Documente Academic
Documente Profesional
Documente Cultură
www.elsevier.com/locate/compgeo
Abstract
This paper concerns the formulation of a macroscopic constitutive relation for multilayered materials with imperfect interface,
which allows taking into consideration the possibility of slipping at the interface of constituents. The model is developed using the
homogenization approach assuming elasto-plastic behavior for constituents. The paper is composed of three parts. The first one
presents the mathematical formulation of the macroscopic constitutive relation. The second describes the numerical implementation
of this relation in a finite element program. The last part concerns the use of this constitutive relation to study the influence of the
interface properties on the behavior of reinforced soils. Results of numerical simulations show the necessity of the consideration of
slipping at the soil-reinforcement interface in the calculation of reinforced earth structures.
# 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Elasto-plastic; Finite element; Homogenization; Interface; Macroscopic; Multilayered; Reinforced earth; Slipping
The interface is considered as a third material with Lint denotes the length of the interface; the symbol
zero thickness (i.e. a surface) which allows tangential stands for the following notation:
M.N. Ensan, I. Shahrour / Computers and Geotechnics 30 (2003) 339–345 341
1 X 2 X
X 2
ðu vÞij ¼ ðui vj þ uj vi Þ ð9bÞ ¼ ahom : E W i Sji : ðep Þj þ Siint
2
i¼1 j¼1
Application of the homogenization approach to peri- : upt nint ð16Þ
odic multilayered materials shows that the micro-stress
and micro-strain tensors are constant in each con-
stituent [1–3]. This result permits the expression of the where ahom denotes the homogenized elasticity tensor,
macro-stress and strain tensors as follows: whose expression is given by:
X
2 X
2
¼ W i i ð10aÞ ahom ¼ Wi SEi ð17Þ
i¼1 i¼1
X
2
Eq. (16) shows that the macroscopic constitutive
E¼ W i ei þ ut nint ð10bÞ relation involves the micro-plastic strains ((ep)i,
i¼1
i=1,2) and the plastic component of the relative
displacement at the interface (upt ) as hardening
The uniformity of the shear stress at the interface parameters.
implies the uniformity of the tangential displacement at
this interface. This result leads to the following condi-
tion on the micro-strain tensors: 3. Numerical implementation
e1 ¼ e2 ; ¼ 1; 3 ð11Þ
The macroscopic constitutive model presented in the
The continuity of the stress at the interface leads to previous section was implemented in the finite element
the following condition: program PECPLAS [14,15] using the classical Newton-
1 2 Raphson scheme generally used in elasto-plastic finite
2 ¼ 2 ¼ 1; 3 ð12Þ
element modeling. The stiffness matrix is calculated
Eqs. (2), (6) and (10)–(12) permit the determination of using the homogenized elasticity tensor ahom. For a
the micro-strain tensors (e1 and e2) in terms of the given loading step, the increment of the macro-strain
macro-strain tensor (E), the plastic microstrain tensors tensor (E) is derived from the field displacement.
((ep)i, i=1,2) and the plastic component of the relative Implementation of the macroscopic constitutive
displacement at the interface (upt ): model required the elaboration of certain subroutines,
X
2 which permit the determination of the model
ei ¼ CEi : E þ Cji : ðep Þj þCiint : upt nint ð13Þ response to the increment of the macro-strain tensor
j¼1 (E). The latter is performed according to the fol-
lowing procedure.
The micro-stress increment is computed assuming an
where CEi represents the strain concentration tensor in elastic behavior for the multilayered material
the elastic domain. Expressions of tensors CEi ; Cji and ( i ¼ SEi : E; int ¼ Kint ut ). Yield criteria were then
Ciint are given in Appendix A. calculated for the two constituents and the interface. If
The micro-stress tensors can be computed from Eqs. any yield criterion is violated, plastic strains are com-
(2) and (13): puted using the consistency condition and the plastic
X
2 flow rules:
i ¼ SEi : E Sji : ðep Þj þ Siint : upt n ð14Þ 8 i
j¼1 < @f : i ¼ 0
>
i ¼ 1; 2
@ ð18Þ
> @f
: int : int ¼ 0
Expressions of tensors SEi ; Sji and Siint are given by:
@
SEi ¼ ai : CEi
8
Sji ¼ ai : ðIij Cji Þ > @gi
< ðep Þi ¼ ðlÞi
@ ð19Þ
Siint ¼ ai : Ciint ð15Þ >
: u ¼ ðlÞ @g
p
int
t int
@
where I and ij denote the fourth order unit tensor and where l is the plastic multiplier.
the Kronecker symbol, respectively. Incorporating Eqs. (2), (5) and (19) into Eq. (18)
The expression of the macro-stress tensor can be leads to the following set of equations for the plastic
deduced from Eqs. (10a) and (14): multipliers:
342 M.N. Ensan, I. Shahrour / Computers and Geotechnics 30 (2003) 339–345
X
2
@f i @g j @f i @gint which can be written in matrix form:
: S ij : lj : S iint : lint nint 2 1 3
j¼1
@ @ @ @ l
@f i ½M :4 l2 5 ¼ ½m ð21Þ
¼ : SiE : E lint
@
X2 where
@fint @g j @fint @gint
: Sij : lj : Siint : lint nint 8
j¼1
@ @ @ @ >
> @f i i @g j
>
> Mij ¼ : Sj : i; j ¼ 1; 2
@fint >
> @ @
: SiE : >
> @f i
@gint
¼ ð20Þ >
>
@ >
> Mi3 ¼ : Siint : nint i ¼ 1; 2
>
> @ @
>
> j
< M3j ¼ @fint : S1 : @g
>
j ¼ 1; 2
Table 1 j
@ @ ð22Þ
>
> @f int 1 @g int
a: Properties of the reinforcement and the soil material >
> M33 ¼ @ : Sint : @ nint
>
>
>
>
Material Elastic C (MPa) ’ (
) (
) W >
> @f i i
>
> mi ¼ : SE : E i ¼ 1; 2
modulus >
> @
(MPa) >
> @f
>
: m3 ¼ int : S1E : E
Reinforcement 10500 0.22 6.0 0.0 0.0 0.008 @
Soil 150 0.3 0.0 30.0 30.0 0.992
The out of equilibrium forces are calculated and dis-
b: Properties of the interface
tributed following the classical method used in elasto-
Eint (MPa) Cint (MPa) int (
) int (
) plastic modeling. Eq. (21) allowed the computation of
5300 0.0 20 4.0
the plastic multipliers, which permit the determination
of the plastic micro-strain tensors ((ep)i, i=1,2) and the
Fig. 2. Response of the reinforced soil to out-of-plane tension loading. (a) Schematic of the configuration, (b) a=30
, (c) a=45
, (d) =60
.
M.N. Ensan, I. Shahrour / Computers and Geotechnics 30 (2003) 339–345 343
plastic relative displacement at the interface (upt ). The be observed that the presence of an imperfect interface
micro-stress and macro-stress increments are calculated largely affects the behavior of the reinforced soil. For the
using Eqs. (14) and (16). The computation of the stress loading direction =30
, failure occurs by plastification
vector at the interface is carried out using Eq. (6). of the soil material and interface slipping, which appear
at the same time. The presence of the imperfect interface
reduces the resistance of the reinforced soil by about 58%.
4. Influence of the interface condition on the behavior However, for other loading directions (=45 and 60
),
of the reinforced soil failure occurs at the interface before plastification of the
soil material. The reduction of the reinforced soil resis-
4.1. Description of simulations tance is about 54% for =45
and 50% for =60
.
The macroscopic model has previously been shown to 4.3. Reinforced earth wall under its own weight
effectively simulate and provide valuable results for
multilayered structures [16,17]. Assuming a perfect bond This section includes analysis of the reinforced earth
between the constituents, good correlation has been illustrated in Fig. 3a under its own weight. The wall is
demonstrated by comparing results obtained by the
macroscopic model to those obtained by finite element
calculations with discretization at the microscopic scale.
The macroscopic model is used to study the influence
of the interface properties on the behavior of reinforced
soils. The study described in this section includes two
numerical simulations. The first concerns the response
of a basic cell to out-of-plane tension loading, while the
second deals with a reinforced earth wall under its own
weight. In both simulations, the response of the rein-
forced earth with an imperfect interface is compared to
that obtained with the perfect interface from references
[16] and [17].
The behavior of the soil is described using the non-
associated Mohr–Coulomb criterion:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
f ¼ ð22 11 Þ2 þ412 2 þ sin’ð þ Þ 2Ccos’
11 22
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
g ¼ ð22 11 Þ2 þ412 2 þ sin ð þ Þ
11 22 ð23Þ