You are on page 1of 6

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

6, 835±840

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 {

KEYWORDS: embankments; landslides; limit state ing Group, 1995). For given element shape
design analysis; numerical modelling and analysis; functions, the set of algebraic equations solved by
plasticity; slopes. FLAC is identical to that solved with the ®nite
element method. However, in FLAC, this set of
equations is solved using dynamic relaxation (Otter
INTRODUCTION et al., 1966), an explicit, time-marching procedure
For slopes, the factor of safety F is traditionally in which the full dynamic equations of motion are
de®ned as the ratio of the actual soil shear strength integrated step by step. Static solutions are ob-
to the minimum shear strength required to prevent tained by including damping terms that gradually
failure (Bishop, 1955). As Duncan (1996) points remove kinetic energy from the system.
out, F is the factor by which the soil shear The convergence criterion for FLAC is the nodal
strength must be divided to bring the slope to the unbalanced force, the sum of forces acting on a
verge of failure. Since it is de®ned as a shear node from its neighbouring elements. If a node is
strength reduction factor, an obvious way of com- in equilibrium, these forces should sum to zero.
puting F with a ®nite element or ®nite difference For this study, the unbalanced force of each node
program is simply to reduce the soil shear strength was normalized by the gravitational body force
until collapse occurs. The resulting factor of safety acting on that node. A simulation was considered
is the ratio of the soil's actual shear strength to the to have converged when the normalized unbalanced
reduced shear strength at failure. This `shear force of every node in the mesh was less than
strength reduction technique' was used as early as 10ÿ3 .
1975 by Zienkiewicz et al. (1975), and has since
been applied by Naylor (1982), Donald & Giam
(1988), Matsui & San (1992), Ugai (1989), Ugai & LIMIT ANALYSIS SOLUTION
Leshchinsky (1995) and others. A limit analysis, upper bound solution for the
The shear strength reduction technique has a stability of a homogeneous embankment was de-
number of advantages over the method of slices rived by Chen (1975), assuming a log-spiral failure
for slope stability analysis. Most importantly, the surface. Chen's solution was extended to include
critical failure surface is found automatically. Ap- the effects of pore pressure by Michalowski
plication of the technique has been limited in the (1995a, 1995b). The solutions are presented in the
past due to the long computer run times required. form of dimensionless stability numbers NS given
But with the increasing speed of desktop com- by
puters, the technique is becoming a reasonable ã
alternative to the method of slices, and is being NS ˆ HC (1)
c
used increasingly in engineering practice. However,
there has been little investigation of the accuracy where H C is the critical height for a slope with
of the technique. In this paper, factors of safety soil unit weight ã and cohesion c. Values of NS
obtained with the shear strength reduction tech- are tabulated for various values of the slope angle
nique are compared to limit analysis solutions for â (from the horizontal), the friction angle ö and
a homogeneous embankment. the pore pressure coef®cient ru . The pore pressure
coef®cient, introduced by Bishop (1954), speci®es
the pore pressure as a fraction of the overburden
THE EXPLICIT FINITE DIFFERENCE METHOD stress. The pore pressure u at a depth z below the
Factors of safety are computed using the expli- ground surface is given by
cit±®nite±difference code, FLAC (Itasca Consult- u ˆ ru ãz (2)

Manuscript received 9 Feb 1999; revised manuscript


accepted 13 Aug 1999.
Discussion on this paper closes 30 June 2000; for further EXAMPLE STABILITY ANALYSIS
details see p. ii. Consider a homogeneous embankment of height
 Dames & Moore, Los Angeles. H ˆ 10 m, sloping at angle ⠈ 458 with friction
{ University of Minnesota, Minneapolis. angle ö ˆ 208, unit weight 㠈 20 kN=m3 , cohe-

835

Downloaded by [ UC San Diego Libraries] on [24/04/17]. Copyright © ICE Publishing, all rights reserved.
836 DAWSON, ROTH AND DRESCHER

sion c ˆ 12:38 kPa and with ru ˆ 0:0. With these 0.5

Normalized unbalanced force


soil properties the slope has a factor of safety of
exactly 1´0, according to the limit analysis solution 0.4
of Chen (1975). 0.3
The embankment is simulated with FLAC in
plane strain, using small-strain mode (the coordi- 0.2
nates of the nodes are not updated according to the
computed nodal displacements). The soil is mod- 0.1
elled as a linear elastic±perfectly plastic material
0.0
with a Mohr±Coulomb yield condition and an
associated ¯ow rule. The numerical mesh, shown 20.1
in Fig. 1, is 20 elements wide and 20 elements 0.80 0.85 0.90 0.95 0.100 1.05 1.10 1.15
high. Horizontal displacements are ®xed for nodes Trial factor of safety
along the left and right boundaries while both
horizontal and vertical displacements are ®xed Fig. 2. Unbalanced force as the trial factor of safety is
increased in small steps
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 bring the slope to failure. As the strength is
c and ö adjusted according to the equations reduced further, the unbalanced force continues to
1 increase, in an approximately linear fashion. Thus,
ctrial ˆ c (3) the strength reduction technique gives a factor of
F trial safety between 1´02 and 1´03. The velocity ®eld at
 
1 collapse (F trial ˆ 1:03) is shown in Fig. 3, along
ötrial ˆ arctan tanö (4)
F trial with the critical log-spiral failure surface from the
limit analysis solution.
Figure 2 shows the normalized unbalanced force The sharp break in the unbalanced force in Fig.
obtained as the shear strength is reduced in small 2 shows that there is no ambiguity in identifying
steps, starting from a value of F trial ˆ 0:8. Note the trial factor of safety at which the slope fails.
that the soil shear strength is decreased by increas- This is a consequence of using a linear elastic±
ing F trial . For trial factors of safety up to 1´02 the perfectly plastic constitutive model, a model with a
simulations converge to equilibrium with an un- sudden transition from elastic to plastic behaviour.
balanced force of the order of 10ÿ6 . However, Some previous applications of the strength reduc-
when F trial is increased to 1´03 the simulations no tion technique, such as that of Matsui & San
longer converge, with the unbalanced force exceed- (1992), have used a hyperbolic constitutive model
ing 10ÿ3 . It might seem somewhat paradoxical that (Duncan & Chang, 1970), a model which exhibits
collapse occurs as F trial is increased. But this a smooth transition from elastic behaviour to plas-
follows from the de®nition of F as the factor by tic behaviour. Identifying the limit state is more
which the soil shear strength must be divided to dif®cult when using such a model.

γ 5 20 kN/m3
φ 5 20°
10 m
c 5 12.38 kPa

45°

Fig. 1. Numerical mesh for homogeneous embankment

Downloaded by [ UC San Diego Libraries] on [24/04/17]. Copyright © ICE Publishing, all rights reserved.
SLOPE STABILITY ANALYSIS 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 differ-
ence between upper and lower brackets was less
than 4 3 10ÿ4 . The factor of safety reported is the
average of the ®nal upper and lower brackets.
Factors of safety computed for ru ˆ 0:0 are pre-
sented in Table 1, while those for ru ˆ 0:25 are
presented in Table 2 and those for ru ˆ 0:5 are
Critical log-spiral presented in Table 3. The strength-reduction factors
surface 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.
Fig. 3. Velocity ®eld at collapse, along with critical As would be expected, the ®ne mesh gives better
log-spiral surface results than the coarse mesh.
Two exceptions to the close agreement with the
limit analysis solutions are for vertical slopes
( ⠈ 908) with ru ˆ 0:25 and ru ˆ 0:50 (see the
The value of F trial at which collapse occurs can be last row of Tables 2 and 3). Here the simulations
found more ef®ciently using bracketing and bisec- fail at the toe of the slope due to the combination
tion. First, upper and lower brackets are estab- of high pore pressure (u ˆ ru ã H) and zero hor-
lished. The initial lower bracket is any F trial for izontal total stress. For instance, for ru ˆ 0:50 the
which a simulation converges. The initial upper pore pressure at the slope face at the toe exceeds
bracket is any F trial for which the simulation does the soil tensile strength.
not converge. Next, a point midway between the For graphical comparison, it is convenient to
upper and lower brackets is tested. If the simula- express the numerical results in terms of stability
tion converges, the lower bracket is replaced by numbers. The stability number corresponding to a
this new value. If the simulation does not con- computed factor of safety can be found using
verge, the upper bracket is replaced. The process is equations (1) and (3). Stability numbers are plotted
repeated until the difference between upper and against soil friction angle in Figs 4 and 5 for the
lower brackets is less than a speci®ed tolerance. ru ˆ 0:0 case (Table 1). Fig. 4 shows strength-
reduction results for the coarse mesh, while Fig. 5
shows results for the ®ne mesh. The strength-
BENCHMARK STUDY reduction results appear to converge to the limit-
To assess the accuracy of the strength reduction analysis solution as the mesh is re®ned.
technique, simulations were performed for a wide
range of parameters. Embankments were simulated
with slope angles â ranging from 158 to 908. The
soil was given values of ö ranging from 108 to 408 CONCLUSION
and values of ru of 0´0, 0´25 and 0´5. For each Slope stability factors of safety computed with
combination of â, ö and ru, the stability numbers the strength reduction technique have been com-
given by Chen (1975) or Michalowski (1995a) pared to an upper-bound limit analysis solution
were used to compute a combination of slope based on a log-spiral failure mechanism. Analyses
height, cohesion and unit weight such that the were performed for a wide range of slope angles,
embankment would have a factor of safety of soil friction angles and pore pressure coef®cients.
exactly 1´0. Strength-reduction factors of safety were within a
In practice, the Mohr±Coulomb yield condition few per cent of the limit analysis solution, provided
is often combined with a tension cut-off (Chen, the numerical mesh was suf®ciently re®ned.
1975). For comparison with the limit analysis solu- Strength-reduction results were generally slightly
tion, a tension cut-off was not used here. Instead, higher than those predicted by limit analysis. The
the Mohr±Coulomb yield condition was assumed close agreement obtained does not, strictly speak-
to be valid in the tensile normal stress domain. In ing, demonstrate the accuracy of the strength re-
other words, the failure envelope intersects the duction technique, since the limit analysis solution
normal stress axis at ÿc=tan ö. If a zero-tension is an upper bound. Nevertheless, the similarity
cut-off is used, such that the uniaxial tensile between the two solutions, obtained by entirely
strength of the soil is zero, tensile failure occurs at different methods, supports the widely held view
the crest of the embankment and the computed that the log-spiral solution is, in effect, an exact
factor of safety is a few per cent lower. solution. If this were true, it would then be reason-
Simulations were run for both a ®ne mesh able to expect the numerical solution to be slightly

Downloaded by [ UC San Diego Libraries] on [24/04/17]. Copyright © ICE Publishing, all rights reserved.
838 DAWSON, ROTH AND DRESCHER

Table 1. Computed factors of safety (ru ˆ 0´0)


Slope Friction Stability number Factor of safety
angle, angle, N S ˆ H C (ã=c)
â ö (from Chen, 1975) Fine mesh Coarse mesh
15 5 14´38 1´002 1´023
10 45´49 1´003 1´027
30 10 13´50 1´006 1´034
15 21´69 1´008 1´027
20 41´22 1´010 1´033
45 10 9´31 1´001 1´019
20 16´16 1´006 1´026
30 35´54 1´006 1´031
40 185´49 1´008 1´008
60 10 7´26 1´003 1´026
20 10´39 1´009 1´035
30 16´04 1´012 1´044
40 28´91 1´011 1´036
75 10 5´80 1´006 1´035
20 7´48 1´014 1´046
30 9´94 1´019 1´061
40 13´97 1´017 1´070
90 10 4´58 1´015 1´051
20 5´50 1´024 1´063
30 6´69 1´031 1´081
40 8´29 1´039 1´105

Table 2. Computed factors of safety (ru ˆ 0´25)


Slope Friction Stability number Factor of Safety
angle, angle, N S ˆ H C (ã=c)
â ö (from Michalowski, Fine mesh Coarse mesh
1995a)
15 10 23´18 1´001 1´026
30 10 10´71 1´002 1´026
20 20´05 1´010 1´030
30 53´90 1´013 1´040
45 10 7´95 1´000 1´019
20 10´94 1´006 1´029
30 15´84 1´010 1´033
40 25´70 1´007 1´035
60 10 6´38 1´001 1´026
20 7´69 1´010 1´039
30 9´26 1´015 1´048
40 11´23 1´012 1´060
75 10 5´18 1´006 1´038
20 5´76 1´014 1´050
30 6´27 1´022 1´067
40 6´65 1´016 1´090
90 10 4´14 1´000 1´047
20 4´35 1´001 1´051
30 4´40 1´013 1´066
40 4´26  
 See comments in text.

Downloaded by [ UC San Diego Libraries] on [24/04/17]. Copyright © ICE Publishing, all rights reserved.
SLOPE STABILITY ANALYSIS 839
Table 3. Computed factors of safety (ru ˆ 0´50)
Slope Friction Stability number Factor of Safety
angle, angle, NS ˆ H C (ã=c)
â ö (from Michalowski, Fine mesh Coarse mesh
1995a)
15 10 14´03 1´004 1´026
30 10 8´70 0´997 1´019
20 12´07 1´008 1´029
30 18´00 1´013 1´038
45 10 6´88 0´998 1´015
20 7´99 1´008 1´030
30 9´32 1´011 1´034
40 11´06 1´013 1´041
60 10 5´68 1´000 1´026
20 6´02 1´001 1´038
30 6´26 1´016 1´051
40 6´43 1´014 1´068
75 10 4´68 1´000 1´034
20 4´66 1´013 1´050
30 4´50 1´021 1´069
40 4´21 1´013 1´092
90 10 3´77 0´999 1´038
20 3´58 0´992 1´033
30 3´26 0´959 1´001
40 2´82  
 See comments in text.

50
β 5 15 Strength reduction

40 β 5 30 Limit analysis

β 5 45
Stability number

30 β 5 60

20

β 5 75
10 β 5 90

0
0 10 20 30 40 50
Friction angle: degrees

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


ru ˆ 0´0 case. Coarse mesh

higher, with the difference decreasing as the mesh initial stress and the stress path to be ignored.
is re®ned. These have no effect on the collapse load for an
An associated plastic ¯ow rule was used in this associated material (Chen, 1975). However, for
paper so that results could be compared to limit more realistic soil models, such as those with non-
analysis solutions. The use of an associated ¯ow associated ¯ow rules, these factors cannot be
rule allowed the effects of elastic constants, the ignored.

Downloaded by [ UC San Diego Libraries] on [24/04/17]. Copyright © ICE Publishing, all rights reserved.
840 DAWSON, ROTH AND DRESCHER
50
β 5 15 Strength reduction

40 β 5 30 Limit analysis

β 5 45

Stability number
30 β 5 60

20

β 5 75
10 β 5 90

0
0 10 20 30 40 50
Friction angle: degrees

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


ru ˆ 0´0 case. Fine mesh

ACKNOWLEDGEMENT analysis of continua, version 3.3. Itasca Consulting


The authors are grateful for the suggestions of Group, Minneapolis, Minnesota, USA.
Dr P. A. Cundall of Itasca Consulting Group, Dr Matsui, T. & San, K. C. (1992). Finite element slope
A. Delnik of Edison International and Dr S. Inel of stability analysis by shear strength reduction tech-
Dames & Moore. nique. Soils and Foundations 32, No. 1, 59±70.
Michalowski, R. L. (1995a). Stability of slopes: limit
analysis approach. Rev. Eng. Geol. 10, 51±62.
Michalowski, R. L. (1995b). Slope stability analysis: a
REFERENCES kinematical approach. GeÂotechnique 45, No. 2, 283±
Bishop, A. W. (1954). The use of pore pressure coef®- 293.
cients in practice. GeÂotechnique 4, 148±152. Naylor, D. J. (1982). Finite elements and slope stability.
Bishop, A. W. (1955). The use of the slip circle in the Numer. Meth. in Geomech., Proc. NATO Advanced
stability analysis of slopes. GeÂotechnique 5, 7±17. Study Institute, Lisbon, Portugal, 1981, pp. 229±244.
Chen, W. F. (1975). Limit analysis and soil plasticity. Otter, J. R. H., Cassell, A. C. & Hobbs, R. E. (1966).
Amsterdam: Elsevier. Dynamic relaxation. Proc. Instn Civ. Engrs 35, Paper
Donald, I. B. & Giam, S. K. (1988). Application of the No. 6986, 633±656.
nodal displacement method to slope stability analysis. Ugai, K. (1989). A method of calculation of total factor
Proceedings of the 5th Australia±New Zealand confer- of safety of slopes by elasto-plastic FEM. Soils and
ence on geomechanics, Sydney, Australia, 456±460. Foundations 29, No. 2, 190±195 (in Japanese).
Duncan, J. M. (1996). State of the art: limit equilibrium Ugai, K. & Leshchinsky, D. (1995). Three-dimensional
and ®nite-element analysis of slopes. J. Geotech. limit equilibrium and ®nite element analyses: a com-
Engn Div. Am. Soc. Civ. Engrs 122, No. 7, 577±596. parison of results. Soils and Foundations 35, No. 4,
Duncan, J. M. & Chang, C. Y. (1970). Nonlinear analy- 1±7.
sis of stress and strain in soils. J. Soil Mech. and Zienkiewicz, O. C., Humpheson, C. & Lewis, R. W.
Found. Div. Am. Soc. Civ. Engrs 96, No. SM5, (1975). Associated and non-associated visco-plasticity
1629±1653. and plasticity in soil mechanics. GeÂotechnique 25,
Itasca Consulting Group (1995). FLAC, fast Lagrangian No. 4, 671±689.

Downloaded by [ UC San Diego Libraries] on [24/04/17]. Copyright © ICE Publishing, all rights reserved.

You might also like