You are on page 1of 11

International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

Incorporation of non-associative ow rules into rock slope stability analysis MARK


a, b b b
S. Melentijevic , A. Serrano , C. Olalla , R.A. Galindo
a
Complutense University of Madrid, Faculty of Geology, C/ Jos Antonio Novis n 2, Madrid 28040, Spain
b
Technical University of Madrid, Spain, ETSI Caminos, C. yP., C/ Profesor Aranguren s/n, Madrid 28040, Spain

A R T I C L E I N F O A BS T RAC T

Keywords: The circular shear failure mechanism in rock slopes occurs in large-scale, heavily jointed or weak rock masses.
Rock slope stability Most rock slope stability analysis problems do not consider the eect of ow rules, i.e., the dilatancy angle. The
Heavily jointed rock mass failure criteria commonly applied are the linear Mohr-Coulomb or the non-linear Hoek and Brown. In general,
Weak rocks all slope stability methods originate from the assumption of the associative ow rule, calculating equivalent
Hoek and Brown failure criterion
values of the cohesive strength and friction angle. In this study, the non-linear Hoek and Brown failure criteria
Flow rules
are applied in the parametric form, which permits the incorporation both the associative and non-associative
Dilatancy
ow rules. A theoretical approach of the limit equilibrium method is performed for circular slip surfaces under
the assumption that there are only self-weight forces acting on the failure mass of the slope, taking the critical
plane failure mode as the foundation. Both the analysis and results are performed in dimensionless form. The
study emphasizes the importance of the value of dilatancy angle and non-conservative character of the
hypothesis of the associative ow law, i.e., the overestimation of the value of the factor of safety. Specic
examples are analysed and compared to the nite dierence method and a real case is presented.

1. Introduction theoretical analysis is originally planned in as general a way as possible,


which means for any shape of the slip surface through the rock mass,
In the assessment of rock slopes in weak and/or heavily jointed there could be any type of failure criterion and ow law, e.g., linear
rock masses subjected to circular shear failure, the factor of safety (FS) variation of the 0 .
is used to quantify how close or far the slopes are from failure. The calculation procedure is performed based on the LEM. At the
Traditional limit equilibrium methods (LEM) are still frequently used. beginning, it involves the same basic assumptions as the Morgenstern
Chart solutions have been presented by many researchers112 and and Price method of slices3 to pass later to dierential equations in the
they are still widely used as design tools in geotechnical engineering for development of mathematical expressions that take the widths of the
preliminary analysis. Rock slope stability charts, such as Hoek and slices to zero (x dx ).
Brays,4 often need to use the equivalent Mohr-Coulomb shear The procedure is applied to the circular slip surfaces with the plane
strength parameters. This conversion produces a signicant dierence failure mode as the foundation and the original Hoek and Brown failure
between the FS of a slope with Hoek and Brown and equivalent Mohr- criterion14 given in the parametric form by Serrano and Olalla.15 This
Coulomb envelopes.6 parametric form relates normal and shear stresses along the failure
Carranza-Torres5 proposed a solution for estimating the shear surface under the hypothesis of the associative and non-associative
strength of rock masses from the Hoek and Brown criterion, which ow laws. The further development of the original Hoek and Brown
was incorporated in the Bishop simplied method for the analysis of failure criterion(1618) due to renements based on experience gained
the rock slope stability. Stability charts for directly estimating rock during its application is not considered because the basic purpose of
mass slopes from the Hoek and Brown parameters were originally this study is to investigate the variation of the FS on ow rules, i.e., the
proposed by Li et al. 6,7 using limit analysis. dilatancy angle.
However, the inuence of the dilatancy on rock masses in engineer- The calculation procedure presented in this study contains math-
ing practice is usually not considered when the global stability of the ematical development of equations for a case of non-associative ow
rock slopes is concerned. The use of associative ow rules over- rule while considering constant values for the dilatancy angle that
estimates the FS. The purpose of this study is to evaluate the variation depend on the rock mass quality.17 A comparison and dierences
in the FS on the dilatancy angle (0 ) for the 2D rock slope stability. The between the results obtained under dierent assumptions for the


Corresponding author.
E-mail address: svmelent@ucm.es (S. Melentijevic).

http://dx.doi.org/10.1016/j.ijrmms.2017.04.010
Received 13 May 2016; Received in revised form 22 February 2017; Accepted 21 April 2017
1365-1609/ 2017 Elsevier Ltd. All rights reserved.
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

associative and the non-associative ow rules are presented and based Combining these two basic intrinsic parameters, the dimensional
on a previously developed study.19 intrinsic length parameter and dimensional intrinsic force parameter
The validity of this study is limited within the homogeneous and can be obtained by expressions (1) and (2). The lengths and forces are
isotropic rock masses for massive and extremely fractured rock masses. made dimensionless by dividing them by Le and Fe .
It can also be applied to homogeneous and isotropic hard soils, soft Le = / R (1)
rocks or poor quality rock masses after failure of the Hoek and Brown
criterion. This means that failure in these rock masses is not structu- Fe = Le2 = 3 / R2 (2)
rally from singular discontinuities within the rock masses.
Previous studies of the inuence of the phenomenon of dilatancy on When working with dimensionless variables, the mathematical
the rock slope stability have evaluated the following: a) the stability of development is simplied. Dierent physical variables with their
the plane slip surface applying the Hoek and Brown failure criterion in corresponding units can be evaluated at the end of the calculation.
the parametric form with the non-associative ow law and constant
values of the 0 , supposing a constant value of stresses along the failure 3. Failure criterion
surface20; b) the stability of the plane slip surface with variable stresses
along the slip surface by applying LEM (Spencer's method of slices), 3.1. Hoek and Brown failure criterion
taking the width of slices to the limit, and solving mathematically
equilibrium conditions with an explicit equation21; and c) the stability The failure criterion applied in this study is the Hoek and Brown
of innite rock slopes (skin instability parallel to the natural slope non-linear criterion in its original version.14 The criterion assumes the
angle) under associative and non-associative ow laws and various behaviour of the rock mass is homogeneous and isotropic and is
hypotheses of simplied seepage ow nets 22,23.. dened by the relationship between the principal stresses at failure:
The scale eect is a very important topic when considering the rock
1 3 3 0.5
mass strength and possible mechanisms of failure in rock slopes. The = m + s
behaviour of a blocky rock mass of a small-scale slope is vastly dierent c c (3)
from a blocky rock mass for large-scale slopes.24,25 The term circular in which 1 and 3 are the major and minor principal stresses at failure;
shear failure is usually assumed for failure in large-scale slopes with a c is the uniaxial compressive strength of the intact rock material; and
heavily and randomly jointed rock mass, i.e., without any clear m and s are constants that depend on the rock mass properties and
structural control and/or in weak rocks. The failure surface need not degree of fracture, where s = 1 for intact rock. Parameters m and s are
be circular, but this collective term is used to denote shear failures with given as a function of the rock type (m 0 ) and geomechanical index RMR
a curved failure surface that probably involve the slip along pre-existing (Rock Mass Rating) or the GSI (Geological Strength Index) 28:
discontinuities coupled with failures through intact rock bridges.26 The
case studies of both real circular shear failures observed in weak rocks
3.2. Relationship between failure criteria
and hard rock slopes are presented in Ref. 26,27.
The condition for a circular failure surface to occur in favour of a
The stresses exerted on the failure surface (Coulomb type failure
plane failure surface should be determined considering that in a highly
criterion) can be parametrically obtained from the Mohr type failure
fractured rock mass, a step path failure along pre-existing disconti-
criterion (envelope of stress circles), setting the value of the instanta-
nuities can follow nearly a straight line from the toe to the crest of the
neous friction angle () as a parameter. These stresses, referred to the
slope; also, the step path failure can follow along a curved failure line
point R, in the dimensionless form are (Fig. 1):
involving a large volume of a rock mass with respect to the length of the

failure surface. With the increase in the slope height, there is a higher * = q* ()cos
probability of a curved failure surface considering that the signicance (4)
of individual discontinuities decreases, while a plane failure surface is n
n* = p* () q* ()cos
more likely for small-scale slopes.27 (5)

2. Hypotheses A general ow rule for rock masses, which is necessary to dene a


failure criterion of the Coulomb type, can be of the linear type:
2.1. Basic hypothesis sin = sin + k (6)
Specic cases of this general ow law are a) the associative with
The following basic hypotheses are adopted: a) there is a 2D LEM;
= 1 and k = 0 that gives and b) the non-associative with a
b) the slip surfaces pass through the toe of the slope; c) there are no
constant value of the dilatancy angle ( 0 ) for = 0 and k = sin 0 .
external forces (seismic forces, rock bolts, anchors, etc.) acting on the
rock mass nor there are internal pore pressures and only the self-
weight of the slope failure mass is considered; d) the material is
coaxial; and e) the rock mass is a homogeneous, isotropic medium
whose behaviour permits the application of the Hoek and Brown failure
criterion.

2.2. Dimensionless calculation

To perform the dimensionless approach of the entire numerical


analysis, all physical variables that inuence the slope stability are
formulated in a dimensionless way. Two fundamental independent
parameters are used for that purpose, which are a) the strength
modulus( ), associated with the rock mass strength obtained from the
Hoek and Brown failure criterion (Eq. (9)) and b) the specic weight of
the rock (R ). Dividing stresses (1; 3; p ; q ; n; ) by the value of ,
dimensionless stress values (1*; 3*; p* ; q*; n*; *) are obtained. Fig. 1. Types of failure criteria (modied from ).

48
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

In this study, cases with constant values of 0 from recommenda- same along the slip surface h = yr (x ) yt (x ). With the purpose of
tions given in Ref. 17 are considered as follows: 0 = 0 for poor quality simplifying the mathematical analysis of equations, the coordinate
rock mass (GSI 30); 0 = 4 for average quality rock mass (GSI = 50) centre is set at the highest point of the slip surface.
and 0 = 11.5 for very good quality hard rock mass (GSI = 75).
However, other ow rules and values of 0 are possible. 4.1. Geometry

3.3. Parametric formulation The hypothesis regarding the critical circular failure surface that
provides the minimum FS of the slope is that it is located close to the
When approaching 2D rock slope stability analysis, the case is of the region of the critical plane failure surface with its minimum FS. That
plane strain. The Hoek and Brown failure criterion can be simplied means the centre of the circular slip surface (C) is situated along the
using Lambe's variables, which is given in dimensionless form in Ref. line that passes perpendicularly through the medium point (M) of the
15: plane failure surface (Fig. 2b).
In the analysis of the plane failure, each studied slip surface forms
q / = 2( p / + ) + 1 1 (7) an angle of i with the axis x , and the one corresponding to the
which becomes ( p* p / and q* q / ): minimum FS has its inclination MIN (Fig. 2b). The slope surface
consists of the following two parts: a) x < Lo in the upper natural part
q* + 1 = 2( p* + ) + 1 (8) of the slope with its horizontal length projection Lo and b) Lo < x < L in
the lower excavated part with its horizontal length projection (L Lo ).
Parameters and , which establish the rock strength, are based on
The height of the slope failure mass (h ) is variable along the slip
the Hoek and Brown's m and s parameters as follows:
surface, taking the maximum value of ho in the point where the slope
mc /8 (9) inclination varies.
Supposing the geometrical data with previously known H , MIN ,
8s / m2 (10)
and , the following geometrical data can be dened: yr (x ), yt (x ), h , L ,
The strength modulus is used to scale the failure criterion. The and Lo :
coecient of toughness represents the relative quality and strength of tan tan MIN
the rock mass, i.e., it inherits the type of rock mass and extent to which Lo = cot H
tan MIN tan (16)
it is fractured and disturbed. It represents the dimensionless isotropic
tensile strength of the rock mass and can be regarded as the tensile tan tan
L= cot H
strength coecient of the rock mass. tan MIN tan (17)
The instantaneous friction angle (), dened as the tangent line of
ho = (1 cot tan MIN )H (18)
the Mohr's envelope, can also be expressed as
sin = dq*/ dp* = 1/(1 + q*) The coordinates of the centre of the circular slip surface are
(11)
L H
Depending on the instantaneous friction angle (), Lambe's dimen- xC = + Dist sin MIN ;yC = T Dist cos MIN
2 2 (19)
sionless variables that follow the Hoek and Brown's criterion are as
follows16: Where Dist is dened between the centre of the circular slip surface and
medium point of the plane slip surface. It is determined by multiplying
p cot2
p* = the length of the plane slip surface (Dist = L2 + HT2 ) with a
2 (12) coecient representing the total height of the sliding mass
1 sin HT = H + L 0 tan .
q
q* = The radius of the circular slip surface is dened for the top point of
sin (13) the sliding mass:
In the case of the non-associative ow law ( ), dimensionless
Radius 2 = xC2 + yC2 = Dist 2 + 0.5 L2 + HT2 (20)
stresses veried on the failure surface, under Coulomb type of criterion,
are In function of the value of the radius, ordinates of the sliding mass
1 sin are established, where the positive sign in front of the square root is
* = cos considered to refer to the point under the centre of circle:
sin (14)
yr = yC (x xC )2 + Radius 2 (21)
n cot2 1 sin
n* = sin
2 sin (15) Geometrical data are dened for two dierent parts of the external
slope surface. For x < Lo :
The ow law given by Eq. (8) provides for the Coulomb type failure
criterion to be expressed only as a function of parameter . In the yt = x tan (22)
specic case of the associative ow law, in Eqs. (14) and (15)
should be introduced, while in the specic case of the non-associative h = yr yt = yC + (x xC )2 + Radius 2 x tan (23)
ow rule with a constant value of the dilatancy angle 0 . and for Lo < x < L :
yt = Lo(tan tan ) + x tan (24)
4. Theoretical analysis of the stability of rock slope
h = yr yt = yC + (x xC )2 + Radius 2 (Lo(tan tan ) + x tan ) (25)
The most general case in the slope stability analysis is the slip
surface of any shape (Fig. 2a). In the slope of the innite height, the The value of the angle of inclination ( ) of the slip surface is
nite excavation Hof the slope is supposed with any shape of external variable in each point of the sliding mass:
yt (x ) and slip yr (x ) surface. The horizontal projection of the slip surface dyr (x xC )
along the axis x is given by the length L . The inclination of the slip tan = =
dx (x xC )2 + Radius 2 (26)
surface is variable in each point of the slip surface and is dened by
angle = arctan(dyr / dx ). The height of the slope failure mass is not the The length of the arc of each slice of the sliding mass (si ) is dened

49
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

Fig. 2. Cross-section of the (a) general slip surface and (b) circular failure surface.

as M + n 0b0 n1(b1 x tan ) t1x n(s 2 /2) = 0 (31)


si = Radius = Radius(i 1 i ) (27) Considering = R / FS , s = x /cos , t = fi ni , n = ni ni1,
t = ti ti1, m = ni 1bi 1 nibi , and only the self-weight, and then
with as the angle of the dierence between the angles of inclination of
transforming the limit equilibrium to dierential equations by taking
the sliding mass in extreme points of the slice.
the width of slices to zero (x dx ), with n(s 2 /2) 0 , Eqs. (29)(31)
become:
4.2. Limit equilibrium analysis
dn
= n tan R
For the slope stability analysis by LEM, the slope failure mass is dx FS (32)
divided into N slices. The same hypotheses of the Morgenstern and dt
Price method13 are assumed. The limit equilibrium equations are = Rh n + R tan
dx FS (33)
transformed to dierential equations by taking the widths of the slices
to zero (x dx ). The forces acting on each slice are given in Fig. 3b. dm
= ni(fi tan )
Each slice is of width x and is inclined by the same angle to the dx (34)
horizontal such that the base of each slice is s = x /cos . The height
To make the calculation dimensionless, all equations are divided by
of each slice is hi , and the distance bi is the point where interslice forces
Rh , resulting in dimensionless results in the form of charts.
act, representing the distance between the failure line and internal
In accordance with the last Eq. (34) the exact value of the
stress line.
parameter of Morgenstern and Price ( ) can be found by considering
Forces acting on each slice are: a) normal (n ) and shear ( ) stresses
the whole sliding mass, i.e., from x = 0 to x = L .
exerted along the slip surface, acting in the midpoint of the base of each
slice; b) horizontal (ni ) and vertical (ti ) interslice forces acting on dm = (f tan )ndx (35)
vertical sides of each slice at the distance bi with respect to the base of x=L
the slice; and c) external forces acting on each slice (P, Q, M ).
In this study, only the self-weight case is analysed with
m(x = L ) = m(x =0) + (f tan )ndx
x =0 (36)
P = Rhx , Q = 0 and M = Px /2 = Rhx 2 /2 0 .
The existing relationship between shear strength of the rock mass Considering that the values of moments in extreme points of the
and shear stress in the rock mass is specied by the value of the safety sliding mass are equal to zero (m(x = L ) = 0) and (m(x =0) = 0), and
factor = R / FS for the particular slip surface. considering Eq. (26), Eq. (36) becomes
Interslice forces that act vertically between each slice presented in x=L x=L

Fig. 3c are given under the rule of tan i = ti / ni . The basic assumption of = ndy f ndx
the Morgenstern and Price method of slices is that the direction of the x =0 x =0 (37)
resultant of interslice forces is determined using a selected function:
tan i = fi (28) 4.4. Calculation procedure

where fi is the function of the distribution of the interslice forces, and


The angle of the planar failure surface MIN that provides the
is the Morgenstern and Price parameter that denotes the percentage of
minimum FS of the slope is known. This allows the centre of the
the function fi applied in the calculation. The usual mathematical forms
circular failure surface to be situated along the line that passes through
of fi are given in Ref. 29: constant, half-sine, clipped-sine, trapezoidal
its medium point M (Fig. 2b). The required steps of the calculation
and previously specied. In this study, the half-sine interslice force
procedure for each slice are:
function f (x ) = sin(x / L ) is used (see Section 4.5).
1st step: It is necessary to determine the value of the instantaneous
4.3. Mathematical formulation of equations friction angle () for each slice. The equations for the equilibrium of
horizontal Eq. (32) and vertical Eq. (33) forces are considered and
For each slice (Fig. 3a), three conditions must be veried: a) the transformed by the procedure described in Appendix A to obtain the
horizontal force, b) the vertical force and c) the moment equilibrium. value of from the algebraic expression (A17).
Q + n 0 n1 s cos + ns sin = 0 (29) 2nd step: Knowing the value of and = 0 , the stresses exerted
along the failure surface can be obtained (n, R ), which are ancillary
P + t0 t1 s sin ns cos = 0 (30) variables necessary to obtain the forces of the 3rd, 4th and 5thsteps.

50
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

Fig. 3. (a) The slice of the circular slip surface; (b) forces acting on each slice; and (c) forces acting on vertical sides of each slice.

3rd step: The value of the horizontal interslice force for each slice: presented in Fig. 4.
The process is repeated for the range of = 0.2 1.0 until the
ni = ni 1 + (n tan R / FS )dx (38)
minimum FS is obtained for = i .

4th step: The value of the vertical interslice force for each slice: 4.5. Validity of the proposed method
ti = nifi (39)
The theoretical method presented in this paper is compared with
the existing software package Geo-Slope,30 based on the LEM, with the
5th step: The value of the interslice moment for each slice: purpose of conrming the validity of the analytical method. The slice

mi = mi 1 + ni(tan fi )dx (40)

6th step: Steps 15 are repeated for each slice of the sliding mass
of the particular slope with dimensionless height (H *), slope angle
( ) and various values of Dist dened in Section 4.1 with
ranging from 0.7 to 5 to obtain FS for each slip surface
(FS1, ... ,FSN ). The minimum value is selected as FSMIN , which
represents the minimum FS of the slope for the critical slip
surface.

To obtain values of parameter FS that satisfy the equilibrium


conditions, values of the horizontal force (nF ) and moment (mF ) at the
toe of the sliding mass are obtained, and auxiliary function U = nF2 + mF2
is dened. The calculation procedure is repeated until the function U is
equal to or approximates zero. The systematic procedure is graphically Fig. 4. Procedure for achieving the minimum FS for the circular failure surface.

51
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

Table 1 Table 3
Geotechnical data introduced in the calculation. Comparison of the proposed analytical and numerical calculation.

Function N 1 2 3 4 5 6 7 8 9 Case N Function N FS (Proposed FS (FLAC 2D) FS (FLAC 2D)


(Table 1) Method) (Coarse mesh: (Finer mesh:
0.001 0.01 0.1 0.001 0.01 0.1 0.001 0.01 0.1 4200 cells) 8400 cells)
0 ()Z 0 0 0 4 4 4 11.5 11.5 11.5
GSI 30 30 30 50 50 50 75 75 75 A1 1 0.72 0.85 0.76
A2 2 0.73 0.94 0.78
A3 3 0.81 0.91 0.87
A4 4 0.75 0.80 0.76
method applied in Geo-Slope is the mathematically rigorous method of
A5 5 0.76 0.81 0.78
Morgenstern and Price13 with various interslice functions. The rock A6 6 0.83 0.94 0.88
properties are dened in the most general way to specify the failure A7 7 0.79 0.90 0.76
envelope, i.e., by the shear/normal strength function model, which is A8 8 0.80 0.91 0.78
used to dene a general data point function as a relationship between A9 9 0.87 0.96 0.87

the shear strength and normal stresses along the failure surface. With
that purpose, the failure criterion given by Eqs. (14) and (15) and
eld satises the lower bound theorem. Also, the deformation eld at
applying the non-associative ow rule with constant values of the
the failure state calculated by FLAC is a kinematically admissible
dilatancy angle ( 0 ) is used. Nine shear/normal function models
deformation (it fullls all the criteria of the upper bound theorem).
based on the geotechnical data given in Table 1 are introduced in Geo-
Thus, one may say that if the calculated FS tends to a limit as the grid
Slope. Geo-Slope computes an equivalent friction angle () and
size is reduced, this limit may be considered to be very close to the
cohesion (c ) intercept for each slice as a function of the normal stress
exact FS for the problem.
at the slice base.
Table 3 shows dierent cases of implemented calculation and FS
A few hypothetical slopes of the geometry given in Table 2 and they
values obtained by dierent methods. The FLAC analysis was per-
are studied under the assumption of a circular slip surface. Dierent
formed using two dierent grid sizes. For the ner mesh size the values
values of the angles of slip surfaces ( ) are considered. A total of 27
of FS are similar to the proposed method solution, while by the wider
cases are analysed with an automatic search.
mesh size the FS is greater being non-conservative.
Several interslice force functions f (x ) as given in Ref. 29 are studied
to obtain its inuence on the value of FS being within the limit of 1%.
5. Results
Therefore, the half-sine interslice force function f (x ) = sin(x / L ) is
chosen.
The results of this study for the circular critical failure surface are
Values of FS obtained for a circular failure surface, by the procedure
given in the form of dierent charts for dierent values of parameters
described in this study (FSM ) and by automatic search by Geo-Slope
that inuence the calculation. All charts are presented in the dimen-
(FSSlope ), are referred to dierent values of a) the Morgenstern and Price
sionless form, i.e., for the dimensionless height of the excavated slope
parameter, b) position of the centre of the circular failure surface and c)
H * = H /( / ).
the corresponding angle of the plane failure surface ( , Dist , MIN ). The
The calculation is performed for the values of the dilatancy angle
dierence between values of the FS evaluated by (FSM FSSlope )/ FSM is
0 = 0; 4; 11.5 recommended in 17 for rock masses of dierent
lower than 5%. This dierence is larger in cases with better quality rock
qualities with GSI = 30; 50; 75, respectively. The values of the coe-
masses, i.e., higher values of .
cient of toughness introduced in the calculation are
To perform a complete and accurate validation of the analytical
= 0.001; 0.01; 0.1. With those values of , from the practical point
method, the numerical modelling is performed by applying the nite
of view, all real possible values of the state of the rock mass are also
dierence method using the FLAC 2D.31 Cases corresponding to a
considered.
particular slope with the following geometrical parameters: H*=10,
For dierent slopes of singular geometry data, i.e., dimensionless
=51, and =0 have been implemented considering the nine shear/
height (H *) and various slope angles ( ), the value of FS is determined
normal function models based on the geotechnical data given in
for all combinations of parameters cited above.
Table 1 (using dierent values of , RMR and 0 ).
All charts provide a quick estimation of the geometry parameters
Therefore, 9 case studies that investigated the critical failure surface
(H *, ) necessary for excavation of a rock slope in particular rock
and FS are analysed. The rock material is modelled as Hoek and Brown
geotechnical data (GSI , , ) and required FS. In general, design charts
type using properties of the rock =25 kN/m3, E=1000kPa, =0.3. The
are valuable tools in providing preliminary design values; however, for
parameters of the case A4 (Table 3) whose failure surface (coarse
the nal comprehensive design analysis, a computer programme
mesh) is the one presented in Fig. 5 corresponding to Hoek and
should be used.
Brown's parameters with 0=4, c=0.36 MPa, m=5.56, s=0.00387.
The strength reduction technique is applied in FLAC in FS
5.1. Comparison of results obtained under hypotheses of associative
calculations by progressively reducing the shear strength of the
and non-associative ow rule
material to bring the slope to a state of limiting equilibrium. FLAC
provides an approximate exact solution to the problem at that state, in
To identify the dierences in results taking into consideration
the sense that local equilibrium may not be satised everywhere at the
dierent ow laws and the non-conservative nature of the associative
boundary between zones, but if the zone size is reduced to zero, local
ow rule ( ), the charts are presented in Figs. 68. They relate the
equilibrium will be satised to the limit. In particular, the limit stress
dimensionless height (H *) and rock slope angle ( ) for the limit
equilibrium (FS = 1.0) and dierent values of = 0.001; 0.01; 0.1.
Table 2
For each value of , the correlation for both associative and non-
Geometry data of the studied slopes.
associative ow rules with constant values of 0 = 0; 4; 11.5 is
Case N H* () () () Function N (Table 1) presented. The correlation for the associative ow rule for each value
of is obtained by Geo-Slope through introducing the failure criterion
A 10 51 0 50, 45, 40, 35, 30 1, 2, 3, 4, 5, 6, 7, 8, 9
in the parametric form as a shear/normal function given by Eqs. (14)
B 1 61 0 60, 55, 50, 45, 40 1, 2, 3, 4, 5, 6, 7, 8, 9
C 2.5 45.5 10 45, 40, 35, 30, 25, 20 1, 2, 3, 4, 5, 6, 7, 8, 9 and (15) and introducing the non-associative ow rule with a constant
value of 0 .

52
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

Fig. 5. Failure surface of case A4 obtained by FLAC 7.0.

Fig. 6. Circular failure relationship of H* and the slope angle comparison of the Fig. 7. Circular failure relationship of H* and the slope angle comparison of
associative and non-associative ow rules for = 0.001. associative and non-associative ow rules for = 0.01.

The signicant dierence between the results for the associative means that the geotechnical parameters do not have much impact on
and non-associative ow laws is observed. The smaller the value of H *, the global stability of slope. However, the eect of 0 is more important
the greater the dierence. The higher the value of H *, the more the for lower values of H *.
correlation lines overlap. The same conclusion can be deduced from Fig. 10, which gives the
The smaller the value of the , the greater the dierence between relationship of the slope angle ( ) and range of values of FS=0.81.0.
results for the associative and the non-associative ow laws. These Three values of the H * = 0.1; 1; 12.5; three values of 0 = 0; 4; 11.5
results conrm the role of as the cohesion of rock masses. By contrast, and the particular value of = 0.01 are considered. The greater the H *,
for a higher , the rock mass can be regarded as frictional with little or the lower the eect of 0 on the results, whereas for lower values of H *,
no cohesion. the value of the slope angle ( ) can dier up to 10 for the required value
All three charts (Figs. 68) show that for elevated values of of FS and dierent values of 0 = 0; 4; 11.5 .
H * > 10 , the results do not dier much about whether the associative
or the non-associative ow law with constant value 0 is assumed.
5.3. Eect of the coecient of toughness ( )
The inuence of the on the results of the circular failure are
5.2. Eect of dilatancy angle (0 ) depicted in Fig. 11 with the correlation between the slope angle ( ) and
Fig. 9 presents the variation of the H * and slope angle ( ) for a a wide range of possible values of FS=0.81.6 for the particular value of
variety of values of = 0.001; 0.01; 0.1,0 = 0; 4; 11.5 and the same the dimensionless height (H * = 1). They are classied for each value of
degree of required FS=1.0. It shows that in the region of H * > 10 , i.e., = 0.001; 0.01; 0.1 in combination with dierent values of
for higher values of H *, the eect of 0 and can be overlooked. This 0 = 0; 4; 11.5 . It can be concluded that the higher the for the

53
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

same value of 0 with the same value of FS, the inclination of the slope
to be excavated tends to be more vertical.

6. Examples

To show the numerical application and provide a quantitative


comparison of the results obtained under dierent hypotheses of ow
laws, two examples of application of charts (a step-by-step procedure
for the preliminary design) and a case history are presented.

6.1. Example of application of charts

Assume a slope of height H = 250m excavated in the sandstone


material (m 0 = 19) of an average quality (GSI = 50) with uniaxial
compressive strength C = 15MPa and specic weight = 25kN / m3.
The intermediate data are as follows: 0 = 4 for GSI = 50 (see
Section 3.2); parameters of the rock mass dened for undisturbed
natural rock medium: m = 19exp[(50 100)/28] = 3.19 and
s = exp[(50 100)/9] = 0.00387; strength modulus = 3.1915/8 = 5.97MPa
(Eq. (9)); coecient of toughness = 80.00387/3.192 = 0.00305 (Eq.
(10)); and dimensionless height H * = 250/(5.97103/25) = 1.046
Fig. 8. Circular failure relationship of H* and the slope angle comparison of (Section 5).
associative and non-associative ow rules for = 0.1. The acquired results of this numerical example, concluded from
charts presented in Section 5 for circular failure surface and indicated
in Table 3, shows that the hypothesis of the associative ow law
provides greater values of the slope angle with respect to the constant
dilatancy non-associative ow rule with 0 = 4 (see Table 4).

6.2. Case study

A real case is analysed by comparing the result obtained from direct


application of design charts provided in this research with the limit
equilibrium calculation using Geo-Slope software.30
The failure criterion is given by Eqs. (14) and (15), introducing the
non-associative ow rule by the shear/normal strength function model,
which is used to dene a general data point function as a relationship
between the shear strength and normal stresses along the failure
surface.
The real case of instability corresponds to a slope of limestone
located in Punta Lucera (Vizcaya) in the central sector of the Basque-
Cantabrian watershed.32 The slope has a height of 200 m and can be
characterized by the following geotechnical parameters: RMR =46,
Fig. 9. Circular failure relationship of H* and the slope angle for the non-associative m0=8, c =10 MPa, and =25 kN/m3. Under these conditions,0 = 4 is
rule with = 0.001; 0.01; 0.1 and 0 = 0; 4; 11.5 .
assigned.
A value of the FS=0.65 is obtained using design charts, as shown in
the previous section, while the calculation with Geo-Slope results in a
value close to 0.65, as shown in Fig. 12.

7. Conclusions

As a summary of this study, the following basic conclusions can be


stated:
Dierent charts are provided for the analysis of stability of a
homogeneous and isotropic rock mass by considering the Hoek and
Brown failure criterion and non-associative ow laws that depend on a
constant value of the 0 for the Morgenstern and Price method.13
Charts can be very useful for a preliminary design analysis and
evaluation of parameters involved in a slope stability study. For every
detailed comprehensive nal design of the slope stability, an analysis
by any widely applied commercial computer programme (e.g., Geo-
Slope or FLAC) is required.
The hypothesis of the associative ow law, which is broadly
accepted for its application in slope stability analysis, provides values
Fig. 10. Circular failure correlation of the slope angle and FS for the non-associative of FS that are evidently higher than values obtained under the
ow rule with H* = 0.1; 1; 12.5, = 0.01 and 0 = 0; 4; 11.5 . assumption of a non-associative ow law. The need for considering

54
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

Fig. 11. Circular failure correlation of the slope angle and FS for the non-associative ow rule with H* = 1, = 0.001; 0.01; 0.1 and 0 = 0; 4; 11.5 .

Table 4 and incorporating a non-associative ow law into slope stability


Results of the example application of charts. analysis is conrmed. The non-conservative character of the hypoth-
for FS = 1.0
esis of the associative ow law is noted. The inuence of 0 , chosen in
this analysis as a function of the GSI index,17 on the value of FS is
non-associative ow law (0 = 4) associative flow law important. The higher the value of 0 , the higher the value of FS.
The analysis method presented in this study is compared with
0.001 48 (Fig. 6) 72 (Fig. 6)
commercial programme Geo-Slope while considering the same hypoth-
0.01 54 (Fig. 7) 78 (Fig. 7)
0.00305 49 (interpolated) 73 (interpolated) esis of analysis. The dierence between FS obtained by this method and
Geo-Slope for the circular failure surface is 5% (in 27 analysed cases).
The additional conrmation is the relationship between the distribu-
tion of interslice forces and stresses with the distance along the failure
surface.
The application of the FDM using the FLAC 2D software presents a
good match between obtained results regarding FS for all studied cases.
The FLAC analysis was performed using two dierent grid sizes. For
the ner mesh size the values of FS are similar to the proposed method
solution, while by the wider mesh size the FS is greater being non-
conservative.
The inuence of the function of distribution of interslice forces
f (x ), i.e., distribution of internal stresses, on the value of FS is
negligible.
The inuence of the parameter of Morgenstern and Price ( ) on the
value of the FS is low.
The character of the coecient of toughness ( ) that plays the role
of cohesion of the rock mass is observed. The higher the (the better
the quality of rock mass), the higher the FS.
The higher the H *, the lower the dierence between the FS under
hypothesis of the associative and non-associative ow laws and the
more minor the dierence for dierent 0 values. However, the higher
Fig. 12. Failure surface of the case study presented in chapter 6.2.
the H *, the more minor the dierence between the FS for dierent
values of .

55
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

Appendix A. Determination of the instantaneous friction angle for each slice

Equations of the equilibrium of horizontal and vertical forces are given as follows:

n = ni ni 1 = n tan R x
FS (A1)


P ti + ti 1 n + R tan x = 0
FS (A2)
The relationship between horizontal and vertical interslice forces (Eq. (38)) is introduced in the Eq. (A2):

P nifi + ni 1fi 1 n + R tan x = 0
FS (A3)
Adding and resting fi 1 ni to the Eq. (A3):

n + R tan x P + ni( fi fi 1) + fi 1 (ni ni 1) = 0
FS (A4)
Introducing Eq. (A1) into Eq. (A4):

n + R tan x P + ni( fi fi 1) + fi 1 n tan R x = 0
FS FS (A5)

n (1 + fi 1 tan ) x + R (tan fi 1) x P + ni( fi fi 1 ) = 0
FS (A6)
P
n (1 + fi 1 tan ) FSR (tan fi 1 ) + x
ni = ( fi fi1)
x (A7)
Setting Eqs. (A1) and (A7) equal to the horizontal interslice force on the left side of each slice results in:
P
n (1 + fi 1 tan ) FSR (tan fi 1 ) +
ni 1 + n tan R x = x
FS ( fi fi1)
x (A8)

n (1 + fi 1 tan ) x
ni 1 = n tan R x
FS ( fi fi 1)
(tan fi 1) x P
R +
FS ( fi fi 1) ( fi fi 1 ) (A9)

1 + f tan f tan P
ni 1 = n i
x + R i x +
( fi fi 1 ) FS ( fi fi 1 ) ( fi fi 1 ) (A10)
hi 1 + hi
Setting the value of P = 2
xand passing to dierential equations by (i) x dx ; (ii) fi 1 = fi = sin(x / L ); (iii)
fi1 = fi = ( / L ) cos(x /L ); (iv) (( fi fi 1 )/ x ) fi1; and (v) ((hi 1 + hi )/2) hi 1, the Eq. (A10) becomes:
1 + f tan f tan h
ni 1 = n i
+ R i + i 1
f
i 1 FS fi1 fi1 (A11)
Considering the Hoek and Brown failure criterion with the non-associative ow rule and constant values of the dilatancy angle (0 ), stresses
exerted along the failure surface (Coulomb type of failure) are given by Eqs. (14) and (15). Introducing them into (A11):

1 sin2 1 sin 1 + fi tan


ni 1 = sin 0 +
2sin
2 sin fi1
1 1 sin f tan h
+ cos 0 i + i 1
FS sin fi1 fi1 (A12)
Multiplying the last Eq. (A12) with fi1
FS2sin2 generates:
ni 1f 2FS + FS(1 + 2 sin 0 2 )(1 + f tan ) +
sin2 i 1 i 1
+
+2 cos 0(fi 1 tan ) 2FShi 1
+ sin [2FS sin 0(1 + fi 1 tan ) 2 cos 0(fi 1 tan )] +
+ FS(1 + fi 1 tan ) = 0 (A13)
Constant values in the Eq. (A13), given as AA, BB and CC , only depend on the constant value of the dilatancy angle (0 ):

56
S. Melentijevic et al. International Journal of Rock Mechanics & Mining Sciences 96 (2017) 4757

AA = ni 1fi1 2FS + FS(1 + 2 sin 0 2 )(1 + fi 1 tan ) +


+ 2 cos 0(fi 1 tan ) 2FShi 1 (A14)

BB = 2FS sin 0(1 + fi 1 tan ) 2 cos 0(fi 1 tan ) (A15)

CC = FS(1 + fi1 tan ) (A16)


Therefore, the Eq. (A13) can be written by following algebraic expression with only the value of the instantaneous friction angle () as a variable:
AAsin2 + BB sin + CC = 0 (A17)

References TAC 2002. Mining Innovation and Technology, Toronto, Ontario; 2002. p. 267-273.
19 Melentijevic S. Estabilidad de Taludes en Macizos Rocosos Con Criterios De Rotura
no Lineales Y Leyes De Fluencia no Asociada [PhD Thesis], Spain: Technical
1 Michalowski RL. Limit analysis and stability charts for 3D slope failures. J Geotech University of Madrid; 2005.
Geoenviron Eng. 2010;136:583593. 20 Serrano A, Olalla C, Perucho A. Planar failure surfaces on rock assuming a non linear
2 Taylor DW. Stability of earth slopes. J Boston Soc Civil Eng. 1937;XXIV(3):337386. strength law and constant dilatancy. In Gama D, Ribeiro eSousa L, ed. . Proc. Int.
3 Baker R. A second look at Taylors stability chart. J Geotech Geoenviron Eng. Symp. on Rock Engineering for Mountainous Regions, EuRock 2002., ISRM,
2003;129(12):11021108. Funchal; 2002, pp.179186.
4 Hoek E, Bray JW. Rock Slope Engineering, 3rd ed., London: Instn.Min. Metall; 1981. 21 Serrano A, Olalla C. Closed form solution of planar rock slope stability under Hoek
5 Carranza-Torres C Some comments on the application of the Hoek-Brown failure and Brown (1980) failure criterion and non-associative plasticityLacerda WA, ed. .
criterion for intact rock and rock masses to the solution of tunnel and slope problems. Proc. of Landslides: Evaluation and Stabilization, Rio de Janeiro., Rotterdam:
In:Barla G, BarlaM,editors. Proceedings of MIR 2004-X Conference on Rock and Balkema; 2004, pp.533537.
Engineering Mechanics,vol.2425. Torino; November 2004. p.285326. 22 Manzanas J. Estabilidad de taludes rocosos innitos con criterios de rotura no
6 Li AJ, Merield RS, Lyamin AV. Stability charts for rock slopes based on the Hoek- lineales y leyes de uencia no asociada [PhD Thesis], Spain: Technical University of
Brown failure criterion. Int J Rock Mech Min Sci. 2008;45(5):689700. Madrid; 2002.
7 Li AJ, Merield RS, Lyamin AV. Eect of rock mass disturbance on the stability of 23 Serrano A, Olalla C, Manzanas J. Stability of highly fractured innite rock slopes with
rock slopes using the Hoek-Brown failure criterion. Comput Geotech. nonlinear failure criteria and nonassociated ow laws. Can Geotech J.
2011;38:546558. 2005;42:393411.
8 Steward T, Sivakugan N, Shukla SK, Das BM. Taylors slope stability charts revisited. 24 Hoek E. Strength of jointed rock masses. Geotechnique. 1983;33(3):187223.
Int J Geomech. 2011;11(4):348352. 25 Mostyn G, Douglas KJ. Strength of intact rock and rock masses. In Proceedings
9 Wyllie DC, Mah C. Rock Slope Engineering: Civil and Mining, 4th ed., NewYork: GeoEng2000: An International Conference on Geotechnical and Geological
SponPress; 2004. Engineering, Melbourne, Australia. Vol. 1: Invited Papers, p. 13891421; 2000.
10 Taheri A, Tani K. Assessment of the stability of rock slopes by the slope stability 26 Hustrulid WA, McCarter MK, Van Zyl DJA. Slope Stability in Surface Mining; ISBN
rating classication system. Rock Mech Rock Eng. 2010;43:321333. 0-87335-194-0; 2000.
11 Baker R, Shukha R, Operstein V, Frydman S. Stability charts for pseudo-static slope 27 Sjoberg J. Analysis of Large Scale Rock Slopes [PhD Thesis], Sweden: Lulea
stability analysis. Soil Dyn Earthq Eng. 2006;26:813823. University of Technology; 1999.
12 Shen J, KarakusM Xu.C. Chart-based slope stability assessmentusing the 28 Hoek E, Brown ET. The Hoek-Brown failure criterion a 1988 update. In Curran JH,
GeneralizedHoek-Browncriterion. Int J Min. 2013;64:210219. editor. In: Proceedings of the 15th Canadian Rock Mechanics Symposium, Toronto,
13 Morgenstern NR, Price VE. The analysis of the stability of general slip surfaces. Ontario, October 1988. University of Toronto Press, Toronto, Ontario; 1988. p. 31-
Geotechnique. 1965;15(1):7993. 38.
14 Hoek E, Brown ET. Empirical strength criterion for rock masses. J Geotech Eng Div 29 Fan K, Fredlund DG, Wilson GW. An interslice force function for limit equilibrium
ASCE. 1980;106(9):10131035. slope stability analysis. Can Geotech J. 1986;23(2):287296.
15 Serrano A, Olalla C. Ultimate bearing capacity of rock masses. Int J Min. 30 Geo-Slope International Ltd. Slope/W for slope stability analysis, user guide, Calgary,
1994;31(2):93106. Alberta, Canada: Geo-Slope International Ltd.; 2012.
16 Hoek E, Wood D, Shah S. A modied Hoek-Brown criterion for jointed rock 31 Itasca Consulting Group, Inc., FLAC, Fast Lagrangian Analysis of Continua, Version
massesHudson JA, ed. . Proceedings Symposium ISRM: Rock Characterization 7.0 Fluid-Mechanical Interaction. Minneapolis; 2012.
(Eurock 92)., London: British Geotechnical Society; 1992, pp.209214. 32 Puell F. Aplicaciones de Criterios de Rotura no Iineales a la Estabilidad General de
17 Hoek E, Brown ET. Practical estimates of rock mass strength. Int J Min. Taludes en Macizos Rocosos [PhD Thesis], Spain: Technical University of Madrid;
1997;34(8):11651186. 2003.
18 Hoek E, Carranza-Torres C, Corkum B. Hoek-Brown failure criterion 2002 Edition.
In Hammah R, Bawden W, Curran J, Telesnicki M, editors. Proceedings of NARMS-

57

You might also like