You are on page 1of 8

Computers and Geotechnics 36 (2009) 1090–1097

Contents lists available at ScienceDirect

Computers and Geotechnics


journal homepage: www.elsevier.com/locate/compgeo

Technical Communication

An embedded bond-slip model for finite element modelling of soil–nail interaction


Y.D. Zhou, C.Y. Cheuk *, L.G. Tham
Department of Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong

a r t i c l e i n f o a b s t r a c t

Article history: Soil nailing has been widely used as a reinforcing technique to retain excavations and stabilise slopes.
Received 25 November 2008 Proper assessment of the interaction between the nails and the surrounding soil is central to safe and eco-
Received in revised form 5 March 2009 nomical design of the composite reinforced soil structure. In this note, a new interface model, denoted as
Accepted 9 March 2009
‘‘embedded bond-slip model”, is proposed to model the soil–nail interaction numerically in a simplified
Available online 11 April 2009
manner. Combining the key features of the embedded element technique and the conventional interface
element method, the proposed plane–strain interface model has the advantages that no special consid-
Keywords:
erations have to be given to the arrangement of the finite element mesh for the soil nails, and that pos-
Soil nail
Embedded bond-slip
sible tangential slippage along the interface can be modelled. The formulation also allows pore water flow
Embedded element across the soil nails to be incorporated into the analysis. The proposed model has been implemented into
Interface element a finite element code and verified by simple element tests under different uni-direction loading condi-
Finite element modelling tions. Using the proposed interface model, back analyses of a field test involving a soil-nailed cut slope
subjected to a rise in groundwater table have been conducted. This note presents the details of the
embedded bond-slip model and the numerical results which demonstrate that the proposed model is
capable of simulating soil–nail interaction conveniently and realistically.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction ‘‘discrete” approaches. In the former approach, the nail is consid-


ered as an additional axial component built into the bulk soil ele-
Soil nailing is an effective reinforcing technique that has been ment such that its nodal displacements are consistent with those
widely used to retain excavations and stabilise slopes. The interac- of the soil element at the same isoparametric positions. As such,
tion between a soil nail and the surrounding soil has important the strengthening effect (i.e., the stiffness contribution) due to
implications on the design methodology, and therefore is of great the presence of the nails is integrated directly into the stiffness ma-
interest to both engineers and researchers. Soil nails used in slope trix of the corresponding soil element. This approach implicitly as-
upgrading works normally consist of a steel reinforcement sur- sumes perfect bonding between the nail and the surrounding soil
rounded by cement grout, and are usually designed as a passive and that the nail force is only effective in the longitudinal direction.
element in that resisting force is mobilised only when slope insta- On the contrary, in the ‘‘discrete” type of model, the soil nails are
bility is triggered by extreme loading. The primary resisting force modelled explicitly. Under plane strain conditions, the nails are
comes from the tensile resistance of the steel reinforcement. The represented either by a continuous plate element [1–3] or a cable
interaction mechanism is characterised by the mobilisation of fric- element [4], both with an equivalent stiffness taking into account
tional forces along the entire length of the inclusion, which conse- the horizontal spacing of the nails. A significant advantage of the
quently results in the generation of tensile forces along the discrete representation of the nails is that possible slippage along
reinforcement. Although the principal mechanism can be qualita- the soil–nail interface can be considered. However, this approach
tively described, the exact stress transfer mechanism is complex, imposes the following restrictions: (1) the finite element mesh
and is dependent on the interfacial behaviour along the soil–nail must be arranged such that a nail element is aligned with the edge
interface which may involve slippage at large displacements. A of a soil element, and (2) a double node is required to represent the
numerical model that can closely mimic the interfacial behaviour relative deformation (i.e. slip) between the nail and the soil. For a
is needed for proper assessment of the soil–nail interaction. reinforced structure with complex geometry, the above restrictions
The representation of a soil nail in a finite element model can be lead to considerable increase in the number of degrees of freedom
broadly classified into two categories, namely the ‘‘embedded” and and complexity of mesh generation. In addition to the above
restrictions, the formulation requires special considerations for
water flow calculations across the structural element as illustrated
* Corresponding author. Tel.: +852 2859 1974; fax: +852 2559 5337.
E-mail addresses: yuande@hkucc.hku.hk (Y.D. Zhou), cycheuk@hkucc.hku.hk in Fig. 1. If the soil nail is modelled explicitly, the surrounding soil
(C.Y. Cheuk), hrectlg@hkucc.hku.hk (L.G. Tham). is separated into two distinct parts by the structural element.

0266-352X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2009.03.002
Y.D. Zhou et al. / Computers and Geotechnics 36 (2009) 1090–1097 1091

Fig. 1. Illustration of local soil–nail interaction and pore water flow around nails.

Special treatment in the numerical model is needed for flow calcu- displacements are compatible with the soil displacements, the nail
lations such that interfacial continuity can be maintained on oppo- stiffness matrix can be assembled together with the soil element
site sides of the soil nail. This deficiency is a major obstacle if pore stiffness matrix to form the total stiffness matrix of the whole sys-
water flow would affect the behaviour of the geotechnical struc- tem. Fig. 2b shows a soil element reinforced by a nail element in
ture under concerned, for example, a soil-nailed slope subjected the embedded bond-slip model. Two user-defined 4-node plane
to rainfall infiltration. interface elements ijj0 i0 and ijj00 i00 are defined along the soil–nail
To overcome the above limitations, a new soil–nail interface interface, of which one side ij represents the nail element and
model is proposed. The new model, denoted as ‘‘embedded bond- the other two sides i0 j0 and i00 ji00 are embedded into the host soil ele-
slip model”, combines the advantages of the embedded element ment. The shear deformation along the soil–nail interface can be
technique and the conventional interface element method. In the represented by the relative displacements in the normal and tan-
model, the soil nail is modelled as a discrete element, but the nodal gential directions of a pair of points, such as P–P0 and P–P00 . The
displacements of the nail are calculated as in the embedded mod- stiffness matrix of the nail and the associated interfaces is estab-
elling approach. Slippage along the interface is allowed by the lished through the following procedures.
introduction of a pair of interface elements. The formulation of As in the conventional embedded element technique, the dis-
the model also implies that no special treatment is needed for placement vectors of the embedded sides, i0 j0 and i00 ji00 , are related
the consideration of pore water diffusive between regions sepa- to the nodal displacements of the host soil element by the follow-
rated by the nail element. Meshing of the entire reinforced struc- ing equations:
ture is convenient and direct, and is independent of the nail
locations. This technical note presents the formulation of the ui0 j0 ¼ N p0 q0 r0 s0 us ð1Þ
new model. A number of verification element tests, together with ui 00 00
j ¼N p00 q00 r00 s00 us ð2Þ
an example application of the model in back analysing a soil-nailed
cut slope subjected to a rise in groundwater table, are presented. where ui0 j0 ¼ ð ui0 x ui0 y uj0 x uj0 y ÞT ; ui00 j00 ¼ ð ui00 x ui00 y uj00 x uj00 y ÞT ;
and the displacement vector of host element us ¼
2. Formulation of the embedded bond-slip model ð u1x u1y u2x u2y u3x u3y u4x u4y ÞT ; (p0 , q0 , r0 , s0 ) and (p00 , q00 ,
r00 , s00 ) are the isoparametric coordinates of the end points of the
Main features of the embedded bond-slip model are described embedded element edges, i0 j0 and i00 ji00 , respectively. The definition
below through a simple example composed of one host soil ele- of N is given by:
ment and one nail element (Fig. 2). In the model, the steel–grout 2 3
N 1 ðp;qÞ 0 N2 ðp;qÞ 0 N 3 ðp;qÞ 0 N 4 ðp;qÞ 0
composite, referred to as the ‘‘nail” hereafter, is modelled as a sin- 6 0 N 1 ðp;qÞ 0 N 2 ðp;qÞ 0 N 3 ðp;qÞ 0 N 4 ðp;qÞ 7
6 7
gle material with equivalent elastic properties; failure along the N ¼6 7
4 N 1 ðr;sÞ 0 N 2 ðr;sÞ 0 N 3 ðr;sÞ 0 N 4 ðr;sÞ 0 5
steel–grout interface is therefore not considered. The nail is mod-
0 N 1 ðr;sÞ 0 N2 ðr;sÞ 0 N 3 ðr;sÞ 0 N 4 ðr;sÞ
elled by a truss element as the major strengthening effect origi-
ð3Þ
nates from the axial resistance [5], whilst the surrounding soil is
modelled by a plane strain 4-node isoparametric element. where Ni(n, g) (i = 1–4) are the shape functions corresponding to the
The conventional approach of embedding a nail element into a four corner nodes in an isoparametric 4-node plane element; N p0 q0 r0 s0
soil element is shown in Fig. 2a. Based on the assumption that end and N p00 q00 r00 s00 correspond to those evaluated at the two edges,

Fig. 2. Modelling of soil nail.


1092 Y.D. Zhou et al. / Computers and Geotechnics 36 (2009) 1090–1097

respectively. For an arbitrary point P0 (P00 ) on the embedded side, i0 j0 where us is the nodal displacement vector of the soil element,
(i00 j00 ), the displacements can be formulated by linear interpolation of un(=uij) is the nodal displacement vector of the nail element; Ps
the two end points according to its isoparametric positions on the and Pn are the nodal external force vectors on the soil and nail
side. For instance, for point P0 , the displacement vector can be writ- element, respectively. The stiffness matrices are given by:
ten as: Z
uP0 ¼ N s ui0 j0 ¼ N s N p0 q0 r0 s0 us ð4Þ K sss ¼ BTs Ds Bs dX
  Xs
Z
1 1
Ns ¼ ð1  tÞI ð1 þ tÞI ð5Þ K iu
ss ¼ N Tp0 q0 r0 s0 N Ts BTu Du Bu N s N p0 q0 r0 s0 dX
2 2 Xu
Z
where uP0 ¼ ð uP0 x uP0 y ÞT ; I is a 2  2 identity matrix; t is the local K id
ss ¼ N Tp00 q00 r00 s00 N Ts BTd Dd Bd N s N p00 q00 r00 s00 dX
isoparametric coordinate of P0 on the side i0 j0 (t e [1, 1]). The dis- Xd
Z
placements at point P00 on the edge i00 j00 can be formulated in same K iu N Tp0 q0 r0 s0 N Ts BTu Du Bu N s dX
sn ¼ 
way. Z Xu
The nodal displacements of the nail element (i.e. ij) are taken as K id N Tp00 q00 r00 s00 N Ts BTd Dd Bd N s dX
sn ¼ 
independent degrees of freedom, from which the displacement at
ZXu
an arbitrary point P can be written as:
K iu
ns ¼  N Ts BTu Du Bu N s N p0 q0 r0 s0 dX
uP ¼ N s uij ð6Þ Xu
Z
T
in which uP ¼ ð uPx uPy Þ ; uij ¼ ð uix uiy ujx ujy Þ . Based on Eqs. T K id
ns ¼  N Ts BTd Dd Bd N s N p00 q00 r00 s00 dX
Xd
(4) and (6), the relative deformation in the upper interface element
K nnn ¼ T T1 K nl T 1
Duup is given by: Z
K iu
nn ¼ N Ts BTu Du Bu N s dX
Duup ¼ uP0  uP ¼ N s N p0 q0 r0 s0 us  N s uij ð7Þ Xu
Z
The relative deformation field in the lower interface element can be K id
nn ¼ N Ts BTd Dd Bd N s dX
Xd
formulated in the same fashion. To establish the traction–displace-
ment relationship along the soil–nail interface, the mechanical where the subscripts ‘‘s” and ‘‘n” indicate the domain of soil element
behaviour of each component, including the nail element, the soil and nail element, respectively; ‘‘u” and ‘‘d” refer to the upper and
element, and the soil–nail interface, needs to be defined and incor- lower interface element domains, respectively; ‘‘iu” and ‘‘id” refer
porated in the global stiffness matrix. to the interaction from upper and lower interface zones, respec-
In the model, it is assumed that the nail element remains elastic tively; the B matrix denotes the strain-deformation transformation
under all loading conditions, and its elemental stiffness matrix K ng matrix for each domain; Xs, Xu and Xd are the integration domain
in the global coordinates can be written as: of soil element, the upper and the lower interface zones,
respectively.
K ng ¼ T T1 K nl T 1 ð8Þ The formulation described above illustrates that the embedded
  bond-slip model combines the advantages of the embedded ele-
AE 1 1
K nl ¼ ð9Þ ment technique and the conventional interface element method.
L 1 1
Since only the degrees of freedom of the nail element and soil ele-
where K nl is the stiffness matrix in the local coordinates x0 oy0 ; E is ment are incorporated (Eq. (11)), the nonlinear bond-slip interfa-
Young’s modulus; A and L are the cross-sectional area and the cial behaviour along the soil–nail interface can be incorporated
length of nail element, respectively; T1 is the transformation matrix through proper definition of Du and Dd. It should also be noted that
under plane conditions: the nodal external force on nail element Pn is normally zero in
  most practical cases due to the passive nature of a soil nail. The de-
cos h sin h 0 0 grees of freedom associated with the nail element can therefore be
T1 ¼ ð10Þ
0 0 cos h sin h condensed out from the element stiffness matrix before it is
assembled into the system stiffness matrix. Moreover, the formula-
Similar to the formulation of the so-called joint elements [6–8], the tion ensures that deformation and possible diffusive behaviour,
stiffness matrix of the interface elements, ijj0 i0 and ijj00 i00 , can be for- such as pore water flow, would be continuous within the soil
mulated directly by the definition of normal and tangential stiffness domain.
components. Alternatively a finite but thin adhesive layer can be The embedded bond-slip model has been implemented into the
introduced between the nail and surrounding soil, as in the formu- finite element package ABAQUS [10] through the user element
lation of the interface elements of finite thickness by Desai et al. [9]. (UEL) subroutine. The model has been verified by a number of sim-
The very thin zone would behave according to a constitutive model ple analyses based on a two-element model subjected to unit dis-
specified for contact and frictional sliding, such as damage or elasto- placement boundary conditions in different directions. An example
plastic models. In the present study, the former approach of directly application of the model to a field test on cut slope reinforced by
defining the normal and tangential stiffness components is adopted. soil nails has also been conducted. The results of these analyses
For simplicity, the stiffness matrices of the upper and lower interfa- are presented in the following sections.
cial elements in the global coordinates are denoted as Du and Dd,
respectively.
3. Verification analysis
Eq. (7) suggests that the relative deformation of the interface
element can be fully described by the nodal displacements of the
3.1. Element tests
soil element and the nail element. Based on the principle of virtual
work, the elemental equilibrium equations can be written as:
A simple exercise composed of two plane strain elements with
" #    overall dimensions of 1 m  2 m (Fig. 3b) is chosen for validating
K sss þ K iu
ss þ K id
ss K iu
sn þ K id
sn us Ps
¼ ð11Þ the implementation of the embedded bond-slip model. A nail, with
K iu
ns þ K id
ns K nnn þ K iu id
nn þ K nn
un Pn
a cross-sectional area of 0.01 m2, is located at a distance of 0.1 m
Y.D. Zhou et al. / Computers and Geotechnics 36 (2009) 1090–1097 1093

bond-slip model are in close agreement with those predicted by


the conventional embedded element technique. Under direct ten-
sile or compressive loading, the stresses mobilised in the soil are
very close for the two interface models. The small discrepancy in
the nail displacement and the corresponding mobilised nail force
can be attributed to the presence of the interface elements which
allows relative movements between the soil and the nail. For the
case of shearing, the differences between the two models are rela-
tively larger. The average vertical stresses r
 y mobilised in the soil
and the nail across the entire height of the model are within 10%
comparing the two models.

3.2. Back analysis of field test on nailed cut slope

3.2.1. Numerical model of the slope


Fig. 3. Element tests under prescribed displacement boundary conditions. A full-scale field test on a cut slope was conducted by the
Department of Civil Engineering of the University of Hong Kong
to study the development of nail forces when the slope was sub-
from the left side and is modelled by two truss elements. The bot-
jected to a rise in groundwater level [11]. The overall performance
tom edge of the model is vertically fixed. Loading is applied
of the slope was monitored through a comprehensive instrumenta-
through specifying a unit displacement (1 mm) at the top bound-
tion system. The field measurements are used to assess the appli-
ary of the model. Three typical loading conditions, namely tension,
cability of the embedded bond-slope model in mimicking
compression, and shear modes, are considered.
mobilisation of nail forces.
An idealised case with perfect bonding along the soil–nail inter-
As shown in Fig. 4, the 10 m high cut slope constructed for the
face is chosen for the verification analysis, such that a direct com-
field test had a steep angle of 55° to the horizontal. The slope
parison can be made with the conventional embedded element
mainly consisted of completely decomposed volcanic (CDV) soil
method (Fig. 3a) which implicitly assumes perfect bonding be-
covered with a thin layer of residual soil and fill material. The
tween soil and nail. When the embedded bond-slip model is used,
width of the cut slope was 8 m, with a moderately gentle sloping
the interface elements are assumed to be purely elastic and the
crest area with an average gradient of 23°. Nine (3  3) soil nails
interfacial stiffness matrix Du and Dd are defined by the normal
with vertical and horizontal spacings of 2 m were installed into
(kn) and shear (ks) stiffness moduli which are of similar magnitude
the slope, at an inclination of 15° to the horizontal. Each nail, com-
to the stiffness of the nail element to approximate perfect bonding
posed of 6 m long high yield steel bar with a diameter of 25 mm,
conditions. A summary of the material parameters adopted in the
was inserted into a drilled hole of 100 mm in diameter. The empty
verification analyses is shown in Table 1.
space along the length of the hole was fully grouted. Grout curtains
Table 2 compares the numerical results obtained from the two
were formed behind, at the bottom and on both lateral sides of the
models. In general, the predictions obtained from the embedded
slope in order to form a barrier so that groundwater level was
raised artificially when water was injected into the slotted PVC in-
let pipes. Further information on the field test can be found in
Table 1 Kwong and Lee [11].
Input parameters for the idealised element tests. During the field test, loading was applied through a 24-hour
Elastic properties supply of water which continued for almost 12 days (from 20th
Soil element E = 20.0 MPa, m = 0.25
to 31st October 2005). The change in the groundwater level during
Truss element (nail) E = 2.0  104 MPa, m = 0.15 the watering process was determined by piezometers. The ground-
Soil–nail element kn = 2.0  109 N/m3, ks = 1.0  109 N/m3 water levels at three chosen moments are shown in Fig. 4, denoted
as Stage I (mid-day of day 5), II (mid-day of day 7), and III (mid-day

Table 2
Results of verification analyses.

Case r (tension)
Soil, ry (KPa) Nail, ry (KPa) Nail top, uy (mm)
Embedded element model 10.7 1.0  104 1.0
Embedded bond-slip model 10.87 0.96  104 0.98
Difference (%) 1.9 4.7 2.0
Case s (compression)
Soil, ry (KPa) Nail, ry (KPa) Nail top, uy (mm)
Embedded element model 10.7 1.0  104 1.0
Embedded bond-slip model 10.87 0.96  104 0.98
Difference (%) 1.9 4.7 2.0
Case t (under shear)
 y a (KPa)
Soil, r Nail, r
 y (KPa) Nail top, ux (mm)
Embedded element model 2.18 217.6 1.0
Embedded bond-slip model 1.96 210.9 1.0
Difference (%) 10.0 3.1 0.0
a
Note: r
 y denotes the average of stress component ry along the entire model height.
1094 Y.D. Zhou et al. / Computers and Geotechnics 36 (2009) 1090–1097

@p0 m 0
K¼ ¼ p ð12Þ
@ eev j
where eev denotes the elastic volumetric strain, m = 1 + e is the spe-
cific volume, e is the void ratio, and j is the slope of the recompres-
sion-unloading line on the m–ln p0 diagram. The Poisson’s ratio, l, is
assumed to be constant, and the shear modulus, G, is inferred from
l and K.
The model parameters describing the mechanical and hydraulic
properties of the soil are obtained from relevant field or laboratory
tests. Table 3 summarizes the parameters adopted in the analysis.
The hydraulic behaviour of the three types of soil in unsaturated
soil state is assumed to be the same. This is characterised by the
permeability function and the soil–water characteristic curve
(SWCC) which are assumed based on laboratory and field measure-
ments (Figs. 6a and b). For the grouted soil which serves as an
impermeable flow boundary, a layer of soil elements with a com-
paratively small hydraulic conductivity of 1  1012 m/s is adopted
to represent its impermeable property, whilst the stiffness and
shear strength properties are assumed to be the same as those of
the original soil.
In the numerical model, the soil nails made up of steel rod and
Fig. 4. Overview of the field test on a cut slope subjected to rise in groundwater grout are idealised as one-dimensional homogeneous bars.
level (Stage I: 12:00 day 5; Stage II: 12:00 day 7; and Stage III: 12:00 day 9). Through the compatibility assumption of the axial deformation
of the rod and surrounding grouts along the longitudinal direction,
the Young’s modulus of the idealised bar is determined by:
of day 9). Field measurements showed that the groundwater table
was finally raised to the bottom of the fill layer, which is approxi- E r Ar þ E g Ag
mately 1.5 m below the slope surface. The measured nail loads in e
E¼ ð13Þ
Ar þ Ag
response to the rise of groundater table are compared with the
numerical predictions. where e E is the equivalent elastic modulus of the nails; Er and Eg are
A plane strain coupled analysis has been conducted to model the elastic moduli of the steel and the grout, respectively; Ar and Ag
the nailed cut slope under the field test conditions. The finite ele- are the cross-sectional area of the steel rod and grout, respectively.
ment mesh is shown in Fig. 5. The soil is modelled by a plasticity The interface elements are modelled by a plasticity model with
model with a modified Mohr–Coulomb failure criterion [12] and a classic Mohr–Coulomb failure criterion. The elastic properties are
a non-associated flow rule. The modified Mohr–Coulomb failure defined by direct input of normal and tangential stiffness compo-
criterion is used to capture the contribution of suction since the nents, which are in the same order as those of the surrounding soil.
field test involved wetting of initially unsaturated soil. For simplic- Due to the absence of nail pull-out test results, the apparent cohe-
ity, the two friction angles describing the failure criterion (ub and sion and friction angle along the soil–nail interface are defined as
u) are assumed to be the same in this study. A smooth flow poten- the same as those of the surrounding soil. A zero dilation angle is
tial function proposed by Menétrey and Willam [13] is adopted in defined for the interfacial behaviour.
the model. The function has a hyperbolic shape in the meridional
stress plane and a piecewise elliptic shape in the deviatoric stress 3.2.2. Results and comparisons
plane. Plastic flows in the deviatoric and meridional planes are To illustrate the merits of the new embedded bond-slip inter-
non-associated in general, and dilatancy can be controlled by the face model, a different set of analysis using the conventional
dilation angle. A perfect plastic hardening law is assumed in the embedded element technique is also carried out for comparison.
calculations. The predicted nail forces at different wetting stages by the two
To capture the stress-dependent stiffness of typical residual soil, types of nail models are compared with the field measurements
the bulk modulus, K, of the soil skeleton is written as a function of in Figs. 7–9. The reported field measurements correspond to the
the mean effective stress, p0 , according to average value for each row of nails. They illustrate that the mobi-
lised nail force increases from Stage I to III due to the change in
pore water pressure distribution and the increase in soil sub-
merged weight.
The two nail models generally predict similar patterns of nail
load distributions in response to the rise of groundwater level.
The new embedded bond-slip model gives better predictions at lar-
ger loads (i.e. higher groundwater levels). With a low groundwater
table in Stage I, the mobilised nail loads are comparatively small
for all rows of nails. The predictions from the two nail models give
similar distribution pattern along the whole nail length, with a
slight under-prediction at the nail head for the bottom nail. In
the Stages II and III, larger nail forces are triggered as the ground-
water level rises. More prominent difference can be seen between
the numerical results produced by the two nail models. In particu-
lar, for the lower two rows of nails, the embedded bond-slip model
gives better predictions along the length of the nail. The discrep-
Fig. 5. 2D Finite element mesh of the cut slope. ancy between the two models is most significant near the nail
Y.D. Zhou et al. / Computers and Geotechnics 36 (2009) 1090–1097 1095

Table 3
Summary of input parameters for the cut lope back analysis.

Initial conditions Elastic property Shear strength Hydraulic properties


3 3 3
Fill qd = 1.4  10 kg/m j = 8.4  10 c = 1 KPa, Variable k – Fig. 6a
Mc0 = 15%, e0 = 1.0 l = 0.25 ub = u = 33°, w = 3° SWCC – Fig. 6b
Residual soil qd = 1.5  103 kg/m3 j = 8.4  103 c = 1 KPa, Variable k – Fig. 6a
Mc0 = 15%, e0 = 1.0 l = 0.25 ub = u = 35°, w = 3° SWCC – Fig. 6b
CDV qd = 1.5  103 kg/m3 j = 8.4  103 c = 6 KPa, Variable k – Fig. 6a
Mc0 = 15%, e0 = 1.0 l = 0.25 ub = u = 33°, w = 3° SWCC – Fig. 6b
Nail – E = 2.2  104 MPa, – –
l = 0.2
Soil–nail interface – kn = 3.0  108 N/m3 c = 6 KPa, –
ks = 1.5  108 N/m3 u = 33°, w = 0°

Note: E, l, j, Mc0, qd, e0, k, w, c and u are Young’s modulus, Poisson’s ratio, the slope of the unloading-reloading line on the m  ln p0 diagram, initial moisture content, dry
density, initial void ratio, permeability coefficient, dilation angle, cohesion intercept and internal friction angle, respectively, and the subscript ‘‘0” denotes the initial value.

ends. This is presumably the location where slippage is likely to The field measurements at the bottom row of soil nails (Fig. 9)
occur. present a large nail force at the nail head in Stage III, in an order of

Fig. 6. Hydraulic properties of the soil.


1096 Y.D. Zhou et al. / Computers and Geotechnics 36 (2009) 1090–1097

Fig. 7. Distribution of nail force at the top soil nail.


N

Fig. 8. Distribution of nail force at the middle soil nail.

3 3 3

3 3 3

Fig. 9. Distribution of nail force at the bottom soil nail.

20 kN. However, the simulation results by either of two nail mod- the advantages of the embedded element technique and those
els give a much smaller prediction, approximately <5 kN. The dif- of the conventional interface element method. Based on the
ference can be attributed to the confining effect of nail head principle of virtual work, the equilibrium equations of the soil
arrangement in the field test, which consisted of a steel plate and and nail composite system are formulated, with respect to the
nut assembly with a concrete pad. The passive pressure induced displacement degrees of freedom associated with the nail and
by the nail head–soil interaction has not been considered in this the soil elements. The new model takes into account possible
numerical study. This aspect of soil–nail interaction deserves fur- tangential slippage along the soil nail. With the new model, fi-
ther investigation. nite element discretisation of the soil domain is not constrained
The comparisons described above illustrate that both the con- by the locations of the soil nails. The new model has been
ventional embedded element technique and the new embedded implemented into a commercial finite element code, and veri-
bond-slip model are capable of modelling the soil–nail interaction. fied by simple element tests. The model has been used in a
However, under extreme loading conditions, the new bond-slip back analysis of a field test conducted on a cut slope. Compar-
model could mimic possible slippage along the soil–nail interface, isons between predictions and field measurements demonstrate
and hence give better predictions on the mobilised nail force. On that the new model gives better nail force predictions over the
the other hand, the conventional embedded element technique as- conventional embedded element technique, especially under ex-
sumes perfect bonding along soil–nail interface and therefore over- treme loading conditions where slippage along soil–nail inter-
estimates nail force mobilised in the soil nail. face may occur.

4. Conclusions Acknowledgements

A new embedded bond-slip model has been developed to The authors acknowledge the support by Grant Nos. 7176/05E
simulate the soil–nail interaction. The new model combines and 7171/06E of the University of Hong Kong.
Y.D. Zhou et al. / Computers and Geotechnics 36 (2009) 1090–1097 1097

References [7] Zienkiewicz OC, Best B, Dullage C, Stagg K. Analysis of nonlinear problems in
rock mechanics with particular reference to jointed rock systems. In:
Proceedings of the second international congress on rock mechanics,
[1] Nagao A, Kitamura T, Mizutani J. Field experiment on reinforced earth and its
Belgrade; 1970.
evaluation using FEM analysis. In: Int geotech sym on theor and pract of earth
[8] Ghaboussi J, Wilson EL, Isenberg J. Finite element for rock joints and interfaces.
reinforc. Fukuoka, Japan; 1988. p 329–34.
J Soil Mech Found Div, ASCE 1973;99(SM10):833–48.
[2] Ehrlich M, Alimeida MS, Lima AM. Parametric analysis of soil nailing system.
[9] Desai CS, Zamman MM, Lightner JG, Siriwardane HJ. Thin layer element for
In: Yasufuku N, Omine V, editors. Earth reinforcement. Rotterdam: Balkema;
interfaces and joints. Int J Numer Anal Methods Geomech 1984;4(8):19–43.
1996. p. 747–52.
[10] ABAQUS Inc. Analysis user’s manual. Version 6.6; 2006.
[3] Yang MZ, and Drumm EC. Numerical analysis of the load transfer and
[11] Kwong AKL, Lee CF. A field test study on instrumented soil nail installed in cut
deformation in a soil nailed slope. Geotechnical special publication, no. 96;
slope. In: Proceedings of sixth international conference on case histories in
2000. p. 102–16.
geotechnical engineering and symposium (CD-ROM), Arlington, VA, USA;
[4] Cheuk CY, Ng CWW, Sun HW. Numerical experiments of soil nails in loose fill
August 11–16, 2008.
slopes subjected to rainfall infiltration effects. Comput Geotech
[12] Fredlund DG, Morgenstern NR, Widger RA. Shear strength of unsaturated soils.
2005;32:290–303.
Can Geotech J 1978;15(3):313–21.
[5] Jewell RA, Pedley MJ. Analysis for soil reinforcement with bending stiffness. J
[13] Menétrey P, Willam KJ. Triaxial failure criterion for concrete and its
Geotech Eng 1992;118(10):1505–28.
generalization. ACI Struct J 1995;92(3):311–8.
[6] Goodman RE, Taylor RL, Brekke TL. A model for the mechanics of jointed rock. J
Soil Mech Found Div, ASCE 1968;94(SM3):637–59.

You might also like