A finite element analysis of the thermo-hydro-mechanically coupled

problem of cohesive deposit using a thermo-elasto-viscoplastic model

F. Oka, S. Kimoto, Y.-S. Kim & N. Takada

Department of Civil & Earth Resources Engineering, Kyoto University, Japan

Y. Higo
Geo-Research Institute, Japan

ABSTRACT: It is well known that viscosity of water-saturated clay is affected by temperature variation as
well as strain rate. In addition, the conservation of energy during deformation should be considered since the
plastic stress power changes the temperature of soil. In the present study, a thermo-hydromechanically coupled
finite element method using a thermo-elasto-viscoplastic constitutive model for clay is newly developed. As
examples, the method was applied to a numerical simulation of thermal consolidation phenomena mainly due
to heat-generated pore water pressure.

1 INTRODUCTION newly introducing the volume change of soil particles

and pore fluid into the previous analysis method.
It is well known that viscosity of water-saturated clay
is affected by temperature variation as well as strain
rate. Boudali et al. (1994) demonstrated the tempera- 2 THERMO-ELASTO-VISCOPLASTIC
ture dependency of clay through experimental results MODEL FOR CLAY
of natural clays. In addition, the plastic stress power
due to inelastic deformation of geomaterials changes Based on the experimental data (Leroueil et al. (1985),
the temperature of the materials. Vardoulakis (2002) Boudali et al. (1994)),Yashima et al. (1998) introduced
displayed that the temperature of soil increases in the the temperature dependency of clay into Adachi-Okas
shear band with very high strain rate and very large elasto-viscoplasticity theory (1982). By use of this
deformation by a dynamic thermo-poro-mechanical theory, Oka et al. (2004) developed an elasto-thermo-
analysis. viscoplastic model for both normally consolidated
In order to address the temperature dependency of clay and overconsolidated clay considering structural
clay and to deal with the temperature change induced changes (Kimoto et al. 2004).
by plastic stress power, Oka et al. (2004) developed In the present study, we propose a thermo-elasto-
a thermo-hydro-mechanically coupled finite element viscoplastic model by introducing the temperature-
analysis method of clay with an elasto-thermo- dependent characteristics of elastic stretching tensor
viscoplastic model through formulating the energy into the model by Oka et al. (2004), in addition to that
balance equation. The elasto-thermo-viscoplastic of a viscoplastic parameter.
model is based on an Adachi-Okas elasto-visco-
plasticity theory (1982) considering the temperature
dependency of viscoplasticity proposed by Yashima
et al. (1998). Then, Oka et al. (2004) numeri- 2.1 Elasto-viscoplastic model for clay
cally simulated thermal consolidation by use of considering structural changes
an empirical assumption (Campanella & Mitchell The overconsolidation boundary surface controls the
1968). shape of the plastic potential function.
In the present study, we propose a thermo-hydro-
mechanically coupled finite element analysis method
of clay with a thermo-elasto-viscoplastic model by

383
Copyright 2005 Taylor & Francis Group plc, London, UK
where is the relative stress ratio defined as When viscoplastic strain does not occur, i.e., z = 0, the
value of Na is equal to 1. On the other hand, when the
value of z becomes large enough, Na approaches n,
and the softening converges due to structural changes.
Adachi and Oka (1982) assumed a Cam-clay type
of static yield function to describe the mechanical
behavior of clay at its static equilibrium state.
in which ij is Terzaghis effective stress tensor, ij is
the stress ratio tensor, Sij is the deviatoric stress tensor,
m is the mean effective stress, ij is Kroneckers delta,
and subscript (0) indicates the state at the end of con-

(s)
solidation, Mm is the value of ij ij at the maximum where the my is the static-hardening parameter and
compression and mb
is the parameter which controls fy = 0 represents the static state in which no viscoplas-
the size of the surface. tic deformation occurs.
Originally, the hardening rule for the overconsoli- Considering volumetric strain hardening, the hard-
(s)
dation boundary surface was defined with respect to ening rule my was originally given as
p
the viscoplastic volumetric strain kk as

Adopting Eq. (8) to Eq. (10) yields

where is the compression index, is the swelling

index, e is the void ratio, and mbi is the initial value

of mb , which is defined as the isotropic consolidation
yield stress. (s) (s)
In order to describe the degradation of the mate- where myi is the initial value of my . The decrease
(s)
rial caused by structural changes, strain softening in my , defined by Eq. (11), leads to the shrinking
is introduced into the hardening parameter with the of the static yield function according to the structural
viscoplastic strain as collapse.
Viscoplastic potential function fp is given as

in which mai and maf are the initial and the final

values of ma , and z is an accumulation of the second
p
invariant of viscoplastic strain rate ij . where M is constant in the NC region, while the value
in the OC region depends on the current stress
of M

and mc as

Substituting ma into Eq. (4), mb is defined by the
hardening rule with respect to the viscoplastic volu-
metric strain and by the softening rule with respect to
structural changes as

where mc denotes the mean effective stress at the inter-
section of the overconsolidation boundary surface and
the m axis as

in which is a parameter which denotes the degrada-

tion rate of ma in Eq. (5).

The ratio of maf to mai , namely, n = maf /mai ,
provides the degree for a possible collapse of the struc-
ture at the initial state. n satisfies the condition of In the case of isotropic consolidation, mc equals mb .
p
0 < n 1. Viscoplastic stretching tensor Dij , based on an
overstress type of viscoplasticity theory, is given as
In addition, the ratio of ma to mai is defined by
Eq. (5) as

384
Copyright 2005 Taylor & Francis Group plc, London, UK
 If the stress ratio during one-dimensional compres-
sion is assumed to be constant and the initial hard-
(s)
ening parameter myi corresponds to the consolidation

where 1 indicates the strainrate sensitivity. Based on yield stress p , the temperature-dependent viscoplastic
the experimental data of the strainrate constant triaxial parameter C() is rewritten from Eq. (18) as
tests, material function 1 is given as

Substituting Eq. (21) into Eq. (22) yields

in which m and C are viscoplastic parameters and

Equations (7), (8), and (11) are used.
where 0 is preconsolidation stress. From Eq. (23),
2.2 Temperature-dependent elastic stretching temperature dependency of viscoplastic parameter C
An additive decomposition of total stretching tensor is obtained as follows:
Dij into elastic stretching tensor Dije and viscoplas-
p
tic stretching tensor Dij is adopted. In addition, we
assumed that expansion of soil particles due to temper-
ature variation affected the volumetric part of elastic in which coefficient of dilatancy M is a function of
stretching tensor Dije , namely, stress ratio in the OC region (see Eq. (13)). However,
thermo-viscoplastic parameter is independently
determined as material constant.

in which is thermal coefficient of expansion for

soil particle, is temperature, the superimposed dot 3 THERMO-HYDRO-MECHANICALLY
denotes the time differentiation, and compression is COUPLED FINITE ELEMENT METHOD
positive.
Elastic stretching tensor due to stress change Dije is When we deal with cohesive soils, it is important
given by a generalized Hooke type of law, i,e, to investigate the effects of temperature on deforma-
tion characteristics of soils, since it is known that
both viscosity and permeability of clay are depen-
dent on temperature. In order to consider temperature
changes of clay due to plastic stress power, Oka
et al. (2004) developed a thermo-hydro-mechanically
where S ij is the deviatoric stress rate tensor, m is
coupled finite element method by formulating the
the mean effective stress rate, G is the elastic shear
conservation of energy in addition to the previous soil-
modulus, e is the void ratio.
water coupled finite element method (e.g., Oka et al.
2002, Higo 2003, and Higo et al. 2004a, 2004b).
2.3 Temperature-dependent viscoplastic In the present study, we have newly introduced
parameter the compressiveness of the pore fluid into continuity
Based on the experimental data by Boudali et al. equation as extension of Oka et al. (2004). Based on
(1994), Yashima et al. (1998) showed the following the finite deformation theory, an updated Lagrangian
relation between consolidation yield stress p and method with the objective Jaumann rate of Cauchy
temperature : stress is used for the weak form of the rate type of equi-
librium equations for the soil-water whole mixture.
As for the element type, an eight-node quadrilat-
eral isoparametric element with a reduced Gaussian
four points integration is used for the displacement,
while the pore water pressure and the temperature are

in which pr is the value of p at the referential defined by a four-node quadrilateral isoparametric ele-
temperature r and is the gradient of the line log ment. In the following, an introduction of governing
p log . equations of the present analysis method is given.

385
Copyright 2005 Taylor & Francis Group plc, London, UK
3.1 Equilibrium equation for the whole mixture 3.3 Conservation of energy
The rate type of equilibrium equations are given as The first law of thermodynamics, namely, conserva-
follows: tion of energy, is given as

where S t is the total nominal stress rate tensor (e.g., where e is the rate of internal energy, Tij is the effective
Yatomi et al. 1989). Cauchy stress tensor, and hi is the heat flux vector. As
for the formulation of hi , we used the Fouriers law,
namely,
3.2 Continuity equation
For describing the motion of pore water, a Biots type of
two-phase mixture theory is used in the analysis with a in which k is the coefficient of heat conductivity, is
(velocity) uw (pore water pressure) formulation. In temperature.
the formulation, we consider compressiveness of the The relation between the elastic stretching tensor Dije
fluid i.e., and the rate of internal energy e is obtained as follows:

where is the density of soil and c is the specific heat

where superscripted s and f indicate solid phase and capacity.
fluid phase, respectively, k is the coefficient of perme- By taking Eq. (19), Eq. (32), and Eq. (34), con-
ability, w is the unit weight of the pore water, Dijs is servation of energy is written by use of viscoplastic
the stretching tensor (Dijs = Dij ), n is porosity, and f vp
stretching tensor Dij as
is fluid density.
The compressiveness of the fluid is defined as
follows:

4 NUMERICAL ANALYSIS OF THERMAL

CONSOLIDATION
in which K f is the bulk mudulus for the fluid.
The assumption that the time derivative of the den-
Using the thermo-hydro-mechanically coupled finite
sity of the fluid depends only on temperature changes
element method with the thermo-elastoviscoplastic
gives the following relation, i.e.,
model, we conducted numerical simulation of thermal
consolidation. Heating watersaturated soils induces
volume expansion of soil particles and pore water.
Unless the heated soils are under fully drained con-
ditions, excess pore water pressure generates. Dis-
where D/Dt indicates the material time differentiation, sipation of the heatgenerated pore pressure induces
V f is volume of the fluid, w is a thermal coefficient consolidation, which is called thermal consolidation
of the fluid, and is temperature. (e.g., Campanella & Mitchell 1968, and Baldi et al.
Using the relation V f /V f = Dii (compression is
f
1988, Delage et al. 2000).
possitive) and Eq. (28), Eq. (27) yields

4.1 Material parameters and models

Material parameters used in the analysis are listed in
Table 1. Note that the parameters related to mechanical
We redefine a thermal coefficient of expansion characteristics are determined by triaxial compression
for water w = n w . Finally, the continuity equation test and consolidation test for Fukakusa clay, while
considering temperature change is given as follows. those related to thermal characteristics are general val-
ues of clay. As for the thermal coefficient of expansion
for water w , we employed an empirical formula by
Baldi et al. (1988).

386
Copyright 2005 Taylor & Francis Group plc, London, UK
Table 1. Material parameters.

Compression index 0.191

Swelling index 0.043
Initial void ratio e0 1.10

Initial mean effective stress me 200 (kPa)

Compression yield stress mbi 200 (kPa)
Coefficient of earth pressure K0 1.0
at rest
Viscoplastic parameter m 24.3
Viscoplastic parameter C(r ) 5.8 1010 (1/s)
Stress ratio at maximum Mm 1.14
compression
Initial elastic shear modulus G0 17700 (kPa)
Coefficient of permeability k 1.63 109 (m/s)
Referential temperature r 15.0 ( C)
Soil density 1.96 103 (kg/m3 ) Figure 2. Settlement at node 321.
Specific heat capacity c 938 (J/kg C)
Coefficient of heat k 1.16 (W/m C)
conductivity Figure 1 depicts the boundary conditions and the
Thermal coefficient of volume 0.35 104 (1/ C) size of the specimen. Isothermal boundary is imposed
expansion for soil particle at the left side of the specimen as an installation of
Thermo-viscoplastic 0.15 heat source, on which temperature increases gradually
parameter from its initial value of 20 C at t = 0 (hour) to 60 C

Structural parameter maf 170 (kPa) at t = 4 (hour). Upper boundary is also isothermal
Structural parameter 10 boundary of the atmospheric temperature 20 C.

4.2 Simulation results of thermal consolidation

Settlement at node 321 is shown in Figure 2.
We can observe around 1.4 (mm) of settlement at
t = 148(hour) and it is seen that deformation does not
converge. The distribution of temperature, the distri-
bution of pore water pressure, and the distribution of
mean effective stress are demonstrated in Figures 3, 4,
and 5, respectively. It can be observed from these fig-
ures that thermal consolidation due to dissipation of
pore water pressure induced by temperature change is
reproduced. In the beginning of heating, temperature
gradually increase near the heat source. Correspond-
ing the temperature elevation, pore water pressure
generates near the heat source. At t = 80 (hour), the
distribution of temperature becomes stationary, while
that of excess pore pressure is different, namely, gen-
erated pore pressure remains at the bottom of the
specimen. It is seen that mean effective stress decrease
Figure 1. Boundary conditions and the size of the specimen. with increased temperature due to heat-generated pore
water pressure.
In the analysis, we consider temperature depen-
dency of coefficient of permeability k, namely,
5 CONCLUSION

In the present study, we proposed a thermo-

hydro-mechanically coupled finite element method
where K is the intrinsic permeability, w is the unit with a thermo-elasto-viscoplastic constitutive model
weight of water, in which temperature dependency of for clay, considering volume expansion of soil particle
w is assumed to be negligible. As for the temperature and pore fluid due to temperature change. Using the
dependency of viscosity of water (), an empirical newly developed analysis method, a numerical simu-
formula by Baldi et al. (1988) is used. lation of thermal consolidation was conducted. It was

387
Copyright 2005 Taylor & Francis Group plc, London, UK
Figure 3. Distribution of temperature for the simulation of thermal consolidation ( C).

Figure 4. Distribution of pore water pressure for the simulation of thermal consolidation (kPa).

Figure 5. Distribution of mean effective stress for the simulation of thermal consolidation (kPa).

confirmed that the analysis method can well reproduce strain localization analysis of fluid-saturated gradient-
the thermal consolidation phenomenon. dependent viscoplastic geomaterial, Int. J. Numerical and
Analytical Methods in Geomechanics, applied.
Kimoto, S., Oka, F., & Higo, Y. 2004, Strain localization
analysis of elasto-viscoplastic soil considering structural
REFERENCES degradation, Computer Methods in Applied Mechanics
and Engineering, 193, 28452866.
Adachi, T. & Oka, F. 1982, Constitutive equations for nor- Leroueil, S., Kabbaj, M., Tavenas, F. 1985, & Bouchard, R.,
mally consolidated clay based on elasto-viscoplasticity, Stress-strain-strain rate relation for the compressibility of
Soils and Foundations, 22, 4, 5770. sensitive natural clays, Gotechnique, 35, 2, 159180.
Baldi, G., Hueckel, T., & Pellegrini, R. 1988, Thermal Oka, F., Higo, Y., & Kimoto, S. 2002, Effect of dilatancy on
volume change of the mineral-water system in low- the strain localization of water-saturated elastoviscoplas-
porosity clay soils, Canadian Geotechnical Journal, 25, tic soil, Int. J. Solids and Structures, 39, 36253647.
807825. Oka, F., Higo, Y., Kim, Y.-S., Imura, Y., & Kimoto, S. 2004,
Boudali, M., Leroueil, S., & Srinvasa Murthy, B.R. 1994, Thermo-hydro-mechanically coupled finite element anal-
Viscous behaviour of natural clays, Proc. 13th Int. Conf. ysis of cohesive soil using an elasto-viscoplastic model,
on SMFE, New Delhi, India, 411416. Computational Mechanics (Abstracts)WCCM VI, Sept.
Campanella, R.G. & Mitchell, J.K. 1968, Influence of tem- 510, 2004, Beijing, China, 266.
perature variations on soil behavior, ASCE J. of SMFE, 94, Yashima, A., Leroueil, S., Oka, F., & Guntoro. I. 1998,
3, 709734. Modelling temperature and strain rate dependent behav-
Delage, P., Sultan, N., & Cui, Y.J. 2000, On the thermal con- ior of clays: one dimensional consolidation, Soils and
solidation of Boom clay, Canadian Geotechnical Journal, Foundations, 38, 2, 6373.
37, 343354. Yatomi, C., Yashima, A. 1989, Iizuka, A., & Sano, I., General
Higo, Y. 2003, Instability and strain localization analysis theory of shear bands formation by a non-coaxial Cam-
of water-saturated clay by elasto-viscoplastic constitutive clay model, Soils and Foundations, 29, 3, 4153.
models, doctoral thesis, Kyoto University, Japan. Vardoulakis, I. 2002, Dynamic thermo-poro-mechanical
Higo, Y., Oka, F., Jiang, M.J., & Fujita, Y. 2004a, Effect analysis of catastrophic landslides, Gotechnique, 52, 3,
of transport of pore water and material heterogeneity on 157171.

388
Copyright 2005 Taylor & Francis Group plc, London, UK