Computers and Georechnics, Vol. 20, No. 1, pp.

I-14, 1997
0 1997 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
PII: SO266-352X(96)00010-9 0266-352X/97 $17.00+ .OO
ELSEVIER

Analysis of Soil Nailed Earth Slope by Discrete Element

Method

J. S. Kim, J. Y. Kim & S. R. Lee

Department of Civil Engineering, Korea Advanced Institute of Science and Technology, Taejon, Korea

(Received 9 August 1995; revised version received 8 July 1996; accepted 24 July 1996)

ABSTRACT

Soil nailing has been widely used during the last two decades, to stabilize
steep excavated slopes in several countries. In this study, the discrete element
method (DE&f) has been applied to evaluate the stability of reinforced slopes.
This method is capable of not only estimating tensile and shear stresses
mobilized in nails but also providing individual safety factors for soil and nails.
It has been assumed that the nailed slope is composed of slices connected
together with elastoplastic Winkler springs. In replacing nails by spring
elements, a function is proposed for evaluating shear and tensile forces in nails
which are induced by relative displacements between nail and adjacent soil.
Taking into account for the sequence of construction, the developed method
can predict the measured tensile forces in nails and hence it properly evaluates
both local and overall stabilities of the reinforced slope. 0 1997 Elsevier
Science Ltd. All rights reserved.

INTRODUCTION

Soil nailing technology is currently adopted as a cut slope retaining system,

generally in nonplastic soils. The design methods that have been mostly used
are Davis [l], German [2], and French methods [3], on which the limit equi-
librium approaches are based [4]. Recently, a kinematical limit equilibrium
design method has been proposed by Juran [5] and Byrne [6]. This method
provides estimated values of maximum tensile and shear forces mobilized in
each reinforcement. In Juran’s method, the normal stress distribution along
the failure surface is evaluated by using Kotter’s equation. Hence, the
method has the advantage that a force developed in each nail can be
obtained from the horizontal force equilibrium of the slice including a nail.
2 S.Kim et al.
.I.

However, there is a possibility of argument on the ambiguity of the boundary

condition in applying Kotter’s equation and the vertical force equilibrium [7].
Besides, all of the above methods cannot appropriately take into account the
construction characteristics of the top-down nailing system.
In the discrete element method [8], a whole slope is considered to be com-
posed of slices that are connected with elastoplastic Winkler springs. This
method considers the condition of compatibility between slices as well as the
equilibrium condition. Thus, in order to obtain the solutions, it is not
necessary to make an assumption as in the limit equilibrium approach which
is an indeterminate problem. The analytical technique adopted in this study
was originally proposed by Chang [9] for non-reinforced slopes. In order to
apply the DEM to reinforced steep slopes, a formulation replacing nails by
elastoplastic springs is presented. The present method is capable of estimat-
ing tensile and shear stresses mobilized in nails and thus providing individual
safety factors of the soil and each nail-reinforcement, especially considering
the realistic sequence of construction.

MODELING OF NAIL BEHAVIOR

Shear force

The shear stress mobilized in the inclusion could be calculated by considering

the equation of an elastic bending of the inclusion and by assuming that the
soil can be represented by a series of elastoplastic springs. Theoretically
the inclusion is considered as infinitely long, and then the solution for the
maximum shear force (fl mobilized at the intersection with failure surface is
obtained as

F = 2EIA3 . u, (1)

where h = 4Jk,Dlo, k, = modulus of lateral soil reaction, D = dia-

meter of the nail, EZ = bending stiffness of the nail and u, = displacement
normal to the direction of nail length.

Tensile force

Soil nails are passive inclusions requiring a soil displacement to efficiently

mobilize the resisting forces. Consequently, it is essential to measure the
frictional properties between soils and nails. The solution for tensile stresses
developed in nails was derived from the following assumption that has been
proposed by Mitachi et al. [lo] in calculating the soil behavior of geo-grids.
 Analysis by discrete element method

Fig. 1. Assumption for frictional behavior between soil and nail.

The frictional property at the interface is assumed to be that shown in Fig. 1.

Combining equations of force equilibrium in the soil-nail system with the
frictional behavior, the following differential equations can be obtained.
d2u
dx2
(24
d2u
_=-
Dq,
dx2 S
Pb)

where S = tensile stiffness of the nail (= EA), u = displacement between

soil and nail in the direction of nail length, D = diameter of the nail, and
= displacement mobilizing peak shear stress at the interface of soil and
UP
nail.
By solving eqn (2) with boundary conditions, the tensile force developed
in the nail can be obtained. According to the magnitude of the mobilized
displacement, two different cases should be separately considered.
(1) ForcaseI:u, < u,atX = L(BC’s:c = OatX = O&U = u,atX = L)
In this case, the magnitude of the mobilized displacement at the assumed
failure surface is less than the magnitude of displacement causing peak shear
stress at the interface of soil and nail. Thus, the distribution of shear stresses
along the nail is assumed to be as shown in Fig. 2(a). Applying the given
boundary conditions to eqn (2a), the following equation is obtained.
FT = m tanh(aL) . uf

where a = ds
(2) For case II: u, > up at X = L (BC’s: E = 0 at X = 0 & u = up at
x = X,)
4 J. S. Kim et al.

The X, is the position in which the peak frictional stress starts to develop.
The distribution of shear stresses along the nail is assumed to be as shown in
Fig. 2(b). The tensile force developed at X = L can be obtained as follows.

FT = m tanh(aX’)u, + Dnt,(L - X,) (4)

Furthermore, the relative displacement at X = L can be calculated by

FT ~DlrtX
u’-up=~- o AdX
EA
J-
(5)
= (LiAxp) JEAD75k tanh(aXp)up + DxrprEi *p)2

where 1 = L - X,. Then, the value of X, can be obtained by a numerical

method provided that the value of ut is given.

MODELING OF GROUND BEHAVIOR

The behavior of normal and shear springs at the sides of slices is assumed to
be elastoplastic and defined by the Mohr-Coulomb criterion. The normal
springs do not yield in compression but they yield in tension. The shear
springs yield when the shear strength is reached for the soil. After the peak
strength is reached, the strength approaches the residual shear strength.

F
~f”‘““sul~
..:
.. ‘. x
.:’
/’
./ I4 X=L
,...
,...‘.....

(4 ut< up 03 Ut’ %

Fig. 2. Distribution of shear stresses along the nail.

 Analysis by discrete element method

METHOD OF ANALYSIS

Modeling of nailed slope by DEM

Discrete elements are discretized as shown in Fig. 3(a). It is assumed that

each discrete element makes rigid body motion and the slice interface can be
modeled by Winkler springs.
The expected failure surface is assumed to be a circle, parabola, or
logarithmic spiral, and the soil mass is divided into active and passive zones
by the expected failure surface. It is assumed that the passive zone does not
induce any deformations.
The constitutive law of Winkler springs is consistent with the Mohr-
Coulomb’s failure criterion. For normal springs, there is no compressive
failure, and some tension cut-off is allowed in cohesive soils, but not in
cohesionless soils. The shear strength is modeled as if for deformations larger
than those at the maximum shear strength (eqn (6)), the residual shear
strength (eqn (7)) was applied according to the ground condition.

(6)
(7)

where cP and & are maximum strength parameters, and c, and & are
residual strength parameters.
The relative displacement between slices including the nail can be trans-
formed from the x-y coordinates system to a specific coordinates system
(normal and tangent to the slope of nail), as shown in Fig. 4. Then, the
displacements in the specific coordinates system are related with the dis-
placements in the x-y coordinate system by

1Sa?]=a+m
a : ground spring
m : ground - nail spring

NORMAL

(4 (W
Fig. 3. Discrete element modeling of the reinforced slope by Winkler springs.
 J. S. Kim et al.

Fig. 4. Relative displacement between slices including a nail

cos( -cz) sin( -a)

- sin(-cz) cos(-_(y) IC >
Ad,
Ad, (8)
where Ad, = d, - d,yi, Ad, = dy - dyi, d_xiand dyi are components of the
initial displacement considered in a sequence of construction, and a is the
intersection angle of the x-coordinate with a nail.
Equation (8) can be written as a matrix form of eqn (9).

29 = BYAd” = Bq(dq -d;) (9)

where the superscript q represents the number of element sides that the nail
crosses the potential failure line.
The formulation of the present method follows a similar procedure to the
previous research by Chang [9]. Let P be the mid point of the interface
between two slices as shown in Fig. 5. The displacement of a slice B relative
to a slice A at point P is then expressed as the following matrix form.

dP = RPhUb _ RPaUa

where RP is a matrix transforming the displacement (u) at the mass center of

a slice to the displacement at the location P. The superscript p represents the
side number of slices.
The relative displacement d in the global coordinate system can be trans-
formed into dP in the local coordinate system for the side face of slices by a
transformation matrix A.
 Analysis by discrete element method

slice A slice B
Fig. 5. Shear and normal stresses between slices due to relative displacement.

&’ = APdP
(11)
Then, the normal and shear stresses at the slice interface are generated in
terms of the normal spring constant (k,), the shear spring constant (k,), and
components of d P. The resultant forces and moment can be obtained by
integrating these stresses over the interface.

For convenience, the side forces in the local coordinate system are trans-
formed into forces in the global coordinate system.

F; = (Ap)TFp (13)

The subscript g represents the effect of ground. The relationship between

the resultant force and the displacement in a nail can be obtained from
eqn (1) and eqns (3) or (4).

$ = K,,$
(14)
where K, is a diagonal matrix for representing the effect of nail and its
components are krr = FT/ur (FT in eqns (3) or (4)) and k22 = 2EIh3. The
tilde (“) symbol stands for the nail coordinates system. With eqn (13) the
nail forces in the nail coordinate system may be transformed into forces in
the global coordinate system.

F,” = (Bq)kq (15)

The forces acting on all sides of a slice should satisfy the equilibrium
requirement given by
 J. S. Kim et al.

j= =y; +fi = 5 -(RpyTF; - (R”yTFz (16)

P

where N = side number of a slice, andf” = body force at the mass center of
element A. Combining eqns (9)-(16) the relationship between force and
moment of a slice and those of a nail can be obtained as follows:

_f” = p Cap(Rpbub- RP’u*) - Daq(Rqaua + d;)

-
P

where
Cap = - (RP") ‘(AP)TK&‘, and Da4 = -(Rr)T(p)TKnp (17)

Simulation of construction sequence

In applying the DEM, the nailed slope is modeled by slices connected toge-
ther with springs as shown in Fig. 3. To consider the installation procedure
of nails according to each excavation step, a nail is installed as the spring of
the excavated face is eliminated. That is, at the initial ground condition, both
self weight and external loads are applied without any nails. After the first
nail is installed, the first spring Si is eliminated as the ground is excavated up to
the level. The simulation process for installing nails is incorporated by eqn (8).
This consideration of real excavation steps is more reasonable since only the
relative displacement of nail-ground after the installation of a nail is allowed
to have an effect on the computation of tensile force developed in the nail.

Selection of tensile force developed in the nail

In the passive zone of soil mass, the resisting force developed in the nail
along the length is considered as a spring force at the expected failure sur-
face. When the resulting displacement is smaller than the pull-out displace-
ment of the nail, the pull-out spring constant is defined as in eqn (3).
However, when it is larger, it is defined as the displacement function of
eqn (14) with eqn (4). Therefore, in obtaining the global nonlinear equation
an iterative numerical computation is required. Furthermore, an inner part
non-linear solution scheme is necessary to calculate the value of XP.

Convergence check criterion

The residual force which is computed during the solution process of the
nonlinear equation by DEM is defined as follows.
 Analysis by discrete element method 9

R =fext
-fint(U) (18)
In a slope reinforced by soil nails, an external loadf,,, is composed of the
self weight of the ground and any external loads applied on the ground sur-
face, and may be assumed to remain constant during an excavation stage.
The internal load vector & is computed by eqn (17). The check on con-
vergence was carried out by setting a criterion as the relative ratio of the
norm of the residual force vector to the norm of the external load vector.

II R II
(19)
IlfextII < &
where generally E= 10e2 - 1O-5 is applied.

Definition of safety factor

In the analysis of a reinforced slope by the conventional limit equilibrium

theory, the overall safety factor concept is applied to both the ground and
the nail. Thus, the local safety evaluation according to the location of each
nail and ground cannot be considered. However, in the analysis of reinforced
slopes by DEM, shear stresses, shear strengths, and forces mobilized in nails
as well as at the interface between any two slices can be computed. There-
fore, the local safety factor for resisting components may be evaluated. The
local and total safety factors for ground and nails are defined as described in
Table 1.

Determination of expected failure surface

In determining the most critical expected failure surface, a plane representing

the minimum safety factor for nails is found by considering a number of
expected failure surfaces. From the results obtained by analyzing a number
of reinforced slopes, it has been found that the ground is almost close to the

TABLE 1
Local and total safety factors

Ground Nail

Local SF.

Total S.F.

a = element No., q = zP or TV,T,,,,, = min [T, (pull-out strength), T, (tensile strength of nail)].
10 J. S. Kim et al.

yield state. Thus, the safety factor for nails is selected as a criterion in
determining the critical plane. However, there are cases where the magnitude
and the location of the maximum stress in nails have been measured in many
real field sites. Therefore, in analyzing these data, the expected failure surface
is selected as the plane on which the sum of tensile forces acting on nails is
the maximum.

COMPARISON OF THEORETICAL PREDICTIONS WITH FIELD

OBSERVED DATA

The proposed DEM approach is evaluated by analyzing the results observed

in real field soil-nailed walls. The following two real field cases and one small
scale test are considered.

Soil nailed wall in Seattle, Washington [6],(11]

The design parameters used were as follows; height of vertical wall = 16 m,

number of nails = 9, nail length = 10.7 m, inclination angle of nail = 15”,
unit weight of soil = 21.6 kN/m3, 4 = 40”, c = 9.6 kN/m2, Young’s
modulus of soil = 30000 kN/m2 (assumed), diameter of nail hole = 0.2 m,
diameter of steel = 0.025 m, horizontal and vertical spaces of nail = 1.8 m
x 1.8 m, maximum frictional resistance of nail = 186 kN/m2, and
% = 0.0127 m [6].
The potential failure surface is chosen to give the maximum sum of tensile
forces in nails. The value of width/height for the most critical failure surface
is about 0.5. Figure 6 illustrates the comparison between the predicted and
measured values of maximum tensile forces.

Cl 35 1 15

TN

Fig. 6. Comparison of tensile forces in nails (Seattle).

 Analysis by discrete element method 11

Soil nailed wall at Cumberland Gap wall [5]

The design parameters used were as follows; height of vertical wall = 16 m,

number of nails = 7, nail length = 9 m, inclination angle of nail = 20”,
unit weight of soil = 18.8 kN/m3, 4 = 38”, c = 4 kN/m2, Young’s
modulus of soil = 45,000 kN/m2, diameter of nail hole = 0.1 m, diameter
of steel = 0.025 m, horizontal and vertical spaces of nail = 1.5 m x 1.5 m,
maximum frictional resistance of nail = 170 kN/m2 (assumed), and
UP
= 0.010 m (assumed).
The comparison between predicted and measured values of maximum
tensile forces is given in Fig. 7.

Small scale test

To have more confidence in using the proposed method a small scale test has
been carried out. An outline of the test is given in Fig. 8. The grain-size-dis-
tribution of the sand used in the small scale test is shown in Fig. 9. To reduce
the wall friction, the bentonite slurry is coated on the wall. The design para-
meters used were as follows; height of vertical wall = 2 m, number of nails
= 4, nail length = 1.5 m, inclination angle of nail = lo”, unit weight of soil
= 18.2 kN/m3, $ = 39”, c = 0, Young’s modulus of soil = 25,000 kN/m2
(assumed), diameter of nail hole = 0.06 m, diameter of steel = 0.01 m,
horizontal and vertical spaces of nail = 0.5 m x 0.5 m, maximum frictional
resistance of nail = 34.9 kN/m2, and up = 0.0065 m. The comparison
between the predicted and measured values of maximum tensile forces is
given in Fig. 10.
In the above examples, predictions for the axial tensile strength display
approximately the same trends of parabolic type as measured. In the case of
Cumberland Gap wall the magnitude of tensile forces is predicted relatively

15
q-N-
Y~WSd3"

Fig. 7. Comparison of tensile forces in nails (Cumberland).

 J. S. Kim et al.

6.0 M
GROUND
I

Fig. 8. An outline of the small scale test.

GRAIN SIZE (mm)

Fig. 9. Grain size distribution of sand used in small scale test.

04
0 05 1 1: 2

TN

Fig. 10. Comparison of tensile forces in nails (small scale test).

 Analysis by discrete element method 13

well. It is mainly due to the consideration of a real construction sequence in

the proposed method which cannot be considered in the other popular
methods. Furthermore, this method of analysis can predict the displacement
at the face of the wall. However, the proposed method assumes that the ele-
ments are rigid, and the design parameters are merely estimated values.
Therefore, it is desirable that the predicted displacement value is used as
auxiliary data.

CONCLUSION

In this study, the DEM which has been used for a general stability analysis
was extended to be applicable to a nail reinforced steep slope. For this pur-
pose, a mechanism to simulate characteristics of interactive behavior of nail
and surrounding ground was established, and it was replaced by elastoplastic
springs. By utilizing the developed method, tensile forces mobilized in nails
could be predicted to some degree of accuracy, and local safety factors can
also be calculated. The results give assistance for the stability analysis of a
reinforced structure. The excavation steps followed for the real sequence of
construction were also considered in the analysis. The main feature of the
developed method is the ability to consider the real sequence of construction
and to appropriately take into account the observed interface behavior
between nails and the adjacent ground.

REFERENCES

1. Michell, J. K., Reinforcement of earth slopes and embankments. NCHRP-290,

Transportation Research Board, 1987.
2. Stocker, M. F. and Riedinger, G., The bearing behaviour of nailed retaining
structures, Geotechnical Special Publication, ASCE, New York, 1990, No. 25,
612628.
3. Schlosser, F., Unterreiner, P. and Plumelle, C., French research program
CLOUTERRE on soil nailing. Geotechnical Special Publication, ASCE, 1992,
2( 12), 739-750.
4. Juran, I. and Elias, V., Soil nailed retaining structures: Analysis of case
histories. Geotechnical Special Publication, ASCE, 1987, New York, No. 12,
232-244.
5. Juran, I., Baudrand, G., Farrag, K. and Elias, V., Kinematical limit analysis for
design of soil-nailed structures. Journal of the Geotechnical Engineering, ASCE,
1990, 116(l), 54-73.
6. Byrne, R. J., Soil nailing: a simplified kinematic analysis. Geotechnical Special
Publication, ASCE, 1992, 2(12), 751-763.
7. Leshchinsky, D., Discussion on kinematical limit analysis for design of soil-
nailed structures. Journal of Geotechnical Engineering, ASCE, 1991, 117(1 l),
1821-1824.
14 J. S. Kim et al.

8. Cundall, P. A., A computer model for simulating progressive, large-scale

movements in blocky rock systems, Nancy, France. Proceedings of the Inter-
national Symposium on Rock Fracture, 1971, Vol. 1, pp. 8-17.
9. Chang, C. S., Discrete element method for slope stability analysis. Journal of the
Geotechnical Engineering, ASCE, 1992, 118(12), 1889-1905.
10. Mitachi, T., Yamamoto, Y. and Muraki, S., Estimation of in-soil deformation
behavior of geogrid under pull-out loading. Proceedings of the International
Symposium on Earth Reinforcement Practice, Japan, 1992, Vol. 1, pp. 121-126.
11. Thompson, S. R. and Miller, I. R., Design construction and performance of a
soil nailed wall in Seattle, Washington. Geotechnical Special Publication AXE,
New York, 1990, No. 25, 629-643.