58 CHUGH

CALCULATIONS FOR FACTOR OF SAFETY (see Fig. 1 for the meanings of the various
The factor of safety in a slope stability analysis symbols).
is defined by The variation in side force inclination in equa-
F= tions (2) and (3) is defined as (Spencer, 1973)
Available shear strength along a shear surface tan 6 = If(x) (4)
Driving shear stress along the shear surface
where f(x) is a predefined characteristic shape
(1) function and i is a scalar factor to be determined.
The available shear strength at a point in a soil This formulation for variable interslice force incli-
deposit depends on the shear strength parameters nation does not increase the number of
(c’, 4’) of the soil and the induced effective normal unknowns (Spencer, 1967).
stress at that location. In the conventional slope Equations (2) and (3) are recurring relations
stability analysis procedures in which the slide and allow the boundary value problem to be
mass is divided into slices, equation (1) is applied solved as an initial value problem for an assumed
for each slice. Typically, the derivation of the value of the factor of safety F, the side force incli-
slope stability equations leads to the following nations 6 and the known boundary conditions at
(Chugh, 1984). (The equations included here are the left-hand side of the shear surface (Chugh,
for the finite formulation of Morgenstern & Price 1982): see Fig. 1.
(1965) adapted to Spencer’s (1967) procedure. The idea of incorporating a variable factor of
However, the ideas presented for a variable factor safety in a slope stability analysis follows very
of safety can be adapted to any other procedure closely the idea used for the variable interslice
based on the limit equilibrium method.) force inclination in that a characteristic shape for
For static equilibrium of forces acting on the its variation along a shear surface is predefined
slice shown in Fig. l(b) and the solution procedure is required to calcu-
2, = 2, late a scalar factor which scales the characteristic.
cos (6, - a)[1 - (l/F) tan (6, - a) tan 4’1 Thus, for equation (1)

x cos (& - a)[1 - (l/F) tan (6, - LX)tan 4’1 T g(x) =

+ [(l/F)c’b set a - W sin c( Available shear strength along a slice base
+ (l/F)(W cos c( - U) tan 4’1 Driving shear stress along the slice base
x {cos (6, - a)[1 - (l/F) tan (6, - LX)tan #I}- ’ (5)

P cos (p - a)[tan (/? - a) + (l/F) tan 4’1 where T is the unknown scalar factor and g(x) is
+ the characteristic shape for the variation in factor
cos (6, - a)[1 - (l/F) tan (6, - a) tan 4’1
of safety along the slip surface. For g(x) = 1,
H, cos a[1 + (l/F) tan a tan 4’1 equations (1) and (5) lead to T = F. The defini-
+ tion of the variable factor of safety according to
cos (6, - a)[1 - (l/F) tan (6, - a) tan 4’1
equation (5) introduces only one unknown-the
H, cos a[1 + (l/F) tan a tan 4’1 same as for the constant factor of safety assump-
- cos (6, - a)[1 - (l/F) tan (6, - a) tan 4’1 tion. The solution procedure for calculating T
from equation (5) is similar to the procedure used
(2)
for calculating F from equation (1). The effect of
For moment equilibrium of the forces acting on using equation (5) in the slope stability analysis is
the slice shown in Fig. l(b) to change the distribution of induced shear stress
2, cos 6, and normal stress along a shear surface.
h, = - -4
z, cos 6,
CHARACTERISTIC SHAPE FOR VARIABLE
1
+b sin (6, - a) FACTOR OF SAFETY
2 cos a cos 6,
The shear stress distribution in materials along
a shear surface, for the constant factor of safety
+ 2 sin (6, - a) assumption, depends on the normal stress dis-
R 1 tribution and the values of the shear strength
parameters, i.e. induced shear stresses are higher
+ 5 cos B set 6,(h, tan /I - e)
R in materials with higher c’, 4’ values-all else
being equal. Thus, if a,’ at two points in two
+ $ set b,(H, h, - H, h5) (3) different materials were the same, the present
R analysis would indicate two different values of
 FACTOR OF SAFETY IN SLOPE STABILITY ANALYSIS 59

Ground surface
Last slice
\BJ

Potential shear surface

of some geometric
conflguratlon

(a)

(b)

Fig. 1. General description of the limit equilibrium analysis of a soil

deposit: (a) slope stability problem; (b) forces acting on a typical
slice

shear stress and their magnitude would depend of safety assumption and a preselected inter-
on the values of c’, 4’ for these materials. Intu- slice force inclination assumption, and finding
itively though, it would be expected that the shear (a,‘, 7)induced at the base of each slice. Now g(x) is
stresses at these points would depend on the coef- redefined as (T/(T,‘)~~~~_.~ and the calculations are
ficient of lateral stress and inclination of the shear repeated-all else being the same. The procedure
surface at these locations. If these were nearly the is repeated until (T/u.,‘)~,,~~~~~ is constant along the
same, the induced shear stresses would be shear surface. If the calculated value of the
expected to be about the same as well. induced shear stress is greater than the available
It has been observed in the analysis results of shear strength then only a shear stress equal to
several problems that the normal stress distribu- the available shear strength need be used in cal-
tion along a shear surface is generally smooth; culating g(x). The numerical procedure is gener-
the difference in normal stress distribution along ally able to achieve convergence in two or three
a shear surface is relatively small for different iterations. The value of T g(x) then defines the
interslice force inclination assumptions, shear factor of safety along the slip surface.
strength parameter values for the soils in a
deposit and methods of analysis. However, the
induced shear stresses differ appreciably along the GEOMETRIC INTERPRETATION OF
shear surface. The results of deformation analysis FACTOR OF SAFETY
by the finite element method indicate that the Constant factor of safety
ratio of induced shear stress to induced effective Equation (1) is the commonly accepted defini-
normal stress along a shear surface is reasonably tion of the factor of safety in slope stability
constant. This observation needs to be further analysis by the limit equilibrium method. By this
confirmed with additional studies. definition, the normal and shear stresses induced
If (7/gn’)induced were to be held constant for fric- along a shear surface are such that their plot
tional materials along a shear surface, then it maintains a constant proportion (equal to the
could be used to define the characteristic shape factor of safety value) to the corresponding
function g(x) in equation (5). The procedure then normal and shear stress strength plots at every
consists of making a slope stability analysis for point along the shear surface. Thus, for the com-
g(x) = I.0 in equation (5), i.e. a constant factor puted F > 1 the (on’, T)~,,,,~~_, plots below the
60 CHUGH

7 = C’/F
tan 2 = (tan @‘)lF

0 @ W$r;glomb strength

/
@@ (a,. r) dlstnbution
along the shear surface

@ IfF> 1.0

@ if F < 1.0

Normal stress u”’

(a)

- -- Induced stresses for

constant factor of safety
Bifurcation pomt for
bllmear strength envelope

Shear strength envelo

Induced stresses for

variable factor of safety

Normal stress 0”’

(b)

Fie. 2. Geometric interpretation of the factor of safety: (a) constant factor of safety; (b) variable
factor of safety

(a,‘, 7)strength data; for computed F < 1 the Variable factor of safety
(Q,‘, r)induaed plots above the (cn’, 7Lgth data; Equation (5) is proposed to define a variable
for computed F = 1 the ((T,‘, 7)induced plots on the factor of safety along a slip surface. According to
(on’, 7),trengtlI data (Fig. 2(a)). this definition a distribution of induced normal
Although F < 1 is routinely calculated in slope and shear stresses is sought along a slip surface
stability analysis of earth slopes for assigned that plots as a single continuous curve in the
values of material strengths and pore pressure (on’, 7) plane (see Fig. 2(b)). The F value at any
conditions, the calculated (o,,‘, 7)induced values are point along the slip surface is still defined as the
not acceptable as these stresses plot in a non- ratio of available shear strength to the driving
admissible stress space according to the theory of shear stress at that point. Since (o,‘, 7)induced
plasticity. The most shear stress that can be sus- values fall on one curve and there may be several
tained by a frictional material is equal to its shear strength curves, one for each material along the
strength for the corresponding induced normal shear surface, the ratio of mobilized shear
stress. Thus, an analysis procedure should not strength to shear stress induced is different, and
calculate F to be less than unity, with F = 1 being hence the factor of safety along the shear surface
interpreted as a failure condition. is varying.
 Mate1 propertIes
Material Unit weight:
c’ lb/v+ @‘: deg
lb/fts

0 135 0 20

134 0 25
130 0 15

130 0 35

Berm constructed in response 140 0 40

to movement along shear 125 0 12
Y surtace under study
. (i, 125 0.75 17
I
Shear stress distribution for
variable factor of safety
i

\ Shear stress dlstrlbutlon for

,-.t\ constant factor of safety

(a) (d)

- Y
g(x) for constant factor of safely
d
2 2.0
g(x) for variable factor of safety e
J= F= l-926 m
___---_--_ 1 ---- ---- -. $ 1.0
J = 5.033 0
I ----7
R 67 :
t 1
0
0) (e)

Essentially same dlstrlbutlon for constant

variable factor of safety
Effective normal stress at
base of shear surface \ Total normal stress

7’ _____ -_---- ------ ---

1
Porewater pressure
I 1 I I I , I
1 . bX
200 400 600 600 1000 1200 200 400 600 800 1000 1200
Projected distance along shear surface: It Profected distance along shear surface, ft
(c) VI

Fig. 3. Sample problem (reservoir water elevation, 540 ft)

 w
l- Maten; xopertles
MateW3
Unit weight:
c’: lb/lr? $‘. deg
Ib/ft3
0 135 0 20
0 134 0 25
0 130 0 15
0- 130 0 35
0 140 0 40
Berm constructed in response
0 125 0 12
Y to movement along shear
sunace unaer stuay
n 125 0.75 17
600 Shear surface under stu Shear stress distrlbutlon for
Drawdown c vanable factor of safety
Shear stress distribution for
constant factor of safety
8
9 400
a,
w
; 500
300 i x x
(a) (d)
Y

g(x) for constant factor of safety

SC 2.0
g(x) for vanable factor of safety 1 F
T= F = 1.545 ;”
__- _--_-_-__- 1 ___ -__--_
m 1.0
T = 3.993 2
I 0.210 ------1
-----_____. _--
E x
v) 0t
(b) (e)
% Essentially same distribution for constant and
Y Y vanable factor of safety
‘c A :
:b zo- Effective normal stress at ,z 4.0-
Total normal stress ool
base of shear surface “Z
2
zz 2.0-
2,J? lo- EL ------ ---------_--- ----
Porewater pressure
200 400 600 600 1000 1200 * x sg m 0 1
200 400 1 .Y
5 o- 600 800 1000 1200
Projected distance along shear surface: ft Projected distance along shear surface: ft
(cl (1)

Fig. 4. Sample problem (reservoir water elevation, 480 ft)

 FACTOR OF SAFETY IN SLOPE STABILITY ANALYSIS
CASE STUDY
Figure 3(a) shows a part of an embankment
dam cross-section and its foundation zones. The
shear strength properties of different materials in
the embankment and its foundations are shown
in the table included in the figure. There was a
slip failure at this site. As a result, the upstream
berm was constructed and the facility put back to
use.
Recently, during the reservoir drawdown, inter-
nal sliding at the old slip surface has been
detected by the instrumentation; some cracking
at the crest of the dam has occurred, but there
has not been a gross movement of the embank-
ment materials or a noticeable change in the
embankment geometry. Thus, the problem needs
to be analysed to seek answers to questions like
(a) is there a failure?
(b) where is the failure localized?
(c) how safe is the upstream portion of the dam?
(d) can the reservoir be filled without any major
rehabilitation work?
The shear surface geometry and the pore pressure
data for use in this study are shown in Fig. 3.
The slope stability analyses were performed
using the computer program SSTAB2 (Chugh,
1981). This computer program is based on the
limit equilibrium method; it satisfies all the statics
equations; it provides for use of a constant or a
variable interslice force inclination and a constant
or a variable factor of safety along the slip surface
according to the ideas presented in this Paper.
The conventional slope stability analysis of the
sample problem, using constant interslice force
inclination and constant factor of safety assump-
tions, yields F = 1.93 for high level steady state
reservoir operation and F = 1.55 for
reservoir drawdown. The normal stress, the shear
stress and the ratio of shear stress to effective
normal stress along the shear surface, as obtained
through these calculations, are shown in Figs
3(ct3(e) and 4(+4(e) for the high level steady
state reservoir operation and reservoir drawdown
conditions respectively. It is not possible to draw
any inference of actual or impending distress
along the shear surface from these results of con-
ventional slope stability calculations.
A similar slope stability analysis of the sample
problem using the variable factor of safety ideas
presented in this Paper, all else being the same,
give the stress distributions shown superimposed
on their counterpart results for the constant F
assumption. There is no appreciable difference in
the normal stress distribution, but the shear stress
distribution is much improved and more as
would be expected considering the shear surface
geometry, i.e. peak shear stress occurs where the
63
shear surface geometry has the sharpest change in
direction. The computed factors of safety along
the shear surface are shown in Figs 3(f) and 4(f)
for the conditions analysed. The computed factor
of safety in the 12” material for the drawdown
condition is 1.01. It is higher in other materials.
The shear surface under the berm has a factor of
safety of 3.33. For this sample problem, the
results indicate that the local shear failure should
occur in the 12” material. Once this happens,
movement should occur in the 12” material which
will cause tensile stresses in the materials above
the back rest portion of the shear surface. Thus,
the tension cracks at the crest of the dam are a
consequence of internal failure in the 12” material
and define the extent of the active wedge. Since
the factor of safety along the shear surface under
the berm area is substantially higher than 1.0,
gross movement along the shear surface could
not have occurred. Filling of the reservoir,
without any repair work, should only improve
the stability of the embankment because of the
buttressing effect of the water.
CONCLUSIONS
The use of a variable factor of safety in slope
stability analysis of problems of practical impor-
tance is generally of considerable interest in
seeking answers to questions that influence design
decisions. The proposed procedure provides a
means to calculate variations in the factor of
safety along a shear surface within the framework
of the limit equilibrium method. The choice of the
characteristic function g(x) defining the relative
safety factor along a shear surface is a dinicult
issue and needs further study. The proposed pro-
cedure of making g(x) equal to the ratio of shear
the stress to effective normal stress for purely fric-
tional materials is reasonable. The resulting shear
stress distribution along the shear surface is gen-
erally smooth and more likely to occur in nature
than that implied by the constant factor of safety
assumption.
REFERENCES
Bishop, A. W. (1967). Progressive failure--with
special
reference to the mechanism causing it. Proc. Geo-
technical Conf: Oslo 2, 142-154.
Bishop, A. W. (1971). The influence of progressive
failure on the choice of the method of stability
analysis. G&technique 21, No. 2, 168-172.
Chowdhury, R. N. (1978). Slope analysis. New York:
Elsevier.
Chugh, A. K. (1981). User informarion manual for slope
stability analysis program SSTABZ. Denver: Engin-
eering and Research Center, US Bureau of Recla-
mation.
Chugh, A. K. (1982). Slope stability analysis for earth-
quakes. Int. J. Numer. Analyt. Merh. Geomech. 6,
No. 3,307-322.
64 CHUGH

Chugh, A. K. (1984). Variable intersliceforce inclination of the stability of general slip surfaces. GCotechnique
in slope stability analysis. Internal Report, Engineer- 15, No. 1, 79-93.
ing and Research Center, Bureau of Reclamation, Sarma, S. K. (1973). Stability analysis of embankments
Denver. and slopes. Giotechnique 23, No. 3,423-433.
Janbu, N. (1973). Soil stability computations. In Spencer, E. (1967). A method of analysis of the stability
Embankment dam engineering, Casagrande volume, of embankments assuming parallel interslice forces.
pp. 47787 (eds R. C. Hirschfeld and S. J. Poulos). G&echnique 17, No. 1, 1 l-26.
New York: Wiley. Spencer, E. (1973). Thrust line criterion in embankment
Morgenstern, N. R. & Price, V. E. (1965). The analysis stability analysis. G&technique 23, No. 1, 85-100.