RockSlopeStabilityAnalysisUsingDiscreteElementMethod

0RKDPPDG5H]D0DOHNL-DYDQ1, Fouad Kilanehei 2, Amir Mahjoob 3

Received: 20.06.2014 Accepted: 15.01.2015

Abstract
Rock slope stability depends very much on the strength features of the rock and the geometrical
and strength characteristics of the discontinuities (e.g., roughness, wall strength and persistence).
Since a rock mass is not a continuum, its behavior is dominated by such discontinuities as faults,
joints and bedding planes. Also, Rock slope instability is a major hazard for human activities and
often causes economic losses, property damage (maintenance costs), as well as injuries or fatali-
ties. A computer program has been developed in this research study to perform the stability analy-
sis of a rock slope using the Discrete Element Method (DEM). The rock in the present model is
treated as some blocks connected together by elasto-plastic Winkler springs. This method, the for-
PXODWLRQRIZKLFKVDWLVHVDOOHTXLOLEULXPDQGFRPSDWLELOLW\FRQGLWLRQVFRQVLGHUVWKHSURJUHVVLYH
IDLOXUHDQGLVDEOHWRQGWKHVOLSVXUIDFHRUXQVWDEOHEORFNV7RGHPRQVWUDWHWKHDSSOLFDELOLW\DQG
XVHIXOQHVVRIWKHPHWKRGVHYHUDOH[DPSOHVKDYHEHHQSUHVHQWHGIRUWKHDQDO\VLVDQGRSWLPL]DWLRQ
of the rock slope stabilization.

Keywords:5RFNVORSHVWDELOLW\GLVFUHWHHOHPHQWPHWKRGOLPLWHTXLOLEULXP

Corresponding author E-mail: amahjoob@ut.ac.ir

1.Assistant Professor, Department & Civil Eng.University of Tehran, Tehran, Iran.
2. Assistant Professor, Department of Civil Engineering, Imam Khomeini International University, Qazvin, Iran.
3. Assistant Professor, Road, Housing & Urban Development Research Center, Tehran. Iran.

199 International Journal of Transpotation Engineering,

Vol.2, No.3, Winter 2015
 RockSlopeStabilityAnalysisUsingDiscreteElementMethod

1. Introduction Discrete Element Method (DEM) can be used

Rock slope stability depends very much on for the numerical analyses of different geo-
the strength features of the rock and the geo- WHFKQLFDOSUREOHPVWRR,WZDVSUHVHQWHGUVW
metrical and strength characteristics of the E\&6&KDQJDQGDV
discontinuities (e.g., roughness, wall strength a new concept to investigate the bearing ca-
DQG SHUVLVWHQFH >/LQ HW DO @ 6LQFH D pacity of foundations and stability of slopes
rock mass is not a continuum, its behavior is DQG UHWDLQLQJ ZDOOV .LP HW DO  XVHG
dominated by such discontinuities as faults, DEM and analyzed nailed earth slopes [Kim
joints and bedding planes. In general, discon- HWDO@.YHOGVYLNHWDOH[SODLQHG
tinuities (presence/absence) have profound the static and dynamic loading conditions in
LQXHQFH RQ WKH VWDELOLW\ RI URFN VORSHV DQG DEM code for high rock slopes [Kveldsvik et
their behavior plays a critical part in a stabil- DO@5DWKRGHWDOXVHG'(0IRU
ity evaluation. Several authors have used the the analysis of static and dynamic response of
numerical discontinuum modeling method dam abutments with a liner Coulomb slip con-
to analyze slope stability problems, [Cund- stitutive model [Rathod, Shrivastava and Rao,
DOO@(DVNLHWDOXVHGWKHDERYH @/LQHWDOFRQGXFWHGWKHG\QDP-
method and constructed a model of a natural ic analysis of rock slope based on practical
slope to observe the instabilities caused by seismic load and performed collapse analysis
H[FDYDWLRQV QHDU WKH WRH >(DVNL HW DO @ of the crack development in a rock slope [Lin
=KDQJHWDOFDUULHGRXWVWXGLHVRQWKH
dynamic behavior of a 120 m-high rock slope
of Chinas Three Gorges Dam ship lock using
the discrete element model. They found good
agreement between the numerical results and
WKH HOG PHDVXUHPHQWV RI WKH UHVLGXDO GLV-
SODFHPHQWV RI WKH URFN VORSH GXULQJ WKH H[-
FDYDWLRQXQORDGLQJVWDJH>=KDQJHWDO@
HWDO@.DLQWKRODHWDODQDO\]HGD
100 m high natural hill slope composed of ba-
salt using the DEM code for dry and saturated
FRQGLWLRQV >.DLQWKROD HW DO @ 6WDELOLW\
analysis of Surabhi landslide in the Dehradun
and Tehri located in India, was simulated nu-
merically using the distinct element method
E\3DOHWDO>3DOHWDO@6KHQDQG$EEDV
+HX]HHWDOLOOXVWUDWHGWKHXVHIXOQHVV
of the discrete element approach for the anal-
ysis of rock mass mechanical behavior dur-
ing wave propagation due to seismic events
or rock blasting and concluded that although
FRQWLQXXPFRGHVDUHTXLWHXVHIXOLQVLPXODW-
ing some ground shock effects, they are not
DGHTXDWHIRUUHSUHVHQWLQJG\QDPLFEORFNPR-
WLRQSURFHVVHV>+HX]HHWDO@
It is to be emphasized that in all the above
cases, use has been made of the Coulomb-slip
constitutive model for joints deformations be-
cause it has been a common practice to use
&DQGSDUDPHWHUVWRPRGHOWKHEHKDYLRURI
discontinuities by a linear Coulomb relation.
International Journal of Transpotation Engineering,
Vol.2, No.3, Winter 2015
(2013) developed and applied the random set
distinct element method in the stability analy-
sis of a rock slope from China [Shen and Ab-
EDV@
In this method, stresses on all blocks interfac-
es are compatible with the deformations and
fully satisfy the stress-displacement relation-
ship without any further assumptions. This
PRGHODVOLJKWH[WHQVLRQRIWKHFRQYHQWLRQDO
OLPLWHTXLOLEULXPDQDO\VLVSHUPLWVDVROXWLRQ
WKDWVDWLVHVDOOHTXLOLEULXPDQGFRPSDWLELOLW\
conditions.
2. Discrete Element Method (DEM)
In this method, the rock is modelled by several
200
 MohammadRezaMalekiJavan
FouadKilanehei
AmirMahjoob

solid slices connected together with Winkler happens at a certain interface, its local fac-
springs (compression, tension and shear) (Fig- WRU RI VDIHW\ LV DVVXPHG DV  DQG WKH H[FHVV
XUHWRHVWDEOLVKDXQLTXHERXQGHGV\VWHP
Normal springs behave elasto-plastically and
induce rotational as well as normal stiffness;
they do not yield in compression, but they
do in tension cut-offs. Shear springs yield at
shear capacity according to Mohr-Coulomb
stresses are redistributed among the neighbor-
ing slices through the iteration process. This
continues until the stresses on all interfaces
are compatible with the deformations and ful-
ly satisfy the stress-displacement relationship.
Failure of rock joints shear springs depends
FRQVWLWXWLYHPRGHO)LJXUH%ORFNVJUDYLW\
forces are applied during the analysis.
,QHDFKFDOFXODWLRQVWHSZKLOHDVVXPLQJHTXLY-
alent secant stiffness for the Winkler springs,
the load is increased until the spring stresses
H[FHHG WKH DOORZDEOH YDOXHV DQG ZKHQ WKLV
on the joints shear resistance properties.
The authors have applied the method to the
stability analysis of rock slopes and developed
a computer program with which they have
ZRUNHG RXW VHYHUDO H[DPSOHV WR GHPRQVWUDWH
the methods applicability.

Figure 1. Winkler springs between two adjacent slices

)LJXUH%HKDYLRXURID:LQNOHUVSULQJ

201 International Journal of Transpotation Engineering,

Vol.2, No.3, Winter 2015
 RockSlopeStabilityAnalysisUsingDiscreteElementMethod

Figure 3. Two adjacent blocks

3. DEM Formulation ' pn 1 0  rybp ' px

&RQVLGHU WZR QHLJKERXULQJ EORFNV $ DQG % p
's

0 1

rybp ' py

>R @^'p `
' p 0 1 p
' Z
(Figure 3). The relative displacement between 1
Z
WKHVOLFHVFDQEHFDOFXODWHGXVLQJ(T (2)
'
p
1 0  r ubp b
1 0  r uap a

Fnp 'n
x y x y x
Kn
p
0 0
'  0 1 r u
>K @^' p `
p bp b ap a
y 0 1 r u
y y y y p 0 'p
' 1 Z 0 1 1 Z Fs
0 1 Ks 0 s
p b a
Z M p
0 0 KZ p
' Z
(1)
ZKHUH XD  XE DQG D E DUH WKH GLVSODFH- (3)
PHQWVDQGURWDWLRQVRIEORFNV$DQG%UHVSHF-
tively, P is a point located at the middle of the where K = knL3/12 ; Ks = ksL ; Kn = knL and
interface of the two blocks, and r is a vector L is the interface length.
connecting block As centre of gravity to point The interface forces in the global coordinates
P. If one of the blocks remains immobile, its V\VWHP FDQ EH GHWHUPLQHG XVLQJ (T  DQG
GLVSODFHPHQWVDUHVHWWREH]HURLQ(T the forces acting on all sides of a block can be
Vector npFRVVLQLVGHQHGDVDQLQZDUG IRXQGXVLQJ(TZKLFKLVGHULYHGIURPWKH
unit vector normal to the face of block A at HTXLOLEULXPHTXDWLRQV
SRLQW3ZKHUHLQLVWKHDQJOHLWPDNHVZLWK
WKH [ D[LV QRUPDO WR LW LV Qp VLQ  FRV F xp cos D  sin D 0 Fn
p
p
sin D cos D 0 F p
1RZ XVLQJ (T  EHORZ WKH GLVSODFHPHQW
Fy s >T @^F `
YHFWRURQWKHOHIWVLGHRI(TFDQEHWUDQV- M p 0 0 1 F p
IRUPHGIURPWKHJOREDOFRRUGLQDWHV\VWHP[
y) to the local coordinate system (n-s). (4)
If kn and ks are respectively the normal and fn
a
 1 0 0 Fx p

shear constants per unit length of the Winkler a

spring, the interface force between the two
fs 0 1 0 Fyp
m a ryap  ryap  1 M p
blocks can be calculated as follows:
(5)

International Journal of Transpotation Engineering, 202

Vol.2, No.3, Winter 2015
 MohammadRezaMalekiJavan
FouadKilanehei
AmirMahjoob

where fan, fas and ma are respectively the body &KDQJ>@VKRZHGWKDWWKH\KDYHLQVLJQLFDQW

forces and the moment acting on a block, and effects on the computed results. The ratio (kn/
FDQ LQFOXGH JUDYLW\ LQHUWLD HDUWKTXDNH DQG ks) plays an important role (similar to that
loading forces. of the soils Young and shear moduli) in the
&RPELQLQJ(TXDWLRQVWRZLOOUHVXOWLQ DQDO\VLVDQGLVHTXDOWRIRULVRWURSLF
(T  EHORZ ZKLFK VKRZV WKH UHODWLRQVKLS elastic materials (usually ranging from 2 to 3).
between the forces and displacements of a
block. 4. Computer Program
%DVHG RQ WKH IRUPXODWLRQ GHVFULEHG WKH DX-
^f `  >R @ >T @ >K @>T @
a a T T

(6) thors have developed a 2-D FORTRAN com-

>R @^u ` >R @^u `
b b a a
It is obvious that for N blocks in the analysis,
ZHKDYH1HTXDWLRQVDQG1XQNQRZQYDUL-
ables (fan, fas and ma for each block).
3.1 Failure of Winkler Springs
)RUHDFKEORFNWZRLQWHUQDODQGH[WHUQDOORDG
vectors are introduced; the latter is established
on the basis of the loads applied to the sys-
tem and the former is induced in the springs
DV D FRQVHTXHQFH RI UHODWLYH GLVSODFHPHQWV
between slices.
Computation convergence, a necessity when
incremental loading procedure is adopted in
each calculation cycle, is accomplished when
puter program that can analyse different-shape
rock slopes with any sets of joints. The pro-
gram has an interface that shows the slopes
joints and the un-deformed shape (the failed
joints are in red); so, the progressive failure
and the deformed slope and its blocks can be
observed in the program interface.
0RGHO9HULFDWLRQ
0RGHOYHULFDWLRQLVGRQHE\FRPSDULQJWKH
program results with those of the analytical
VROXWLRQ RI WKH VLPSOH WZREORFN H[DPSOH
Figure 3 shows a sliding surface that is joined
ZLWK DQRWKHU DWWHU VXUIDFH DW WKH WRH RI WKH
slope. The strength reserve in the toe (the pas-
VLYH UHJLRQ UHVWLQJ RQ D UHODWLYHO\ DW VOLG-
LQWHUQDO DQG H[WHUQDO ORDG YHFWRUV EHFRPH
HTXDO $V PHQWLRQHG HDUOLHU ZKHQ D VSULQJ
fails, its stiffness is changed by the secant
method; therefore, the system does not con-
YHUJHLQWKHUVWLWHUDWLRQ7KHH[FHVVVWUHVV
appearing as the difference between the inter-
QDODQGH[WHUQDOORDGYHFWRUVLVUHGLVWULEXWHG
among the neighbouring slices. The iterative
procedure continues until the stresses in the
interfaces of all blocks are compatible with
the deformations and fully satisfy the stress-
displacement relationship.
3.2 Winkler Spring Stiffness
LQJVXUIDFHLVRYHUFRPHE\WKHH[FHVVIRUFH
transmitted from the upper region (the active
block that cannot remain at rest by the friction
along its basal surface alone).
Analysis of the force system shown in Figure
3 yields:
W1 sin(G 1  M1 ) cos(G 2  M 2  M 3 ) 
W2 sin(G 2  M 2 ) cos(G 1  M1  M 3 ) (7)
F
cos(G 2  M 2 ) cos(G 1  M1  M 3 )
where M1 , M 2 , M 3 are the friction angles related
to slip along the upper, lower and vertical sur-
IDFHVUHVSHFWLYHO\1DQG2 are the inclina-
$VVKRZQLQ(TWKHPHWKRGUHTXLUHVNQ tions of the upper and lower slip surfaces,
and ks (Winkler springs stiffness values); respectively, W1 and W2 are respectively the

203 International Journal of Transpotation Engineering,

Vol.2, No.3, Winter 2015
 RockSlopeStabilityAnalysisUsingDiscreteElementMethod

weights of the active and passive blocks per
XQLWRIVOLSZLGWK)LVWKHWKUXVWUHTXLUHGLQ
WKHSDVVLYHEORFNWRUHDFKOLPLWLQJHTXLOLEUL-
um (stable, if F > 0 and unstable if F < 0).
Assuming W1 200 t,
, , and , we mod-
HOOHGWKHH[DPSOHE\WKH'(0SURJUDPFRP-
pared the results with those of the analytical
solutions and found good agreement between
them (Figs. 1 and 2). The red colour in each
joint indicates that shear failure has occurred
in the joint; this gives a good insight on the
condition of each joint and helps in optimiz-
ing the slope stabilization. Furthermore, the
normal and shear stresses in each joint can
, EH GHQHG DQG EORFNV GLVSODFHPHQWV FDQ EH
observed in the DEM method where stresses
in all blocks interfaces are compatible with
their deformations and fully satisfy the stress-
displacement relationship without any further
assumptions.

Figure 4. Model for a two-block stability analysis

Figure 5. , , F 2.6 t or , , F 1.2 t

International Journal of Transpotation Engineering, 204

Vol.2, No.3, Winter 2015
 MohammadRezaMalekiJavan
FouadKilanehei
AmirMahjoob

Figure 6. , , F .6 t or , , F .2 t

6. Rock Slope Analysis using the DEM
5HVXOWVRIVHYHUDOH[DPSOHVDUHSUHVHQWHGWR
show the DEMs applicability for the stability
analysis of rock slopes and its usefulness in
RSWLPL]LQJ WKH VORSH VWDELOL]DWLRQ ([DPSOHV
include the stability analyses of a slope with
two sets of perpendicular joints, a slope with
inclined layers, and a toppling mode.
6.1 Stability Analysis of a Slope with Two
Sets of Perpendicular Joints
7KLVH[DPSOHFRQVLGHUVDVORSHZLWKWZRVHWV
of perpendicular joints with the following
data:
H=5 m c=0 t/m2  WP3
Figure 7 shows the rock slope model and the
joints directions. The red colour in each joint
indicates that shear failure has occurred at the
joint.

Figure 7. Stability analysis of a slope with two sets of perpendicular joints

205 International Journal of Transpotation Engineering,

Vol.2, No.3, Winter 2015
 RockSlopeStabilityAnalysisUsingDiscreteElementMethod

Figure 8. Rock blocks displacements in a slope with two sets of perpendicular joints

)LJXUH6WDELOLW\DQDO\VLVRIWKHVORSHZLWKWKHUHPRYDORIXQVWDEOHEORFNVVORSHLVVWDEOH
Figure 8 shows blocks displacements and in-
GLFDWHVWKDWIDLOXUHKDVRFFXUUHGLQWKLVH[DP-
ple. Figs. 7 and 8 help us decide how we can
stabilize the slope by the removal of unstable
EORFNV7KHVL[XSSHUEORFNVKDYHSURGXFHGD
failure surface and have large displacements,
so we can easily decide to remove them and
UHH[DPLQHWKHVORSHVWDELOLW\)LJXUHVKRZV
that blocks removal has caused the slope to
become stable.

6.2 Stability Analysis of a Slope with In-
clined Layers
The rock slope in this analysis is the same as
that in the previous section, but with inclined
layers. During the analyses, blocks weights
were applied in steps. The rock slope model
and joints directions are shown in Figure 10
and blocks displacements (in the steps) in Fig-
ure 11.
%ORFNUHPRYDO VWDELOL]DWLRQ ZDV H[DPLQHG
ZLWKWKUHHH[DPSOHVWKHUVWRQH)LJXUH

International Journal of Transpotation Engineering, 206

Vol.2, No.3, Winter 2015
 MohammadRezaMalekiJavan
FouadKilanehei
AmirMahjoob

shows that the slope is still unstable, the sec- %ORFNVGLVSODFHPHQWVGXULQJVHYHUDOVWHSVLQ

ond one (Figure 13) implies that block remov-
al has caused the slope to become stable, and
the third one (Figure 14) shows the effects of
a berm on the slope and reveals how applying
a berm can increase the stability of the slope.
the analyses are shown in Figure 15. Compar-
ison reveals that the slope in Figure 14 needs
more block removal than the one in Figure 13
to become stable.

Figure 10. Rock slope model and joints directions (red colour indicates that shear failure has oc-
curred at the joint)

Figure 11. Rock blocks displacements

207 International Journal of Transpotation Engineering,

Vol.2, No.3, Winter 2015
 RockSlopeStabilityAnalysisUsingDiscreteElementMethod

Figure 12. Stability analysis of slope with block removal (the slope is still unstable)

Figure 13. Stability analysis of slope with block removal (the slope is stable)

Figure 14. Stability analysis of a bermed slope with inclined layers

International Journal of Transpotation Engineering, 208

Vol.2, No.3, Winter 2015
 MohammadRezaMalekiJavan
FouadKilanehei
AmirMahjoob

Figure 15. Displacements of rock blocks in a bermed slope with inclined layers

6.3 Toppling Mode Stability Analysis
The rock slope for the modelling of the top-
pling mode is similar to that in Section 6.1,
but with near vertical joints. Figure 16 shows
the model and joints directions and Figure 17
shows the blocks displacements and indicates
WKDW IDLOXUH KDV RFFXUUHG LQ WKLV H[DPSOH ,W
is evident from Figure 16 that block removal
cannot stabilize the slope because shear fail-
ure has occurred in all the joints.

Figure 16. Stability analysis of a toppling mode

209 International Journal of Transpotation Engineering,

Vol.2, No.3, Winter 2015
 RockSlopeStabilityAnalysisUsingDiscreteElementMethod

Figure 17. Displacements of rock blocks in the stability analysis of a toppling mode

7. Conclusions od for bearing capacity analysis, Computers

In this study, the Discrete Element Method and Geotechnics, 12, pp. 273-288.
(DEM) was adopted for the stability analyses
of rock slopes. To demonstrate its applicabil-
ity and usefulness in analyzing and optimiz-
LQJ WKH VORSH VWDELOL]DWLRQ VHYHUDO H[DPSOHV
including the stability analyses of a slope with
inclined layers, a slope with two sets of perpen-
 &KDQJ & 6  'LVFUHWH HOHPHQW
PHWKRG IRU VORSH VWDELOLW\ DQDO\VLV -RXUQDO
of Geotechnical Engineering, 118(12), pp.

dicular joints, and a toppling mode have been
presented and its advantages over the conven-
WLRQDOOLPLWHTXLOLEULXPPHWKRGKDVDOVREHHQ
discussed. Progressive failure with which the
VOLSVXUIDFHRUXQVWDEOHEORFNVFDQEHGHQHG
&KDQJ&6'LVFUHWHHOHPHQWDQDO-
ysis for active and passive pressure distribu-
tion on retaining walls, Computers and Geo-
WHFKQLFVSS
is a subject considered in this method; there-
fore, the method helps in optimizing the rock
slope stabilization. The proposed method is
theoretically rigorous and simple; it can eas-
ily treat such more complicated problems as
a slope with inhomogeneous properties and
forces acting or foundations resting on a slope.
 &XQGDOO 3 $  'LVWLQFW HOHPHQW
models of rock and soil structure, Analytical
and Computational Methods in Engineering
Rock Mechanics. George Allen and Unwin,
/RQGRQSS
 (DVNL 7 -LDQJ< %KDWWDUDL 7 1 0DH-
GD 7 1R]DNL$ DQG 0L]RNDPL 7 
References Modeling jointed rock masses and prediction
&KDQJ&6'LVFUHWHHOHPHQWPHWK- of slope stabilities by DEM, Proceedings of

International Journal of Transpotation Engineering, 210

Vol.2, No.3, Winter 2015
 MohammadRezaMalekiJavan
FouadKilanehei
AmirMahjoob

the 37th U.S. Rock Mech. Symp., Vail, Colo- - Lin, Y., Zhu, D., Deng, Q. and He, Q. (2012)
UDGRSS Collapse analysis of jointed rock slope based
on UDEC software and practical seismic
+HX]H)(:DOWRQ250DGGL['0 load, International Conference on Advances
6KDIIHU 5- DQG  %XWNRYLFK 7 5  in Computational Modelling and Simulation,
$QDO\VLV RI H[SORVLRQ LQ KDUG URFNV WKH 31, pp. 441-416.
power of the discrete element modeling, Me-
FKDQLFV RI -RLQWHG DQG )DXOWHG 5RFN 3URF 3DO6.D\QLD$%KDVLQ$0DQG3DXO
,QW &RQI 9LHQQD %DONHPD 5RWWHUGDP SS 5.(DUWKTXDNHVWDELOLW\DQDO\VLVRI
 rock slopes: A case study, Rock Mechanics
 .DLQWKROD$ 6LQJK 3 . :DVQLN$ %
and Singh, T. N. (2012) Distinct element
modelling of Mahabaleshwar Road cut hill
slope, Geomaterials, 2, pp. 105- 113
DQG5RFN(QJLQHHULQJSS
- Rathod, G. W., Shrivastava, A. K. and Rao,
K. S. (2011) Distinct element modelling for
high rock slopes in static and dynamic con-
ditions: A case study, GeoRisk, ASCE, pp.
.LP-6/HH65DQG.LP-< 
Analysis of soil nailed earth slope by discrete
element method, Computers and Geotech- - Shen, H., Abbas, S. M. (2013) Rock slope
QLFVSS reliability analysis based on distinct element
method and random set theory, International
- Kveldsvik, V., Kaynia, A. M., Nadim, F., -RXUQDORI5RFN0HFKDQLFVDQG0LQLQJ6FL-
%KDVLQ5 %MUQ 1 DQG (LQVWHLQ + + ences, 61, pp. 15-22.
 '\QDPLF GLVWLQFW HOHPHQW DQDO\VLV
of the 800 m high Aknes rock slope, Interna- =KDQJ&3HNDX2.)HQJ-DQG*XDQ-
WLRQDO-RXUQDORI5RFN0HFKDQLFVDQG0LQLQJ JOXQ:$SSOLFDWLRQRIGLVWLQFWHOH-
6FLHQFHVSS ment method in dynamic analysis of high rock
slopes and blocky structures, Soil Dynamic
(DUWKTXDNH(QJLQHHULQJSS

211 International Journal of Transpotation Engineering,

Vol.2, No.3, Winter 2015