Carranza-Torres, C. and E. Hormazabal (2018).

Computational tools for the

determination of factor of safety and location of the critical failure surface
for slopes in Mohr-Coulomb dry ground. In Proceedings of the International
Symposium Slope Stability 2018. Seville, Spain. April 11–13, 2018.
Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

Computational tools for the determination of factor of safety and

location of the critical circular failure surface for slopes in
Mohr-Coulomb dry ground

C. Carranza-Torres1∗ & E. Hormazabal2

1
Department of Civil Engineering, University of Minnesota, Duluth Campus, Minnesota, USA
2
SRK Consulting Chile, Santiago, Chile
∗
[email protected] (corresponding author’s E-mail)

Abstract
This paper presents computational tools for the quick estimation of factor of safety and location of
the critical circular failure surface for simple slope problems. The analysis presented here applies
to slopes of arbitrary height and inclination angle excavated in homogeneous/isotropic dry ground,
assumed to obey the Mohr-Coulomb shear failure criterion, and characterized by arbitrary values of
unit weight, cohesion and internal friction angle. The proposed procedure is based on ideas originally
laid out in the classical book Rock Slope Engineering by Hoek and Bray (1981) and more recently
in the books by Read and Stacey (Guidelines for Open Pit Slope Design) and by Wyllie (Rock Slope
Engineering, Civil Applications) from 2009 and 2018, respectively. Development of the proposed
procedure involved computation of approximately 3,400 selected cases of slopes using the Bishop
Method in the limit equilibrium software SLIDE by Rocscience. Results obtained from the analysis
are summarized in dimensionless graphical representations that not only allow factors of safety and
location of the critical circular failure surface to be estimated, but also important, to put light into
the fundamental problem of establishing the concept of mechanical similarity of slopes excavated
in Mohr-Coulomb ground with regard to factors of safety and position of the critical circular failure
surface. In addition to the graphical representation of factor of safety, the paper also provides an
equation to compute the factor of safety of slopes based on these representations. Considering that
in the current practice of geotechnical engineering design the use of computer spreadsheets may be
preferred over the use of dimensionless charts, the proposed procedure is implemented in a simple
to use EXCEL workbook that is freely available to readers, and that allows determination of factor
of safety and location of the failure surface, as it would be obtained with the Bishop Method in the
limit equilibrium method software SLIDE. Finally, to illustrate the application of the proposed tools, a
practical example involving the analysis of stability of a slope in an actual open pit mine is provided.

Keywords: Slope stability, limit equilibrium, method of slices, Bishop method, factor of safety, Mohr-
Coulomb.

1 Introduction

Computation of factor of safety against failure and location of the critical failure surface forms the
basis of the current practice of stability analysis of slopes in civil and mining engineering, whether
carried out in a deterministic or in a probabilistic manner (see, for example, Vanmarcke 1980; Juang
et al. 1998; Steffen et al. 2008). This is particularly true for slopes in soils or in weak rocks, or for
slopes in highly jointed rock masses, for which the assumption of material continuity and isotropy
can be regarded as valid (Read & Stacey 2009; Wyllie 2018). For the case of slopes in soil and rock
(to be referred indistinctly using the term ground in the reminder of this paper), when the ground is
relatively homogeneous and isotropic, it is common practice to assume that the critical failure surface

Slope Stability 2018, Sevilla, Spain 1

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

has a circular shape (see, for example, Abramson et al. 2002; Duncan et al. 2014).
The Mohr-Coulomb shear failure criterion, characterized by internal friction angle and cohesion pa-
rameters, is perhaps the simplest material model that can be used to carry out a slope stability
analysis (see, for example, Coduto et al. 2011; Verruijt 2012; Das & Sobhan 2018). Although
the Mohr-Coulomb shear strength model is extensively used in slopes stability analyses in homo-
geneous/isotropic soils, the Hoek-Brown shear failure criterion is the material model that is gener-
ally preferred in slope stability analyses in (assumed homogeneous/isotropic) rock masses (Hoek &
Brown 1980; Hoek & Brown 1997). Conveniently, it is always possible, to compute equivalent Mohr-
Coulomb frictional parameters from Hoek-Brown parameters for the characteristic range of stresses
in the problem to be analyzed (Hoek et al. 2002).
There exists various methods for computing the factor of safety and location of the critical circular
failure surface for a slope. Among them, limit equilibrium methods, limit analysis methods and full
numerical methods (Potts & Zdravkovic 1999). The so-called strength reduction technique, normally
implemented in finite element or finite difference software, is another method for computing the
stability of a slope that has gained popularity in recent years (see, for example, Matsui & San 1992;
Dawnson et al. 1999; Griffiths & Lane 1999; Hammah et al. 2007). All these existing methods of
analysis require use of computer software for their implementation, as they are compute-intensive.
With this regard, it is remarkable that in contrast with other problems in geotechnical engineering,
there are no rigorous analytical closed-form equations available for computing the factor of safety
and position of the critical circular failure surface for slopes of arbitrary inclination angle, even for the
simplest case of a ground that obeys the Mohr-Coulomb shear failure criterion.
Despite of its introduction in the middle of the last century, the method of slices, which is a particular
formulation of the limit equilibrium method, is still the most popular method to estimate the factor of
safety and location of the critical failure surface for slopes (see, for example, Abramson et al. 2002;
Duncan et al. 2014). Within the method of slices, there exist various formulations, including the
Bishop, Janbu, Spencer and Morgenstern-Price methods (see Bishop 1955; Janbu 1954b, 1954a;
Spencer 1967; Morgenstern & Price 1965, respectively). The Bishop method is a popular one since
it gives similar results (i.e., factors of safety and location of the failure surface) as those obtained
with more elaborated methods (e.g., the Spencer and the Morgenstern-Price methods) with less
computer effort (Abramson et al. 2002). Implementation of the Bishop method of slices to compute
factor of safety and location of the critical circular failure surface for a slope requires use of specialized
computer software. There exist various commercial packages of this type, for example GEO05 (Fine
Inc. 2016), SLIDE (Rocscience Inc. 2015), SLOPE/W (Geo-Slope Inc. 2012), XSTABL (Interactive
Software Designs Inc. 2007).
Considering the fact that there is no closed-form solution or that compute-intensive methods are
required for the determination of factor of safety and location of critical failure surface for slopes in
Mohr-Coulomb frictional ground, many authors have proposed dimensionless charts as a means of
summarizing and conveying the results of slope stability analyses (e.g., obtained from application
of limit equilibrium or limit analysis methods) in a format that can be readily used in practice (see,
for example, Taylor 1948; Bishop & Morgenstern 1960; Spencer 1967; O’Connor & Mitchell 1977;
Cousins 1978; Hoek & Bray 1981; Michalowski 2002; Baker 2003; Steward et al. 2010). It also
has to be mentioned that other authors have gone a further step and have combined existing or
new dimensionless charts with observations of actual slope stability cases in the field, to produce
empirical dimensionless (stability) charts (see, for example, Haines et al. 1991; Sjoberg 2000; Kotze
& Bosman 2015).
Among the various proposed dimensionless stability charts for slopes in Mohr-Coulomb ground ob-
tained from limit equilibrium models, the ones presented in the classical book by Hoek & Bray (1981),
and more recently included in the books by Read & Stacey (2009) and Wyllie (2018), deserve particu-
lar attention. These stability charts use a particular form of dimensionless scaling originally proposed
by Bell (1966) that allows a compact (perhaps the most compact) representation of factor of safety for

Slope Stability 2018, Sevilla, Spain 2

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

slopes of planar face of arbitrary inclination angle and height, in a ground of arbitrary values of unit
weight, internal friction angle and cohesion. Although not explicitly mentioned in the book, the repre-
sentation proposed by Hoek & Bray (1981) forms the basis for establishing a concept of mechanical
similarity of factor of safety for slopes in Mohr-Coulomb ground.
This paper builds on and extends the dimensionless representations proposed by Hoek & Bray (1981)
for slopes in Mohr-Coulomb dry ground. In addition to re-casting the dimensionless representation
of factor of safety to be able to provide a (best fit) closed-form equation to compute factor of safety for
slopes, the paper expands on the analysis by Hoek & Bray (1981) by providing dimensionless repre-
sentations that allow to define the location of the critical circular failure surface, as obtained with the
Bishop method of slices with the software SLIDE (Rocscience Inc. 2015). These proposed dimen-
sionless representations allow the concept of mechanical similarity of factor of safety and location of
the critical circular failure surface of slopes in Mohr-Coulomb ground to be established.
Considering that in modern geotechnical engineering practice computer spreadsheets or simple
equations are preferred over graphical dimensionless representations, this paper also presents a
computer EXCEL (Microsoft 2016) workbook that summarizes the results of the proposed dimen-
sionless stability representations and that can be conveniently used to quickly determine factors of
safety and position of critical circular failure surfaces for slopes in Mohr-Coulomb dry ground.

2 Problem statement

The problem considered in this study is shown in Figure 1. A slope of inclination angle α, and height
H, is excavated in an assumed homogeneous/isotropic dry ground of unit weight γ, that obeys the
Mohr-Coulomb shear failure criterion, and that is characterized by a cohesion c and an internal
friction angle φ. The origin of a system of cartesian coordinates (x,y ) is assumed to be located at
the toe of the slope (point O in Figure 1). When the strength of the ground is affected by the factor of
safety, the slope is assumed to be in a limit state of equilibrium with a critical circular failure surface
of radius R and a center of coordinates xc and yc (point C in Figure 1). The starting point of the
failure surface (point A in the figure) has coordinates xA and yA , while the ending point of the failure
surface (point B in the figure) has coordinates xB and yB . To simplify the problem, no explicit tension
crack is considered at the ending point of the failure surface.
According to the method of slices of slope stability analysis, when failure of the slope occurs at the
critical circular failure indicated in Figure 1, the shear strength of the ground is fully mobilized on the
failure surface (see, for example, Coduto et al. 2011; Verruijt 2012; Das & Sobhan 2018). For the
case of the Mohr-Coulomb shear failure criterion considered in this study, this condition is written as
follows,
τs = σn tan φ + c [1]

where τs and σn are the shear and normal stresses (respectively) at the base of an arbitrary slice on
the failure surface and φ, and c are the internal friction angle and cohesion of the ground (respec-
tively).
In the problem considered in this study, the most common definition of factor of safety (FS) is
adopted. This states that the factor of safety is the ratio of the shear strength of the ground on
the failure surface and the shear stress required for equilibrium (see for example, Abramson et al.
2002; Coduto et al. 2011; Das & Sobhan 2018). Of the various formulations available for computing
the factor of safety, the Bishop Method (as implemented in the software SLIDE) is employed in this
study. The reader is referred to the widely available literature on slope stability analysis with limit
equilibrium methods and the method of slices for more details about the implementation of the stan-
dard methods used in this study (see, for example, Duncan 1996; Abramson et al. 2002; Duncan
et al. 2014; Huang 2014).

Slope Stability 2018, Sevilla, Spain 3

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

Figure 1. Slope excavated in dry ground assumed to obey the Mohr-Coulomb shear failure criterion.

3 Dimensionless representations of factor of safety and critical circular

failure surface

For the slope introduced in Section 2 and following the analysis presented Hoek & Bray (1981), all
three mechanical variables (γ, φ and c) and the single length-dimension variable (H) are conveniently
grouped into a single dimensionless factor, referred to in this study as X , as follows

γ H tan φ
X = [2]
c
The advantage of using the dimensionless factor X is that the ratio of factor of safety and tangent of
the internal friction angle depends on this ratio and on the inclination angle of the slope (α) only, i.e.,

FS
= fFS (X , α) [3]
tan φ

In the equation above, fFS is a function that can be ‘traced’ (i.e., reconstructed) by solving a series
of selected slope cases for properly chosen values of the variables X and α, as it will be explained
later on in this section.
It is worth noting that equations [2] and [3] are the inverse of the dimensionless factors used by Hoek
& Bray (1981). This is because this form permits a wider range of slopes to be studied, as it is also
discussed in this section.
Additionally, all ratios of variables that characterize the position of the critical circular failure surface
for the slope in Figure 1, depend on X and α only. For example, the scaled abscissa and ordinate of
the center of the critical circular failure is defined as
xc
= fxc (X , α) [4]
H
and
yc
= fyc (X , α) [5]
H
The scaled abscissa and ordinate of the starting point of the critical circular failure surface (point A

Slope Stability 2018, Sevilla, Spain 4

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

in Figure 1) is defined as
xA
= fxA (X , α) [6]
H
and
yA
= fyA (X , α) [7]
H
As with the case of the function fFS in equation [3], the functions fxc , fyc , fxA and fyA in equations [4]
through [7] can also be reconstructed by solving a series of selected slope cases for properly chosen
values of the variables X and α.
Using equations [4] through [7], the scaled radius of the critical circular failure surface (R/H) can be
computed as the distance between the points C and A in Figure 1, i.e.,
s
R xc xA 2 yc yA 2
   
= – + – [8]
H H H H H

As needed, the coordinates of the ending point B of the failure surface in Figure 1 can also be defined
as the intersection of the critical circular failure surface and the assumed horizontal plane at the crest
of the slope.
It is worth noting that equations [4] through [8] have not been provided in Hoek & Bray (1981); these
equations are included here to complete the representations originally proposed by those authors
and to be able to define the position of the critical circular failure surface.

To reconstruct the functions of the dimensionless variables in equations [3] through [7], the limit
equilibrium software SLIDE (Rocscience Inc. 2015) was employed. A total of 3,402 cases were
evaluated using the Bishop method of slices implemented in the software. The input variables in the
models were chosen so to obtain 81 equally spaced (in logarithm base-10 scale) slope cases of the
factor X , given by equation [2], between 10–2 and 100. Slopes with slope face angles α between 20◦
and 80◦ with increments of 10◦ were chosen. The factors of safety and corresponding critical circular
failure surfaces were determined for all cases using the ‘Autorefine Search’ option implemented in
the software SLIDE. Appendix A provides a detailed discussion on the characteristics of the limit
equilibrium models considered in this study.

To illustrate the reconstruction of these functions, Figure 2 shows the graphical representation of the
function fFS in equation [3], as obtained with SLIDE. The diagram defines the relationship between
the scaled factor of safety FS/ tan φ (vertical axis) and the ratio X (horizontal axis) for different face
slope angles α (the various curves in the diagram). The small dots on the curves represent the
actual SLIDE cases that were evaluated. Moving towards the left side of the horizontal axis in Figure
2 (i.e., as X decreases), the ground is predominantly cohesive; moving towards the right side (i.e.,
as X increases), the ground is predominantly frictional. It can be shown than in the limit, when X
tends to infinity, all different curves become asymptotic towards the scaled factor of safety predicted
by the analytical solution of an infinite slope in purely frictional ground, i.e., FS/ tan φ = 1/ tan α (see
for example, Abramson et al. 2002; Coduto et al. 2011; Das & Sobhan 2018). Figure 2 shows the
positive influence of the cohesion of the ground on the factor of safety: for a fixed value of friction
angle, the factor of safety increases significantly with the increase in cohesion. Figure 2 also shows
the effect of the slope face angle on the factor of safety: for a fixed value of factor X , the factor of
safety decreases with increase of the slope angle.

The results presented in Figure 2 were compared with the results summarized in the dimensionless
chart for dry ground in Hoek & Bray (1981) and good agreement was found. In this regard it has
to be noted that the dimensionless representations in Hoek & Bray (1981) allows proper readings of
scaled factor of safety only for the range of the factor X from 0.25 to 100 —i.e., the representation in
Figure 2, which considers a range of X from 0.01 to 100, extends further into the region where the

Slope Stability 2018, Sevilla, Spain 5

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

Figure 2. Dimensionless stability diagram for the estimation of the scaled factor of safety.

ground is predominantly cohesive.

A multiple regression analysis of the results represented in Figure 2 was carried out to obtain an
algebraic equation to predict values of scaled factor of safety (FS/ tan φ) as a function of the factor
X and the angle α. The proposed equation and details of the regression analysis are presented in
Appendix B.

For space reasons, no diagrams for the functions fxc , fyc , fxA and fyA in equations [4] through [7] are
provided in this paper. Nevertheless, these diagrams can be easily constructed using the information
provided in the EXCEL workbook, discussed later in Section 4.

To illustrate the application of the equations [3] through [7], the following case of slope geometry and
ground properties is considered:
– Height of slope, H = 300 m
– Angle of slope, α = 52◦
– Unit weight of the ground, γ = 25 kN/m3
– Internal friction angle of the ground, φ = 37◦
– Cohesion of the ground, c = 667 kPa
For the given properties, the dimensionless variable X = γ H tan φ /c results to be ∼ 8.473.
Using the values of functions fFS , fxc , fyc , fxA and fyA in equations [3] through [7], as defined by the
SLIDE models (and as computed by the interpolation scheme implemented in the EXCEL workbook
discussed later on in Section 4), the following results are obtained:

Slope Stability 2018, Sevilla, Spain 6

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

– Scaled factor of safety, FS / tan φ ≈ 2.07 (FS ≈ 1.56 for φ = 37◦ ); see Point E
in Figure 2
– Scaled abscissa of the center of the critical failure surface, xc /H ≈ – 0.42
(xc ≈ –127 for H = 300 m)
– Scaled ordinate of the center of the critical failure surface, yc /H ≈ 1.45
(yc ≈ 436 m for H = 300 m)
– Scaled abscissa of the starting point of the critical circular failure surface, xA /H = 0
(note that the ordinate of the starting point of the critical failure surface is always zero)
– Scaled radius of the critical circular failure surface, R/H ≈ 1.51
(R ≈ 454 m for H = 300 m)
The scaled factor of safety can also be estimated using the proposed equation presented in Appendix
B (see equations [B-1] and [B-3]). This gives FS / tan φ = 2.02 (FS = 1.52 for φ = 37◦ ), which implies
there is a negative error of ∼ 2%, if the solution obtained with the dimensionless representations
(which is derived from SLIDE models) is assumed to be the correct one —see Appendix B for more
details.

Another important aspect of the functions of the dimensionless variables in equations [3] through
[7] is that they allow to reveal the nature of mechanical similarity of stability conditions for slopes,
particularly the similarity of scaled factors of safety and location of critical circular failure surface.
To illustrate the concept, Figure 3 includes a diagram similar to the one represented in Figure 2 but
considering only three curves corresponding to slope angles equal to 30◦ , 50◦ and 70◦ . The six slope
sketches included in the diagram show the corresponding failure surfaces defined from equations [4]
through [8] (with the EXCEL workbook discussed in Section 4) for the three slope inclination angles
and for arbitrarily selected cases of the factor X equal to 0.05, 1 and 25. Figure 3 confirms the known
fact that the critical circular failure surface is deeper for predominantly cohesive ground and shallower
for predominantly frictional ground. The diagram defines the scaled factor of safety and shape of the
critical circular failure surface for slopes characterized by particular values of X and α.
The concept of similarity is further illustrated by the following analysis. Table 1 lists the cases of
five slopes with the same inclination angle but otherwise significantly different values of slope height,
ground unit weight, friction angle and cohesion. The Case 4 in the table corresponds to the same
example described above for which the height of the slope is 300 m (note that Cases 1 and 5 in the
table, which correspond to slopes of unrealistic heights, are included for illustration purposes only).
The five different cases listed in Table 1 are characterized by the same values of similarity constants
X and α, and therefore by the same values of the functions ‘f ’ in equations [3] through [7]. Because
of this, and as shown in Figure 4, all five cases will have the same values of scaled factor of safety
and scaled coordinates that define the position of the critical circular failure surface. Therefore, the
cases can be said to be mechanically similar with regard to stability conditions.

Table 1. Example of slopes displaying mechanical similarity of factor of safety and critical failure surface location.

Case α [◦ ] H [m] γ [kN/m3 ] φ [◦ ] c [kPa] X = γH tan φ /c [-]

1 0.3 25 45 8.852×10–1
2 3 19 15 1.803
3 52 30 24 35 5.95×101 8.473
4 300 25 37 6.67×102
5 3,000 27 8 1.344×103

Slope Stability 2018, Sevilla, Spain 7

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

Figure 3. Dimensionless stability diagram showing similarity of scaled factor of safety and critical failure surface location.

1000
Cases 1 through 5 in Table 1, as represented by point E in this
figure and in Figure 2, all have the same values of dimensionless
functions in equations [2] through [7]; therefore they all have the
same scaled factor of safety and the same critical failure surface.
The cases can be said to be mechanically similar.

100

1.45

10
1.51
E
~2.07

1 52

0.1
0.01 0.1 1 ~8.47 10 100

Predominantly cohesive Predominantly frictional

ground ground

Figure 4. Dimensionless stability diagram showing similarity of the factor of safety and critical failure surface location for
the slope cases in Table 1.

Slope Stability 2018, Sevilla, Spain 8

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

4 Computer implementation of the dimensionless stability representations

Although dimensionless graphical representations or dimensionless charts as discussed in the pre-

vious sections have been the traditional means of summarizing and sharing results of procedures
for obtaining stability results by various authors, in modern geotechnical design practice, engineers
typically prefer the use of computer spreadsheets over the use of these graphical dimensionless
representations (Hoek 2000).
With this idea in mind, another computational tool provided in this paper (supplementary to the di-
agrams introduced in Section 3 and the equations provided in Appendix B) is an EXCEL (Microsoft
2016) workbook called ‘Slope Stability Calculator for Mohr-Coulomb Dry Ground’.
This EXCEL workbook contains the tabulated values of the functions fFS , fxc , fyc , fxA and fyA in
equations [3] through [7], as obtained with the software SLIDE (see Appendix A for details), and
implements a polynomial interpolation scheme that allows the user to obtain estimates of factor of
safety and location of the circular critical failure surface for any slope, provided the resulting factors
X and angles α fall within the ranges discussed in the previous section.
The main worksheet in this EXCEL workbook is shown in Figure 5. The user defines the geometry of
the slope and the properties of the ground in the section entitled ‘Input Data’. Results are displayed
in the section ‘Results’. If specified in the input data in the main worksheet, another worksheet
presents a basic plot of the properly scaled slope, showing the position of the critical circular failure
surface and a list of input data and results. Figure 6 shows a view of the worksheet with the graphical
representation of the problem, including the summary of input data and results.
Note that the particular input values and results displayed in Figures 5 and 6 correspond to the very
same slope case discussed in Section 3.
The EXCEL file corresponding to the workbook presented in this section can be freely downloaded
from the first author’s web site at www.d.umn.edu/∼carranza/SLOPE18.
Note that although computation of results is implemented through formulas without the use of macros,
the EXCEL workbook uses a VBA (Visual Basic for Applications) macro for constructing the scaled
plot of the slope problem (if asked by EXCEL, the user should allow the software to activate the
macro in order for the plot to be generated).

5 Practical application example

Figure 7 shows a view of a slope in a sector of an open pit mine in Chile that was analyzed using the
proposed method. The slope has 10 double-benches with a bench height of 30 m and a bench width
of approximately 13 m. The bench face angle is 68◦ . This results in a total slope height, H = 300 m,
and an overall slope angle (measured from toe to crest), α = 52◦ . For the open pit slope shown in
Figure 7, at the scale of the full height of the slope, the joint spacing is small enough relative to the
slope height that the assumption of ground continuity and isotropy can be considered valid (Read &
Stacey 2009; Wyllie 2018).
The rock mass is characterized by the following Hoek-Brown properties:
– Geological Strength Index, GSI = 50
– Disturbance Factor, D = 0.7
– Unconfined Compressive Strength (intact rock), σci = 50 MPa
– Hoek-Brown constant, mi = 12

The rock mass is assumed to have an average unit weight, γ = 25 kN/m3 .

Using the procedure described in Hoek et al. (2002), considering an interval of minor principal
stresses in the range 0 to 2.5 MPa, the Mohr-Coulomb internal friction angle and cohesion were
estimated to be φ = 37◦ and c = 667 kPa, respectively.

Slope Stability 2018, Sevilla, Spain 9

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

Figure 5. View of the main worksheet in the EXCEL workbook for implementation of stability computations.

Figure 6. View of the worksheet showing the graphical representation of the slope problem in the EXCEL workbook for
implementation of stability computations.

Slope Stability 2018, Sevilla, Spain 10

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

As it can be seen, this slope problem corresponds to the very same slope case discussed already in
Section 3 (see point E in Figure 2; see also Case 4 in Table 1, and Figure 4) and in Section 4 (see
Figures 5 and 6).
The factor of safety for the slope and the critical circular failure surface were already estimated to be
as follows (see, for example, Figures 5 and 6):
– Factor of safety, FS = 1.56
– Abscissa of the center of the critical circular failure surface, xc = –127.40 m
– Coordinate of the center of the critical circular failure surface, yc = 435.50 m
– Radius of the critical circular failure surface, R = 453.76 m

To confirm the validity of these results, a section of the slope shown in Figure 7 was solved with the
software SLIDE (Rocscience Inc. 2015), in one instance considering the slope face with the actual
benches, and in the other instance considering a planar face linking the toe and the crest of the
actual slope. Figure 8 shows a view of the results obtained with the software SLIDE for the first case
in which benches are considered. The figure includes the different circular failure surfaces analyzed
by the software, using the same procedure for locating the critical circular failure surface described
in Appendix A. Figure 8 also shows the critical circular failure surface corresponding to the minimum
factor of safety for the slope —this is indicated with a thick continuum line in Figure 8. The factor of
safety obtained with the software is 1.572 ∼ 1.57 which is approximately equal to the one obtained
with the proposed method (i.e., 1.56). Figure 8 also includes the location of the critical failure surface
obtained with the proposed method (as discussed in the example in Section 3) and with SLIDE, when
the same slope case was solved with SLIDE but with a planar slope face —this is indicated with a
dashed line in Figure 8. The location of the critical failure surfaces in both instances is practically the
same.

Figure 7. View of a benched slope in an open pit mine used to illustrate the application of the proposed stability analysis
tools.

Slope Stability 2018, Sevilla, Spain 11

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

Figure 8. View of the SLIDE (Rocscience Inc. 2015) analysis of a section of the benched slope in Figure 7.

6 Discussion

The tools presented in this paper, including graphical representations, equations and computer EX-
CEL workbook, can be used to make a quick determination of factor of safety and critical circular
failure surface location for simple slopes excavated in Mohr-Coulomb dry ground. The application of
these tools could be particularly useful in the stage of pre-design of projects involving slopes, when
different slope angles or heights, or different ground properties need to be evaluated.
It needs to be emphasized that the adoption of the limit equilibrium method and the Bishop formu-
lation, in particular, used to develop the tools presented in this paper was due to efficiency reasons
only —i.e., to be able to generate thousands of models and obtain results in the fastest way pos-
sible. Similar representations could have been developed with the limit analysis or shear strength
reduction methods discussed in Section 1 —although in the latter case, an impractical amount of
time would have been required to process all models needed to develop these tools. It is important
to remark that for the simple slope cases addressed in this paper (i.e., slopes with planar face in
homogeneous/isotropic frictional-cohesive material) no significant differences in values of factor of
safety or location of the critical failure surface could be expected when applying limit equilibrium,
limit analysis or shear strength reduction methods (see, for example, Baker 1980; Yu et al. 1998;
Dawnson et al. 1999; Cheng et al. 2007; Leshchinsky 2013). In any case, due to the non-existence
of a rigorous closed-form solution for the problem of computing the factor of safety and location of the
critical failure surface, all of the methods available to carry out stability analyses of the slopes men-
tioned above (i.e., limit equilibrium, limit analysis and strength reduction technique) will be expected
to give an approximate solution to the problem only.
Because the results obtained with the proposed tools are approximate in nature, caution must be
exercised when applying these tools. In the context of using the limit equilibrium method as done
in this paper, the approximate nature of the results can be further explained by the fact that slightly
different values of factor of safety and/or slightly different failure surface locations could result from

Slope Stability 2018, Sevilla, Spain 12

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

application of limit equilibrium (SLIDE) models if they are re-evaluated with other methods of factor
of safety evaluation (e.g., different from the Bishop method), or different critical search method (e.g.,
other than the ‘Autorefine Search’ method) or with some other different features (e.g., number of
slices different from 50, no implicit inclusion of a tensile crack at the crest of the slope and others).
In every case, when the tools are applied in the pre-design stage, it is recommended to always carry
out further validation of the stability conditions with other methods, including other stability analysis
methods and empirical methods.
The analysis presented in this paper could be extended further to account for other ground conditions.
One of these conditions is the case in which the ground is purely cohesive (the internal friction
angle is zero). For this case, due to the occurrence of the internal friction angle of the ground in
the denominator of equation [3], computation of factors of safety and location of the critical circular
failure surface will not be possible (it would still be possible to obtain results if a relatively small value
of internal friction angle is used in the computational tools presented in this paper, although the
results will be prone to numerical errors). Particular dimensionless representations that account for
purely cohesive ground conditions could be developed.
Another condition of interest and of practical importance is that of water presence in the ground. Hoek
& Bray (1981) presented dimensionless representations for five different cases of phreatic surface
behind the slope face. The analysis presented in this paper could be extended to those cases as
well, and similar tools as introduced here could be developed to account for the presence of water in
the ground.

7 Acknowledgements

The authors would like to thank Dr. Evert Hoek for the critical review and recommendations for im-
provement provided during development of the first draft version of this paper. The authors would
also like to express their gratitude to Mr. Felipe Sanz Lizt at SRK Consulting Chile, for assistance
with the development of the EXCEL worksheet presented in Section 4 of this paper.

References
Abramson, L. W., Lee, T. S., Sharma, S., M., G. & Boyce 2002. Slope stability and stabilization methods (2nd ed.). John
Wiley & Sons. New York.
Baker, R. 1980. Determination of the critical slip surface in slope stability computations. International Journal for Numerical
and Analytical Methods in Geomechanics 4(4), 333–359.
Baker, R. 2003. A second look at Taylor’s stability chart. Journal of Geotechnical and Geoenvironmental Engineer-
ing 129(12), 1102–1108.
Bell, J. M. 1966. Dimensionless parameters for homogeneous earth slopes. Journal of Soil Mechanics & Foundations Div.
ASCE SM 5(GT9), 51–65.
Bishop, A. W. 1955. The use of the slip circle in the stability analysis of slopes. Géotechnique 5(1), 7–17.
Bishop, A. W. & Morgenstern, N. R. 1960. Stability coefficients for earth slopes. Géotechnique 10(1), 129–150.
Chapra, S. C. & Canale, R. P. 2015. Numerical methods for engineers (7th ed.). Mc Graw Hill. New York.
Cheng, Y. M., Lansivaara, T. & Wei, W. B. 2007. Two-dimensional slope stability analysis by limit equilibrium and strength
reduction methods. Computers and Geotechnics 34(3), 137–150.
Coduto, D. P., Yeung, M. C. & Kitch, W. A. 2011. Geotechnical engineering. Principles and practices (2nd ed.). Pearson.
Cousins, B. F. 1978. Stability charts for simple earth slopes. Journal of the Geotechnical Engineering Division 104(2),
267–279.
Das, B. M. & Sobhan, K. 2018. Principles of Geotechnical Engineering (9th ed.). Cengage.
Dawnson, E. M., Roth, W. H. & Drescher, A. 1999. Slope stability analysis by strength reduction. Geotechnique 49(6),
835–840.
Duncan, J. M. 1996. State of the art: limit equilibrium and finite-element analysis of slopes. Journal of Geotechnical
Engineering 122(7), 577–596.

Slope Stability 2018, Sevilla, Spain 13

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

Duncan, J. M., Wright, S. G. & Brandon, T. L. 2014. Soil Strength and Slope Stability. John Wiley & Sons.
Fine Inc. 2016. GEO5. Slope stability analysis software based on the limit equilibrium method. Praha, Czech Republic.
Geo-Slope Inc. 2012. SLOPE/W Version 2012. Slope stability analysis software based on the limit equilibrium method.
Calgary, Canada.
Griffiths, D. V. & Lane, P. A. 1999. Slope stability analysis by finite elements. Geotechnique 49(3), 387–403.
Haines, A., Terbrugge, P. J., Carrieri, P. J. & Others 1991. Preliminary estimation of rock slope stability using rock mass
classification systems. In 7th ISRM Congress. International Society for Rock Mechanics (ISRM).
Hammah, R. E., Yacoub, T. E. & Curran, J. H. 2007. Serviceability-based slope factor of safety using the shear strength
reduction (SSR) method. In L. R. e Sousa, C. Olalla, & N. Grossmann (Eds.), Proceedings of the 11th Congress of the
International Society for Rock Mechanics, Lisbon.
Hoek, E. 2000. Personal communication.
Hoek, E. & Bray, J. 1981. Rock slope engineering (3rd ed.). CRC Press. London.
Hoek, E. & Brown, E. T. 1980. Empirical strength criterion for rock masses. J. Geotech. Eng. Div. ASCE 106(GT9), 1013–
1035.
Hoek, E. & Brown, E. T. 1997. Practical estimates of rock mass strength. International Journal of Rock Mechanics and
Mining Sciences 34(8), 1165–1186.
Hoek, E., Carranza-Torres, C. & Corkum, B. 2002. Hoek-Brown failure criterion – 2002 edition. In Proceedings of NARMS-
TAC 2002, Mining Innovation and Technology. Toronto – 10 July 2002, pp. 267–273. University of Toronto. (Available for
downloading at Hoek’s Corner, www.rocscience.com).
Huang, Y. H. 2014. Slope stability analysis by the limit equilibrium method. Fundamentals and methods. American Society
of Civil Engineers (ASCE) Press. Reston, Virginia.
Interactive Software Designs Inc. 2007. XSTABL. Slope stability analysis software based on the limit equilibrium method.
Moscow, Idaho.
Janbu, N. 1954a. Application of composite slip surfaces for stability analysis. In European Conference on Stability of Earth
Slopes. Stockholm, Sweden, Volume 3, pp. 43–49.
Janbu, N. 1954b. Stability analysis of slopes with dimensionless parameters. Thesis for the Doctor of Science in the Field
of Civil Engineering. Harvard University, Soil Mechanics Series, No. 46.
Juang, C. H., Jhi, Y. Y. & Lee, D. H. 1998. Stability analysis of existing slopes considering uncertainty. Engineering Geol-
ogy 49(2), 111–122.
Kotze, G. & Bosman, J. 2015. Towards expediting large-scale slope design using a re-worked design chart as derived from
limit equilibrium methods. In In Slope Stability 2015: International Symposium on slope stability in open pit mining and
civil engineering., pp. 341–362. Cape Town Convention Centre, Cape Town, 12–14 October 2015. Johannesburg: The
Southern African Institute of Mining and Metallurgy.
Leshchinsky, B. 2013. Comparison of limit equilibrium and limit analysis for complex slopes. In Geo-Congress 2013: Sta-
bility and Performance of Slopes and Embankments III, pp. 1280–1289.
Matsui, T. & San, K. C. 1992. Finite element slope stability analysis by shear strength reduction technique. Soils and
foundations 32(1), 59–70.
Michalowski, R. L. 2002. Stability charts for uniform slopes. Journal of Geotechnical and Geoenvironmental Engineering.
ASCE 128-4(GT9), 351–355.
Microsoft 2016. Excel Software. Version 2016. Microsoft. Redmond, Washington.
Morgenstern, N. R. & Price, V. E. 1965. The analysis of the stability of general slip surfaces. Géotechnique 15(1), 77–93.
O’Connor, M. J. & Mitchell, R. J. 1977. An extension of the Bishop and Morgenstern slope stability charts. Canadian
Geotechnical Journal 14(1), 144–151.
Potts, D. M. & Zdravkovic, L. 1999. Finite element analysis in geotechnical engineering. Theory. Thomas Telford. London:
Thomas Telford.
Read, J. & Stacey, P. 2009. Guidelines for open pit slope design. CSIRO Publishing. Melbourne, Australia.
Rocscience Inc. 2015. SLIDE Version 7. Slope stability analysis software based on the limit equilibrium method. Toronto,
Canada.
Sjoberg, J. 2000. Failure mechanisms for high slopes in hard rock. In W. A. Hustrulid, M. K. McCarter, & D. J. A. V. Zyl
(Eds.), Slope Stability in Surface Mining, pp. 71–80. Society for Mining, Metallurgical and Exploration (SME). Littleton,
Colorado.
Spencer, E. 1967. A method of analysis of the stability of embankments assuming parallel inter-slice forces.
Géotechnique 17, 11–26.
Steffen, O. K. H., Contreras, L. F., Terbrugge, P. J. & Others 2008. A risk evaluation approach for pit slope design. In The
42nd US Rock Mechanics Symposium (USRMS). American Rock Mechanics Association (ARMA).

Slope Stability 2018, Sevilla, Spain 14

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

Steward, T., Sivakugan, N., Shukla, S. K. & Das, B. M. 2010. Taylor’s slope stability charts revisited. International Journal
of Geomechanics 11(4), 348–352.
Taylor, D. W. 1948. Fundamentals of Soil Mechanics. Wiley. New York.
Vanmarcke, E. H. 1980. Probabilistic stability analysis of earth slopes. Engineering Geology 16(1-2), 29–50.
Verruijt, A. 2012. Soil Mechanics. Delft University of Technology, Delft, The Netherlands (Available for free dowloading at
http://geo.verruijt.net/).
Wyllie, D. C. 2018. Rock slope engineering. Civil Applications. CRC Press. Taylor & Francis.
Yu, H. S., Salgado, R., Sloan, S. W. & Kim, J. M. 1998. Limit analysis versus limit equilibrium for slope stability. Journal of
Geotechnical and Geoenvironmental Engineering 124(1), 1–11.

Appendix A. Characteristics of the limit equilibrium models used to define the

dimensionless graphical representations

This appendix provides details on the 3,402 slope cases solved with the software SLIDE (Rocscience
Inc. 2015), used to produce the dimensionless graphical representations and the EXCEL workbook
presented in the main text.
For most practical problems, the range of values of the variable X (see equation [2]) was determined
to be in the interval 10–2 to 100. This range was determined by application of a Monte-Carlo simu-
lation that evaluated the variable X for thousands of non-correlated random values generated from a
uniform distribution for the following ranges of slope height and ground properties:
– Slope height, H = 1 m to 500 m
– Unit weight of the ground, γ = 15 kN/m3 to 27 kN/m3
– Internal friction angle of the ground, φ = 5◦ to 75◦
– Cohesion of the ground, c = 100 kPa to 500 kPa

Seven different cases of slope inclination angle ranging from 20◦ to 80◦ , in increments of 10◦ , were
considered. Vertical slopes were not considered, as they led to too thin critical failure surfaces,
particularly, for predominantly frictional ground.
For each slope face angle, a total number of 81 slope cases for values of X equally spaced (when
plotted in logarithm base-10 scale) in the range 10–2 to 100 were first evaluated. For each of these
cases, the slope height and ground properties were obtained randomly, from the range of values
listed above. The computation of factor of safety and location of the critical failure surface was
repeated three times (for three different combinations of random input values) to check consistency
in the results, bringing the total number of cases analyzed with SLIDE to 1,701 (i.e., 7×81×3). The
search process of the critical failure surface in SLIDE was implemented in two stages (i.e., each
model was solved twice), as explained later on in this appendix, thus bringing the grand total number
of cases analyzed with SLIDE to 3,402.
Each SLIDE model was constructed using the following geometrical characteristics: denoting the
length of the slope face as L, the left and right boundaries of the slope model were both located at a
horizontal distance of 5L from the toe and crest of the slope, respectively; the lower boundary of the
model was located at a vertical distance of 2.5L from the toe of the slope.
The factor of safety was computed using the Bishop method implemented in SLIDE, assuming 50
slices above the failure surface, and a factor of safety tolerance of 0.005 (with a maximum of 75
iterations) in the iterative implementation of the method.
The search process of the critical circular failure surface was done using the ‘Autorefine Search’
option implemented in the software SLIDE (the documentation of the software states that this method
is the most efficient search method available in the software), using default input parameters (i.e.,
divisions along slope, circles per division, and number of iterations, all equal to 10; divisions in next
iteration equal 50%). As mentioned above, the search process was performed in two stages. In

Slope Stability 2018, Sevilla, Spain 15

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

the first stage, the ‘Autorefine Search’ option was applied to the full upper surface of the model. In
the second stage, the search was limited to intervals for the starting and ending points of the failure
surface (see points A and B, respectively in Figure 1), centered around the starting and ending
points for the critical failure surface obtained in the first stage; in this more localized search, the
search intervals for starting and ending points of the failure surface both had a length of 50% the
height of the slope (for example, Figure 8 shows the second stage of searching of the critical failure
surface; note that the length of the search segments defined by the ‘triangular’ marks near the toe
and crest of the slope is 50% the height of the slope).
The results from all models solved with SLIDE were finally scaled according to equations [3] through
[8], and incorporated into the EXCEL workbook discussed in Section 4.

Appendix B. Proposed equation to compute factor of safety of slopes in Mohr-Coulomb

dry ground

A multiple regression analysis of the results represented in Figure 2 was carried out to obtain an
algebraic equation to predict values of scaled factor of safety, FS/ tan φ, as a function of the factor X
and the angle α. With this purpose the method of minimization of the sum of square of the estimate
residuals was used (Chapra & Canale 2015). The fitting equation is an inverse power equation of the
following form:
FS 1 g (α) g2 (α)
= + 1 + g (α) [B-1]
tan φ tan α X X 3
where the functions g1 (α), g2 (α) and g3 (α) are cubic polynomials that show a discontinuity of the
first and higher derivatives at α = 50◦ , the approximate value of slope angle above which the critical
circular failure surface will always have a starting point that coincides with the toe of the slope —see
point A in Figure 1.
These functions are defined as follows:
If α ≤ 50◦

g1 (α) = 5.524 – 1.222 × 10–2 (α – 50) – 1.569 × 10–3 (α – 50)2 – 3.097 × 10–5 (α – 50)3 [B-2]
–2 –3 2 –5 3
g2 (α) = 1.345 – 4.633 × 10 (α – 50) + 1.037 × 10 (α – 50) + 1.152 × 10 (α – 50)
–1
g3 (α) = 3.746 × 10 – 8.727 × 10–3 (α – 50) – 6.384 × 10–5 (α – 50)2 + 1.808 × 10–6 (α – 50)3

If α ≥ 50◦

g1 (α) = 5.524 – 4.383 × 10–2 (α – 50) + 1.985 × 10–4 (α – 50)2 – 5.952 × 10–6 (α – 50)3 [B-3]
–2 –4 2 –5 3
g2 (α) = 1.345 – 1.035 × 10 (α – 50) – 4.092 × 10 (α – 50) + 1.044 × 10 (α – 50)
g3 (α) = 3.746 × 10–1 – 7.652 × 10–4 (α – 50) – 2.134 × 10–4 (α – 50)2 + 4.164 × 10–6 (α – 50)3

It has to be emphasized that the regression analysis was done considering that the slope inclination
angle α in the equations above is expressed in degrees and not in radians (e.g., when using these
equations to compute the scaled factor of safety for a slope with inclination angle 45◦ , the equations
must consider the variable α to be 45 and not π/4).
With regard to equation [B-1], the reason for choosing the particular form of an inverse power equa-
tion is to be able to recover the solution of an infinite slope when the ground is purely frictional —i.e.,
FS/ tan φ = 1/ tan α when X → ∞ (see for example, Abramson et al. 2002; Coduto et al. 2011; Das
& Sobhan 2018).
As with any regression analysis, application of equations [B-1] through [B-3] to compute the scaled
factor of safety can be expected to show some error with respect to the original results obtained with
the limit equilibrium models; this error is an intrinsic consequence of the fitting process.

Slope Stability 2018, Sevilla, Spain 16

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

The reminder of this appendix presents a discussion on the expected error of the equations [B-1]
through [B-3] and shows that these equations can either overestimate or underestimate the scaled
factor of safety within a margin of approximately ±5%.
Figure B-1 shows a similar representation as in Figure 2, but obtained with the proposed equations.
At first sight, the diagrams in Figure B-1 and Figure 2 are identical. Nevertheless, because equations
[B-1] through [B-3] are best fit equations obtained from polynomial regression analysis, there will be
some error associated with the estimation of scaled factor of safety.
To quantify this error, the dimensionless representation of scaled factor of safety of SLIDE models
given by equation [3] (with the function fFS as shown in Figure 2) can be considered to be the true
solution of the problem. In such case, the absolute error can be computed as follows

FS FS/ tan φ (with Eq. [3]) – FS/ tan φ (with Eq. [B-1])
Error [%] = [B-4]
tan φ FS/ tan φ (with Eq. [3])

Figure B-2 is the graphical representation of equation [B-4]. The figure suggests that the error asso-
ciated with application of equations [B-1] through [B-3] can be positive or negative (i.e., the scaled
factors of safety can be either overestimated or underestimated, respectively), and that for all cases
of the stability factor X considered in this study, the absolute value of the error is not larger than ap-
proximately 5%. The authors recommend caution when applying the equations [B-1] through [B-3],
because there is a certain degree of error associated with the (function) fitting process; also, and as
discussed in Section 6, the dimensionless limit equilibrium results from which these equations have
been derived, are understood to be approximate in nature.

Figure B-1. Dimensionless stability diagram for the estimation of the scaled factor of safety obtained with equations [B-1]
through [B-3].

Slope Stability 2018, Sevilla, Spain 17

Pre-Print Version of the Manuscript C. Carranza-Torres & E. Hormazabal

20

10

-5

-10

-20
0.01 0.1 1 10 100

Predominantly cohesive Predominantly frictional

ground ground

Figure B-2. Graphical representation of scaled factor of safety error computed with equation [B-4].

Slope Stability 2018, Sevilla, Spain 18