You are on page 1of 6

Dawson, E. M., Roth, W. H. & Drescher, A. (1999). Geotechnique 49, No.

6, 835840

Slope stability analysis by strength reduction


E . M . DAW S O N,  W. H . ROT H  a n d A . D R E S C H E R {
ing Group, 1995). For given element shape
functions, the set of algebraic equations solved by
FLAC is identical to that solved with the nite
element method. However, in FLAC, this set of
equations is solved using dynamic relaxation (Otter
et al., 1966), an explicit, time-marching procedure
in which the full dynamic equations of motion are
integrated step by step. Static solutions are obtained by including damping terms that gradually
remove kinetic energy from the system.
The convergence criterion for FLAC is the nodal
unbalanced force, the sum of forces acting on a
node from its neighbouring elements. If a node is
in equilibrium, these forces should sum to zero.
For this study, the unbalanced force of each node
was normalized by the gravitational body force
acting on that node. A simulation was considered
to have converged when the normalized unbalanced
force of every node in the mesh was less than
103 .

KEYWORDS: embankments; landslides; limit state


design analysis; numerical modelling and analysis;
plasticity; slopes.

INTRODUCTION

For slopes, the factor of safety F is traditionally


dened as the ratio of the actual soil shear strength
to the minimum shear strength required to prevent
failure (Bishop, 1955). As Duncan (1996) points
out, F is the factor by which the soil shear
strength must be divided to bring the slope to the
verge of failure. Since it is dened as a shear
strength reduction factor, an obvious way of computing F with a nite element or nite difference
program is simply to reduce the soil shear strength
until collapse occurs. The resulting factor of safety
is the ratio of the soil's actual shear strength to the
reduced shear strength at failure. This `shear
strength reduction technique' was used as early as
1975 by Zienkiewicz et al. (1975), and has since
been applied by Naylor (1982), Donald & Giam
(1988), Matsui & San (1992), Ugai (1989), Ugai &
Leshchinsky (1995) and others.
The shear strength reduction technique has a
number of advantages over the method of slices
for slope stability analysis. Most importantly, the
critical failure surface is found automatically. Application of the technique has been limited in the
past due to the long computer run times required.
But with the increasing speed of desktop computers, the technique is becoming a reasonable
alternative to the method of slices, and is being
used increasingly in engineering practice. However,
there has been little investigation of the accuracy
of the technique. In this paper, factors of safety
obtained with the shear strength reduction technique are compared to limit analysis solutions for
a homogeneous embankment.

LIMIT ANALYSIS SOLUTION

A limit analysis, upper bound solution for the


stability of a homogeneous embankment was derived by Chen (1975), assuming a log-spiral failure
surface. Chen's solution was extended to include
the effects of pore pressure by Michalowski
(1995a, 1995b). The solutions are presented in the
form of dimensionless stability numbers NS given
by

NS HC
(1)
c

where H C is the critical height for a slope with


soil unit weight and cohesion c. Values of NS
are tabulated for various values of the slope angle
(from the horizontal), the friction angle and
the pore pressure coefcient ru . The pore pressure
coefcient, introduced by Bishop (1954), species
the pore pressure as a fraction of the overburden
stress. The pore pressure u at a depth z below the
ground surface is given by

THE EXPLICIT FINITE DIFFERENCE METHOD

Factors of safety are computed using the explicitnitedifference code, FLAC (Itasca Consult-

u ru z

Manuscript received 9 Feb 1999; revised manuscript


accepted 13 Aug 1999.
Discussion on this paper closes 30 June 2000; for further
details see p. ii.
 Dames & Moore, Los Angeles.
{ University of Minnesota, Minneapolis.

EXAMPLE STABILITY ANALYSIS

(2)

Consider a homogeneous embankment of height


H 10 m, sloping at angle 458 with friction
angle 208, unit weight 20 kN=m3 , cohe835

836

sion c 12:38 kPa and with ru 0:0. With these


soil properties the slope has a factor of safety of
exactly 10, according to the limit analysis solution
of Chen (1975).
The embankment is simulated with FLAC in
plane strain, using small-strain mode (the coordinates of the nodes are not updated according to the
computed nodal displacements). The soil is modelled as a linear elasticperfectly plastic material
with a MohrCoulomb yield condition and an
associated ow rule. The numerical mesh, shown
in Fig. 1, is 20 elements wide and 20 elements
high. Horizontal displacements are xed for nodes
along the left and right boundaries while both
horizontal and vertical displacements are xed
along the bottom boundary.
To perform slope stability analysis with the
shear strength reduction technique, simulations are
run for a series of trial factors of safety F trial with
c and adjusted according to the equations
1
c
F trial


1
arctan
tan
F trial

ctrial

(3)

trial

(4)

Figure 2 shows the normalized unbalanced force


obtained as the shear strength is reduced in small
steps, starting from a value of F trial 0:8. Note
that the soil shear strength is decreased by increasing F trial . For trial factors of safety up to 102 the
simulations converge to equilibrium with an unbalanced force of the order of 106 . However,
when F trial is increased to 103 the simulations no
longer converge, with the unbalanced force exceeding 103 . It might seem somewhat paradoxical that
collapse occurs as F trial is increased. But this
follows from the denition of F as the factor by
which the soil shear strength must be divided to

Normalized unbalanced force

DAWSON, ROTH AND DRESCHER

0.5
0.4
0.3
0.2
0.1
0.0
20.1
0.80

0.85

0.90

1.10

1.15

Fig. 2. Unbalanced force as the trial factor of safety is


increased in small steps

bring the slope to failure. As the strength is


reduced further, the unbalanced force continues to
increase, in an approximately linear fashion. Thus,
the strength reduction technique gives a factor of
safety between 102 and 103. The velocity eld at
collapse (F trial 1:03) is shown in Fig. 3, along
with the critical log-spiral failure surface from the
limit analysis solution.
The sharp break in the unbalanced force in Fig.
2 shows that there is no ambiguity in identifying
the trial factor of safety at which the slope fails.
This is a consequence of using a linear elastic
perfectly plastic constitutive model, a model with a
sudden transition from elastic to plastic behaviour.
Some previous applications of the strength reduction technique, such as that of Matsui & San
(1992), have used a hyperbolic constitutive model
(Duncan & Chang, 1970), a model which exhibits
a smooth transition from elastic behaviour to plastic behaviour. Identifying the limit state is more
difcult when using such a model.

5 20 kN/m3
5 20

10 m

0.95 0.100 1.05

Trial factor of safety

c 5 12.38 kPa
45

Fig. 1. Numerical mesh for homogeneous embankment

SLOPE STABILITY ANALYSIS

Critical log-spiral
surface

Fig. 3. Velocity eld at collapse, along with critical


log-spiral surface

The value of F trial at which collapse occurs can be


found more efciently using bracketing and bisection. First, upper and lower brackets are established. The initial lower bracket is any F trial for
which a simulation converges. The initial upper
bracket is any F trial for which the simulation does
not converge. Next, a point midway between the
upper and lower brackets is tested. If the simulation converges, the lower bracket is replaced by
this new value. If the simulation does not converge, the upper bracket is replaced. The process is
repeated until the difference between upper and
lower brackets is less than a specied tolerance.
BENCHMARK STUDY

To assess the accuracy of the strength reduction


technique, simulations were performed for a wide
range of parameters. Embankments were simulated
with slope angles ranging from 158 to 908. The
soil was given values of ranging from 108 to 408
and values of ru of 00, 025 and 05. For each
combination of , and ru, the stability numbers
given by Chen (1975) or Michalowski (1995a)
were used to compute a combination of slope
height, cohesion and unit weight such that the
embankment would have a factor of safety of
exactly 10.
In practice, the MohrCoulomb yield condition
is often combined with a tension cut-off (Chen,
1975). For comparison with the limit analysis solution, a tension cut-off was not used here. Instead,
the MohrCoulomb yield condition was assumed
to be valid in the tensile normal stress domain. In
other words, the failure envelope intersects the
normal stress axis at c=tan . If a zero-tension
cut-off is used, such that the uniaxial tensile
strength of the soil is zero, tensile failure occurs at
the crest of the embankment and the computed
factor of safety is a few per cent lower.
Simulations were run for both a ne mesh

837

(60 3 60 elements), and a coarse mesh (20 3 20


elements). Factors of safety were computed by
successive bracketing and bisection until the difference between upper and lower brackets was less
than 4 3 104 . The factor of safety reported is the
average of the nal upper and lower brackets.
Factors of safety computed for ru 0:0 are presented in Table 1, while those for ru 0:25 are
presented in Table 2 and those for ru 0:5 are
presented in Table 3. The strength-reduction factors
of safety are generally within a few per cent of the
limit analysis solutions. The difference is greatest
for steeper slopes and for higher friction angles.
As would be expected, the ne mesh gives better
results than the coarse mesh.
Two exceptions to the close agreement with the
limit analysis solutions are for vertical slopes
( 908) with ru 0:25 and ru 0:50 (see the
last row of Tables 2 and 3). Here the simulations
fail at the toe of the slope due to the combination
of high pore pressure (u ru H) and zero horizontal total stress. For instance, for ru 0:50 the
pore pressure at the slope face at the toe exceeds
the soil tensile strength.
For graphical comparison, it is convenient to
express the numerical results in terms of stability
numbers. The stability number corresponding to a
computed factor of safety can be found using
equations (1) and (3). Stability numbers are plotted
against soil friction angle in Figs 4 and 5 for the
ru 0:0 case (Table 1). Fig. 4 shows strengthreduction results for the coarse mesh, while Fig. 5
shows results for the ne mesh. The strengthreduction results appear to converge to the limitanalysis solution as the mesh is rened.

CONCLUSION

Slope stability factors of safety computed with


the strength reduction technique have been compared to an upper-bound limit analysis solution
based on a log-spiral failure mechanism. Analyses
were performed for a wide range of slope angles,
soil friction angles and pore pressure coefcients.
Strength-reduction factors of safety were within a
few per cent of the limit analysis solution, provided
the numerical mesh was sufciently rened.
Strength-reduction results were generally slightly
higher than those predicted by limit analysis. The
close agreement obtained does not, strictly speaking, demonstrate the accuracy of the strength reduction technique, since the limit analysis solution
is an upper bound. Nevertheless, the similarity
between the two solutions, obtained by entirely
different methods, supports the widely held view
that the log-spiral solution is, in effect, an exact
solution. If this were true, it would then be reasonable to expect the numerical solution to be slightly

838

DAWSON, ROTH AND DRESCHER

Table 1. Computed factors of safety (ru 00)


Slope
angle,

Friction
angle,

Stability number
N S H C (=c)
(from Chen, 1975)

Factor of safety
Fine mesh

Coarse mesh

15

5
10

1438
4549

1002
1003

1023
1027

30

10
15
20

1350
2169
4122

1006
1008
1010

1034
1027
1033

45

10
20
30
40

931
1616
3554
18549

1001
1006
1006
1008

1019
1026
1031
1008

60

10
20
30
40

726
1039
1604
2891

1003
1009
1012
1011

1026
1035
1044
1036

75

10
20
30
40

580
748
994
1397

1006
1014
1019
1017

1035
1046
1061
1070

90

10
20
30
40

458
550
669
829

1015
1024
1031
1039

1051
1063
1081
1105

Table 2. Computed factors of safety (ru 025)


Slope
angle,

Friction
angle,

Stability number
N S H C (=c)
(from Michalowski,
1995a)

15

10

30

Factor of Safety
Fine mesh

Coarse mesh

2318

1001

1026

10
20
30

1071
2005
5390

1002
1010
1013

1026
1030
1040

45

10
20
30
40

795
1094
1584
2570

1000
1006
1010
1007

1019
1029
1033
1035

60

10
20
30
40

638
769
926
1123

1001
1010
1015
1012

1026
1039
1048
1060

75

10
20
30
40

518
576
627
665

1006
1014
1022
1016

1038
1050
1067
1090

90

10
20
30
40

414
435
440
426

1000
1001
1013


1047
1051
1066


 See comments in text.

839

SLOPE STABILITY ANALYSIS

Table 3. Computed factors of safety (ru 050)


Slope
angle,

Friction
angle,

Stability number
NS H C (=c)
(from Michalowski,
1995a)

15

10

30

10
20
30

45

Factor of Safety
Fine mesh

Coarse mesh

1403

1004

1026

870
1207
1800

0997
1008
1013

1019
1029
1038

10
20
30
40

688
799
932
1106

0998
1008
1011
1013

1015
1030
1034
1041

60

10
20
30
40

568
602
626
643

1000
1001
1016
1014

1026
1038
1051
1068

75

10
20
30
40

468
466
450
421

1000
1013
1021
1013

1034
1050
1069
1092

90

10
20
30
40

377
358
326
282

0999
0992
0959


1038
1033
1001


 See comments in text.

50
5 15
5 30

40
Stability number

Strength reduction
Limit analysis
5 45

30

5 60

20
5 75
10

5 90

0
0

10

20
30
Friction angle: degrees

40

50

Fig. 4. Strength-reduction and limit-analysis stability numbers for the


ru 00 case. Coarse mesh

higher, with the difference decreasing as the mesh


is rened.
An associated plastic ow rule was used in this
paper so that results could be compared to limit
analysis solutions. The use of an associated ow
rule allowed the effects of elastic constants, the

initial stress and the stress path to be ignored.


These have no effect on the collapse load for an
associated material (Chen, 1975). However, for
more realistic soil models, such as those with nonassociated ow rules, these factors cannot be
ignored.

840

DAWSON, ROTH AND DRESCHER

50
5 15
5 30

40
Stability number

Strength reduction
Limit analysis
5 45

30

5 60

20
5 75
10

5 90

0
0

10

20
30
Friction angle: degrees

40

50

Fig. 5. Strength-reduction and limit-analysis stability numbers for the


ru 00 case. Fine mesh

ACKNOWLEDGEMENT

The authors are grateful for the suggestions of


Dr P. A. Cundall of Itasca Consulting Group, Dr
A. Delnik of Edison International and Dr S. Inel of
Dames & Moore.

REFERENCES
Bishop, A. W. (1954). The use of pore pressure coefcients in practice. Geotechnique 4, 148152.
Bishop, A. W. (1955). The use of the slip circle in the
stability analysis of slopes. Geotechnique 5, 717.
Chen, W. F. (1975). Limit analysis and soil plasticity.
Amsterdam: Elsevier.
Donald, I. B. & Giam, S. K. (1988). Application of the
nodal displacement method to slope stability analysis.
Proceedings of the 5th AustraliaNew Zealand conference on geomechanics, Sydney, Australia, 456460.
Duncan, J. M. (1996). State of the art: limit equilibrium
and nite-element analysis of slopes. J. Geotech.
Engn Div. Am. Soc. Civ. Engrs 122, No. 7, 577596.
Duncan, J. M. & Chang, C. Y. (1970). Nonlinear analysis of stress and strain in soils. J. Soil Mech. and
Found. Div. Am. Soc. Civ. Engrs 96, No. SM5,
16291653.
Itasca Consulting Group (1995). FLAC, fast Lagrangian

analysis of continua, version 3.3. Itasca Consulting


Group, Minneapolis, Minnesota, USA.
Matsui, T. & San, K. C. (1992). Finite element slope
stability analysis by shear strength reduction technique. Soils and Foundations 32, No. 1, 5970.
Michalowski, R. L. (1995a). Stability of slopes: limit
analysis approach. Rev. Eng. Geol. 10, 5162.
Michalowski, R. L. (1995b). Slope stability analysis: a
kinematical approach. Geotechnique 45, No. 2, 283
293.
Naylor, D. J. (1982). Finite elements and slope stability.
Numer. Meth. in Geomech., Proc. NATO Advanced
Study Institute, Lisbon, Portugal, 1981, pp. 229244.
Otter, J. R. H., Cassell, A. C. & Hobbs, R. E. (1966).
Dynamic relaxation. Proc. Instn Civ. Engrs 35, Paper
No. 6986, 633656.
Ugai, K. (1989). A method of calculation of total factor
of safety of slopes by elasto-plastic FEM. Soils and
Foundations 29, No. 2, 190195 (in Japanese).
Ugai, K. & Leshchinsky, D. (1995). Three-dimensional
limit equilibrium and nite element analyses: a comparison of results. Soils and Foundations 35, No. 4,
17.
Zienkiewicz, O. C., Humpheson, C. & Lewis, R. W.
(1975). Associated and non-associated visco-plasticity
and plasticity in soil mechanics. Geotechnique 25,
No. 4, 671689.

You might also like