Professional Documents
Culture Documents
& CHAMEAU,
J.-L. (1982). Gtotrchniqur 32, No. 1. 3140
A general method of three-dimensional slope stability limit equilibrium methods. The failure is assumed
analysis using the limit equilibrium concept is proposed. to occur along a known surface and a global factor
The failure mass is assumed to be symmetrical and of safety is then obtained by comparing the shear
divided into many vertical columns. The inter-slice forces
strength available with the shear strength required
have the same inclination throughout the mass, and the
for equilibrium. This simple theoretical approach
inter-column shear forces are parallel to the base of the
column and function of their positions. Force and has been used for several decades. It takes into
moment equilibria are satisfied for each column as well as account all of the major factors that influence the
for the total mass. The analysis is illustrated for several shearing resistance of an earth mass and is con-
slope angles, soil parameters and pore water pressure sidered to be reliable by most practitioners.
conditions. The factors of safety are compared with those Most stability methods are two-dimensional
obtained with two-dimensional stability analyses and the and assume plane strain conditions. Among these
ordinary method of columns. In addition, a three- methods, the methods of slices (Bishop, 1955;
dimensional finite element computer program is
Fellenius, 1936; Janbu, 1957; Morgenstern &
presented. This program can calculate the local factors of
Price, 1965; Spencer, 1967) are the most commonly
safety at selected points along a failure surface as well as
the mean factor of safety for a chosen failure mass. The used because they can handle complex geometries
results obtained by the two methods, limit equilibrium and variable soil and water conditions. Although
and finite element, are compared. there are more than a dozen methods of slices, only
L’article propose une mkthode g&n&ale pour I’analyse
a few three-dimensional methods have been
tridimensionnelle de la stabilitt: des talus baste sur I’idie proposed to study the end-effects which occur in
d’kquilibre limit,. On admet que le volume de rupture est actual slides (Baligh & Azzouz, 1975; Hovland,
symmitrique et divist-e en beaucoup de tranches 1977; Sherard, Woodward, Gizienksi & Clevenger,
verticales. Les efforts entre les tranchCs ont la m&me 1963). These methods are limited to simple
inclinaison dans la masse entiere, tandis que les efforts de problems with uncomplicated geometry and soil
clsaillement entre les tranches sont parall&les li la base de and water conditions.
la tranche et fonction de leurs positions. Les Cquilibres The work presented in this Paper provides
des efforts et des moments sont satisfaits pour chaque
engineers with a general methodology for three-
tranche, aussi bien que pour le volume total. L’analyse
est illustrCle pour plusieurs angles de talus. paramttres du
dimensional slope stability analysis. A three-
sol et conditions de pression de l’eau interstitielle. Les dimensional limit equilibrium analysis which
coefficients de st-curitt sont cornparts avec ceux obtenus relaxes the simplifying assumptions made by
par des analyses bidimensionelles de stabilitk et la previous investigators is presented. The analysis is
mCthode ordinaire des tranches. Un programme illustrated for several slope angles, soil parameters
d’ordinateur li elements finis tridimensionnels est aussi and pore-water pressure conditions. In addition, a
prCsentt. Avec ce programme on peut calculer les three-dimensional finite element computer
coefficients locaux de skcuritt en des points selectionn&
program is proposed to analyse the stability of
le long d’une surface de rupture, aussi bien que le
slopes and embankments. The results obtained by
coefficient moyen de s&curit& pour un volume de rupture
choisi. On fait une comparaison entre les rtsultats the limit equilibrium and finite element methods
obtenus par les deux mCthodes (kquilibre limite et are compared.
t-lbments finis).
THREE-DIMENSIONAL LIMIT EQUILIBRIUM
METHOD
INTRODUCTION The failure mass above the failure surface is
The stability of embankments, cut slopes, excava- divided into several vertical columns (Fig. 1). Fig. 2
tions and natural slopes is commonly analysed by shows the free body diagram of a column taken out
from the failure mass. The parameters included in
the analysis are the normal and shear forces acting
Discussion on this Paper closes on 1 June 1983. For
further details see inside back cover. on the sides (X-Y plane), ends (Y-Z plane) and
CCC article-fee code: base of the column, the points of application of
0016~8505,‘83/010031-4OS2.00. these forces and the overall factor of safety F.
*National Taiwan University, Taipei. Several assumptions are necessary to make the
i- Purdue University, West Lafayette, Indiana. problem determinate.
32 CHEN AND CHAMEAU
m columns 0)
Y ___.. _b
1
/G
.
h
v I /
II/
b4,
Fig. 2. Free body diagram of a column Fig. 3. Force system in a column (side view)
THREE-DIMEMKINAL LIMITEQUILIBRIUM ANALYSI~~FSLOPES 33
the cohesion intercept of the soil, h is the width of a examined for a mixed type of failure mass
soil column and r is the inclination of the shear composed of a cylinder in the central portion
surface with respect to the horizontal. attached to two semi-ellipsoids at the ends. The
It is further assumed that the inter-slice forces roller type of failure or spoon-shaped failure are
acting on the column sides have the same particular cases of this mixed type of failure.
inclination 0 throughout the whole failure mass The mathematical expression for an ellipsoid is
tan 0 = Xi, j/Ei. j = Xi_ ,,,/Ei_ ,, j (7)
(8)
where Xi, j and Xi_, j are the vertical shear forces
acting on ends i and i - 1 of a column, respectively, Each cross-section in the X-Y plane is assumed
and Ei, j and Ei_ i, j are the horizontal normal circular (a = h). Fig. 4 shows the free body
forces acting on ends i and i- 1 of a column, diagram of a column taking into account the
respectively. The resultant of these interslice forces previous assumptions. The subscripts c, c$, E, F, 1
is termed Q. The inclination of Q with respect and 2 stand for cohesion, friction, embankment,
to the horizontal is 0 (Fig. 3). . foundation, side 1 and side 2, respectively. The
With these assumptions the remaining method allows for different materials in the em-
unknowns are the horizontal normal stresses E bankment and the foundation. The vertical
acting on the ends of the columns, the normal dimensions of the column above and below the
forces N acting on the base of the columns, the horizontal plane (ground level) are h, (height of
factor of safety F and the angle of inclination 0. embankment) and h,, respectively. The force
The following procedure shows how the equations polygon of the column is given in Fig. 5. Resolving
of equilibrium are solved to obtain the factor of all forces in the X-Y plane leads to
safety F. N’tan b,‘+(c’/F) Ab-Q cos(x-0)
In narrow rotational slides the underlying
-Wsinr+R,cos(r,--)
surface of rupture, together with the exposed
scarps, is spoon-shaped. However, if the slide -R,cos(a-CQ) = 0 (9)
extends for a large distance along the slope the N’+uA,+Qsin(r-O)-IVcosa
failure mass may approach the shape of a cylinder
with axis parallel to the slope (roller type of + R, sin (x2 - rl)
failure). In the present study, the stability is +R, sin(r-x,) = 0 (IO)
hE2
IQ=0 (14)
Fig. 5. Force polygon of a column
The whole system is also in equilibrium with
respect to moment equilibrium. Thus the overall
where c’ is the effective cohesion intercept of the moment about any point 0 must be equal to zero
soil at the base of the column, A, is the base area of (Fig. 6)
the column, N’ is the effective normal force acting
~Qcos(O-~)(.~-/rhp~~~~) = 0 (15)
on the base of the column, u is the pore-water
pressure acting on the base of the column, W is the or
weight of the column, Q is the resultant of the
1 (Qrcos(O-x)-QphQcosrcos(O-a)) = 0
inter-slice forces X and E, R, and R, are the shear
forces acting on the two sides 1 and 2 respectively, (16)
9, ’ is the mobilized effective friction angle, 2, 2, where the value of Qha can be obtained by
and x2 are the inclinations of the base in the X-Y summing all the moments in a column at the centre
plane at the central section, side 1 and side 2 of the base of that column. The derivation of Qh, is
respectively, 0 is the inclination of the resultant Q given in Appendix I.
and F is the factor of safety. In equations (14) and (16) the only unknowns
The effective normal forces N’ can be obtained are the inclination of inter-slice force 0 and the
from equation (10) and substituted into equation factor of safety F. The problem is now determinate
and can be solved by using the secant method for
non-linear equations (Wolfe, 1959). A computer
program, LEMIX, has been developed to perform
the above operations.
,,, 1.16
t \
Embankment
2’ Axisof rotation .?
6 0.96 c
Fig. 7. Front view of a mixed type of failure surface Fig. 9. Ratio FJF, (slope 25/l, r, = 0)
Table 1. Comparison between the factors of safety given by the ordinary method of columns and LEMIX
-I--
set of OMC - LEMIX
L/H
strength
parameter
Relative difference between factors of safety: per cent
! LEMIX
xl00
>
T” = 0.0 r, = @50
1,/H = 0.5 1,/H= 1 1,/H = 2 1,/H = 4 1,/H= 0.5 1,/H = 1 1,/H= 2 1,/H = 4
T
0.5 1 - 16.2 - 13.5 -11.8 - 10.8 -3.1 -0.1 1.9 3.0
2 -9.2 -8.3 -1.6 -7.3 - 3.6 t - 6.0 -5.8 - 5.1
3 -7.5 -7.6 - 7.8 -8.0 0.2 -0.7 - 1.5 -2.2
4 -2.0 -3.1 -4.3 -4.9 3.8 2.2 0.8 -0.2
5 0.1 - 1.7 -3.3 -4.5 4.5 2.3 0.1 - 1.8
41 2
-9.8
-6.1
-9.8
-6.2
-9.6
-6.3
-9.8
-6.6
3.8
-4.3
4.0
-4.7
4.1
-4.7
4.2
-4.9
3 -6.9 - 7.0 -1.2 -1.6 - 1.0 -1.2 - 1.6 - 1.9
4 -3.8 -4.1 -4.4 -4.1 0.7 0.6 0.2 -0.3
defined as
F, = (c + oNtan d)/~~ (18)
where g’N is the normal stress and ~~ the shear
stress. The normal stress, shear stress and local
factors of safety are given in Fig. 1I for different
sections of the failure surface (as a function of the Z
co-ordinate). Fig. 11 shows that the normal stress
is higher in the central portion of the embankment
and is very small at the two ends. The shear stress
distribution is similar to the normal stress
distribution. The maximum shear stresses are only
about 207” of the maximum normal stresses. As the
section is farther away from the centre line, both
the normal and shear stresses decrease at the same
rate and the local factor of safety increases. The
observed discontinuities of F, are due to the
2.2 discontinuity of stresses at the toe.
UC To compare the finite element results with the
,z
m factors of safety obtained with the limit
% 1.8- equilibrium methods, a mean factor of safety F is
6 defined from the local factors of safety as
6 -
ti
f 1.4- F _ C(c+%tan4)dA (19)
2 - Toe
Cs,dA
l.O-. where the summation 1 is over the whole failure
surface and dA is the bottom area of a vertical
column.
Distance
from toex: m
The results are presented in Table 3. The limit
Fig. II. Normal stress, shear stress and local factor of
safety
equilibrium methods, Spencer’s method and LEMIX
yield factors of safety F, and F, of 1.59 and 1.90
respectively. The two-dimensional finite element
values are then used in the next increment. The computer program ISBILDgives a mean factor of
program is capable of handling embankments on safety F, of 1.62; FESPON leads to a mean factor of
rigid or compressible foundations. safety F, of 2.01. The ratio FJF, is 1.20 for the
The stability analyses were performed for an limit equilibrium methods and 1.24 for the finite
embankment of height 9m, and slope of 1,5/l in element solutions. The factors of safety obtained
as-compacted condition. The soil was assumed from limit equilibrium analyses are smaller than
homogeneous in both embankment and those from finite element analyses. The agreement
foundation. The foundation and embankment is good in this case with differences of 1,8’i,, and
were divided into two and eight layers respectively. 5.50, in two-dimensional and three-dimensional
The parameters of the hyperbolic soil model used cases respectively.
in FESPoN were generated from the results of
conventional triaxial and consolidation tests on CONCLUSIONS
highly plastic St Crois clay (Table 2). Local factors This study was directed at developing
of safety were computed along a spoon-shaped techniques of three-dimensional slope stability
failure surface defined by the critical circle analysis and comparing the results obtained using
obtained by STABLZand a minor axis of length these techniques with those given by conventional
12.2 m (40 ft). The local factor of safety F, is two-dimensional methods. The computer program
LEMIXbased on the limit equilibrium method was
Table 2. Hyperbolic parameters for as-compacted Table 3. Comparison of factors of safety obtained with
condition* the finite element and limit equilibrium methods
Spencer, E. (1967). A method of analysis of the stability of Wolfe, P. (1959). The secant method for simultaneous
embankments assuming parallel interslice forces. nonlinear equations. Communs Ass. Comput. Mach. 2,
GPotechnique 17, No. 1, 1 l-26. 12-13.
Wilson. E. G. (1971).SOLID SAP. a static ana/vsis orouram Wright, S. G., Kulhawv, F. H. & Duncan, J. M. (1973).
fir three-dimksional s&l structures.’ Structural Accuracy of equilibrium slope stability analysis. J.
Engineering Laboratory, University of California, Soil Mech. Fdns Div. Am Sot. Cit. Enyrs 99, SM 10,
SESM Report 71-19. 783-791.
Wilson, E. L., Taylor, R. L., Doherty, W. P. &
Ghaboussi, J. (1973). Incompatible displacement
modes. In Numerical and computer methods in struc-
tural mechanics (eds S. T. Ferves et al.), pp. 43357.
London: Academic.