See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/266475982

A theory of plasticity with generalized hardening for natural

geomaterials under mechanical and environmental loading:
Constitutive modeling and numerical impleme....

Conference Paper · September 2014

DOI: 10.13140/2.1.3724.6726

CITATION READS

1 216

2 authors:

Claudio Tamagnini Matteo Oryem Ciantia

Università degli Studi di Perugia University of Dundee
102 PUBLICATIONS   1,192 CITATIONS    47 PUBLICATIONS   212 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Macroelement models for shallow and deep foundations View project

Theory of plasticity with generalized hardening for mechanical and "environmental" loading conditions View
project

All content following this page was uploaded by Matteo Oryem Ciantia on 06 October 2014.

The user has requested enhancement of the downloaded file.

 Computer Methods and Recent Advances in Geomechanics – Oka, Murakami, Uzuoka & Kimoto (Eds.)
© 2015 Taylor & Francis Group, London, ISBN 978-1-138-00148-0

A theory of plasticity with generalized hardening for natural geomaterials

under mechanical and environmental loading: Constitutive modeling and
numerical implementation

C. Tamagnini
Department of Civil and Environmental Engineering, University of Perugia, Italy

M.O. Ciantia
Department of Geotechnical Engineering and Geosciences, Universidad Politècnica de Cataluna (UPC),
Barcelona, Spain

ABSTRACT: An extended theory of plasticity with generalized hardening is proposed to describe the response
of geomaterials under both mechanical and environmental processes. It is assumed that the coupling between
mechanical and environmental processes takes place at two levels: a) as an additional direct contribution to
the constitutive stress changes, taking place even for purely elastic processes; and, b) as a consequence of the
evolution of the internal state variables of the material with the changes in the environmental process variables.
This last effect is incorporated through a set of generalized hardening rules. The general formulation is then
specialized to the case of calcarenite rocks under unsaturated conditions, which can experience significant
degradation processes due to plastic deformations, changes in degree of saturation (short–term debonding) and
chemical dissolution of the bond material and the solid grains (long–term debonding). The resulting model,
implemented in a FE code through the GBE algorithm extended to incorporate the full formalism of plasticity
with generalized hardening, demonstrate the practical relevance of environmental degradation effects in practical
applications, over periods of time comparable with the life cycle of most geotechnical structures.

1 INTRODUCTION (or suction), temperature, concentrations of chemical

As pointed out by Gens (2010) in his Rankine Lecture,
a number of recent developments in geomechanics
have been focused on the modification of classical
approaches to describe the effects of the interactions
between the soil (or rock) mass and the environment –
intended in a very broad sense to include all the pos-
sible coupled phenomena involving changes in pore
fluid pressures and degree of saturation (hydraulic
coupling), temperature (thermal coupling) and chem-
ical composition of the solid and liquid phases (chemi-
cal interaction) – and on the formulation of appropriate
constitutive theories capable of quantitatively repro-
ducing the observed behavior under both mechanical
and environmental loading conditions.
In particular, a number of modifications to the clas-
sical theory of hardening elastoplasticity have been
proposed in the context of geomechanics to describe a
number of practically relevant aspects of the mechan-
ical behavior of geomaterials. A common, distinctive
feature of those constitutive theories is that the size and
shape of the yield locus, as well as its evolution with the
loading process are assumed to depend to accumulated
plastic strains and on the evolution of other environ-
mental process variables, such as degree of saturation
species in the liquid phase. In particular, we recall:
i) the elastoplastic models for unsaturated soil (for-
mulated in terms of average skeleton stress and
suction) in which an explicit dependence of the
size of the yield surface on the degree of saturation
or suction (desaturation or suction hardening) is
assumed to reproduce the phenomenon of col-
lapse upon wetting in unsaturated soils (see, e.g.,
Jommi 2000,Tamagnini 2004, DellaVecchia et al.
2013);
ii) the thermoplastic models for rocks (Nova 1986)
and soils (Gens et al. 2002, Cekerevac and
Laloui 2004, Nova et al. 2004, Gens et al. 2005,
Uchaipichat and Khalili 2009), in which the
internal variables depend on the temperature T
(thermal softening);
iii) the extension of classical elastoplasticity advo-
cated by Nova (2000), Tamagnini et al. (2002),
Nova et al. (2003) to model chemical degradation
effects in cemented granular soils or weak rocks,
in which some bonding–related internal variables
are subject to both mechanical and chemical
degradation, or the chemoplastic models pro-
posed by Hueckel (1997), Loret et al. (2002) to

81
 describe chemo–mechanical coupling effects in
fine–grained soils.
It is worth noting that, in most cases, the effect of
the changes in environmental variables on the elas-
tic domain of the material is introduced by extending
the hardening laws of the material to allow the inter-
nal variables to change with the degree of saturation
(or suction), temperature, or concentration of chem-
ical species in the pore fluid, other than with the
accumulated plastic strains.
The primary goal of this paper is to show that a gen-
eral theory of plasticity with generalized hardening –
which includes as a special case most of the afore-
mentioned works – can be constructed based on the
hypothesis that the coupling between mechanical and
environmental processes takes place at two levels: a)
as an additional direct contribution to the constitutive
stress changes (e.g., suction–induced stress relax-
ation; thermal expansion/contraction effects, chemical
damage effects) taking place even for purely elastic
processes; and, b) as a consequence of the evolution
of the internal state variables of the material with the
changes in the environmental process variables. This
last effect is incorporated through a set of generalized
hardening rules.
The application of the constitutive models devel-
oped within this theoretical framework to practical
engineering problems requires the development of
accurate and robust numerical procedures to integrate
(at the Gauss point level, in a typical displacement–
based FE implementation) the constitutive equations
in rate form along the prescribed loading history.
Since the early works on metal plasticity, summa-
rized in Hughes (1984), a number of fundamental
studies have been published on this subject. A fairly
complete account of the current state of the art in com-
putational plasticity is provided by Simo and Hughes
(1997). Due to its properties of stability and accu-
racy for large time steps, one of the most widely
used integration schemes is the so–called General-
ized Backward Euler (GBE) method. Applications of
the GBE method to complex elastoplastic models for
geomaterials are reported, among others, by Borja and
Lee (1990, Alawaji et al. (1992, Jeremić and Sture
(1997). Tamagnini et al. (2002) have extended the GBE
method to a chemo–plastic hardening model for mod-
eling weathering processes in cemented granular soils
and weak rocks. In the second part of the work, the
work of Tamagnini et al. (2002) is extended to incor-
porate the full formalism of plasticity with generalized
hardening.
In the third and final part of the work, the general
formulation is specialized to the case of unsaturated,
soft porous rocks such as calcarenites, chalks and
porous sandstones, quite common in the Mediter-
ranean region. In such materials, bond degradation can
occur due plastic deformations, changes in degree of
saturation (short–term debonding) and chemical dis-
solution of the bond material and the solid grains
(long–term debonding), as reported by Ciantia and
82
Hueckel (2013), Ciantia et al. (2014) and Ciantia
and di Prisco (2014). A number of numerical sim-
ulations of initial–boundary value problems where
short– and long–term debonding of calcarenites are
finally presented in order to demonstrate the predic-
tive capabilities of the model as well as the accuracy
and efficiency properties of the adopted stress–point
algorithm.
2 PLASTICITY WITH GENERALIZED
HARDENING: GENERAL FRAMEWORK
The mathematical formulation of a theory of plastic-
ity with generalized hardening, capable of describing
the response of (possibly unsaturated) geomaterials
under both mechanical and “environmental” load-
ing conditions of various kind can be constructed as
follows.
Let ! ∈ Sym and α ∈ Rn be the strain tensor. Within
the framework of linearized kinematics, we consider
the total strain rate additively decomposed into an
elastic, reversible part, !˙ e , and a plastic, irreversible
part, !˙ p :
In addition, let alpha ∈ Rm be a vector collecting all
the non–mechanical (environmental) process variables
whose changes may affect the mechanical response of
the material:
where Sr is the degree of saturation (to be considered
for unsaturated states), T is the temperature, and ck
are the concentrations of nc chemical species present in
the solid skeleton or dissolved in the pore liquid which
may be responsible for chemo/mechanical processes,
such as dissolution, precipitation, cation exchange, etc.
In a unsaturated state, the description of the
mechanical response of the material requires the use
of two independent constitutive stress quantities. As
observed by Gens (2010), there are several possible
alternative choices for the first constitutive stress vari-
able, the most common being the net stress tensor σ =
σ − ug 1, adopted, e.g., byAlonso et al. (1990), Wheeler
and Sivakumar (1995), Cui and Delage (1996), and the
average skeleton stress σ̂ = σ − {(1 − Sr )ug + Sr u! }1,
used, e.g., by Jommi and Di Prisco (1994), Bolzon
et al. (1996), Jommi (2000), Sheng et al. (2004), Borja
(2006), Della Vecchia et al. (2013). In the following,
we denote with the symbol σ ∗ ∈ Sym the constitutive
stress adopted as a first constitutive variable, and adopt
the suction s = ug − u! , where ug and u! are the pore
gas and liquid pressures respectively, as the second
constitutive stress variable.
The reversible response of the material is defined
by postulating a (hypo)elastic relation between σ̇ ∗ and
∈˙ e of the form:
where De is the elastic tangent stiffness tensor, and
M is a coupling term accounting for such effects
as changes in suction, thermal expansion/contraction
and deformations induced by modifications in the
chemical composition of the pore liquid or by grain
dissolution.
Irreversibility of the material response is introduced
by requiring that the state of the material, defined in
terms of the stress tensor σ ∗ and the internal variable
pseudo–vector q ∈ Rn , to lie in the convex set:
where f : Sym × Rn % → R is the yield function of the
material. Note that, according to eq. (4), the actual
shape and size of the yield locus may be affected by
the previous loading history as well as by the evolution
of the non–mechanical process variables through the
internal variables q.
The evolution of plastic strains is defined by means
of a standard flow rule:
where g : Sym × Rn % → R is the plastic potential, and
γ̇ is the plastic multiplier, subject to the standard
Kuhn–Tucker complementarity conditions:
stating that plastic deformations may occur only for
states on the yield surface.
The evolution of the internal variables is provided
by the following generalized hardening law:
where: h ∈ Rn and N ∈ Rn × Rm are suitable harden-
ing functions. The first term on the RHS of eq. (7)
describes the changes in the internal variables associ-
ated with plastic deformations, while the second term
accounts for all non–mechanical hardening/softening
processes induced by a change of environmental
process variables α.
This formulation of the evolution equations for
the internal variables incorporate, as special cases,
a number of well–known experimentally observed
features of the hardening/softening processes expe-
rienced by natural geomaterials upon environmental
loading, such as desaturation hardening, thermal soft-
ening, chemo–mechanical coupling in fine–grained
soils and chemical degradation effects in bonded
granular materials or soft rocks.
The consistency condition (γ̇ f˙ = 0) applied to a
plastic loading process allows to determine the plas-
tic multiplier as a function of the rates of the process
variables. Let:
83
denote the gradients of the yield f with respect to σ ∗
and q. By exploiting the elastic constitutive equation
(3) and the generalized flow rule (6), the following
expression for the plastic multiplier is obtained:
where 'x( >: = (x + x)/2 denotes the ramp function,
and:
the scalar Hp being the hardening modulus of the
material. The derivation of eq. (8) requires that Kp
is strictly positive. As discussed, e.g., by Simo and
Hughes (1997), this last assumption is crucial in
the establishment of the correct formulation of the
loading/unloading conditions in presence of softening.
It is worth noting that eq. (8) can be interpreted as
follows:
with:
While γ̇mec coincides with the plastic multiplier of
classical elastoplasticity (see, e.g., Simo and Hughes
1997), γ̇env accounts for the effect of environmental
processes. These affect the response of the material
in two different ways: by contributing directly to the
evolution of the constitutive stress through the “relax-
ation” term M α̇ in eq. (3), and by contributing to the
evolution of the internal variables through the term
N α̇ in eq. (7). For plastic loading to occur, the sum
of γ̇mec and γ̇env needs to be positive, while one of the
two terms may still be negative. In particular, plastic
strains may occur even if the mechanical trial stress
rate (σ̇ ∗ )trα=const. is pointing inwards the current yield
locus (γ̇mec < 0), provided that the changes in environ-
mental variables α determine a reduction in size of
the elastic domain sufficiently large to keep the plas-
tic multiplier positive, as it occurs, for example, in the
case of chemical degradation (Tamagnini et al. 2002).
Substituting the expression eq. (8) for the plastic
multiplier in the evolution equations eq. (3) – where
the flow rule (5) is used to express the plastic strain
rate – and (7), we obtain, after some algebra:
where:

and H(γ̇) is the Heaviside step function, equal to one
if γ̇ > 0 and zero otherwise.
Finally, the formulation is completed by provid-
ing the evolution equation for the second constitutive
variable, namely the suction s:
where k # and k α are suitable constitutive functions
which generalize the concept of soil–water reten-
tion curve used in unsaturated soil modeling to more
complex environmental loading conditions.
3 IMPLEMENTATIONAL ISSUES
For the applications of the theory discussed in the pre-
vious section within the context of the FE method,
a suitable computational strategy should be defined
to integrate numerically (at the Gauss point level)
the evolution equations (13) and (14) in time, from
a given initial state. In the following, the Generalized
Backward Euler (GBE) algorithm proposed by Tam-
agnini et al. (2002) for bonded geomaterials subject
to mechanical and chemical degradation is extended
to the general formulation outlined in Sect 2. In the
following, it is further assumed that the constitutive
equation (3) can be derived from an elastic potential
function ψ(ine , α), so that:
and:
In this way, the constitutive stress tensor σ ∗ can be
replaced by ∈e in the set of state variables.
!
Let I = Nn=0 [tn , tn+1 ] be a partition of the time
interval of interest into time steps. It is assumed that
at time tn ∈ I the state of the material (σ ∗n , qn , αn ) is
known at any quadrature point in the adopted finite
element discretization. The computational problem to
be addressed is thus the updating of the state variables
(∈e , q) at time tn+1 for any given history of the pro-
cess variables ∈ and α, through the integration of the
differential–algebraic system given by eq. (3), (5) and
(7), subject to the constraints (6).
84
The additive structure of the rate equations (3) and
(7) suggests the following extension to the classical
elastic–plastic operator split of the original problem
of evolution (OR):
OR ≡ EP + PC
!˙ = ∇ s (u̇) !˙ = ∇ s (u̇) !˙ = 0
α̇ = dα/dt α̇ = 0 !˙ = 0
!˙ e = !˙ − γ̇Q !˙ e = !˙ !˙ e = −γ̇Q
q̇ = γ̇h + N α̇ q̇ = N α̇ q̇ = γ̇h
From the above decomposition, a product formula
algorithm is constructed in a standard fashion as fol-
lows. First, the elastic predictor (EP) problem is solved
and a so–called trial elastic state is obtained. Then,
the constraints (6) are checked for the trial state, and
if they are violated, the trial state is taken as the initial
condition for the plastic corrector (PC) problem.
The solution of the predictor stage in terms of elas-
tic strains is trivially given by !e,tr e
n+1 = !n + !n+1 − !n .
As for the internal variables, the possibility of obtain-
ing their trial values qtrn+1 by closed–form integration
of the EP evolution equation depends crucially on the
nature on the constitutive function N . When this is not
the case, a numerical integration procedure to obtain
the trial state is required. The state of stress at the
trial state, σ ∗,tr
n+1 , is then obtained by mere function
evaluation from (!e,trn+1 , αn+1 ), via eq. (20).
The trial state is then checked for consistency with
the yield locus and, if fn+1 tr
:= f (σ ∗,tr tr
n+1 , qn+1 ) > 0, the
process is declared plastic, and consistency is restored
by solving the plastic corrector problem. In the GBE
method, this last task is performed by means of a fully
implicit Backward Euler algorithm, which yields the
following nonlinear system of algebraic equations in
the unknowns (∈en+1 , qn+1 , &γn+1 ):
typically solved via a local Newton iteration. Remark-
ably, the plastic corrector equations are identical
to those of standard elastoplasticity, as all non–
mechanical effects are fully accounted for in the
predictor stage. In particular, a closed form lineariza-
tion of the algorithm is possible. following the same
strategy outlined, e.g., in Tamagnini et al. (2002), and
omitted here for brevity.
4 APPLICATION: ENVIRONMENTAL
DEGRADATION OF CALCARENITE ROCKS
As an example of application of the general theory to
a specific class of geomaterials of relevant practical
interest, we consider in this section the particular case
of unsaturated, soft porous rocks such as calcarenites,
chalks and porous sandstones, quite common in the
Mediterranean region. The progressive loss of strength
caused by degradation of interparticle bonds of vari-
ous kind has potentially dangerous consequences for
the anthropic environment, associated to slope stabil-
ity problems or collapse of underground cavities (see,
e.g., Parise and Lollino 2011).
The mechanical behavior of calcarenite rocks
has been intensively investigated by Ciantia (2013),
Ciantia and Hueckel (2013), Ciantia et al. (2013),
Ciantia et al. (2014). From a detailed microstruc-
tural investigation of Gravina calcarenite, Ciantia et al.
(2013) have identified two distinct types of inter-
granular bonds, characterized by different strengths
and formation mechanisms: strong diagenetic bonds,
made of bridges of calcite formed through a diagenetic
process, and weak depositional bonds, formed by a
mixture of the calcite “powder” and seawater salts that
precipitates as sediment into a meniscus–shaped bond
when water evaporates, see Fig. 1. The degradation
processes affecting these bonds are of two different
kinds (Ciantia and Hueckel 2013, Ciantia et al. 2014):
a) short–term debonding (STD), associated to the
increase of the degree of saturation of the material,
due to the rapid destruction of depositional bonds
as the micrograins of calcite returns in suspension
in the pore water;
b) long–term debonding (LTD), due to the chemical
dissolution of the diagenetic bonds of calcite, taking
place when the rock is flooded by water for a long
period of time, under fully saturated conditions.
The chemical reactions between the species dis-
solved in the pore liquid and the individual grains
of the solid skeleton can also be responsible for the
occurrence of significant grain dissolution, which, in
turn, can be responsible for chemically–induced vol-
umetric deformations of the solid skeleton. The three
degradation processes mentioned are characterized by
progressively increasing time scales. In this respect,
Figure 1. Typical microstructure of a calcarenite rock.
85
it is worth noting that even the slowest degradation
processes may produce significant effects over time
periods comparable to the life cycle of most geotech-
nical engineering structures, and are thus of significant
practical relevance.
From the constitutive modeling point of view,
restricting our attention to isothermal problems, the
environmental process variables which are of rele-
vance to describe bond degradation and grain dissolu-
tion in a unsaturated calcarenite rock are the degree of
saturation Sr and a normalized measure of the chemi-
cal dissolution process, ξd , defined as the ratio between
the mass of calcite dissolved in the pore water and the
initial solid mass per unit volume:
The evolution of ξ with time is governed by the calcite
dissolution reaction rate (Ciantia and Hueckel 2013).
The vector α of eq. (2) is therefore reduced to:
In the description of the mechanical behavior of the
rock, the average skeleton stress is adopted. In view of
the large dimensions of the macropores and of the high
stiffness of the solid skeleton, a simple van Genuchten
soil–water characteristic curve is adopted, assuming
no coupling with the mechanical deformation of the
solid skeleton and the chemical dissolution processes.
As in Tamagnini et al. (2002), the elastic response
of the material is defined by means of a suitable stored
energy function,
where ψ0 (!e )(1 + δb ) is the stored energy function
of the intact material, with ψ0 given by Tamagnini
et al. (2002); D is a non–mechanical damage vari-
able which accounts for the short– and long–term
debonding effects on the elastic response of the mate-
rial, and δb is a material constant. A similar approach
has been used by Saetta et al. (1998) in the context
of damage mechanics of concrete subject to chemical
degradation.
According to eq. (26), the elastic tangent stiffness
and coupling operators are obtained as:
When the chemical dissolution process had progressed
to the point that all the diagenetic bond material has
been dissolved (ξd = ξd,cr ), the dissolution of the cal-
cite grains gives rise to additional chemically–induced
volumetric deformations of the solid skeleton which,
to a first approximation, can be evaluated as:

and C a material constant. In this case, the operator M

in eq. (28) must be replaced by:

as the chemical strain rate has to be subtracted to the

total strain rate in the elastic constitutive equation.
As for the yield function and plastic potential,
the expressions originally employed by Lagioia et al.
(1996) have been adopted:

where:
Figure 2. Yield surface of calcarenite: a) representa-
tion in q : p plane for axisymmetric compression (the
bond–permitted stress space is outlined in grey); b) view in
principal stress space.

negligible for calcarenite rocks, the functions Nsw and

Ntξ (components of the first row of N) can be set to
zero. For the function hs (first component of vector h),
the following expression is assumed (Tamagnini et al.
2002, Nova et al. 2003):

where:

and ρs and ζs are material constants. The evolution

equations for the bonding–related variable pt (linked
to the rock tensile strength) includes both mechan-
ical and environmental hardening/softening through
the functions ht (first component of vector h), Ntw and
Ntξ (components of the second row of N), given by:
In the above expressions, the quantities af , mf , ag , mg
and k are material constants; p (mean stress), q (devia-
tor stress) and θ (Lode angle) are the three invariants of
the stress tensor σ ∗ ; the scalar quantities q = {ps , pt }
are the hardening variables describing the effects of
previous (mechanical and non–mechanical) loading
history, and p̃c is a dummy parameter. An illustration of
the yield surface in the q : p plane and in the principal
stress space is given in Fig. 2.
As the internal variable ps is intended to describe the
mechanical hardening/softening processes induced by in which ρt and ζt are material constants. Eqs. (44)
plastic deformations in a completely unbonded mate- and (45), supported by a large number of experimen-
rial, and suction hardening effects can be considered tal results (Ciantia 2013, Ciantia et al. 2013, Ciantia

86
et al. 2014), are based on the assumption that the envi-
ronmental process variables affect bond degradation
through the single scalar damage variable Dd .
Explicit expressions for the damage coefficients
D(Sr , ξd ) and Dd (Sr , ξd ) have been proposed by
Ciantia (2013) and Ciantia and di Prisco (2014), as a
result of the upscaling of a suitable micromechanical
model of the the calcarenite rock defined for a ideal-
ized meso–REV, and are not reported here for brevity.
The reader is referred to the aforementioned works for
details.

5 NUMERICAL EXAMPLES

The model discussed in Sect. 4 has been implemented

in the FE code GeHoMadrid (Fernandez-Merodo et al.
2007) which incorporates a fully coupled chemo–
hydro–mechanical formulation for the deformation,
flow and reactive transport of chemical species in
porous media. Two representative numerical simula-
tions are discussed in the following to demonstrate
the capabilities of the model and to show the impact
of environmental loads on the behavior of realistic
geotechnical structures.
The first simulation refers to a wetting/chemical
degradation test performed on a cylindrical specimen
of calcarenite rock (Ciantia et al. 2014). The test is
intended to simulate, at small scale, the degradation
processes which might occur in a rock pillar within
an underground cavity, when the rock is first soaked
by water and then subjected to chemical weathering.
The testing layout and an illustration of the different
stages of the test are shown in Fig. 3a and 3b, respec-
tively. Fig. 3c shows the problem geometry, the spatial
discretization adopted in the FE simulation and the
assumed boundary conditions.
In the first stage of the test (OA), the specimen is
loaded in uniaxial conditions, starting from a dry state.
Then the specimen is saturated at constant axial stress
by imposing an upward water flow (stage AB), up to
full saturation (stage BC) during a period of 2 min.
In the final stage of the test (CD), the specimen is
brought to failure by a long term exposure to a weak
acid solution on the lateral surface.
The isochrones of the computed normalized verti-
cal displacements along the specimen axis are shown
in Fig. 4a. Fig. 4b shows the time evolution of the
normalized vertical displacement at the top of the spec-
imen. The response of the material during the first
loading stage – performed in steps – is almost linear
elastic. Small additional deformations are produced
by the subsequent STD during the wetting stage. On
the contrary, the progressive chemical dissolution of
intergranular bonds and grains give rise to progres-
sively increasing deformations with time at constant
stress, eventually leading to the failure of the speci-
men when the reduction in uniaxial rock strength due
to debonding is such that the axial load cannot be
sustained any longer. This behavior is correctly cap-
tured by the FE simulations, as demonstrated by the
Figure 3. Uniaxial loading test: a) experimental setup; b)
description of testing stages; c) geometry, spatial discretiza-
tion and boundary conditions.
comparison between experimental observations and
numerical predictions.
The computed isochrones of ξd and of axial stress
along a radius at the specimen mid–height are shown in
Fig. 5a and Fig. 5b, respectively. As the acid flows into
the specimen from the lateral boundary, the values of
ξd are not uniform along the section, but minimum at
the specimen axis and maximum at the boundary. The
inhomogeneous degradation process resulting from
the horizontal acid flow causes a progressive weak-
ening of the external portion of the specimen which
induces a reduction in axial stress close to the lateral
boundary and a corresponding stress increase at the
specimen axis, due to stress redistribution within the
specimen.
This phenomenon is qualitatively similar to the
one observed in the pillars of abandoned mines when
the intact rock is exposed to long–term chemical
degradation (Castellanza et al. 2002).
The second problem examined is a strip footing,
2.44 m in width, resting on a layer of saturated weak
calcarenite rock (Fig. 6). In the simulation, the foot-
ing is first loaded up to 300 kPa; then, under constant
loading conditions, a chemical dissolution process is
simulated by imposing a constant acid concentration
at the upper surface of the rock layer and modeling the
subsequent diffusion process along the layer depth.
This results in a progressive evolution with depth of
the chemical process variable ξd , and an associated
reduction of bond strength, in this case of diagenetic
origin only (no STD).

87
Figure 4. Uniaxial loading test: a) isochrones of normal- Figure 5. Uniaxial loading test: a) isochrones of ξd with
ized vertical displacements with depth; b) time evolution of distance from pillar axis; b) isochrones of axial stress with
normalized vertical displacement at the top of the specimen. distance from pillar axis.

When the contraction of the elastic domain associ-

ated to the long–term debonding is sufficient to bring
the actual stress state in the foundation rock on the
yield surface, plastic deformations should occur to
restore consistency. This results in a progressive devel-
opment of strains and displacements within the soil,
the magnitude of which is significantly affected by
the stress distribution within the rock layer.
Fig. 7a shows the computed isochrones of ξd along
depth at footing centerline, for different time stations.
It can be observed how, for the particular case at hand,
the chemical process variable increases up to a maxi-
mum value of 30% in a relatively short time span (440
days). During this period, the thickness of the rock
layer affected by the debonding process is about 2 m.
The computed isochrones of the vertical deforma-
tion along depth at footing centerline are shown in Fig.
7b. The first isochrone, at t = 10 days, corresponds Figure 6. Strip footing on a calcarenite rock layer: geometry,
to the deformation profile produced by the footing spatial discretization and loading conditions.
load (no chemically induced deformations, as ξd = 0
throughout the entire layer). The deformation profiles solid symbols). It can be observed how the deforma-
reported in the figure for larger elapsed times refer to tions induced by grain dissolution, which occur as
two different simulations, performed ignoring chemi- ξd > ξd,cr , could be of the same order of magnitude
cal deformations due to grain dissolution (C = 0, open of the plastic deformations induced by the debonding
symbols) and taking account grain dissolution (C > 0, process.

88

Figure 7. Strip footing on a calcarenite rock layer: a)
isochrones of ξd along depth at footing centerline; b)
isochrones of vertical strain along depth at footing centerline.
The relevance of grain dissolution effects can also
be appreciated from Fig. 8, showing the time evolution
of vertical strains close to the top of the layer, at the
foundation centerline (point A in Fig. 6) and far from it
(point B in Fig. 6). As expected, the bond–degradation
induced vertical deformations are larger beneath the
footing, where the stress state is significantly affected
by the footing load. However, the evolution in time of
additional deformations caused by grain dissolution
(starting at about 300 days) show a similar trend at both
points A and B, since in the proposed formulation !c is
not affected by the stress state. Further experimental
investigations are required to assess if this assumption
can be considered realistic in practical applications.
6 CONCLUDING REMARKS
In this paper, starting from the works of Tamagnini
et al. (2002) and Nova et al. (2003), a general approach
has been proposed to extend the classical theory of
hardening plasticity to a wide range of (saturated or
89
Figure 8. Strip footing a calcarenite rock layer: time evolu-
tion of vertical strain at points A and B (see Fig. 3).
unsaturated) natural geomaterials subject to “envi-
ronmental” loading conditions. In this approach, the
coupling between the different physical processes is
taken into account in two ways: by the relaxation
terms in the stress rate associated to strain rates of
non–mechanical origin, and by the effects that envi-
ronmental process variables can have on the internal
variables (environmental hardening), described by a
generalized hardening function.
Remarkably, the additive structure of the general-
ized hardening law, coupled with the additive decom-
position of strain rates allows a simple extension of the
standard closest–point projection algorithms (GBE),
widely used in computational mechanics for their
excellent properties of stability and accuracy, to the
present case.
The application examples reported, relative to a spe-
cific model for unsaturated calcarenite rocks subject
to complex debonding process of different origin, have
demonstrated the practical relevance of environmental
degradation effects in engineering applications, over
periods of time comparable with the life cycle of most
geotechnical structures.
ACKNOWLEDGEMENTS
The authors would like to warmly acknowledge the
fundamental role played by their mentor and friend
Roberto Nova in inspiring this work. The authors
would also like to thank Claudio di Prisco and Riccardo
Castellanza for many fruitful discussions, and José
Antonio Fernandez–Merodo for the support provided
with the FE code GeHoMadrid.
REFERENCES
Alawaji, H., K. Runesson, & S. Sture (1992). Implicit integra-
tion in soil plasticity under mixed control for drained and
 undrained response. Int. J. Num.Anal. Meth. Geomech. 16,
737–756.
Alonso, E. E., A. Gens, & A. Josa (1990). A constitutive
model for partially saturated soils. Géotechnique 40(3),
405–430.
Bolzon, G., B. Schrefler, & O. Zienkiewicz (1996). Elasto-
plastic soil constitutive laws generalized to partially
saturated states. Géotechnique 46(2), 279–289.
Borja, R. I. (2006). On the mechanical energy and effec-
tive stress in saturated and unsaturated porous continua.
International Journal of Solids and Structures 43(6),
1764–1786.
Borja, R. I. & S. R. Lee (1990). Cam–clay plasticity, part I.
implicit integration of elastoplastic constitutive relations.
Comp. Meth. Appl. Mech. Engng. 78, 49–72.
Castellanza, R., R. Nova, & C. Tamagnini (2002). Mechanical
effects of chemical degradation of bonded geomaterials
in boundary value problems. Revue française de génie
civil 6(6), 1169–1192.
Cekerevac, C. & L. Laloui (2004). Experimental study
of thermal effects on the mechanical behaviour of a
clay. International Journal for Numerical and Analytical
Methods in Geomechanics 28(3), 209–228.
Ciantia, M., R. Castellanza, & C. di Prisco (2013). Chemo–
mechanical weathering of calcarenites: experiments and
theory. In Coupled phenomena in environmental geotech-
nics. Taylor & Francis Group, London, pp. 541–548.
Ciantia, M. O. (2013). Multiscale hydro-chemo-mechanical
modelling of the weathering of calcareus rocks: an Exper-
imental, Theoretical and Numerical Study. Ph. D. thesis,
Politecnico di Milano.
Ciantia, M. O., R. Castellanza, & C. di Prisco (2014).
Experimental study on the water-induced weakening of
calcarenites. Rock Mechanics and Rock Engineering,
1–21.
Ciantia, M. O. & C. di Prisco (2014). Extension of plastic-
ity theory to debonding, grain dissolution and chemical
damage of bonded materials. (in preparation).
Ciantia, M. O. & T. Hueckel (2013). Weathering of stressed
submerged calcarenites. Géotechnique 63, 768–785.
Cui, Y. & P. Delage (1996). Yielding and plastic behaviour
of an unsaturated compacted silt. Géotechnique 46(2),
291–311.
Della Vecchia, G., C. Jommi, & E. Romero (2013). A
fully coupled elastic–plastic hydromechanical model for
compacted soils accounting for clay activity. Interna-
tional Journal for Numerical and Analytical Methods in
Geomechanics 37(5), 503–535.
Fernandez-Merodo, J., R. Castellanza, M. Mabssout, M. Pas-
tor, R. Nova, & M. Parma (2007). Coupling transport of
chemical species and damage of bonded geomaterials.
Comp. & Geotechnics 34(4), 200–215.
Gens, A. (2010). Soil–environment interactions in geotech-
nical engineering. Géotechnique 60(1), 3–74.
Gens, A., L. d. N. Guimaraes, & S. Olivella (2002). Coupled
chemomechanical analysis for saturated and unsaturated
soils. Environmental geomechanics, 109–123.
Gens, A., L. d. N. Guimarães, & S. Olivella (2005).
Thmc coupling in partially saturated geomaterials. Revue
européenne de génie civil 9(5-6), 747–765.
Hueckel, T. (1997). Chemo–plasticity of clays subjected to
stress and flow of a single contaminant. Int. J. Num. Anal.
Meth. Geomech. 21, 43–72.
Hughes, T. J. R. (1984). Numerical implementation of con-
stitutive models: rate–independent deviatoric plasticity. In
90
View publication stats
S. Nemat-Nasser, R. Asaro, and G. Hegemier (Eds.), The-
oretical Foundations for Large Scale computations of Non
Linear Material Behavior, Horton, Greece, pp. 29–57.
Martinus Nijhoff Publisher, Dordrecht.
Jeremić, B. & S. Sture (1997). Implicit integration in elasto-
plastic geotechnics. Mech. Cohesive–Frictional Materi-
als 2, 165–183.
Jommi, C. (2000). Remarks on the constitutive modelling of
unsaturated soils. Volume 153. eds A. Tarantino and C.
Mancuso. Rotterdam: Balkema.
Jommi, C. & C. Di Prisco (1994). A simple theoretical
approach for modelling the mechanical behaviour of
unsaturated soils. In Proceedings of the Conference “Il
ruolo dei fluidi nei problemi di Ingegneria geotecnica”,
Volume 1, pp. 167–188. (in Italian).
Lagioia, R., A. M. Puzrin, & D. M. Potts (1996). A new ver-
satile expression for yield and plastic potential surfaces.
Comp. & Geotechnics 19, 171–191.
Loret, B., T. Hueckel, & A. Gajo (2002). Chemo–mechanical
coupling in saturated porous media: elastic–plastic
behaviour of homoionic expansive clays. International
Journal of Solids and Structures 39(10), 2773–2806.
Nova, R. (1986). Soil models as a basis for modelling the
behaviour of geophysical materials. Acta Mechanica 64,
31–44.
Nova, R. (2000). Modelling the weathering effects on the
mechanical behaviour of granite. In D. Kolymbas (Ed.),
Constitutive Modelling of Granular Materials, Horton,
Greece. Springer, Berlin.
Nova, R., R. Castellanza, & C. Tamagnini (2003). A consti-
tutive model for bonded geomaterials subject to mechan-
ical and/or chemical degradation. International Journal
for Numerical and Analytical Methods in Geomechan-
ics 27(9), 705–732.
Nova, R., R. Castellanza, & C. Tamagnini (2004). A con-
stitutive model for mechanical and thermal loading of
bonded geomaterials based on the concept of plasticity
with extended hardening. Proc. NUMOG 9.
Parise, M. & P. Lollino (2011). A preliminary analysis of
failure mechanisms in karst and man-made underground
caves in southern italy. Geomorphology 134(1), 132–143.
Saetta, A., R. Scotta, & R. Vitaliani (1998). Mechanical
behavior of concrete under physical-chemical attacks.
Journal of engineering mechanics 124(10), 1100–1109.
Sheng, D., S. Sloan, & A. Gens (2004). A constitutive model
for unsaturated soils: thermomechanical and computa-
tional aspects. Computational Mechanics 33(6), 453–465.
Simo, J. C. & T. J. R. Hughes (1997). Computational
Inelasticity. Springer Verlag, New York.
Tamagnini, C., R. Castellanza, & R. Nova (2002). A gener-
alized backward euler algorithm for the numerical inte-
gration of an isotropic hardening elastoplastic model for
mechanical and chemical degradation of bonded geoma-
terials. Int. J. Num. Anal. Meth. Geomech. 26, 963–1004.
Tamagnini, R. (2004). An extended cam-clay model for
unsaturated soils with hydraulic hysteresis. Géotech-
nique 54(3), 223–228.
Uchaipichat, A. & N. Khalili (2009). Experimental inves-
tigation of thermo-hydro-mechanical behaviour of an
unsaturated silt. Geotechnique 59(4), 339–353.
Wheeler, S. & V. Sivakumar (1995). An elasto-plastic critical
state framework for unsaturated soil. Géotechnique 45(1),
35–53.