You are on page 1of 8

MATHEMATICAL COMPUTER MODELLING

PERGAMON

Mathematical

and Computer

Modelling

37 (2003) 623-630 www.elsevier.nl/locate/mcm

A Strain-Hardening Elastoplastic Model for Sand-Structure Interface under Monotonic and Cyclic Loading
M. BOULON UniversitZ: Joseph Fourier de Grenoble BP 53-38041 Grenoble Cedex 9, France marc.boulon(Dhmg.inpg.fr V. N. GHIONNA AND G. MORTARA Faculty of Engineering, University of Reggio Calabria Via Graziella, Feo di Vito, 89060 Reggio Calabria, Italy <ghionna><mortara>@ing.unirc.it
Abstract--Frictional interaction between granular materials and solid surfaces is encountered in many geotechnical applications. Such an interaction develops through a very thin layer of soil called interface. This paper deals with the formulation of a nonassociated elastoplastic model for interface behaviour under monotonic and cyclic conditions. Two plastic surfaces are employed: an external surface subjected to isotropic hardening and an internal one subjected to a purely kinematic rotational hardening. Flow rule is similar to that used in Cam Clay model but differs if hardening or softening conditions are involved. @ 2003 Elsevier Science Ltd. All rights reserved.

Keywords-I+iction,

Soil-structure interface, Plasticity, Cyclic loading.

1. INTRODUCTION
The study tures of the frictional characteristics of the interfaces the behaviour between granular materials works. and strucThis interacis fundamental for understanding of many geotechnical

tion develops inside a thin layer of sand called interface whose behaviour sand characteristics and roughness of the inclusion surface. Furthermore, different when static or cyclic conditions are examined. These features can be observed through experimental tests performed

principally depends on this behaviour is very with the modified shear at to

box apparatus, that allows one to perform tests at constant constant normal stiffness (CNS). In the latter tests the stress the shear according surface is allowed to vary (linearly) with to a constant stiffness coefficient

normal load (CNL) and tests acting in the direction normal normal

the displacement

to the interface,

K,

where on and u are the stress and the displacement normal to the interface, way, the volumetric behaviour of the interface defines the sliding resistance

respectively. In this of the inclusion in a


Typeset by A@- QX

0895-7177/03/$ - see front matter @ 2003 Elsevier Science Ltd. All rights reserved. PII: SO895-7177(03)00056-6

decisive ciated

manner. constitutive

The conceptual is to explain elastoplastic

model that defines the aforementioned the mechanical model behaviour explain behaviour under static two surface based on isotropic

is shown in Figure through and rotational

1 [I!

The aim of this paper The model

of the interface

a nonasso hardening where t lw

is able to predict respectively.

interface

and cyclic conditions,

role of dilatancy of the inclusion,

and sand densification

the increase

and decrease

of the shear resistam(,

grouted body (rough surface) I

adjacent soil (idealised by spring)

Traction t

axial displacement of anchor

\ <
sz

-a*-

amount of dilation Vr = compression d spring

initial state
Figure 1. Restraining effect of the soil on the interface

strained state
dilation [I].

Boulon and,

and Nova [2] observe of these

the similarities analogies, adapt

between

triaxial

and direct

shear

tests

on sand that

as a consequence

for the interfaces

an elastoplastic

model

was previously formulated for the triaxial behaviour of sands [3], substituting opportunely the variables involved in the model. The present work follows this approach and is formulated in terms of normal and shear stresses and displacements, that is cn, r, u, and w.

2. PLASTIC
The isotropic placement plastic

FUNCTIONS
function of the model [4]
wn =

AND
is based

HARDENING
on the definition

LAWS
of the normalized dis-

w, defined

by the equation

J t

~2, dt, displacement

(2)

where ti, at failure

is the ratio between the plastic shear displacement rate tip and the plastic to the peak value of the stress ratio np, wpr corresponding ti,=--. tip
WP

From definition (2) it follows that the locus of the stresses at failure corresponds function for w, = 1 (Figure 2). The equation of the family of the plastic functions is f = 0 - Q1un = 0,

to the plastic of the model (4)

Strain-Hardening

Elsstoplastic

Model

625

experimental data at failure

0
0

I 200

I 400

I 600

normal stress CL (kPa)


Figure 2. Experimental Mortara [4]). data at failure and plastic function for W, = 1 (data from

where different

~1:is a function values of wn.

of the normalized function.

displacement

that

defines

the rule of expansion envelopes utilized

and for to

contraction Figure interpolate

of the plastic

Function between a(~,)

CYrepresents

the slope of the stress and function

3 shows the comparison oeXP values.

experimental is described

cr values

cr(w,)

The function 4%)

by the equation + oCr according (5) to the

= o, [(ww, + 1) - 23 exp { -Lw,M} of the model

where

CQ, w, and $ are parameters

and L and M are derived

conditions

a(w, = 1) =
In (6), or, is the maximum conditions is obtained

CQ,

&(w_ 1) = = 7l
function. subjected

0.

value of the a(~,)

The extension to kinematic

of the model rotational

to cyclic hardening

by inserting

a cyclic surface

1.0

0.8

g ._ I! 12 8

0.6

0.4 a.np 0.2 ab.) = = =

1.0

2.0 w.

3.0

4.0

normalized displacement
Figure 3. Comparison between experimental

and model cx values

M. BOULON et ~11.

*
normal stress 4

Figure 4. Plastic

surfaces

of the tiodel.

inside surface

the isotropic is described

plastic

surface.

Figure

4 shows the plastic - Q()dn = 0. system

surfaces

of the model.

Cyclic (7)

by equation f(J = v@

Stresses

8,

and 7 are the stresses

in the solid reference

with the surface

(7) rotated

by

an angle 8 with respect

to axis T = 0, 8,
=c~cos~+Ts~~I~, -o,sine+7.c0se. (8)

7 =

Constant

CYO determines amplitude

the size of the cyclic surface of the isotropic arctan surface

and is assumed

as a constant

proportion

of

the maximum

o0 = C, arctan cyp, 4 is given by

(9)

where C, is a parameter
c

of the model.

The angle 0,. shown in Figure - sgn arctan L. cn

er = arctano for plastic state

(10) is given by an interpolation [5] (11)

The hardening function

modulus

inside the external

surface

which is formally

the same as in the Dafalias H=H,+h------, 6.

and Popov

model

eTo - 8, mechanism and 6&c represents h is given by

where H,

is the hardening

modulus

relative

to the isotropic occurs. Function

the

value of the angle 6$, when a load inversion h=

(12)
displacement. displacement,

where y, b, n, and Xh are parameters of the model while se is an arbitrary reference Function (12) includes the dependency of the hardening modulus on normalized densification of sand, and relative position of the two surfaces. Quantity Rae is R ae = 2(arctancr - arctanac)

e TO

(13)

Strain-Hardening

Elastoplaetic

Model

627

Rae is important follows only cycle.

in the simulation occurs

of tests that

with

unsymmetric

cycles

and condition

Rae # 1

if reloading

without

the external

surface

is activated

in the previous

3. The relation components between

THE

PLASTIC

POTENTIAL incremental displacement vector and the

the components

of the plastic

of the stress vector

is given by ~P=j(-_-

aff, 7

ag

,P=ia9

a71
the relation between the stress

(14
and i is the plastic ratio 77

where UP and tip are the plastic multiplier. and dilatancy The plastic potential

rate of the normal g is obtained respectively, as

and shear displacements

through

d which are defined,

Figure

5 shows the stress-dilatancy

relation

for an interface

test.

As one can observe,

the flow in

rule can be considered as bilinear but is not unique Figure 5 can be described for hardening and softening

as for triaxial tests. Flow rule shown conditions, respectively, as

(16)

0.8

t5 ;s :
0
G

0.6

0.4

0.2 0 *
0 *0
<A

*0

-0.3

-0.2

-0.1

0.0

0.1

dilatancy d=ip/);p
Figure 5. Flow rule of an interface test (data from Mortara [4])

Equations different than

(16) differ from the Cam Clay flow rule only by the choice of the slopes al and a2 -1. Flow rules (16) converge vp = at the peak of the direct
mc =

shear test

(w, = 1)

aldmax+

4max +

rlc,

(17)
is to

where rip and Q are the stress ratios for peak and ultimate conditions, respectively, and d,,, the maximum value of dilatancy. The parameter p is the ratio between the stress ratios relative

628

M. BOULON

et al.

_ _ __ __ __ __ __ _

sof&ning

normal stress 0 Figure 6. Plastic potential for failure conditions.

K = 1000 kPa/mm 500 400 3 B ; f kr E


100
A 1

expi

r o
I. A

GO= 100 kPa onno= 200 kPa w= 300 kPa model

*AAAAAAAAAAAAAAAAA

300 200

6 (mm)

10

shear displacement

Figure 7. Comparison between monotonic experimental data and model predico and rough aluminium plate, K = tion for CNS tests. Toyoura sand (DR = 859') 1000kPa/mm (data from Mortara [4]).

condition d = 0 in hardening the flow rule is

and softening

phases.

The plastic

potential

obtained

by integrating

pdP-

Aun

[l+a($l+a) =o, a]

(18)

where a and b are the generic slope and intercept of the flow rule in the v--d diagram and uC is the value of the normal stress for d = 0. Plastic potential (18) is formally the same of the Nova and Wood model and is shown in Figure 6 for failure conditions. When subjected to cyclic loading, the interface exhibits a progressive densification that, in constant normal stiffness tests, determines For increasing numbers of cycles, the cyclic the cyclic degradation of the inclusion resistance. flow rule undergoes a progressive comparison to the static one variation that can be schematized as a parallel translation in

q=ad+b+Ab=ad+b+Xb*

UP

so

(19)

where Xb is a parameter of the model. The direction of the incremental plastic displacement still remains to be defined when an inversion of load occurs. The hypothesis made is similar to that made by Pastor et al. [6] and consists of changing the sign of the deviatoric component of

Strain-Hardening

Elastoplastic

Model

629

-80 -, 0 2 4 6 8 0 2 4 6 8 shear displacement (mm) aluminium (b) Toyoura sand (DR plate K = 500 kPa/mm. 160 / 4 120 = 85%)-rough aluminium

shear displacement (mm) (a) Toyoura sand (DR plate, K = 500 kPa/mm. 160 120 = 85%)-rough

-80 0

I I x I I
40 80 120 160 normal stress (kPa)

I
200

-80 0

I 40

I 80

I 120

I 160

-I 200

normal stress (kPa)

Cc)
Figure 8. Comparison from Vita [7]).

(4
between cyclic experimental tests and model predictions (data

the gradient plastic

of the plastic

potential.

Under

general d

conditions,

therefore,

the gradient

v of the

potential

is given by v= [ afo ( z >

1
. normal load stress stiffnesses tests. and shear elastic

(201

4. MODEL
The parameters elastic data behaviour and model to be K; of the model is assumed predictions are derived as linear,

PREDICTIONS
only from constant The are identified

and the normal

= lO Pa/m

and Kz = O.lKz. for three static

Figure 7 shows the comparison between experimental CNS tests K = 1000 kPa/mm) with different initial

normal stresses g,-,c. Figure 8 shows instead the comparison between experimental data and model predictions for two cyclic CNS tests. Stress paths in Figure 8 evidences the importance of shear stress degradation for interface behaviour during cyclic loading (first part of the test) a remaining capacity of dilatancy resulting from monotonic shearing (end of the test). and

5. CONCLUSIONS
This paper has dealt with the formulation of a sand-structure interface model. The model is based on isotropic and kinematic hardening and utilizes a flow rule different according to whether hardening or softening conditions are involved. The model is able to predict interface behaviour

630

M. BO~JLONet al.

under both static and cyclic conditions, and comparisons with experimental tests show a good accuracy level.

REFERENCES
1. E. Wernick, Skin friction of cylindrical anchors in non-cohesive soils, In Symposzum on Soil Reinforcing and Stab&sing Techniques in Engineering Practice, Sydney, Australia, October 16-19, 1978, pp. 201-219. 2. M. Boulon and R. Nova, Modelling of soil-structure interface behaviour, a comparison between elastoplastic and rate type laws, Computers and Geotechnics 9 (l/2), 21-46, (1990). 3. R. Nova and D.M. Wood, A constitutive model for sand in triaxial compression, International Journal for Numerical and Analytical Methods in Geomechanics 3 (3), 255-278, (1979). interface behaviour under monotonic and cyclic 4. G. Mortara, An elastoplastic model for sand-structures loading, Ph.D. Thesis, Dipartimento di Ingegneria Strutturale e Geotecnica, Politecnico di Torino, (2001). 5. Y.F. Dafalias and E.P. Popov, Plastic internal variables formalism of cyclic plasticity, ASME, Journal of Applied Mechanics 43 (4), 645-651, (1976). 6. M. Pastor, O.C. Zienkiewicx and K.H. Leung, Simple model for transient soil loading in earthquake analysis. II. Non-associative model for sands, InternationaE Journal for Numerical and Analytical Methods in Geomechanics 9 (5), 477-498, (1985). 7. G.P. Vita, Comportamento delle interfacce tra terreni sabbiosi ed inclusioni solide in campo dinamico, Degree Thesis, Dipartimento di Meccanica e Materiali, Universit& di Reggio Calabria, (1998).

You might also like