You are on page 1of 9

PLASTICITY MODELS AND VARIATIONS OF G/Gmax WITH SHEAR

STRAIN FOR SAND UNDER MONOTONIC LOADING


Kallol Sett
ABSTRACT
In this report the effectiveness of plasticity based models in predicting the reduction of shear modulus of sand was compared. Predictions of Drucker-Prager perfectly plastic model, Drucker-Prager
isotropic hardening model and cam clay model were compared with Hardin Drnevich equation and
experimental results for sand. Though there are lot of scatter in experimental data cam clay model
predictions matched both the experimental data and Hardins equation closely.

Keywords: G/Gmax , plasticity models, hardening, Drucker-Prager, cam clay, K 0 triaxial test
INTRODUCTION
Much progress has been made in recent years, motivated by the development of computational power and numerical methods for boundary value problem, in the development of
analytical models to predict constitutive behavior of soil and these plasticity based theories
provide a consistent framework in which the bahavior of soil can be accurately understood and
predicted.
For the solution of many soil dynamics problems, reliable and accurate shear stress-strain
relations are necessary. In the past through comprehensive laboratory investigation it was found
that the elastic shear moduli of soil, Gmax , depend largely on the effective principal stresses. It
was also found that the shear stress-strain relationship of soil is highly nonlinear and hence the
shear moduli of soil, G, depend largely on the magnitude of strain. The other important factors affecting the shear moduli were found to be void ratio (e), overconsolidation ratio (OCR),
and the effective stress strength parameters (c and ). Based on these observation, Hardin et
al. (Hardin and Black 1968) proposed an emperical equation relating elastic shear modulus
(Gmax ) with mean stress, (m ), void ratio (e), and overconsolidation ratio (OCR) and considering hyperbolic stress-strain relationship Hardin and Drnevich (Hardin and Drnevich 1970)
related plastic shear modulus (G) with shear strain () with the concept of reference strain ( r ).
Seed and Idriss (Seed and Idriss 1970) coupled the Gmax equation with the G relationship
and proposed a parameter K2 , a function of void ratio (e) and strain amplitude () and presented design curves for plastic shear modulus (G). Later Roesler (Roesler 1979) pointed out
that the elastic shear modulus (Gmax ) of sand is dependent only on the two principal stresses
in a shear plane and independent of the normal stress perpendicular to the plane. Based on
Roesler (1979) study, Hardin (Hardin 1980) revised the early model and the nal form of the
model was proposed by Hardin and Blandford (Hardin and Blandford 1989).
In this paper plasticity models with and without hardening have been compared with the hyperbolic stress-strain relationship. K0 triaxial tests were simulated numerically using different

plasticity models and the inuences of void ratio (e), strength parameters (c and ) and overconsolidation ratio (OCR) were investigated and compared with those obtained from Harden
and Drnevich (1972) expressions.
THEORY
General Formulation
The total strain increment is assumed to be the sum of the elastic strain increment and
plastic strain increment (Chen and Han 1988),
dij = deij + dpij

(1)

The plastic strain increment is obtained from the associated ow rule,


dpij = d
where
d

f
ij

(2)

= 0 when f < 0 or f = 0 but df < 0;


> 0 when f = 0 and df = 0.

Hence, from Hookes law (dij = Cijkl ekl ),


dij

= Cijkl (dkl dpkl )

(3)

f
= Cijkl dkl dCijkl
kl

(4)

Now, the tensor of elastic moduli, Cijkl , in Eq. (4) can be expressed in terms of shear modulus,
G, and bulk modulus, K, as,

2
Cijkl = K G ij kl + G (ik jl + il jk )
(5)
3
and hence Eq. (4) reduces to,
dij = 2Gdij + Kdkk ij d

2
K G
3

f
f
mn ij + 2G
mn
ij

(6)

Elastic-Perfectly Plastic Model


The general expresiion of a yield surface for perfectly plastic material has the form
f (ij ) = 0

(7)

and hence, the consistency condition is assumed as


df =

f
dij = 0
ij

(8)

which ensures that the stress state (ij + dij ) existing after the incremental change dij has
taken place still satises the yield criterion f , given by,
f (ij + dij ) = f (ij ) + df = f (ij )
2

(9)

Drucker-Prager Model
Now, for a Drucker-Prager material model, yield condition,f, is given by
p
f = J2 + I1 k = 0
where,

2 sin
3(3sin )
tan 2
(9+12 tan )

6c cos
3(3sin )
3c

(9+12 tan2 )

and

(10)

for triaxial condition;


for plane strain condition.

for triaxial condition;


for plane strain condition.

c and being the cohesion and friction angle respectively. From Eqs. (6), (8), and (10), after
some tensor algebra, it can be shown that the most general form of Drucker-Prager elasticperfectly plastic constitutive relationship as,

G
Sij + 3Kij G
2
J2
Smn + 3Kmn dmn
dij = 2Gim jn + K G ij mn
3
G + 9K2
J2
(11)

Elastic-Plastic Model with Hardening Parameters


The general expression of a yield surface for a work-hardening material has the form,
f (ij , pij , k) = 0

(12)

and hence the consistency condition is assumed as


df =

f
f
f
dij + p dpij +
dk = 0
ij
k
ij

(13)

which ensures that in a plastic loading process, the subsequent stress and defromation states
remain on the subsequent yield surface.
Drucker-Prager Isotropic Hardening Model
The general form of the yield function for isotropic-hardening Drucker-Prager material can
be expressed as,

(14)
f (ij , p ) = (p )I1 + J 2 k(p ) = 0
Please note that in Eq. (10), the yield function for a perfectly plastic material has been presented
with =constant and k =constant.
Herein, for simplicity, we shall assume that the slope of the

loading surface in the I1 J 2 space is a constant, (p ) = , so that the hardening behavior


of the material can be uniquely determined by a single uniaxial stress-strain relation through
the hardening parameter k(p ),

f (ij , p ) = I1 + J 2 k(p ) = 0
(15)

With the Appendix giving the details of derivation, the general form of Drucker-Prager isotropichardening elastic-plastic constitutive relationship can be written as,

Sij + 3Kij

G
2
J2
Smn + 3Kmn

dij =
2G

+
K

ij
mn
im
jn

dmn

2
3
J2
1
2

G + 9K + + 3 Hp

(16)
which is exactly the same equation obtained for Drucker-Prager elastic-perfectly plastic material (Eq. (11)) except the hardening term in the plastic tangent stiffness tensor.
Cam Clay Model
For a cam clay material model, the yield function, f, can be written in terms of stress
invariants p and q as
f = M 2 p2 M 2 p0 p + q 2 = 0
(17)
where
p=

I1
3

3
q = oct
2
M=

6 sin
3 sin

and p0 is the value of p at the intersection of the yield cap with the p-axis. Here p 0 is treated as
a variable which has a unique value for any yield surface and is the hardening parameter.
In terms of stress invariants I1 and J2 , Eq. (17) can be written as
f = M 2 I12 M 2 I01 I1 + 27J2 = 0

(18)

where I01 is analogous to p0 (hardening parameter) and is assumed to depend on the plastic
volumertic strain. Now, at yield as f = f [J2 , I1 , I01 (pv )], we have the consistency condition
as,
f
f
f
df =
(19)
dJ2 +
dI1 + p dpv = 0
J2
I1
v
Now, if we dene
F
Aij =
ij

J2
I1
f
f
+
=
J2 ij
I1 ij
f
f
=
Sij +
ij
J2
I1

= 2M 2 I1 M 2 I01 ij + 27Sij

(20)

using Eqs. (2) (4) (18), it can be shown that the most general form of the cam clay constitutive
relationship can be written as,
"
#
Cijkl Akl Amn Cmnrs
dij = Cijrs
drs
(21)
2 I (1+e )
01
0
Amn Cmnrs Ars (2M I12 M 2 I01 ) M I1

G/GM AX VERSUS SHEAR STRAIN RELATIONSHIP


It can be observed from any three dimensional elastic-plastic constitutive relationship that
ep
for an initially istropic material the elatic-plastic stiffness (Cijmn
) always composed of two

terms: the elastic stiffness tensor, Cijmn (=2Gim jn + K 23 G ij mn ) and a subractive


p
term, Cijmn
, called the plastic tangent stiffness, which represents the degradation of the stiffness of the material due to plastic ow. The elastic stiffness tensor is dened in terms of two
elastic constants, G and K (or E and ), which can be obtained from triaxial test. The plastic
tangent stiffness tensor is dened by one or two more constants other than G and K, depending upon perfectly plastic and work-hardening material model, with the exception of cam clay,
which requires more parameters. Some plasticity parameters can be obtained from strength
parameters of soil, c and , (e.g. and k for Drucker-Prager perfectly plastic model, M for
cam clay model) and others require specialized testing (e.g. the plasticity modulus, H p , for
Drucker-Prager isotropic hardening model, and for cam clay model).
These three dimensional plasticity models can be simplied to any particular direction of
interest to model the degradation of modulus with strain. The degradation of shear modulus
with shear starin can be computed by substituting i = m = 1 and j = n= 2 in the general incremental elatic-plastic constitutive ralationships() e.g. in the 1-2 direction the cam clay model
reduces to
"
#
729G2
d12 = G

2J2

729G +

9K(2M I12

M 2 I01 )2 (2M I12 M 2 I01 ) M

2I

1 I01 (1+e0 )

S12 S12 d12

(22)
where G is the elastic shear modulus and the term inside the bracket is the elastic-plastic shear
modulus. Eq. (22) clearly shows that the elastic-plastic shear modulus decreases with shear
strain. Similarly elastic-plastic shear modulus can be obtained for other plasticity based models
by substituting i = m = 1 and j = n = 2 in the general forms.
The elastic shear modulus, G, was computed based on Hardin and Blandford (1989) equation as follows,
OCRa Bij
P 1n (i0 j0 )n/2
(23)
Gij =
F (e)2(1 + ) a
where, Gij is the shear modulus in the shear plane containing principal effective stresses, i0
ans j0 ; Bij is the dimensionless elastic stiffness coefcient; a is the soil constant depending
on the plasticity index of soil; Pa is the atmospheric pressure; n is the soil constant equal to 0.5
for sand; is the Poissons ratio, F (e) = 0.3 + 0.7e2 ; and e is the void ratio.
Different plasticity models were used to numerically simulate the soil behavior under K 0
triaxial condition. K0 loading condition was imposed numerically on a 8-node brick element
and the variations of G/Gmax with shear strain for different plasticity models were computed
using OpenSees and compared with the hyperbolic stress-strain relationship and reference
strain (r ) based model(), described as follows,
G=

Gmax
1 + r

r =

max
Gmax

where

(24)

FIG. 1. Variation of Shear Modulus with Shear Strain for Sands

and
max =

"

1 + K0 0
v sin 0 + c0 cos 0
2

1 K0 0
v
2

2 #1/2

Fig. 1 shows the variation of G/Gmax with shear strain as predicted by Hardin-Drenvich
equation (Hardin and Drnevich 1972) and by different plasticity models with previously published experimental observation (Seed and Idriss 1970). It can be observed that though all the
plasticity based models predicted stiffness degradtion, the cam clay model did it more closely
to the experimental observation. Also all the plasticity based models sensed the small strains
as measured and reported by Vucetic et al. (Lanzo et al. 1997)
CONCLUSIONS AND FUTURE PLAN
Based on the analytical and numerical study the following conclusions can be advanced:
1. Though all the plasticity-based models predicted the reduction of shear modulus with
shear strain the cam clay model predicted the behavior closely with the experimental observation.
2. The cam clay model did not predict the reduction of shear modulus at small strain
level well. This might be because the yield surface assumed for the cam clay model was for
isotropic consolidation and the experimental data were for K 0 consolidation. Yield surface for
K0 consolidation might work well as it will early yielding and will be studied in the future.
REFERENCES
6

Chen, W. and Han, D. (1988). Plasticity for Structural engineers. Springer-Verlag.


Hardin, B. O. (1980). Discussion of anisotropic shear modulus due to stress anisotropy. Journal of Geotechnical Engineering Division, ASCE, 106(8), 956958.
Hardin, B. O. and Black, W. L. (1968). Vibration modulus of normally consolidated clay.
Journal of Soil Mechanics and Foundations Division, ASCE, 94(2), 353379.
Hardin, B. O. and Blandford, G. E. (1989). Elasticity of particulate materials. Journal of
Geotechnical Engineering, 115(6), 788805.
Hardin, B. O. and Drnevich, V. P. (1970). Shear modulus and damping in soils - ii: Design
equations and curves. Technical Report UKY 27-70-CE3, Soil Mechanics Series No. 2, University of Kentucky.
Hardin, B. O. and Drnevich, V. P. (1972). Shear modulus and damping in soils; design equations and curves. Journal of the Soil Mechanics and Foundation division, ASCE, 98(SM7),
667692.
Lanzo, G., Vucetic, M., and Doroudian, M. (1997). Reduction of shear modulus at small
strains in simple shear. Journal of Geotechnical and Geoenvironmental Engineering,
123(11), 10351042.
Roesler, S. K. (1979). Anisotropic shear modulus due to stress anisotropy. Journal of the
Geotechnical Engineering Division, ASCE, 105(GT7), 871880.
Seed, H. B. and Idriss, I. M. (1970). Soil moduli and damping factors for dynamic factors for
dynamic response analysis. Technical Report EERC Report, 70-10, University of California,
Berkeley.
APPENDIX
The increment of the isotropic hardening parameter, dk in Eq. (15) can be obtained as
dk =

dk
dp
dp

(25)

Now, the isotropic hardening parameter, k, can be expressed in the simple form,
q
dp = C dpij dpij

(26)

and hence,

dk
C
dk =
dp

f f
d
ij ij

(27)

Now, dpij can be obtained from the associated ow rule (Eq. (2)) and d ij can be obtained
from the Hookes law (Eq. (4)) and hence, the consistency condition (Eq. (13)) reduces to the
form
f
df =
Cijkl dkl hd = 0
(28)
ij
where,
f
f dk
f
f f
Cijkl

h=

C
ij
kl pij ij
k dpij

f f
ij ij

(29)

From Eq. (28), the scalar function d, can be solved as,


d =

1 f
1
Cijkl dkl = Hkl dkl
h ij
h
7

(30)

where, the second-order tensor Hkl associated with the yield function, f , is dened as,
Hkl =

f
Cijkl
ij

(31)

Now, for Drucker-Prager yield function for isotropic hardening (Eq. (15)),
f
1
= ij + Sij
ij
2 J2

(32)

Hence, using the elastic stiffness tensor in terms of G and K (Eq. (5)) we obtain (since the
f
yield function is not expressed as a function of pij explicitly, we have
p = 0),
ij

Hkl = Cijkl

f
G
= 3Kkl + Skl
ij
J2

(33)

dk
and
Now, to determine the scalar function h dened in Eq. (29), we also need to obtain d
p
the parameter C. We can expressk as follows
(since
the
effective
stress,

,
for
Drucker-Prager
e

3(I1 + J2 )

),
material can be written as e =
1+ 3

p
1 + 3

k = I1 + J2 =
e
(34)
3

from which we obtain,

dk
1 + 3 de
1 + 3

=
=
Hp
dp
3 dp
3
where, Hp is determined from a uniaxial stress-strain curve, d = Hp d. Now,
q
kl dpkl
dWp
=
= C dpij dpij
dp =
e
e

(35)

(36)

From Eq. (36), we have the parameter C as,

C=q

f
kl
kl
f f
st st e

(37)

Substituting Eqs. (5), (32), (35), and (36) into Eqs. (??) we can express, the scalar function, h,
as,

I1 + J2
h = G + 9K2 +
(1 + 3)2 Hp
3k

1 2
2
Hp
(38)
= G + 9K + +
3

Hence, the plastic strain increment , dpij , is given by,


dpij

f
ij
f p
1 f
Cmnst
d
h mn
ij st
1
f p
Hst
d
h
ij st

= d
=
=

(39)

and from Hookes law (Eq. (4)), the stress increment, dij , can be determined as

f
1 f
Cmnst
dst
dij = Cijkl dkl
h mn
kl

f
1 f
Cmnst
dst
= Cijkl sk tl
h mn
kl

f
1 f
dst
Cmnst Cijkl
=
Cijst
h mn
kl

1
=
Cijst Hst Hij dst
h

(40)

Hence the most general form of Drucker-Prager isotropic-hardening elastic-plastic constitutive relationship can be written as,

Sij + 3Kij

2
G
J2
dmn

2G

+
K

S
+
3K
dij =
G

im
jn
mn
mn
ij
mn

2
3
J
2
1
G + 9K2 + + 3 Hp
(41)

You might also like