388

Two- and three-dimensional analyses of

translational slides in soils with nonlinear failure
envelopes
Hisham T. Eid

Abstract: This paper aims to assess the effect of nonlinearity of the soil shear strength envelope on the sliding mass con-
figurations and the minimum two- and three-dimensional factors of safety for slopes susceptible to translational failure in
both static and seismic conditions. An extensive parametric study was conducted using slope models and nonlinear soil
strength envelopes that simulate translational failure case histories. Analysis of the results led to the development of stabil-
ity charts that do not require an iterative procedure when determining the factors of safety. The introduction of a two-
dimensional stability number and three-dimensional stability function, the values of which depend on the degrees of nonli-
nearity of the involved soils’ strength envelopes, has made it possible to develop such charts. These charts give the prac-
ticing geotechnical engineer a reliable and fast method to analyze translational failures. Using available software in such
an analysis can be complicated if shear resistance along the sliding mass vertical sides and nonlinearity of soil shear
strength envelopes are considered. Numerical examples and a case history are given to verify the reliability and illustrate
the different applications of these charts.
Key words: nonlinear shear strength failure envelope, slope stability analysis, stability charts, translational failure, three-
dimensional analysis, two-dimensional analysis.
Résumé : Cet article évalue l’effet de la non linéarité de l’enveloppe de résistance au cisaillement du sol sur les configura-
tions des masses glissantes, ainsi que les facteurs de sécurité minimaux en deux et trois dimensions pour des pentes sus-
ceptibles de rupture en translation en conditions autant statiques que sismiques. Une vaste étude paramétrique a été
effectuée avec des modèles de pentes et des enveloppes de résistance non linéaires qui simulent des cas historiques de rup-
ture en translation. L’analyse des résultats a permis de développer des chartes de stabilité qui ne nécessitent pas une procé-
dure itérative pour déterminer les facteurs de sécurité. Le développement de ces chartes a été possible grâce à
l’introduction du nombre de stabilité en deux dimensions et de la fonction de stabilité en trois dimensions, dont les valeurs
dépendent du degré de non linéarité des enveloppes de résistance des sols impliqués. Les chartes offrent aux praticiens
une méthode rapide et fiable pour l’analyse des ruptures en translation. L’utilisation des logiciels disponibles pour une telle
analyse peut être compliquée si l’on considère la résistance au cisaillement le long des bords verticaux de la masse glis-
sante et la non linéarité des enveloppes de résistance au cisaillement des sols. Des exemples numériques et une historique
de cas sont présentés afin de vérifier la fiabilité des chartes ainsi que pour illustrer les différentes applications de ces char-
tes.
Mots-clés : enveloppe de rupture en résistance au cisaillement non linéaire, analyse de stabilité de pente, chartes de stabi-
lité, rupture en translation, analyse en trois dimensions, analyse en deux dimensions.
[Traduit par la Rédaction]

Introduction three-dimensional (3D) slope stability analysis in which

One of the main inputs in any slope stability analysis is shearing along shallower surfaces near, and at, the sliding
the shear strengths or failure envelopes of the involved soils. mass sides is considered (Jiang et al. 2003).
It has long been recognized that failure envelopes of many Contrary to the rotational mode of slope failure that usu-
soils are significantly nonlinear, especially at a low normal ally occurs in homogeneous soils, failure of slopes in the
stress range (Terzaghi and Peck 1948). This range is usually translational mode involves a high shear strength soil that is
relevant to two-dimensional (2D) slope stability analysis be- underlain by a low shear strength one or a pre-existing slip
cause the critical slip surface must intersect the slope sur- surface passing near the toe (Fig. 1). Several case histories
face. Such relevancy is more pronounced in the case of a describing soil slopes that failed in the translational mode
have been reported in the literature (e.g., Gretton embank-
ment failure (Chandler and Pachakis 1973); Maymont slide
Received 16 June 2008. Accepted 10 September 2009. Published (Krahn et al. 1979); Gardiner Dam movement (Jasper and
on the NRC Research Press Web site at [Link] on 17 March Peters 1979); Denholm landslide (Sauer and Christiansen
2010. 1987); Portuguese Bend slide (Ehlig 1992); Oceanside
H.T. Eid. Civil Engineering Department, College of Manor slide (Stark and Eid 1992); Cincinnati landfill slide
Engineering, Qatar University, Doha, P.O. Box 2713, Qatar (e- (Eid et al. 2000)). Analyses of the reported case histories
mail: heid@[Link]). showed that translational failure often involves a drained

Can. Geotech. J. 47: 388–399 (2010) doi:10.1139/T09-110 Published by NRC Research Press
Eid 389

Fig. 1. Types of translational sliding mode investigated in this study.

shearing condition, i.e., long-term failure condition. Sliding Fig. 2. Upper Lias clay drained shear strength failure envelopes and
in which residual strength is mobilized along a pre-existing their reported linear approximations. Subscripts f, p, and r represent
shear surface underlying stiff clay that mobilizes peak or fully softened, peak, and residual failure envelopes, respectively.
fully softened shear strength is a typical translational failure
case (Eid 1996, 2004). For soils usually involved in transla-
tional failures, envelopes that represent shear strength at
such conditions and states exhibit significant degrees of non-
linearity. Figure 2 shows examples of this stress-dependant
behavior in the case of drained peak, fully softened, and re-
sidual states.
The conventional approach in slope stability analysis has
been to approximate the actual curved drained failure enve-
lope to the linear Mohr–Coulomb relationship in terms of
cohesion intercept (c’) and friction angle (f0 ) as shown in
Fig. 2. A study of many case histories of rotational slides
showed that the use of these parameters in slope stability
analysis could lead to the determination of a considerably
deeper critical slip surface and an associated higher factor
of safety than those observed and determined using the ac-
that simulate frequently encountered field conditions. Slopes
tual, i.e., nonlinear envelopes (Mesri and Abdel-Ghaffar
with total heights (H) of 5, 10, and 15 m were analyzed in
1993; Popescu 2000). This is because linear approximation
this study. Slope inclination (i:1) was taken to be either 2:1,
usually overestimates shear strength values at low normal
3:1 or 4:1. Total unit weights of the upper and lower soils
stresses. The effect of degree of curvature of the shear
(g) were assumed to be the same and equal to 18 or
strength envelope on the critical slip surface location and
21 kN/m3. The soil shear strength and pore-water pressure
factor of safety against rotational failure has also been
representation, 2D and 3D slope models, and methods of
studied by many researchers (e.g., Charles 1982; Charles
stability analysis needed for the parametric study were
and Soares 1984; Jiang et al. 2003). However, similar stud-
adapted based on the following considerations.
ies for translational failure are not available in the literature.
The existing translational failure studies only utilized shear
strength envelopes that are linear and pass through the ori- Nonlinear shear strength envelopes
gin, i.e., c’ = 0 (Chen 1981; Stark and Eid 1998; Arellano To avoid overestimation of shear strength values at low
and Stark 2000; Qian et al. 2003; Eid et al. 2006). normal stresses due to approximation of the nonlinear shear
The main aim of the research presented in this paper is to strength envelope to a linear one with c’ and f0 parameters,
assess the significance of considering nonlinearity of the in- many researchers have suggested utilizing strength functions
volved soils’ strength envelopes when determining the crit- or relations between the effective normal stress (s’) and the
ical sliding mass configurations and the associated associated shear strength (t) in slope stability analysis.
minimum 2D and 3D factors of safety (F2D, F3D) for slopes These include bilinear functions (Lefebvre 1981), trilinear
susceptible to translational failure, and consequently to functions (De Mello 1977), and power-law relations (De
present the research results in form of stability charts. This Mello 1977; Charles and Watts 1980; Charles and Soares
was achieved through conducting an extensive parametric 1984; Collins et al. 1988; Maksimovic 1989; Perry 1994;
study that considers — in addition to nonlinearity of the Baker 2004). The most commonly used relation has the gen-
soil strength envelope — the effect of slope geometrical pa- eral form of t ¼ Aðs 0 Þb , where A and b are constants. The
rameters and soil physical properties. magnitude of t at any value of s’ is directly proportional to
the value of A. In contrast, the degree of envelope curvature
is inversely proportional to the value of b. Curve-fitting soft-
Parametric study ware should be used to determine the values of A and b
The parametric study was conducted on proposed slopes needed to form the power-law relation that best represents

Published by NRC Research Press

390 Can. Geotech. J. Vol. 47, 2010

the curved shear strength failure envelope. In addition, these Fig. 3. Parameters used to describe nonlinearity of a shear failure
constants depend on the units used and have no physical envelope.
meaning. As a result, a different method was adapted to de-
scribe the nonlinear failure envelopes utilized in conducting
the analysis and developing the stability charts presented in
this paper. The envelopes were described using a methodol-
ogy similar to those presented by Skempton (1985) and
Mesri and Abdel-Ghaffar (1993) in which the curvature of
the envelope is expressed in terms of the ratio of the secant
friction angle at a given stress to that at a standard stress. To
fit the purposes of this study, those two angles are defined
as (i) the secant angle (fm0 ) at the maximum effective stress
acting on the critical slip surface and (ii) the tangent of the
soil envelope at the origin or the initial friction angle (f00 ). Fig. 4. Nonlinear strength failure envelopes used in the parametric
Based on these definitions, a parameter (m) that defines the study.
degree of nonlinearity of a soil shear strength failure enve-
lope is taken for the analysis of a certain slope as the ratio
of tanfm0 to tanf00 . Consequently, m is always less than unity
for nonlinear failure envelopes and inversely proportional to
their degree of nonlinearity (Fig. 3).
Seven nonlinear failure envelopes were used in the analy-
sis presented in this paper. These envelopes were formed to
represent the realistic shapes of soil nonlinear shear strength
envelopes and the associated wide range of initial friction
angles and different degrees of nonlinearity (Fig. 4). To rep-
resent nonplastic soils, the shear strengths of which are prac-
tically stress independent, linear envelopes that pass through
the origin and have friction angles of 308, 358, 408, and 458 Fig. 5. Average and equivalent pore-water pressure ratios for pro-
were also considered in the analysis. Envelopes with f00 ‡ posed slip surface in a 2(horizontal):1(vertical) slope using soil
308 were assigned to the upper stratum while those with peak and residual failure envelopes and their linear approximations
f00 £ 308 were assigned to the lower one (Figs. 1 and 4). As shown in Fig. 2.
a result — and to keep the shear strength of the lower stra-
tum always less than that of the upper one — a total of 37
strength envelope combinations were utilized in the study.

Pore-water pressure
In slope stability analysis, pore-water pressures are usu-
ally represented using two alternatives. The first introduces
water table elevations or peizometric surfaces; the second
uses pore-water pressure ratio (ru) as described by Bishop
and Morgenstern (1960). Pore-water pressure ratio at a point
is defined as the magnitude of pore-water pressure divided
by the total stress at this point. As a result, the ratio will
not be constant over the slip surface and some form of aver-
aging should be carried out. Utilizing different values of c’
and f 0 , Barnes (1999) showed that using an average ru value
to represent the global pore-water pressure condition may
lead to significant errors in the slope stability analysis of cir-
cular failures. A similar conclusion was presented by Eid et
al. (2006) for translational failures through soils with linear
envelopes that pass through the origin.
The technique used by these researchers was adapted in
this study to determine if the same conclusion could be drawn
for translational failures through soils with nonlinear shear
strength envelopes. For an example slope of 2:1 inclination,
values of the ratio of water height to slope height (HW/H) are
plotted in Fig. 5 against the corresponding values of both the stratum, respectively. It can be seen that, except for values of
average ru proposed by Bishop and Morgenstern (1960) and HW/H equal to 0 and 1.0, using one average value for ru either
the equivalent ru that gives the same factor of safety. The overestimates or underestimates the pore-water pressures de-
peak and residual strength envelopes shown in Fig. 2 were pending on the value of HW/H. To avoid this error and its
used to represent the shear strength of the upper and lower consequences on stability analyses, pore-water pressures

Published by NRC Research Press

Eid 391

were represented in the slope models used in the present para- pressure and a vertical side produces the minimum area of
metric study by introducing water table elevations. shear surface.
The reported sliding mass configurations lead to describing
2D slope model a slope failure scenario that starts with having sections with
The 2D slope model utilized in the present paper — F2D less than 1.0. For failure to occur, the back scarp of a
which also represents the critical cross section of the mass including these sections must extend laterally to de-
adapted 3D model described in the next section — is shown crease the end effect relative to the driving forces until a crit-
in Fig. 6 along with the associated friction angles and pa- ical mass width is reached and F3D equal to 1.0 is mobilized.
rameters used to represent nonlinearities of the upper and This means that failure through regular slopes should have
lower soil envelopes (mU and mL, respectively). The section almost the same sliding mass width that is equal to the crit-
illustrated in Fig. 6 was formed to simulate critical sections ical mass width. This conclusion along with associated geo-
that have been frequently reported for translational failures metrical features, e.g., vertical sides and rounded-end back
(e.g., Skempton and Petley 1967; Chandler et al. 1973; Cru- scarp, is supported by the width regularity and shape of slid-
den et al. 1991) with a sliding surface that passes below the ing masses described in case histories such as the landslides
slope toe and has a curved back scarp, the bottom of which at Herne Bay, UK (Bromhead 1986), as shown in Fig. 7.
is not necessarily an orthogonal projection of the slope crest Considering these commonly reported configurations of
line. The sliding surface at the utilized section is taken to the sliding mass, a 3D slope model was designed and em-
consist of a curved back scarp that starts at a distance (T), ployed to conduct the parametric study needed for the inves-
or top width of the sliding mass, behind the crest going tigation of the effects of soil strength envelope nonlinearity
down through the upper soil at an inclination of 45 + on the stability of slopes susceptible to translational failure
(f00 )U/2 with the horizontal, which is connected at an incli- and the development of stability charts. Figures 8a, 8b, and
nation of 45 + (fm0 )U/2 with the horizontal to the bottom 8c show an isometric view, critical cross section, and near-
part that extends 0.1 m into the lower soil and runs parallel side cross section of this model, respectively. The model rep-
resents a sliding mass — with several and identical critical
to its surface until it rises at 45 – (f00 )U/2 — starting from a
cross sections (Fig. 8b) — that is rounded at the ends of its
point directly below the toe — to daylight. It can be noticed
back scarp, creating a group of noncritical near-side cross
that the inclinations of the sliding surface parts that pass
sections and bounded by two vertical sides. The model is dis-
through the upper soil reflect development of the active and
tinguished from those utilized in previous translational fail-
passive stress conditions in the back and toe zones, respec- ure studies (e.g., Chen 1981; Montés et al. 1996; Stark and
tively. The friction angles used to define such inclinations Eid 1998; Arellano and Stark 2000; Qian et al. 2003) in sev-
have been chosen to represent the shear strengths mobilized eral aspects that make it more representative of field case his-
at the effective normal stresses acting on the points or zones tories and more generic. In 3D configurations, the model
where the slip surface passes. incorporates vertical sides, rounded back corners, and curved
The weaker soil surface has been assumed to pass through back scarps. In both 2D and 3D configurations, it considers
or below the toe having an inclination (d) of 08 and 68 to sim- the possibility of locating the weak material at a depth H2 be-
ulate natural bedding planes or commonly encountered pre- low the slope toe and having a top width for the sliding mass.
existing slip surfaces, the existence of which usually leads to The model differs from that presented by Eid et al. (2006) in
translational failures (Eid et al. 2006). Values of 0 and 0.25 that it has a curved back scarp due to consideration of the
are assigned for H2/H1 in this study, where H1 and H2 are de- nonlinearity of the soil shear strength envelope. As a result,
fined as the vertical distances from the crest and toe, respec- the stability charts presented in this paper can be considered
tively, to the point where the surface of the weaker material as the general form of those developed by Eid et al. (2006).
passes beneath the toe. Groundwater tables having linearly
decreasing heights, maintaining a constant value of HW/H Method of analysis
(i.e., equal values of HW1/H1 and HW2/H2) are used for the The slope stability program UTEXAS3 (Wright 1992)
slope model. HW1 and HW2 are defined as the water heights was utilized in the 2D stability analysis described in this
near the slope crest and toe, respectively. Values of HW/H of paper, while the 3D analysis was conducted using the
0, 0.4, and 0.8 were utilized in the analysis. CLARA 2.31 program (Hungr 1988). In these analyses,
each nonlinear envelope was modeled using 19 shear and
3D slope model normal stress combinations and the power-law form when
Several field case histories of translational failures were utilizing the UTEXAS3 and CLARA 2.31 programs, respec-
studied to design the 3D slope model, i.e., form the lateral tively. Stability analysis procedures used to determine the
extensions of the previously described 2D model, that simu- F2D values and the corresponding F3D/F2D ratios are similar
lates the configurations of the sliding mass developed in to those described in Eid et al. (2006).
such failures. The study revealed that the mass usually com- The technique utilized by Eid et al. (2006) in quantifying
prises several similar cross sections at the middle and sec- and incorporating the end effects was modified to estimate
tions with a steeper back scarp near the sides. The mass is shear resistance mobilized along the vertical sides of masses
also rounded at the ends of its back scarp and is bounded failing through soils with nonlinear strength envelopes. The
by two vertical sides at a distance W (slope width). In trans- resistance along each side was included by imposing a group
lational failures, the vertical sides provide the minimum of external horizontal and vertical forces (FX and FZ, respec-
amount of shear resistance because the effective normal tively) that are the components of the shear resistance forces
stress acting on these sides is only due to the lateral earth (F). The horizontal and vertical force components were im-

Published by NRC Research Press

392 Can. Geotech. J. Vol. 47, 2010

Fig. 6. Schematic drawings showing parameters used in this study point ‘‘n’’ as ko = 1 – sinfn0 . Because the nonlinearity of the
to describe (a) critical cross section and (b) nonlinear failure envel- upper soil strength envelope was taken into considerion, val-
opes of involved soils. Subscripts U and L represent the upper and ues of fn0 and consequently those of the corresponding ko var-
lower soil envelopes, respectively. W.L., water level; gw, water unit ied with the location of each point (Fig. 9).
weight. Refer to text for definitions of H1, H2, HW1, and HW2. Seismic loads acting on the slope model were considered
in the stability analyses by including a pseudostatic force
due to seismic acceleration. Forces were imposed at the cen-
ter of gravity of the predetermined critical sliding mass. Val-
ues of 0.1 and 0.2 were assigned to the coefficient Kh, which
represents the intensity of horizontal acceleration as a frac-
tion of the gravitational acceleration (g).
To develop charts for determining location of the critical slip
surface or configurations of the sliding mass and values of the
corresponding F2D and F3D/F2D in both static and seismic con-
ditions, these configurations and values were correlated to
slope geometrical parameters and shear strength parameters
that reflect nonlinearity of the involved soils’ strength enve-
lopes. All of the presented charts are in a dimensionless form
and do not require iterative procedures as shown in the next
sections.

Location of critical slip surface

As shown in Figs. 6a and 8, the location where the crit-
ical slip surface daylighted behind the slope crest, i.e., the
top width of critical sliding mass or the graben width T,
should be determined to define the configuration of the
slope model and consequently, the failure extent. Having a
relationship or chart that helps in estimating T is useful for
Fig. 7. Landslides in London clay at Herne Bay, UK (adapted from locating the critical slip surface for simple slopes as a neces-
Bromhead 1986). sary step to determine the minimum factor of safety against
translational failure. In addition, it provides the slip surface
location that is needed as an initial input to start a more so-
phisticated automatic search for locating the critical slip sur-
face of complicated slopes susceptible to translational
failure. However, charts for locating critical slip surfaces
for slopes failing in the translational mode are limited in
the literature.
Through measurements of graben width and site explora-
tion of the depth to the slip surface of 10 translational land-
slides in central Alberta, Canada, Cruden et al. (1991)
suggested that the graben width is about 90% of the depth
to the slip surface. They also related the significant scatter
in data to complicating factors that influence the proposed
simple relation. It should be noted that the effect of shear
strength values mobilized along the scarp and base of the
slip surface were not included in this relation. Montés et al.
(1996) included the undrained shear strength values in deter-
posed at the centroids of their corresponding areas of the verti- mining the location of the critical trilinear slip surface of
cal sides of the sliding mass (Fig. 9). As the predominant slopes in cohesive stratified deposits. However, the use of
movement direction in translational failures is parallel to the the resulting charts is restricted to short-term stability analy-
surface of the lower soil, the shear resistance forces were as- sis as they were developed for an undrained failure condi-
signed this direction. Each shear resistance force was calcu- tion. In their analysis using linear shear strength envelopes
lated by multiplying the corresponding earth pressure force for the involved soils, Eid et al. (2006) concluded that the
(E) acting perpendicular to the vertical sides by the tangent of magnitude of T is insensitive to values of the upper soil fric-
the secant friction angle — of the upper soil strength envelope tion angle and lower soil surface inclination. They also indi-
— at the average earth pressure acting on the corresponding cated that this magnitude increases with increasing slope
area of the vertical side. Each earth pressure force was esti- angle and depth of lower soil beneath the toe, and decreas-
mated as the volume of the at-rest earth pressure diagram im- ing values of the lower soil friction angle.
posed on the corresponding area of the slope’s vertical side. A supplementary 2D parametric study utilizing the 2D crit-
The coefficient of earth pressure at rest (ko) was taken at any ical cross section shown in Fig. 6a and the group of nonlinear

Published by NRC Research Press

Eid 393

Fig. 8. Slope model used in the analysis: (a) 3D view; (b) section A–A’ representing critical cross sections; (c) section B–B’ representing the
vertical sides of the model.

and linear strength envelopes combinations described earlier shown in Fig. 10. It can be noted that, for the proposed con-
revealed that the value of T is insensitive to the absolute val- figurations and shear strength conditions, the commonly used
ues of (f00 )U and (fm0 )U, which mainly control the back-scarp simple bilinear critical sliding surface that passes through the
inclination when the failure progresses down through the slope crest and toe can only be developed if the lower soil
upper layer (Figs. 6 and 8b). The study also showed that T is passes through the slope toe, i.e., H2 = 0, and the upper soil
slightly affected by the absolute values of (f00 )L and mL. A has a linear strength envelope that passes through the origin,
simple correlation between the value of T and the governing i.e., mU = 1.0. Development of T is directly proportional to
slope geometrical and soil parameters was developed and is the value of H2 because a larger driving force needs to be

Published by NRC Research Press

394 Can. Geotech. J. Vol. 47, 2010

Fig. 9. Forces imposed on vertical sides of slopes in the 3D stability analysis.

Fig. 10. Relationship between top width of the critical sliding mass Results of two-dimensional stability analysis
and geometrical and soil parameters of the slope. b, slope angle.
A parametric study that was conducted utilizing the 2D
critical cross section shown in Fig. 6a revealed that F2D de-
pends on the ratio of H2 to H1 rather than on their absolute
values. The magnitude of F2D increases with increasing val-
ues of H2/H1, (f00 )L, (f00 )U, mL, and mU, and decreasing val-
ues of HW, d, and the slope angle. The study also showed
that while F2D increases with an increase in both mL and
mU, it is more sensitive to the value of mL or the lower soil
envelope degree of nonlinearity. Based on these results, 2D
stability charts for estimating F2D were developed and are
shown in Fig. 11. While the use of software has superseded
most graphical methods, charts for 2D slope stability analy-
sis are still routinely used in practice. The charts presented
in Fig. 11 were developed to help in the 2D analysis for
slopes susceptible to translational failures.
It should be noted that F2D in these charts is presented as
a function of an introduced stability number for slopes sus-
ceptible to translational failure through soils with nonlinear
strength envelopes (Ntn). This stability number is defined as
 0 
0:4 1:2 tanðf0 ÞL
½1 Ntn ¼ mU mL
tanðf00 ÞU
Consequently, determining F2D does not require any itera-
tive procedure because Ntn is independent of the safety fac-
tor that is equal for both (f00 )L and (f00 )U. This concept of
utilizing a parameter — the numerator and denominator of
which include the safety factor — in such a way that leads
built up by shifting the back scarp a certain distance behind to cancellation of the safety factor was first introduced by
the slope crest to counter the enhanced passive resisting Bell (1966) for developing charts for rotational failure of
forces created by increasing H2. slopes in which the safety factor is applied to both c’ and
It can be also seen that the value of T is inversely propor- tanf0 .
tional to mU, i.e., directly proportional to the degree of non- The charts in Fig. 11 show that the effect of flattening the
linearity of the upper soil strength envelope. This is because slope to increase the factor of safety decreases in the case of
shifting the back scarp behind the slope crest imposes higher a low Ntn value. This is indicated by the converging nature
normal stress on the scarp surface, which leads — due to of F2D at low values of Ntn. The charts also show that the
nonlinearity of the upper soil strength envelope — to the existence of high water levels decreases the sensitivity of
mobilization of the lower friction angles that in turn reduces F2D to changes in these values. This is indicated by the flat-
the 2D factor of safety. This reduction continues with scarp tening nature of F2D at high values of HW/H.
shifting until T reaches a critical value at which point the
frictional force mobilized along the increasing length of the
Results of three-dimensional stability
sliding surface base, i.e., along the lower soil, counters such
a reduction and then starts to increase the factor of safety. A analysis
similar mechanism can interpret the change in the effect of Slopes failing in a translational mode exhibit the most
tan (fm0 )L/ tanb on the value of T with the changing magni- pronounced difference between F2D and F3D (Chen and Cha-
tude of H2/H1. meau 1983; Leshchinsky et al. 1985; Stark and Eid 1998).

Published by NRC Research Press

Eid 395

Fig. 11. Two-dimensional stability charts for slopes susceptible to translational failure for (a) d = 08 and Hw/H = 0.0; (b) d = 08 and Hw/H =
0.4; (c) d = 08 and Hw/H = 0.8; (d ) d = 68 and Hw/H = 0.0; (e) d = 68 and Hw/H = 0.4; ( f ) d = 68 and Hw/H = 0.8.

As a result, the 2D analysis may lead to overdesigned slopes Leshchinsky and Baker (1986) used a stability function,
and (or) overestimation of back-calculated shear strength pa- the value of which depends on c’, tanf0 , g, and slope height,
rameters of materials involved in slopes that failed in a in producing 3D stability charts for uniform slopes suscepti-
translational mode. A parametric study utilizing the 3D ble to rotational failures. Adapting a similar concept to
model and side forces shown in Figs. 8 and 9, respectively, develop stability charts to estimate values of F3D/F2D for
was conducted to check the sensitivity of the ratio F3D/F2D the translational mode, the influential parameters of the
to the involved soil properties and slope geometrical param- upper soil strength, i.e., (f00 )U and mU, were represented by
eters. The value of F2D used in this study is the one that cor- one dimensionless stability function (l) that is defined as
responds to the most critical cross section, i.e., the cross
0
section with the minimum F2D. The study showed that d ½2 l ¼ m0:5
U tanðf0 ÞU
and HW have a slight effect on F3D/F2D values. It also re-
vealed that F3D/F2D is not sensitive to the shear strength of The developed 3D stability charts are shown in Fig. 12. It
the lower soil. This conclusion agrees with data presented can be seen that the 3D effect, i.e., the F3D/F2D value, in-
by Arellano and Stark (2000) and Eid et al. (2006) and can creases with decreasing W/H1 and b and increasing H2/H1
be interpreted in terms of the overwhelming effect of the values.
upper soil strength (over that of the lower soil) on the value Utilizing the charts shown in Figs. 10, 11, and 12 leads to
of F3D, which mainly increases because of the inclusion of a complete definition of the 3D geometrical configuration of
the shear resistance along the parallel vertical sides of the the potential critical sliding mass if the parameters b, (f00 )U,
sliding mass. The magnitude of this resistance is directly (fm0 )U, (f00 )L, (fm0 )L, d, H1, H2, and HW are known for the
proportional to the values of (f00 )U and mU. critical 2D section that has F2D < 1.0. The length of this crit-

Published by NRC Research Press

396 Can. Geotech. J. Vol. 47, 2010

Fig. 12. Three-dimensional stability charts for slopes susceptible to Fig. 13. Three-dimensional stability charts for slopes susceptible to
translational failure considering static loads for (a) H2/H1 = 0.0 and translational failure considering static and seismic loads for (a) H2/
(b) H2/H1 = 0.25. H1 = 0.0 and (b) H2/H1 = 0.25.

ical mass can be calculated once the value T is determined sensitive to the shear strength of the upper soil, i.e., values
using Fig. 10. The value of F2D is then estimated utilizing of l. As a result, curves of different l for each slope incli-
Fig. 11, which is used in determining the width of the crit- nation used in Fig. 12 were combined in Fig. 13 in one
ical mass utilizing Fig. 12 by assigning a value of 1.0 to curve due to the seismic effect consideration. This means
F3D. that, for the range of parameters considered in this study,
The ratio of the 3D factor of safety considering seismic F3D(seismic)/F2D value is geometrically related and practically
loads (F3D(seismic)) to the critical cross-section factor of safety does not depend on soil properties.
without considering them (F2D or F2D(static)) was used for the
charts presented in Fig. 13 to illustrate the combined 3D and
Evaluation of study results
seismic effects on the stability of slopes. Considering the
seismic effect led to a calculated F3D(seismic)/F2D that is in- The method and reliability of using charts presented in

Published by NRC Research Press

Eid 397

this paper, as well as the importance of considering nonli- tan248/ tan348 = 0.66. Using values of l = (0.66)0.5 
nearity of the soil strength envelope in the analysis of trans- tan348 = 0.55 and W/H1 = 4 we read F3D/F2D = 1.21 from
lational slides, are illustrated utilizing four hypothetical Fig. 12a. As F3D = 1.0 at failure, F2D = 1.0/1.21 = 0.83.
examples and a case history as shown below. With the given (f00 )U and i, we read Ntn = 0.11 from
Fig. 11b, which gives (f00 )L = fL0 = 5.18 assuming a linear
Example 1 envelope passing through the origin for the lower soil. It
A 5 m high soil slope with g = 19 kN/m3 and a nonlinear should be noted that if the 3D effect is not considered in
shear strength envelope similar to that of the peak strength this example, the back-calculated fL0 will be overestimated
shown in Fig. 2 placed on a horizontal pre-existing shear to 7.18, which is a result of using Fig. 11b and F2D = 1.0.
surface that mobilizes a nonlinear shear strength envelope
similar to that of the residual condition also shown in Example 4
Fig. 2. From Fig. 2 we read (f00 )U = 438 and (f00 )L = 168. If the failure described in example 3 occurs due to seismic
With s m0 = 19  5 = 95 kN/m2 we also read (fm0 )U = 328 forces that can be represented by Kh = 0.1, back-calculation
and (fm0 )L = 11.58. This leads to mL = tan11.58/ tan168 = of fL0 requires the use of Fig. 13a to read F3D(seismic)/F2D =
0.71 and mU = tan328/ tan438 = 0.67. Determining the slope 0.686. As F3D(seismic) = 1.0, F2D = 1.0/0.686 = 1.46. With
inclination that allows for F2D = 1.5 against translational F2D/tan(f00 )U = 1.46/ tan348 = 2.16 we read Ntn = 0.27 from
failure requires the use of Fig. 11a. Using values of Ntn = Fig. 11b, which gives fL0 = 128.
(0.67)0.4(0.71)1.2(tan168/ tan438) = 0.174, F2D/ tan(f00 )U =
1.5/ tan438 = 1.61, and H2/H1 = 0 we read i = 2.48 or a Field case history
slope inclination of 228. It should be noted that using In his analysis of failures in clay slopes, James (1970)
UTEXAS3 for solving this problem yielded slope inclina- presented the Peterborough slope failure as a case of sliding
tions of 21.38 and 338 when nonlinear envelopes and their through Oxford clay. As shown in Fig. 14a, the slide has a
linear approximations (Fig. 2) were utilized, respectively. horizontal base passing through Oxford clay and a back
Such a large difference in allowable slope inclination can scarp that passes through the overlaying Callow layer. Cal-
be attributed to a relatively small slope height and conse- low is highly weathered and remoulded Oxford clay. The re-
quently, low effective stresses acting on the slip surface. ported analyses for this case (e.g., James 1970; Mesri and
The envelope linear approximation significantly overesti- Shahien 2003) revealed that the residual and fully softened
mated shear strength values at these low normal stresses shear strengths of Oxford clay were developed along the
and as a result led to a considerable overestimation of the base and back scarp of the sliding surface, respectively. The
allowable slope inclination. shear strength envelopes of Oxford clay at these two states
are shown in Fig. 14b based on data presented by Petley
Example 2 (1966), James (1970), and Mesri and Shahien (2003).
A 6 m high 3:1 slope cut in a 8 m thick layer of soil with The location and factor of safety of the critical slip sur-
g = 20 kN/m3 and a shear strength envelope similar to that of face for this case history were estimated using charts pre-
the fully softened condition shown in Fig. 2. The cut layer is sented in this study. The friction angles required for this
underlain by an extended one with a shear strength envelope estimation are shown in Fig. 14b. These values lead to mL =
similar to that of the residual strength also shown in Fig. 2 0.79 and mU = 0.75. HW/H1 was assigned a value of 0.8 that
and an inclined surface of 68 downslope. From Fig. 2 we read corresponds to the reported ru = 0.37 (Fig. 5). Using
(f00 )U = 348 and (f00 )L = 168. With s m0 = 20  8 = 60 kN/m2 we tan(fm0 )L/ tanb = 0.96, mU = 0.75, and H2/H1 = 0, we read
also read (fm0 )U = 238 and (fm0 )L = 108. This leads to mL = from Fig. 10 that T/H1 = 0.17. This gives T = 0.17  12.5 =
0.62 and mU = 0.63. Evaluation of F2D that is the same for 2.13 m. Using Ntn = (0.75)0.4(0.79)1.2(tan178/ tan328) = 0.33,
the shear strengths of both the upper and lower layers does we read that F2D/ tan(f00 )U = 1.67 from Fig. 11c. This yields
not require an iterative procedure. Using H2/H1 = 2/8 = 0.25 F2D = 1.67  tan328 = 1.04. The closeness in locations of
and Ntn = (0.63)0.4(0.62)1.2  (tan168/ tan348) = 0.2, we read the determined critical slip surface and the observed slip
in Fig. 11d that F2D/ tan(f00 )U = 2.0. This yields F2D = 2.0  surface (Fig. 14a), along with the agreement between the
tan348 = 1.35. The critical slip surface is located utilizing estimated F2D (1.04) and that mobilized at failure (1.0),
Fig. 10 and the values tan(fm0 )L/ tanb = tan108/ tan18.448 = support the reliability of the results and charts presented in
0.53, mU = 0.63, and H2/H1 = 0.25 to read T/H1 = 0.44. This this paper.
yields T = 0.44  9 = 3.96 m. The top width of the critical slip
surface and the associated F2D were determined using
UTEXAS3 to be 4.05 m and 1.4, respectively. Conclusions
An extensive parametric study was conducted on slope
Example 3 models designed to simulate translational slides in soils fail-
A slope failed in translational mode with the following pa- ing under a drained condition and possessing nonlinear shear
rameters: i = 4, H1 = 9 m, H2/H1 = 0, W/H1 = 4, HW/H1 = strength envelopes. The effects of this nonlinearity along
0.4, d = 0, g = 18 kN/m3, and an upper soil shear strength with other soil properties, slope geometrical parameters, and
envelope similar to the fully softened strength one shown in loading conditions on the stability of slopes susceptible to
Fig. 2. Back-calculation of the average friction angle mobi- translational failure were investigated. Analysis of the re-
lized at the lower soil (fL0 ) requires the use of Fig. 12a and sults showed that the critical sliding mass configurations are
then Fig. 11b. With s m0 = (18  9) – (gw)(3.6) = 126.7 kN/m2 significantly affected by nonlinearity of the upper soil
we read (fm0 )U = 248 from Fig. 2. This leads to mU = strength envelope. However, the 2D factor of safety is more

Published by NRC Research Press

398 Can. Geotech. J. Vol. 47, 2010

Fig. 14. Peterborough slope failure case history: (a) cross section; tio ru in slope stability analysis. Ground Engineering, 32(8): 38–
(b) strength envelopes of the involved soil. 40.
Bell, J.M. 1966. Dimensionless parameters for homogenous earth
slopes. Journal of the Soil Mechanics and Foundation Engineer-
ing Division, ASCE, 92(5): 51–65.
Bishop, A.W., and Morgenstern, N.R. 1960. Stability coefficients
for earth slopes. Géotechnique, 10(4): 129–150. doi:10.1680/
geot.1960.10.4.129.
Bromhead, E.N. 1986. The stability of slopes. Chapman and Hall,
N.Y.
Chandler, R.J. 1972. Lias clay: weathering process and their effect
on shear strength. Géotechnique, 22(3): 403–431. doi:10.1680/
geot.1972.22.3.403.
Chandler, R.J. 1974. Lias clay: the long-term stability of cutting
slopes. Géotechnique, 24(1): 21–38. doi:10.1680/geot.1974.24.1.
21.
Chandler, R.J., and Pachakis, M. 1973. Long-term failure of a bank
on a solifluction sheet. In Proceedings of the 8th International
Conference on Soil Mechanics and Foundation Engineering,
Moscow, 6–11 August 1973. Strojizdat, CIS, Moscow. Vol. 2,
pp. 45–51.
Chandler, R.J., Pachakis, M., Mercer, J., and Wrightman, J. 1973.
Four long-term failures of embankments founded on areas of
landslip. Quarterly Journal of Engineering Geology, 6(3–4):
405–422. doi:10.1144/[Link].1973.006.03.17.
Charles, J.A. 1982. An appraisal of the influence of a curved fail-
ure envelope on slope stability. Géotechnique, 32(4): 389–392.
doi:10.1680/geot.1982.32.4.389.
sensitive to nonlinearity of the lower soil strength envelope, Charles, J.A., and Soares, M.M. 1984. The stability of slopes in
while the ratio between the 3D and 2D factors of safety is soils with nonlinear failure envelopes. Canadian Geotechnical
influenced by nonlinearity of the upper soil envelope only. Journal, 21(3): 397–406. doi:10.1139/t84-044.
Charles, J.A., and Watts, K.S. 1980. The influence of confining
Charts were produced for determining the configurations of
pressure on the shear strength of compacted rockfill. Géotechni-
the critical sliding mass and estimating the associated 2D
que, 30(4): 353–367.
and 3D factors of safety without requiring an iterative proc- Chen, R.H. 1981. Three-dimensional slope stability analysis. Joint
ess. Utilizing a simple methodology to define the shear Highway Research Project 81–17. Purdue University, West La-
strength envelope nonlinearity by a single parameter m fayette, Ind.
along with the introduction of a dimensionless 2D stability Chen, R.H., and Chameau, J.L. 1983. Three-dimensional limit equi-
number, Ntn, and 3D stability function, l, have made it pos- librium analysis of slopes. Géotechnique, 33(1): 31–40.
sible to develop such charts. The charts provide a fast and Collins, I.F., Gunn, C.I.M., Pender, M.J., and Wang, Y. 1988.
reliable method for translational failure analysis. Using Slope stability analyses for materials with a nonlinear failure en-
available slope stability software in such an analysis can be velope. International Journal for Numerical and Analytical
cumbersome and time-consuming if the shear resistance Methods in Geomechanics, 12(5): 533–550. doi:10.1002/nag.
along the sliding mass vertical sides and the nonlinearity of 1610120507.
the soil shear strength envelope are considered. Cruden, D.M., Thomson, S., and Hoffman, B.A. 1991. Observation
of graben geometry in landslides. In Slope Stability Engineering:
Acknowledgement Developments and Applications, Proceedings of the International
Conference on Slope Stability, Isle of Wight, UK, 15–18 April
The author gratefully acknowledges Dr. Edward N. Brom-
1991. Edited by R.J. Chandler. Thomas Telford Publications,
head of Kingston University, UK, for providing information
London. pp. 33–35.
and photographs for the Herne Bay landslides. De Mello, V.B.F. 1977. Reflections on design decisions of practical
significance to embankment dams. Géotechnique, 27(3): 281–
References 354.
Arellano, D., and Stark, T.D. 2000. Importance of three- Ehlig, P.L. 1992. Evolution, mechanics and mitigation of the Portu-
dimensional slope stability analysis in practice. In Proceedings guese Bend landslide, Palos Verdes Peninsula, California. In En-
of Slope Stability 2000, Specialty Conference, Denver, Colo., gineering geology practice in southern California. Edited by
5–8 August 2000. Geotechnical Special Technical Publication B.W. Pipkin and R.J. Proctor. Star Publishing Co., Belmont, Ca-
No. 101. Edited by D.V. Griffiths, G.A. Fenton, and T.R. Mar- lif. pp. 531–553.
tin. American Society of Civil Engineers, Reston, Va. pp. 18– Eid, H.T. 1996. Drained shear strength of stiff clays for slope sta-
32. bility analyses. Ph.D. thesis, University of Illinois at Urbana-
Baker, R. 2004. Nonlinear Mohr envelopes based on triaxial data. Jour- Champaign, Urbana, Ill.
nal of Geotechnical and Geoenvironmental Engineering, 130(5): Eid, H.T. 2004. Shear strength for locating critical slip surface in
498–506. doi:10.1061/(ASCE)1090-0241(2004)130:5(498). stiff clay slopes. In Proceedings of the 9th International Sympo-
Barnes, G. 1999. A commentary on the use of the pore pressure ra- sium on Landslides, Rio de Janeiro, Brazil, 28 June – 2 July

Published by NRC Research Press

Eid 399

2004. A.A. Balkema, Rotterdam, the Netherlands. Vol. 2, Montés, P., Silvestri, V., and Soulié, M. 1996. Slope stability ana-
pp. 1037–1044. lysis and charts for non-circular slip surface. In Proceedings of
Eid, H.T., Stark, T.D., Evans, W.D., and Sherry, P. 2000. Munici- the 7th International Symposium on Landslides, Trondheim,
pal waste slope failure I: Waste and foundation soil properties. Norway, 17–21 June 1996. A.A. Balkema, Rotterdam, the Neth-
Journal of Geotechnical and Geoenvironmental Engineering, erlands. Vol. 2, pp. 1305–1310.
126(5): 397–407. doi:10.1061/(ASCE)1090-0241(2000) Perry, J. 1994. A technique for defining non-linear shear strength
126:5(397). envelopes and their incorporation in a slope stability method of
Eid, H.T., Elleboudy, A.M., Elmarsafawi, H.G., and Salama, A.G. analysis. Quarterly Journal of Engineering Geology, 27(3): 231–
2006. Stability analysis and charts for slopes susceptible to 241. doi:10.1144/[Link].1994.027.P3.04.
translational failure. Canadian Geotechnical Journal, 43(12): Petley, D.J. 1966. The shear strength of soils at large strains. Ph.D.
1374–1388. doi:10.1139/T06-095. thesis, University of London, London.
Hungr, O. 1988. CLARA: Slope stability analysis in two or three Popescu, M. 2000. Linear versus nonlinear failure envelopes in
dimensions. Version 2.31 [computer program]. O. Hungr Geo- LEM and FEM slope stability analysis. In Proceedings of the
technical Research, Inc., Vancouver, B.C. 8th International Symposium on Landslides, Cardiff, UK, 26–
James, P.M. 1970. Time effects and progressive failure in clay 30 June 2000. Edited by E. Bromhead, N. Dixon, and M.L. Ib-
slopes. Ph.D. thesis, University of London, London. sen. Thomas Telford, London. Vol. 3, pp. 1227–1234.
Jasper, J.L., and Peters, N. 1979. Foundation performance of Gardi- Qian, X., Koerner, R.M., and Gray, D.H. 2003. Translational fail-
ner Dam. Canadian Geotechnical Journal, 16(4): 758–788. ure analysis of landfills. Journal of Geotechnical and Geoenvir-
doi:10.1139/T79-083. onmental Engineering, 129(6): 506–519. doi:10.1061/(ASCE)
Jiang, J.-C., Baker, R., and Yamagami, T. 2003. The effect of 1090-0241(2003)129:6(506).
strength envelope nonlinearity on slope stability computations. Sauer, E.K., and Christiansen, E.A. 1987. The Denholm landslide,
Canadian Geotechnical Journal, 40(2): 308–325. doi:10.1139/ Saskatchewan, Canada, an update. Canadian Geotechnical Jour-
t02-111. nal, 24: 163–168. doi:10.1139/t87-017.
Krahn, J., Johnson, R.F., Fredlund, D.G., and Clifton, A.W. 1979. Skempton, A.W. 1985. Residual strength of clays in landslides,
A highway cut failure in cretaceous sediments at Maymont, Sas- folded strata and the laboratory. Géotechnique, 35(1): 3–18.
katchewan. Canadian Geotechnical Journal, 16: 703–715. doi:10. doi:10.1680/geot.1985.35.1.3.
1139/T79-079. Skempton, A.W., and Petley, D.J. 1967. The strength along struc-
Lefebvre, G. 1981. Fourth Canadian Geotechnical Colloquium: tural discontinuities in stiff clays. In Proceedings of the Geo-
Strength and slope stability in Canadian soft clay deposits. Ca- technical Conference on Shear Strength of Natural Soils and
nadian Geotechnical Journal, 18(3): 420–442. doi:10.1139/t81- Rocks, Oslo, Norway, 19–22 September 1967. Norwegian Geo-
047. technical Institute, Oslo. Vol. 2, pp. 29–46.
Leshchinsky, D., and Baker, R. 1986. Three-dimensional slope sta- Stark, T.D., and Eid, H.T. 1992. Comparison of field and labora-
bility: end effects. Soils and Foundations, 26(4): 98–110. tory residual shear strengths. In Proceedings of the ASCE Speci-
Leshchinsky, D., Baker, R., and Silver, M. 1985. Three dimen- alty Conference on Stability and Performance of Slopes and
sional analysis of slope stability. International Journal for Nu- Embankments – II, Berkeley, Calif., 29 June – 1 July 1992.
merical and Analytical Methods in Geomechanics, 9(2): 199– Geotechnical Special Technical Publication No. 31. American
223. doi:10.1002/nag.1610090302. Society of Civil Engineers (ASCE), Reston, Va. Vol. 1,
Maksimovic, M. 1989. Nonlinear failure envelope for soils. Journal pp. 876–889.
of Geotechnical Engineering, 115(4): 581–586. doi:10.1061/ Stark, T.D., and Eid, H.T. 1998. Performance of three-dimensional
(ASCE)0733-9410(1989)115:4(581). slope stability methods in practice. Journal of Geotechnical and
Mesri, G., and Abdel-Ghaffar, M.E.M. 1993. Cohesion intercept in Geoenvironmental Engineering, 124(11): 1049–1060. doi:10.
effective stress-stability analysis. Journal of Geotechnical Engi- 1061/(ASCE)1090-0241(1998)124:11(1049).
neering, 119(8): 1229–1249. doi:10.1061/(ASCE)0733- Terzaghi, K., and Peck, R.B. 1948. Soil mechanics in engineering
9410(1993)119:8(1229). practice. 1st ed. John Wiley and Sons, Inc., New York.
Mesri, G., and Shahien, M. 2003. Residual shear strength mobilized Wright, S.G. 1992. UTEXAS3: A computer program for slope sta-
in first-time slope. Journal of Geotechnical and Geoenvironmen- bility calculations. Geotechnical engineering software GS86–1,
tal Engineering, 129(1): 12–31. doi:10.1061/(ASCE)1090- version 3 [computer program]. Department of Civil Engineering,
0241(2003)129:1(12). University of Texas, Austin, Tex.

Published by NRC Research Press

Copyright of Canadian Geotechnical Journal is the property of NRC Research Press and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.