation soil being twice as strong as the

embankment soil, see Figure 6.19)soil will be the strongest.

However, normally the embankment

The conclusion is that the simplified Janbu method cannot be relied

upon where deep failure surfaces intersect the ground surface at the top

of the slope at angles steeper than a

= 60

coincident with a relative

large total strength* in the zone of intersection.

In the cases of circular shaped failure surfaces, F q provides a

reliable solution.

For this reason the STABL program was modified to

)

include the option of using the modified Bishop method (F

failure surfaces, see Chapter IV.
* In case

for circular

Since F

is much less sensitive to

of effective stress analysis, a relatively large strength intercept c'

POO

variations in a^

it is recommended that the modified Bishop method is

.

used for circular failure surfaces in general

A sliding block search with STABL gave a factor of safety F = I.I42,

almost identical with the NYSTAB result.

This confirms that the Janbu

procedure normally used in the STABL program is still reliable for

failure surfaces where a
ing block has
ct

is smaller than approximately 60

The slid-

= h^

However, STABL with the simplified Janbu procedure should not be

exclusively relied upon for analyses of non-circular failure surfaces,

intersecting the ground surface at the top of the slope with a steep
angle, coincident with a relatively large total strength in the zone of

intersection.

Results of the different analyses are summarized in Table 6.h.

6.5

CONCLUSION OF COMPARATIVE ANALYSIS

The comparisons between STABL results and results obtained by

other methods of stability analysis show that STABL generally provides

solutions that are somewhat conservative.
However, the difference is

usually less than

10/5.

This is acceptable when compared to the much

larger variation in the factor of safety that may result from variable
soil strength data, see Chapter III.
In one case, however, a STABL sliding block surface gave a result

3h% lower than a circular surface analysed with the modified Bishop

method.

This result could not be verified, as the Spencer procedure

did not converge for this failure surface, and the Bishop method only
applies to circular surfaces.

'

201

o
,c pq

ft

03

ro

H
T3

t3
co

o e
-j

p
cd
-i

H o
cd

3 o

> '5
cd
1-5

On
-3

On

CM

T3

cd

<
Ch

c
(0

W
H
(0

C 3

-H
CO

o
Jto

o
LT\ -HII

>.

e
CO

o
,G P

rH ,0

3 5
o
CO

Ha-

H
1

vs.

CO

M
-3PL,
*

^ P
C
<Vh

(1>

p
3

ft

vo

o
>A CO pq -h PQ Eh CO

H
&
id

CO

>> TJ

0) (h

c o
LP>

rQ

CO

P
CO
<D

<H H

<

rn
LTN

UA

H -a
H

m
-=t

H
fes.

p
3 O
CO

rH
<1J

Eh

rH

rH

rH

rH

>
H P>
cd

h O
>, p
CD

o E 3 3

O M
H CJ
CO cd CO
cd

^ H
CO >H

u
cd

P j
E-i

^
P
O

B*

E o

<M
a)

^
CO

H
CM CM

Cn
-3-

O
t
rH

o
Cm

-a-

-3

CM

cj

CO
H

ft H
CO

CM

VI
rH

T3
(1)

U 3
to
fjj

O
u o p o
cd

M
cd

^-^

ft

o
pq

o o

&

to

&
rH

,c -h Eh Pq CO >H -d

<

en

s Cm
OJ

to

-=H
feS.

(I)

H\

3
rH
erj

C E i> 3 E
to
CD CD

P>

bo

o
E

> E E
H
,C
CJ

* P
10

W
CD

C
-H H

co CJ cd

-C
CJ

Ch

U 3
CO CO

* *

U
cd

H
cd OJ

*
Ch"

C H
Cm

H H
C_>

cd co

E a
fi)

cd

-p
CD

rH

CM

CO

CO
1

co
1*

CD

C M
(D

CO CJ

fc
0)

&
rH -H

m
H

H o
h
H

CO

H
CJ

CO

w
Ml o 02

CM

M
H
C_>

W CJ

H 13

rH

-p

O m

Cm m "3

<2
Eh

H o

s
pq
1

CO

pq

ctj

202

An exception to the general conservative tendency are the cases of

deep failure surfaces in soils with strength defined mainly in terms of
a strength intercept c (or
c
1

).

In such cases the simplified Janbu

method used in GTABL may give erroneous and un conservative results

and should not be used. For the general conservative solutions, Janbu (1956) has introduced a correction factor for the factor of safety obtained by the

simplified Janbu method, see Figure 6.20.

These correction factors

were applied to the results presented in Chapter II, Table 2.3.

showed reasonable good agreement
,

They

although the corrected factor of

safety seemed to be slightly on the unconservative side compared to

the Bishop solution.

However, for the non homogeneous soil conditions

analysed in Chapter V and in this chapter, the correction factors did

not seem to work.
A more thorough analysis is required to determine the relationship

between type of failure surface, soil profile and strength properties,

and the conservatism of the simplified Janbu method of slices.

203

shear surface

F-f
F

F.

Factor of safety by Janbu's procedure

Corrected factor of safety

FIGURE 6.20

CORRECTION FACTORS FOR THE SIMPLIFIED JANBU PROCEDURE (AFTER JANBU et al., 1956)

20*4

VII

SUMMARY AND RECOMMENDATIONS FOR FUTURE WORK

7.1

WORK AND RESULTS

The present study is a continuation of a previous study by Siegel

(1975), dealing with computerized slope stability analysis.

All inves-

tigations are related to a computer program, STABL, developed by Siegel.

STABL performs analyses of general slope stability problems.

7.1.1

Par ametric Study

A parametric study (sensitivity analysis) was performed on the

STABL program to assess the precision with which input data need to be
defined.

The results of this study show that variations in the soil

c

parameters

(c') and

1 (J)

((J)

have the largest influence on the stability

of a slope, in homogeneous soils, as well as for more variable soil

conditions.

7.1.2

Modifications of STABL
In addition to the three original surface generators, which are

the circular arc, the irregular, and the sliding block type surface, a new sliding block surface was introduced, which generates the active and passive portions of the surface according to Rankine's theory,
(1*5

<f>'/2)

with respect to the horizontal for the active, and (U5

<J>'/2)

with respect to the horizontal for the passive portion.

The new

sliding block is recommended for slopes with relatively simple soil

profiles.

205

Also, the water surface option is extended, so that it is now

possible to define different piezometric surfaces for different soil

layers.

This is convenient in the analysis of problems encountering

It may also be

artesian water conditions or a perched water table.

used to approximate excess pore water pressures in low permeable soil

layers due to embankment construction, and to analyse reservoir embank-

ments for rapid drawdown.

The modified Bishop factor of safety was introduced for circular

failure surfaces, in addition to the simplified Janbu method.

This was

done because of erroneous results obtained with the Janbu method for deep circular failure surfaces in soils with
<|>

= 0.

The Bishop method

satisfies overall moment equilibrium, which in the case of a soil with

<f>

provides a mathematically correct solution, whereas the Janbu

method with overall horizontal force equilibrium represents an approximate solution.

The Bishop method can be applied only to circular

shaped failure surfaces.

7.1.3

Comparative Study
Results of analyses performed with STABL (simplified Janbu method)

were compared with results obtained by other methods of slope stability

analysis.

Results by STABL are usually conservative, an exception being

The differences

the deep circular failure surfaces mentioned above.

are normally less than 10$, but a value as high as 3h% was obtained.

The conservatism seems to be larger for sliding block type failure surfaces, than for failure surfaces of circular shape.

206

7.?

SUGGESTIONS FOR FUTURE WORK

GTARIi has

been used routinely by the Indiana State Highway Commis-

sion for the last two years, during which period (1976-77) more options

have been added to the program, and comparisons have been made with

other methods of slope stability analysis, as described above.

The result is a program, even more versatile than the original
STABL program.

However, the program is also rather bulky and for many

problems, only a fraction of the program is actually used, even though the total program has to be loaded into the computer memory. This is

inefficient use of computer space, and could be avoided by means of an

overlay structure.

Such a structure divides a program into overlays,

and only those overlays that are needed for a particular problem are

loaded into the central core memory of the comt>uter.

is recommended for the STABL program.

This improvement

A more thorough analysis is required to determine the relationship

between type of failure surface, soil profile and strength properties,

and the conservatism of the simplified Janbu method of slices.

Janbu

did provide correction factors to adjust the simplified solutions to

more accurate results.

However, these correction factors seem to apply

only to slopes in homogeneous soil.

The task still remains to develop an analytical model for slope stability analysis that combines the simplicity of the Janbu method

with the higher accuracy of, for example, the Morgenstern and Price
procedure, and does not have the problems related to misleading factors
of safety.

LIST OF REFERENCES

207

LIST OF REFERENCES

Note:

ASCE

American Society of Civil Engineers

Athanasiou-Grivas, D. S. (1976), "Reliability of Slopes of Particulate Materials", Ph.D. Thesis Purdue University, West Lafayette, Indiana, 1976.
,

Baligh, M. M. and Azzouz, A. S. (1975), "End Effects on Stability of Cohesive Slopes", Journal of the Geotechnical Engineering Division , ASCE, Vol. 101, No. GT11, Proc. Paper 11705, November, 1975, pp. 1105-1117.

Bishop, A. W. (1955), "The Use of the Slip Circle in the Stability Analysis of Slopes", Geotechnique , Vol. 5, No. 1, March, 1955,
pp. 7-17.

Bishop, A. W. and Morgenstern, N. (i960), "Stability Coefficients for Earth Slopes", Geotechnique , Vol. 10, No. 4, December, i960, pp. 129-150.

BJerrum, L. (1972), "Embankments of Soft Ground", Proc. ASCE Specialty Conference on Performance of Earth and Earth-Supported Structures , Vol. 2, Purdue University, June, 1972, pp. 1-54.
Carter, R. K. (1971), "Computer Oriented Slope Stability Analysis by Method of Slices", MSCE Thesis , Purdue University, West Lafayette, Indiana, 1971.

Catalan, J. M. and Cornell, C. A. (1976), "Earth Slope Reliability by a Level-Crossing Method", Journal of the Geotechnical Engineering Division , ASCE, Vol. 102, No. GT6, Proc. Paper 12214, June, 1976, pp. 591-604.
Chen, W. F. and Giger, M. W. (1971), "Limit Analysis of Stability of Slopes", Journal of the Soil Mechanics and Foundations Division , ASCE, Vol. 97, No. SMI, Proc. Paper 7828, January, 1971, pp. 19-26.

Collin, A. (1846), "Landslides in Clays", Translated by W. R. Schriever, University of Toronto Press, 1956.

Coulomb, C. A. (1776), "Essai sur une application des regies des maximis et minimis a quelques problemes de statique", Memoires Academie Royale des Sciences , Vol. 7, Paris, 1776.

208

Pvinoff, A. H. and Harr, M. E. (1971), "Phreatic Surface Location after Drawdown" Journal of the Soil Mechanics and Foundations Division ASCE, Vol. 97, Mo. SMI, Proc Paper 7806 January, 1971, pp. J+7-58
,
, .

Eellenius, W. (1926), "Erdstatische Berechnungen mit Reibung und Kohasion", Ernst und Sohn Berlin, 1926.
,

Janbu, N. (l95Ua), "Application of Composite Slip Surfaces for StabilityAnalysis", Proceedings of the European Conference on Stability of Earth Slopes Stockholm, 195 1*, Vol. 3, pp. +3- *9.
1 1 ,

Janbu, N. (l95^b), "Stability Analysis of Slopes with Dimensionless Parameters", Harvard Soil Mechanics Series No. H6, 195**.
,

Janbu, N., BJerrum, L. and KJaernsli, B. (1956), "Veiledning ved lasning av fundamenteringsoppgaver - 2. Stabilitetsberegning for fyllinger, skjaeringer og naturlige skraninger" (Soil Mechanics Applied to Some Engineering Problems - Chapter 2. Stability Calculations for Embankments, Cuts and Natural Slopes), Publication 16, Norwegian Geotechnical Institute, Oslo, 1956, pp. 17-26.
Ladd, C. C. and Foott R. (197 *), "New Design Procedure for Stability of Soft Clays", Journal of the Oeotechnical Engineer i ng Division ASCE, Vol. 100, No. 0T7, Proc. Paper 1066U July, 197 *, pp. 763-786.
,
1

Ladd, C. C. (1975), "Foundation Design of Embankments Constructed on Connecticut Valley Varved Clays", Massachusetts Institute of Technology (MIT), Research Report R75-7 , Oeotechnical Publication 3^3, June, 1975.
Ladd, C. C.
(1976), Personal Correspondence.

Lowe, J. and Karafiath, L. (i960), "Stability of Earth Dams upon Drawdown" , Proceedings of First Pan American Conference on Soil Mechanics and Foundation Engineering Mexico City, Vol. 2, i960 pp. 537-552.
,

Lowe, J. (1967), "Stability Analysis of Embankments", Journal of the Soil Mechanics and Foundations Division ASCE, Vol. 93, No. SM^, Proc. Paper 9305, July, 1967, pp. 1-33.
,

Lundgren, H. and Brinch Hansen, J. (1965), "Geoteknik" Teknisk Forlag, Kerbenhavn, 1965, P- 180.

2.

udgave,

Maunsell & Partners Pty. Ltd., Consulting Engineers, Melbourne, Victoria, Australia, 1977: Correspondence.
Morgenstern, N. R. and Price, V. E. (1965), "The Analysis of the Stability of General Slip Surfaces", Geotechnique , Vol. 15, No. 1, March, 1965, pp. 79-93.

209

Newmark, N. M. (1965), "Effects of Earthquakes on Dams and Embankments", Oeotechnique Vol. 15, No. 2, June, 1965, pp. 139-159,

Perloff, W. H. and Baron, W. (1976), "Soil Mechanics, Principles and Applications", The Ronald Press Company, New York, 1976, Chapter 12. Stability of Slopes.
te Mars 19l6", Petterson, K. E. (1916), "Kajraset i Goteborg den 5 Teknisk Tidskrift , Vol. U6, 1916, pp. 281-287, 289-291.

Rankine, W. J. M. (1857), "On the Stability of Loose Earth", Philosophic Transactions of the Royal Society London, Vol. lU7, 1857, pp. 9-27.
,

Romani, F. (1970), "Dependence of Stability of Slopes on Initiation and Progression of Failure", Ph.D. Thesis , Purdue University, West Lafayette, Indiana, 1970. Romani, F., Lovell C. W. and Harr M. E. (1972), "Influence of Progressive Failure on Slope Stability" Journal of the Soil Mechanics and Foundations Division ASCE, Vol. 98, No. SM11, Proc Paper 93^9, November, 1972, pp. 1209-1223.
, , , ,
.

Seed, H. B. (1966), "A Method for Earthquake Resistant Design of Earth Dams", Journal of the Soil Mechanics and Foundations Division ASCE, Vol. 92, No. SMI, Proc. Paper U6l6, January, I966, pp. 13-Ul.
,

Seed, H. B. and Lee, K. L. (1967), "Undrained Strength Characteristics of Cohesionless Soils", Journal of the Soil Mechanics and Foundations Division , ASCE, Vol. 93, No. SM6, Proc. Paper 56l8, November, 1967, pp. 333-360.

Lee, K. L., Idriss, I. M. and Makdisi, F. I. (1975), "The Seed, H. B. Slides in the San Fernando Dams during the Earthquake of February 9, 1971", Journal of the Geotechnical Engineering Division , ASCE, Vol. 101, No. GT7, Proc. Paper IIUU9 July, 1975, pp. 651-688.
, ,

Siegel, R. A. (1975a), "Computer Analysis of General Slope Stability Problems", MSCE Thesis Purdue University, West Lafayette, Indiana,
,

1975.

Siegel, R. A. (1975b), "STABL User Manual", Technical Report, Joint H ighway Research Project, No. 75-9 , Purdue University, West Lafayette, Indiana, June, 1975Singh, A. (1970), "Shear Strength and Stability of Man-made Slopes", Journal of the Soil Mechanics and Foundations Division , ASCE, Vol. 96, No. SM6, Proc. Paper 7652, November, 1970, pp. 1879-1892.

Skempton, A. W. (I9U6), "Alexandre Collin (I808-I890) and his Pioneer Work in Soil Mechanics", Foreword to the English Translation of Landslides in Clays by Alexandre Collin (see Collin).

210

Spencer, E. (1967), "A Method of Analysis of the Stability of Embankments Assuming Parallel Inter-Slice Forces", Geotechnique , Vol. 17, No. 1, March, 1967, pp. 11-26.

Taylor, D. W. (1937), "Stability of Earth Slopes", Contributions to Soil Mechanics 1925-19^0, Boston Society of Civil Engineers , Boston, 19^0.

Taylor, D. W. (I9U8), "Stability of Slopes", Fundamentals of Soil Mechanics , John Wiley and Sons , Inc , New York, 19^8, pp. Uo6-*+79
.

Whitman, R. V. and Bailey, W. A. (1967), "Use of Computers for Slope Stability Analysis", Journal of the Soil Mechanics and Foundations Division , ASCE, Vol. 93, No. SMU, Proc Paper 5327, July, 1967, pp. U75-H98.
.

Wright, S. G. (1969), "A Study of Slope Stability and the Undrained Shear Strength of Clay Shales", Ph.D. Thesis , University of California, Berkeley, 1969. Wright, S. G., Kulhawy, F. H. and Duncan, J. M. (1973), "Accuracy of Equilibrium Slope Stability Analysis", Journal of the Soil Mechanics and Foundations Division , ASCE, Vol. 99, No. SM10, Proc. Paper 10097, October, 1973, pp. 783-791.
Wu, H. T. and Kraft, L. M. (1970), "Safety Analysis of Slopes", Journal of the Soil Mechanics and Foundations Division , ASCE, Vol. 9%~, No. SM2, Proc. Paper Ulk, March, 1970, pp. 609-630