You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273749164

Assessment of 3D Slope Stability Analysis Methods Based on 3D Simplified Janbu


and Hovland Methods

Article  in  International Journal of Geomechanics · April 2012


DOI: 10.1061/(ASCE)GM.1943-5622.0000117

CITATIONS READS
17 512

3 authors, including:

Aly Ahmed

37 PUBLICATIONS   454 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Durability of self compacting concrete produced with recycled aggregate exposed to agressive environment View project

All content following this page was uploaded by Aly Ahmed on 01 June 2016.

The user has requested enhancement of the downloaded file.


Assessment of 3D Slope Stability Analysis Methods Based
on 3D Simplified Janbu and Hovland Methods
Aly Ahmed1; Keizo Ugai2; and Qing Qing Yang3

Abstract: An assessment of three-dimensional slope stability analysis methods in terms of safety factors using several idealized sliding
Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

masses composed of plane sliding surfaces was made. Three-dimensional safety factors were calculated and compared for different study
cases considered in this study on the basis of the exact solution methods, the Hovland method, and the 3D simplified Janbu method. Param-
eters investigated in this study included the effect of water pressure, horizontal seismic force, the changing gradient of a sliding surface, the
changing lateral gradient of a sliding surface, and anchor force. Results showed that the Hovland method gives smaller safety factor values
compared to the exact solutions, especially in cases of narrow failure width and high water pressure along sliding surfaces whereas the 3D
simplified Janbu method gives the same safety factor values as the exact solutions DOI: 10.1061/(ASCE)GM.1943-5622.0000117. © 2012
American Society of Civil Engineers.
CE Database subject headings: Slope stability; Three-dimensional analysis; Safety.
Author keywords: Slope stability; Three-dimensional; Safety factor; Hovland method; 3D simplified Janbu method.

Introduction However, it has three shortcomings: (1) the lateral forces acting
on the lateral sides of subdivided columns are assumed to be zero,
Recently, many researchers have studied the slope stability of soil (2) not all equilibrium conditions of forces and moments acting on
and rock slope failure with 3D behavior using three-dimensional the sliding mass can be satisfied, and (3) the 3D safety factor is
methods (Michalowski 2010; Wu et al. 2005; Wang et al. 2004; lower than 2D ones in certain circumstances. The main objective
Yang and Yin 2004; Wang and Yin 2002). Slope stability analysis of this study is to clarify the shortening of the 3D slope stability
is the main key to determining the degrees of slope failure potential analysis method and to provide a better understanding of the results
for both artificial and natural slopes, especially in mountainous of 3D slope stability analysis methods. For this purpose, eight ideal
areas. The limit equilibrium methods (LEMs) are popular methods three-dimensional sliding masses consisting of planar slip surfaces
used for 2D and 3D slope stability analysis. They are easy to apply are proposed. Safety factors are calculated and compared with the
and have enjoyed widespread use. Of these methods, the slice Hovland and exact solution methods as well. Furthermore, the re-
method used with a circular failure surface is the most popular be- sults are also compared with the 3D simplified Janbu method,
cause circles are convenient for analysis and often approximate the which is more accurate than the Hovland method.
observed failure surface. However, there are many cases where the
popular slip circle method may not be appropriate, and noncircular
failure surfaces or nonlimit equilibrium methods must be em- Background
ployed. Generally, in the case of landslide stability analysis, some
elements are three-dimensional, such as topography, shape of slip LEMs in 2D or 3D analysis are considered traditional approaches to
surface, geological and terrene structure, and distribution of water evaluating the stability performance of soil slopes as well as earth-
pressure on the sliding surface. To obtain a much more accurate work structures. Many investigators have used limit equilibrium
safety factor, it is necessary to use a three-dimensional slope sta- analysis to study the stability of earthwork structures and soil
bility analysis method. As a type of 3D slope stability analysis slopes under static and seismic conditions (Basha and Babu
method, the Hovland method is used widely in Japan. It is easy 2010; Chen et al. 2004; Wang et al. 2004; Wang and Yin 2002).
to deduce the derivation and formula for the Hovland method. Three-dimensional LEMs represent straightforward extensions of
the 2D methods of slices. Therefore, any errors resulting from
1
Associate Professor, Civil Engineering Dept., Beni-Suef Univ., Shark the assumptions of 2D methods may be extended to 3D methods.
El-Nile, New Beni Suef City, Beni-Suef, Egypt. Formerly, Researcher, Baligh and Azzouz (1975) presented a method that extended the
Civil Engineering Dept. Graduate School of Engineering, Gunma Univ., concepts of the 2D circular arc shear failure method to 3D slope
Tenjin-1, Kiryu-shi, Gunma, 376-8515, Japan (corresponding author). stability problems. The method was appropriate only for slopes
E-mail: aly_76@hotmail.com in cohesive soil. The results obtained by the method showed that
2
Professor, Civil and Environmental Engineering Dept., Graduate the 3D effects could lead to a 4 to 40% increase in the safety factor.
School of Engineering, Gunma Univ., Japan. Hovland (1977) proposed a general 3D method for cohesion-
3
Ph.D. Candidate, Civil and Environmental Engineering Dept., frictional soils. The method was an extension of the 2D ordinary
Graduate School of Engineering, Gunma Univ., Japan.
method of slices (Fellenius 1927). Not all intercolumn forces and
Note. This manuscript was submitted on November 16, 2009; approved
on February 9, 2011; published online on February 11, 2011. Discussion pore-water pressures were taken into account. Two special cases
period open until September 1, 2012; separate discussions must be sub- were analyzed: a cone-shaped slip surface on a vertical slope and
mitted for individual papers. This paper is part of the International Journal a wedge-shaped slip surface. It was shown that the 3D safety factors
of Geomechanics, Vol. 12, No. 2, April 1, 2012. ©ASCE, ISSN 1532- were generally higher than the 2D ones, and the ratio of safety fac-
3641/2012/2-81–89/$25.00. tor in 3D to that in 2D was quite sensitive to the magnitudes of

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012 / 81

Int. J. Geomech., 2012, 12(2): 81-89


cohesion and friction angles and to the shape of slip surface in three mobilization on those contacts. The overall safety factor was cal-
dimensions. Chen and Chameau (1982) extended Spencer’s 2D culated based on the force equilibrium of the individual block and
method to 3D (Spencer 1967). The sliding mass was assumed to the entire block system as well. Due to the assumed interblock
be symmetrical and divided into several vertical columns. The in- boundary pattern, the method was not fully applicable for dense
tercolumn forces had the same inclination throughout the mass, and sands or overly consolidated material under drained conditions.
the shear forces were parallel to the base of the column. It was Though there has been much more success in extending the meth-
shown that the configuration of a sliding mass in 3D had significant ods of slices of 2D limit equilibrium analysis to column methods
effects on the safety factor when the length of the sliding mass was for analyzing 3D problems of slope stability, many assumptions
small, that for gentle slopes the dimensional effects were significant must be made to render the problem statically determinate in
for soils with high cohesion and low friction angles, and that in the slice methods. Lam and Fredlund (1993) balanced the number
certain circumstances the 3D safety factor for cohesion-less soils of equations that could be established from physical and mechani-
may be slightly less than the 2D one. cal requirements with the number of unknowns involved in equa-
Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

Hungr (1987) directly extended Bishop’s simplified 2D method tions. They found that, for a failure mass divided into n rows and m
of slices to analyze the slope stability in three dimensions. The columns, a total of 8 mn assumptions are required. Chen and Ugai
method was derived based on two key assumptions: the vertical (2008) introduced the development of recent studies on a 3D slope
forces of each column were neglected and both the lateral and stability analysis method based on LEMs.
the longitudinal horizontal force equilibrium conditions were To sum up, there are still shortenings of unified understanding
neglected. Hunger et al. (1989) presented a comparison of 3D for LEMs methods, and the applications of the aforementioned
Bishop and Janbu simplified methods with other published limit
methods need to be analyzed, although the LEMs have been
equilibrium solutions. It was concluded that Bishop’s simplified
steadily refined and numerous methods have been proposed. To
method might be conservative for some slopes with nonrotational
provide a better understanding of these methods, 3D safety factors
and asymmetric slip surfaces. The method appeared reasonably
of idealized slopes composed of plane sliding surfaces were calcu-
accurate in the important class of problems involving composite
lated precisely and compared based on the 3D simplified Janbu
surfaces with weak basal planes. Ugai (1988) proposed new 3D
slice methods to analyze 3D slope stability by extending the method and Hovland method.
Fellenius method, the simplified Bishop method, the simplified
Janbu method, and Spencer’s method to three dimensions. Results
showed that the proposed 3D analytical methods predicted the fail-
Safety Factor Formulas Based on Hovland and
ure conditions and the positions of the 3D failure surfaces more
Janbu Methods
accurately than the conventional 2D methods.
A 3D sliding situation is illustrated in Fig. 1. The topography of the
Using the column method, Lam and Fredlund (1993) extended
slope may be complex, and a simpler one is assumed. The sliding
the 2D general limit equilibrium formulation (Fredlund and Krahn
direction of the sliding mass is assumed to be in y  z plane
1977) to analyze a 3D slope stability problem. Intercolumn force
(perpendicular to the x-axis). The sliding mass is divided into sev-
functions of an arbitrary shape were proposed to simulate various
eral columns by many planes parallel to the y  z or x  z plane.
directions for the intercolumn resultant forces. All the intercolumn
shear forces acting on the various faces of the column were as- A single column element and the forces acting on it are shown in
sumed to be related to their respective normal forces by the inter- Fig. 2. ΔW is the weight of the column element, ΔT and ΔN
column force functions. A geostatistical procedure (i.e., the kriging are respectively the shear force and the total normal force acting
technique) was used to model the geometry of a slope, the stratig- on the base of the column element, and ΔQ is the summation
raphy, the potential slip surface, and the pore-water pressure con- of intercolumn forces acting on the sides of the column element.
ditions. It was found that the 3D safety factors determined by the In Fig. 2, αxz and αyz are respectively the inclinations of the base
method (Lam and Fredlund 1993) were relatively insensitive to the of the column with respect to the x- and y-axes. More details about
form of the intercolumn force functions used in the method. Huang the 3D geometries of the slope masses, falling mass and falling
and Tsai (2000) proposed a 3D slope stability analysis method with surface were illustrated by Ugai (1988). The directions of the ap-
the extension of a simplified 2D Bishop method. The method took plied forces of ΔW, ΔT, and ΔN are given below by the following
into account 2D moment equilibrium. With this method, the pos- unit vectors:
sible direction of sliding for a semispherical failure surface or a
composite failure surface was a part of the analytical solution,
rather than an assumption. As a result, the possible errors associ-
ated with assuming a plane of symmetry in 3D stability analyses
were eliminated. The tedious work on the transformation prior to
the analysis could also be excluded. It was shown that the method
could generate safety factors comparable to those obtained using
other existing methods for symmetrical rotational failure surfaces.
Chang (2002) developed a 3D method of analysis of slope sta-
bility on the basis of the sliding mechanism observed in the 1988
failure of the Kettleman Hills landfill slope and the associated
model studies. By adopting a limit equilibrium concept, the method
assumed the sliding mass as a block system in which the contacts
between blocks were inclined. The lines of intersection of the block
contacts were assumed to be parallel, which enabled the sliding
kinematics. In consideration of the differential straining between
blocks, the shear stresses on the slip surface and the block contacts
Fig. 1. Division of a 3D sliding mass into columns
were evaluated on the basis of the degree of shear strength

82 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012

Int. J. Geomech., 2012, 12(2): 81-89


cJΔA þ ΔN tan φ
ΔT ¼ ð3Þ
F

where c the cohesion and φ the angle of friction.

Hovland Method

In the Hovland method, equilibrium of all forces acting on the en-


tire soil mass is achieved by taking the summed scalar length along
the slip surface (not all equilibrium conditions are satisfied). The
safety factor formula by the Hovland method is shown as Eq. (4):
Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

PP
fcJΔA þ ðΔW=J  uJΔAÞ tan φg
Fh ¼ PP ð4Þ
Fig. 2. Force acting on a divided column ΔW sin αyz

   qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
∂z
ΔW : ð0; 0; 1Þ; ΔT : 1=J 0 ; 0; J0 ; where ΔA ¼ Δx:Δy, ΔW ¼ γzΔA, J ¼ 1 þ tan2 αxz þ tan2 αyz ,
∂x
   z is the vertical coordinate (positive upward), x and y are the hori-
∂z zontal coordinates, and the positive direction of the y-axes is the
ΔN : J 0 ; 1=J 0
∂x reverse of the sliding direction.
A case is referred to as the Hovland method (submerged weight
method) if part of the weight of the column is treated as submerged
where
weight. The relation of these two methods is the same as the sim-
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
∂z 2 plified and modified simplified methods. On the basis of the
J0 ¼ 1þ ð1Þ Hovland method (submerged weight method) (Hovland 1977),
∂x the safety factor formula is shown as follows, as presented by Ugai
(1988):
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
∂z2 ∂z2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
J¼ 1þ þ ¼ 1 þ tan2 αxz þ tan2 αyz ð2Þ P Ph ðΔW  uΔAÞ tan φi
∂x ∂y cJΔA þ
F hw ¼ PP J ð5Þ
ΔW sin αyz
Z(X, Y) is the equation of the sliding surface.
The direction of ΔQ is assumed on the basis of slice method, as qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
shown in Figs. 3(a)–3(c), where ΔQ1 and ΔQ2 are respectively the where ΔA ¼ Δx:Δy, ΔW ¼ γzΔA, J ¼ 1 þ tan2 αxz þ tan2 αyz .
projections of ΔQ on the y  z and x  z planes. Fig. 3(a) shows
the assumption of a 3D simplified column method, where the in-
clinations of ΔQ1 and ΔQ2 are respectively and with respect to Three-Dimensional Simplified Janbu Method
the y- and x-axes. That is, in the 3D simplified column method,
ΔQ is assumed to be parallel to the base of the column element. The 3D simplified Janbu method is explained here, which can be
Fig. 3(b) shows the assumption of the Hovland method, where ΔQ applied to any shape of slip surface. The safety factor F is deter-
is assumed to be zero, which means there are no side forces in the mined by the simultaneous solution of equilibrium equations in the
3D Hovland method. Fig. 3(c) shows the assumption of the 3D vertical and horizontal directions in the 3D simplified Janbu
simplified Janbu method, where ΔQ1 is parallel to the x-axes method. According to previously calculated cases, it is known that
and ΔQ2 is inclined by tan1 ðη tan αxz Þ with respect to the x-axes. is guaranteed to obtain sufficiently accurate results in practice even
η is an unknown constant. On the basis of these assumptions, ΔT if η ¼ 0 is substituted into the 3D simplified Janbu method (the
and ΔN, the failure condition, can be derived from the equilibrium approximate formula) to calculate the safety factor under the con-
of the forces in the direction perpendicular to the shadowed plane dition that the inclination of the slope is below 45° or that a sandy
comprised of ΔQ1 , ΔQ2 , and ΔQ. The failure condition, in com- slope is present. Thus, the safety factor formula is described accord-
bination with the definition of the safety factor F, is written as ing to Ugai (1988) as follows:

Fig. 3. Assumptions of a 3D column method: (a) 3D simplified column method; (b) 3D Hovland method (Q = 0); (c) 3D simplified Janbu method

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012 / 83

Int. J. Geomech., 2012, 12(2): 81-89


P Pn½ðcu tan φÞΔAþΔW tan φo Case C (Half Width of Sliding Mass of Case A)
cos αyz mα
Fj ¼ PP ð6Þ Case C has the same condition as Case A except that its width is
ΔW tan αyz
one half that of Case A. The plan view is shown in Fig. 4(d).
1 sin αyz tan φ
where mα ¼ þ . Case D (Case C with Water Pressure)
J Fj
The plan view is shown in Fig. 4(e). The position of the ground-
water level is the same as in Case B.
Application to Idealized 3D Sliding Mass
Case E (Changing Gradient of Sliding Surface of
As was mentioned previously, eight ideal sliding soil masses are Case A)
proposed whose slip surfaces consist of several planes. Safety fac-
tors are calculated and compared by the exact solution, the Hovland Section and plan views of Case E are shown in Fig. 4(f). As shown
in the section view, the gradient of the sliding surface on the y  z
Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

method, and the 3D simplified Janbu method. Characteristics for


the suggested eight cases are illustrated and presented below. Also, plane varies from 1:4 to 1:2. The factor of water pressure is not
a brief description of the sliding surface, friction angle, and state of taken into account. The shear resistance can be neglected, which
groundwater of soil mass used are summarized and presented in is on the plane parallel to the x  z plane via the dashed line
Table 1 for different case studies. The scales of cross-sections (PQ lines) in the section view.
and plan views may not match in each graph. All the Unit weights
Case F (Changing Lateral Gradient of Sliding Surface
of the sliding soil masses, γ, are 18kN=m3 .
of Case A)
Section and plan views of Case F are shown in Fig. 4(g). As shown
Descriptions for Suggested Case Studies in the plan view, the gradient of the sliding surface varies from 1:2
to 1:1, with a distance being 25 m in the x-direction to the center
Case A (3D Wedge-Shaped Slip Surface) sections on the x  z plane. Consideration of the water pressure is
excluded. The entire soil mass is assumed to move downward be-
An example of a 3D planar wedge-shaped slip surface is shown in cause of the shear strength on the plane parallel to the y  z plane
Fig. 4(a). The cross-section is shown on the left; the plan view is through to the dashed line (RS lines) in the section view.
shown on the right. A 3D view is shown in Fig. 4(b). Case A is
treated as the standard example of this study in order to explain Case G (Case A with a Horizontal Seismic Force)
how to calculate the safety factor. This example shows left-right
Case G is assumed with a horizontal seismic force added to Case A,
symmetry; therefore, the force equilibrium of the sliding soil mass
and the horizontal seismic coefficient K h is assumed to be 0.2 g.
of the right half is chosen to be considered only as illustrated in
Section and plan views are shown in Fig. 4(h), and forces acting on
Fig. 4(a) (to the right of Γ). The volume of the sliding soil mass
the subdivided column are illustrated in Fig. 4(i).
(right half) is 50 × 25 × 1=2 × 50 × 1=2 ¼ 10;417 m2 , so the
weight of the sliding mass is 10;417 × 18 ¼ 187;506 kN, where Case H (Case A with an Anchor Force)
18kN=m3 is the unit weight as mentioned previously. The follow-
ing results can be obtained: ∂ z =∂ y ¼ 12:5=50 ¼ 0:25; ∂ z =∂ x ¼ The case with anchor force P added to Case A is shown in Fig. 4(j),
where P is assumed to be 10,000 kN and is inclined by tan1 ð2Þ
12:5=50 ¼ 0:25 on the right side of the sliding surface, and
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi with respect to the x  y plane.
J¼ 1 þ tan2 αxz þ tan2 αyz ¼ 1:061. Therefore, the area of one
side of the slip surface is A ¼ 50 × 50 × J ¼ 2;652:5 m2 .
Calculation of Safety Factor for Case A Using
Case B (Case A from Water Pressure) Different Methods
Case B differs from Case A in terms of the assumed water pressure.
Safety Factor on Basis of Exact Solution Method for
Section and plan views are shown in Fig. 4(c). The dashed line in
Case A
the section view shows the groundwater level. The water table is
parallel to the sliding surface and located at the center plane be- In what follows, external forces and their directions are described as
tween the slope shoulder and the slip surface. For simplicity, the vectors acting on the sliding mass (right half).
water pressure on the slip surface is treated as the product of Weight: ΔWð0; 0; 1Þ
the unit weight of water and the depth from the water table to Force of the x-direction acting on the Γ cross-section:
the slip surface. ΔQð1; 0; 0Þ

Table 1. Characteristics of Soil Mass for Different Case Studies


Case Sliding surface Friction angle Water pressure
A Standard wedge-type sliding surface 13.56° —
B Case A with water pressure 13.56° Available
C Half the width of the sliding mass of Case A 13.56° —
D Case C with water pressure 13.56° Available
E Changing gradient of sliding surface for Case A 13.56° —
F Changing lateral gradient of sliding surface for Case A 24.71° —
G Case A with a horizontal seismic force 24.71° —
H Case A with an anchor force 13.56° —

84 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012

Int. J. Geomech., 2012, 12(2): 81-89


Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. (a) 3D wedge-shaped sliding surface (Case A); (b) 3D view of wedge-shaped sliding surface; (c) 3D wedge-shaped sliding surface (with water
pressure) (Case B); (d) 3D wedge-shaped sliding surface; (e) 3D wedge-shaped sliding surface; (f) changing gradient of sliding surface of Case A
(Case E); (g) changing lateral gradient of sliding surface of Case A (Case F); (h) 3D wedge-shaped sliding surface (Case G); (i) forces acting on a
divided column adding a horizontal force; (j) 3D wedge-shaped sliding surface (Case H)

Normal force acting on sliding surface: Values are assigned:


    17:15 ¼ c þ 70:6 tan φ ð8Þ
∂z=∂x ∂z=∂y 1 0:25 0:25 1
ΔN  ; ; ¼ ΔN  ; ; Safety factor F would be 1.000 when factors c and ϕ satisfy
J J J 1:061 1:061 1:601
Eq (8). An example of such a combination is shown as follows.
Shear resistance acting on sliding surface:
  c ¼ 17:15 kN=m2 ; tan φ ¼ 0 :
4 4
ΔT 0; pffiffiffiffiffi ; pffiffiffiffiffi ð9Þ
17 17 when friction angle φ ¼ 0
Three equilibrium equations are derived in the x-, y-, and
z-directions acting on the sliding mass (right half): c ¼ 0 kN=m2 ; tan φ ¼ 0:2429 :
x-direction: ΔQ  0:25=1:061ΔN p ¼ffiffiffiffiffi
0 ð10Þ
y-direction: 0:25=1:061ΔN þ 4= 17p ΔT c¼0
ffiffiffiffiffi ¼ 0
when cohesion
z-direction: ΔW þ 1=1:061ΔN þ 4= 17ΔT ¼ 0
From the three equations mentioned above:
ΔT ¼ 45;477 kN, ΔN ¼ 187;242 kN, ΔQ ¼ 44;120 kN Safety Factor on Basis of Hovland Method for Case A
If a critical sliding state (F ¼ 1) is assumed, the following equa- Values are assigned to the parameters in Eq. (4), and then the fol-
tion is derived: lowing relation can be obtained:

ΔT ¼ cA þ ΔN tan φ ð7Þ F h ¼ 0:05828c þ 3:887 tan φ ð11Þ

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012 / 85

Int. J. Geomech., 2012, 12(2): 81-89


Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. (Continued).

Some results are obtained by substituting Eqs. (9) and (10) into For different case studies from B to F, the calculation method is
Eq. (11), respectively: the same as in Case A. Therefore, calculation details for these cases
are not presented herein. All calculated values of safety factors for
c ¼ 17:15 kN=m2 ; tan φ ¼ 0 → F h ¼ 1:000; different case studies using different solution methods including
c ¼ 0; tan φ ¼ 0:2429 → F h ¼ 0:944 Hovland, Hovland (submerged weight), and the 3D simplified
Janbu are tabulated in Table 2. For Case B as shown in Table 2,
According to the results, the effect of the degree of friction angle the accuracy of the safety factor is somewhat limited when it is
on the safety factor is more than the cohesion c. A slightly lower calculated with water pressure by the Hovland method; it is the
safety factor is given by the Hovland method in the case of c = 0; same using the 3D simplified Janbu method and the exact solution.
therefore, the situation of c ¼ 0 is only taken into account in the
following cases. The safety factor is 0.972 in the 2D case, and it is
irrational because it is higher than the 3D case. Table 2. Safety Factors for Different Cases Studies

Safety Factor on Basis of 3D Simplified Janbu Method Solution method


for Case A Hovland 3D
Case Hovland (submerged weight) simplified Janbu
Assign values to the parameters in Eq. (4). Then the following re-
lation can be obtained among F j , c, and φ: A 0.944 (F 2 ¼ 0:972) Same as left 1.00 (F 2 ¼ 0:972)
B 0.924 0.966 1.00
F j ¼ 0:05833c þ 4:118 tan φ ð12Þ
C 0.809 (F 2 ¼ 0:899) Same as left 1.00 (F 2 ¼ 0:899)
Some results are obtained by substituting Eqs. (9) and (10) into D 0.737 0.827 1.00
Eq. (12), respectively: E 0.948 Same as left 1.00
F 0.934 Same as left 1.00
c ¼ 17:15 kN=m2 ; tan φ ¼ 0 → F j ¼ 1:000; G 0.944 Same as left 1.00
c ¼ 0; tan φ ¼ 0:2429 → F j ¼ 1:000 H 0.944 Same as left 0.997
Note: Safety factor F using the exact solution is the same as in Case A.
It is obvious from this result that the 3D simplified Janbu The results are shown in this table merely considering the case of cohesion
method gives the same result as the exact solution. c ¼ 0 on a sliding surface.

86 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012

Int. J. Geomech., 2012, 12(2): 81-89


It is clear from this result that the Hovland method (submerged Safety Factor on Basis of Hovland Method for Case G
weight) gives a slightly higher safety factor than the Hovland
The descriptions of external forces and their directions are pre-
method in this case, whereas in Case C, the precision of results
sented herein as vectors acting on the sliding mass as described
diminishes under the Hovland method because of its narrow width,
previously in the case of the exact solution method.
as presented in Table 2.
Weight: ΔWð0; 0; 1Þ
In the case of narrow width and water pressure as in Case D, as
Horizontal seismic force: K h ΔWð0; 1; 0Þ
shown in Table 2, the safety factor of 0.737 decreases significantly
Force perpendicular to sliding surface:
under the Hovland method and that of 0.827 also decreases by the
Hovland method (submerged weight). On the other hand, in Case D    
∂z=∂x ∂z=∂y 1 0:25 0:25 1
also, the 3D simplified Janbu method gives the same safety factor ΔN  ; ; ¼ ΔN  ; ;
J J J 1:061 1:061 1:601
as the exact solution and the same results are obtained in both Cases
E and F, as shown in Table 2. The equilibrium equation is derived as follows, acting on the
Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

plane perpendicular to the sliding surface:


Calculation of Safety Factor for Case G Using
Different Methods ΔW kh ΔW × ð∂z=∂yÞ
ΔN ¼  ð17Þ
J J
Safety Factor Based on Exact Solution Method for
Case G According to the definition of safety factor F, F is the ratio of
the summation of strength to the summation of mobilized strength,
As mentioned previously, a horizontal seismic force is considered that is:
in Case G, and the horizontal seismic coefficient K h is assumed to P
be 0.2g. External forces and their directions are described below as available resistance
vectors acting on the sliding mass (right half). F¼P
mobilized resistance
Weight: ΔWð0; 0; 1Þ
Horizontal seismic force: K h ΔWð0; 1; 0Þ Thus, the following equation is derived:
Force of x-direction acting on Γ cross-section: ΔQð1; 0; 0Þ
Normal force acting on sliding surface: ðcJΔA þ ΔN tan φÞ
F 0h ¼ ð18Þ
    ΔT
∂z=∂x ∂z=∂y 1 0:25 0:25 1
ΔN  ; ; ¼ ΔN  ; ; The following equation is derived by substituting Eq. (17) into
J J J 1:061 1:061 1:601
Eq. (18):
Shear resistance acting on sliding surface:  
PP ΔWð1  K h tan αyz Þ
  cJΔA þ
J
4 1
ΔT 0; pffiffiffiffiffi ; pffiffiffiffiffi F 0h ¼ P P ð19Þ
ΔWðsin αyz þ K h cos αyz Þ
17 17

Three equilibrium equations are derived in the x-, y-, and Assign values to the parameters in the expression of the safety
z-directions acting on the sliding mass (right half). factor using the Hovland method. Then the following relation can
x-direction: ΔQ  0:25=1:061ΔN ¼ 0 pffiffiffiffiffi be obtained:
y-direction: K h ΔW  0:25=1:061 þ 4 17ΔT ¼ 0
pffiffiffiffiffi F 0h ¼ 0:032404c þ 2:05097 tan φ ð20Þ
z-direction: ΔW þ 1=1:061 þ 1 17ΔT ¼ 0
From the foregoing three equations: Some results are obtained by substituting Eqs. (15) and (16) into
Eq. (20), respectively:
ΔT ¼ 81;858 kN; N ¼ 177;979 kN; Q ¼ 41;913 kN
c ¼ 30:86 kN=m2 ; tan φ ¼ 0 → F 0h ¼ 1:000
If the critical sliding state (F ¼ 1) is assumed, the following
equation would be derived as c ¼ 0; tan φ ¼ 0:4602 → F 0h ¼ 0:944

ΔT ¼ cA þ ΔN tan φ ð13Þ It can be observed that the result is the same as the case without a
horizontal seismic force by the Hovland method.
Values are assigned as follows:
Safety Factor on Basis of 3D Simplified Janbu Method
32:74 ¼ c þ 71:15 tan φ ð14Þ for Case G
As mentioned previously, the result retains a high precision under
The safety factor is F ¼ 1 when the factors c and meet the the 3D simplified Janbu method. Here, an approximate formula
Eq. (14). What follows is an example of such a combination. [Eq. (21)] is used to simplify the calculation as follows:
c ¼ 32:74 kN=m2 , tan φ ¼ 0 when the friction angle  
P P ðc  u tan φÞΔA þ ΔW tan φ
φ¼0 ð15Þ cos αyz :mα
F 0j ¼ PP ð21Þ
ΔWðtan αxz þ K h Þ
c ¼ 0; tan φ ¼ 0:4602 when cohesion
Assign values of the parameters in Eq. (21) to the expression of
c¼0 ð16Þ the safety factor using the 3D simplified Janbu method. Then the

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012 / 87

Int. J. Geomech., 2012, 12(2): 81-89


following relation can be obtained among F 0j , c, and φ: Three equilibrium equations are derived in the x-, y-, and
z-directions acting on the sliding mass (right half):
F 0j ¼ 0:032404c þ 2:173011 tan φ ð22Þ
0:25
x direction: ΔQ  ΔN ¼ 0
Some results are obtained by substituting Eqs. (15) and (16) into 1:061
Eq. (22), respectively: 1 0:25 4
y direction:  pffiffiffi P  ΔN þ pffiffiffiffiffi ΔT ¼ 0
c ¼ 30:86 kN=m2 ; tan φ ¼ 0 → F j ¼ 1:000 5 1:061 17
c ¼ 0; tan φ ¼ 0:4602 → F 0j ¼ 1:000 2 1 1
z direction: ΔW  pffiffiffi P þ ΔN þ pffiffiffiffiffi ΔT ¼ 0
5 1:061 17
From this result, one can conclude that the 3D simplified Janbu
method gives the same result as the exact solution in this case. From the preceding three equations
ΔT ¼ 43;308 kN; N ¼ 197;289 kN=; Q ¼ 46;487 kN
Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

Calculation of Safety Factor for Case H Using If a static sliding state (F ¼ 1) is assumed, the following equa-
Different Methods tion is derived:
ΔT ¼ cA þ ΔN tan φ ð23Þ
Safety Factor on Basis of Exact Solution Method for
Case H Values are assigned as follows:
As mentioned previously, this case deals with the external forces,
16:33 ¼ c þ 74:4 tan φ ð24Þ
and their directions are described below as vectors acting on the
sliding mass (right half). The safety factor F is 1.000 when c and φ satisfy Eq. (24). What
Weight: ΔWð0; 0; 1Þ follows is an example in this condition:
Force of x-direction acting on cross-section: ΔQð1; 0; 0Þ
Normal force acting on sliding surface: c ¼ 16:33 kN=m2 ; tan φ ¼ 0 ð25Þ
   
∂z=∂x ∂z=∂y 1 0:25 0:25 1 when the friction angle φ ¼ 0
ΔN  ; ; ¼ ΔN  ; ;
J J J 1:061 1:061 1:601
c ¼ 0; tan φ ¼ 0:2195 ð26Þ
Shear resistance acting on sliding surface:
when cohesion c ¼ 0.
 
4 1
ΔT 0; pffiffiffiffiffi ; pffiffiffiffiffi Safety Factor on Basis of Hovland Method for Case H
17 17
Anchor force: Pð0; cos θ;  sin θÞ
Anchor force: Pð0; cos θ;  sin θÞ  tan α  tan αyz 1
xz
  Perpendicular to sliding surface: ΔN  ; ;
1 2 J J J
¼ P 0; pffiffiffi ;  pffiffiffi According to the safety factor formula of the Hovland method
5 5 with anchor reinforcement,

    XX  
PP ΔW tan αyz cos θ sin θ
cJΔA þ  uJΔA tan φ þ P þ tan φ
J J J
F 00h ¼ PP PP ð27Þ
ΔW sin αyz  P cos θðαyz þ θÞ

Assign the values of the parameters in the expression of the Safety Factor on Basis of 3D Simplified Janbu Method
safety factor using the Hovland method. Then the following rela- for Case H
tion can be obtained:
In the 3D simplified Janbu method, the equilibrium equation of
forces in the vertical direction is derived as follows:
F 00h ¼ 0:06125c þ 3:862 tan φ ð28Þ

ΔN
Some results are obtained by substituting Eqs. (25) and (26) into ΔW þ P sinðαyz þ θÞ cos αyz ¼ þ ΔS sin αyz ð29Þ
Eq. (28), respectively: J

c ¼ 16:33 kN=m2 ; tan φ ¼ 0 → F h ¼ 1:000


From the definition of the 3D safety factor:
c ¼ 0; tan φ ¼ 0:2195 → F h ¼ 0:944

cJΔA þ ΔN tan φ
Obviously, the results are the same as in the case without an ΔS ¼ ð30Þ
anchor force under the Hovland method. F 00j

88 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012

Int. J. Geomech., 2012, 12(2): 81-89


From Eqs. (27) and (28): 1 sin αyz tan φ
where mα ¼ þ
1 h c i j F 00j
ΔN ¼ ΔW þ P sinðαyz þ θÞ cos αyz  00 ΔAJ sin αyz
mα Fj
From the global equilibrium of horizontal forces:
ð31Þ
XX XX X X ΔN tan αyz
1 ΔS cos αyz þ P cos θ ¼ ð33Þ
ΔS ¼ fcΔA þ ½ΔW þ P: sinðαyz þ θÞ cos αyz  tan φg J
mα :F 00j
The safety factor can be drawn by substituting Eqs. (31)
ð32Þ and (32) into (33):

PP
Downloaded from ascelibrary.org by University of Western Ontario on 06/01/16. Copyright ASCE. For personal use only; all rights reserved.

1
cos αyz :mα fcΔA þ ½ΔW þ P sinðαyz þ θÞ cos αyz  tan φg
F 00j ¼ PP ð34Þ
½ΔW tan αyz  P cos αyz cosðαyz þ θÞ

where mα ¼ 1J þ sin αyz tan φ=F 00j . lines of intersection and differential straining of block contacts.” Can.
Assign values to the parameters in the expression of the safety Geotech. J., 39(4), 799–811.
factor using the 3D simplified Janbu method Then the following Chen, J., Yin, J. H., and Lee, C. F. (2004). “Rigid finite element method for
relation can be obtained: upper bound limit analysis of soil slopes subjected to pore water pres-
sure.” J. Eng. Mech., 130(8), 886–893.
F 00j ¼ 0:06107c þ 4:554 tan φ ð35Þ Chen, Z., and Ugai, K. (2008). “Limit equilibrium and finite element
analysis-A perspective of recent advances.” Proc., 10th Int. Symp.
When the factors c and φ satisfying Eqs. (25) and (26) are as- on Landslides and Engineered Slopes, Xian, China.
signed to Eq. (35), some results would be obtained as follows: Chen, R., and Chameau, J. L. (1982). “Three-dimensional limit equilibrium
analysis of slopes.” Geotech. J., 33(1), 31–40.
c ¼ 16:33 kN=m2 ; tan φ ¼ 0 → F j ¼ 0:997 Fellenius, W. (1927). “Erdstatische Berechnungen mit Reibung und
Kohasion.” Ernst, Berlin (in German).
c ¼ 0; tan φ ¼ 0:2195 → F 00j ¼ 1:000 Fredlund, D. G., and Krahn, J. (1977). “Comparison of slope stability meth-
ods of analysis.” Can. Geotech. J., 14(3), 429–439.
Generally, it can be stated that the 3D simplified Janbu method Hovland, H. J. (1977). “Three-dimensional slope stability analysis
almost gives the same value as the exact method. method.” J. Geotech. Engrg. Div., 103(9), 971–986.
Huang, C. C., and Tsai, C. C. (2000). “New method for 3D and asymmet-
rical slope stability analysis.” J. Geotech. Geoenviron. Eng., 126(10),
Conclusions 917–927.
Hunger, O., Salgado, F. M., and Byrne, P. M. (1989). “Evaluation of a
Ideal 3D sliding masses were assumed that consisted of several pla- three-dimensional method of slope stability analysis.” Can. Geotech.
nar slip surfaces, and their safety factors were calculated and com- J., 26(4), 679–686.
pared on the basis of the exact solution, the Hovland method, and Hungr, O. (1987). “An extension of Bishop’s simplified method of slope
the 3D simplified Janbu method. The results showed that the safety stability analysis to three dimensions.” Geotechnique J., 37(1),
factor is small under the Hovland method with low cohesion c and 113–117.
large friction angle, especially in cases of narrow failure width and Lam, L., and Fredlund, D. G. (1993). “A general limit equilibrium model
for three-dimensional slope stability analysis.” Can. Geotech. J., 30(6),
high water pressure on the sliding surface. The results also indi-
905–919.
cated that it was irrational that the 3D safety factor was smaller Michalowski, R. L. (2010). “Limit analysis and stability charts for 3D slope
than the 2D safety factor under the Hovland method. Furthermore, failures.” J. Geotech. Geoenviron. Eng., 136(4), 583–593.
it was shown that the 3D simplified Janbu method has as high a Spencer, E. (1967) “A method of analysis of the stability of embankments
precision as the exact solution. assuming parallel interslice forces.” Geotech. J., 17(1), 11–26.
Note: Safety factor F using the exact solution is the same as in Ugai, K. (1988). “Three-dimensional slope stability analysis by slice meth-
Case A. The results are shown in this table considering only the ods.” Proc., 6th Int. Conf. on Numerical Methods in Geomechanics,
case of cohesion c ¼ 0 on the sliding surface. F 2 is the 2D safety A.A. Balkema, Rotterdam, The Netherlands, 1369–1374.
factor in the center section. Wang, Y. J., and Yin, J. H. (2002). “Wedge stability analysis considering
dilatancy of discontinuities.” Rock Mech. Rock Eng. J., 35(2), 127–137.
Wang, Y. J., Yin, J. H., Chen, Z. Y., and Lee, C. F. (2004). “Analysis of
References wedge stability using different methods.” Rock Mech. Rock Eng. J.,
37(2), 127–150.
Baligh, M. M., and Azzouz, A. S. (1975). “End effects on stability of Wu, J. H., Ohnishi, Y., Shi, G. H., and Nishiyama, S. (2005). “Theory of
cohesive slopes.” J. Geotech. Engrg. Div., 101(11), 1105–1118. three-dimensional discontinuous deformation analysis and its applica-
Basha, B. M., and Babu, G. L. S. (2010). “Seismic rotational displacements tion to a slope toppling at Amatoribashi, Japan.” Int. J. Geomech., 5(3),
of gravity walls by p (I.c) seudodynamic method with curved rupture 179–195.
surface.” Int. J. Geomech., 10(3), 93–105. Yang, X. L., and Yin, J. H. (2004). “Slope stability analysis with nonlinear
Chang, M. (2002). “A 3D slope stability analysis method assuming parallel failure criterion.” J. Eng. Mech., 130(3), 267–273.

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MARCH/APRIL 2012 / 89

View publication stats Int. J. Geomech., 2012, 12(2): 81-89

You might also like