Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213

Journal of Rock Mechanics and GeotechnicalEngineering

Journal of Rock Mechanics and Geotechnical

Engineering
journal homepage: www.rockgeotech.org

Joints in unsaturated rocks: Thermo-hydro-mechanical formulation and

constitutive behaviour
E.E. Alonso ∗ , M.T. Zandarín, S. Olivella
Department of Geotechnical Engineering and Geosciences, UPC, Barcelona 08034, Spain

a r t i c l e i n f o a b s t r a c t

Article history: A formulation for the coupled analysis of thermo-hydro-mechanical (THM) problems in joints is first
Received 26 March 2013 presented. The work involves the establishment of equilibrium and mass and energy balance equations.
Received in revised form 5 May 2013 Balance equations were formulated taking into account two phases: water and air. The joint element
Accepted 16 May 2013
developed was implemented in a general purpose finite element computer code for THM analysis of
porous media (Code Bright). The program was then used to study a number of cases ranging from labo-
Keywords:
ratory tests to large scale in situ tests. A numerical simulation of coupled hydraulic shear tests of rough
Rock joints
granite joints is first presented. The tests as well as the model show the coupling between permeability
Thermo-hydro-mechanical (THM)
behaviour
and the deformation of the joints. The experimental investigation was focused on the effects of suction on
Finite elements the mechanical behaviour of rock joints. Laboratory tests were performed in a direct shear cell equipped
Suction controlled shear tests with suction control. Suction was imposed using a vapour forced convection circuit connected to the cell
Constitutive model and controlled by an air pump. Artificial joints of Lilla claystone were prepared. Joint roughness of varying
Numerical simulations intensity was created by carving the surfaces in contact in such a manner that rock ridges of different
tip angles were formed. These angles ranged from 0◦ (smooth joint) to 45◦ (very rough joint profile). The
geometric profiles of the two surfaces in contact were initially positioned in a “matching” situation. Sev-
eral tests were performed for different values of suctions (200, 100, and 20 MPa) and for different values
of vertical stresses (30, 60, and 150 kPa). A constitutive model including the effects of suction and joint
roughness is proposed to simulate the unsaturated behaviour of rock joints. The new constitutive law
was incorporated in the code and experimental results were numerically simulated.
© 2013 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Elsevier B.V. All rights reserved.

1. Introduction and appear when there is a change in the characteristics of the

Discontinuities of the rock mass are the result of the origin of
the rock and the subsequent deformations imposed, in most cases,
by tectonic activity. According to Jennings and Robertson (1969),
two sets of discontinuities could be typically defined as major and
minor or secondary. Major discontinuities include bedding planes,
faults, contacts and dykes, while the minor ones are joints of limited
length, i.e. cross joints in sedimentary rocks.
Taking into account their origin, joints can be classified as:
(1) bedding planes, which are associated with sedimentary rocks
∗ Corresponding author. Tel.: +34 934016862.
E-mail address: eduardo.alonso@upc.edu (E.E. Alonso).
Peer review under responsibility of Institute of Rock and Soil Mechanics, Chinese
Academy of Sciences.
1674-7755 © 2013 Institute of Rock and Soil Mechanics, Chinese Academy of
Sciences. Production and hosting by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jrmge.2013.05.004
deposited materials; (2) stress relief joints, which develop as a
result of erosion of weathered rock; (3) tension joints, which are
the result of cooling and crystallization of igneous rock; and (4)
faults, which result in a plane or band of shear failure that exhibits
obvious signs of differential movement of the rock mass on each
side of the central plane as well as rock degradation. Usually, faults
are linked to the movement of tectonic plates.
The characteristics of the planar surfaces constituting a joint
depend on the geological history of the rock mass. They are the
result of mechanical, hydraulic, depositional, chemical and other
processes. The void structure of discontinuities has a dominant
effect on its hydro-mechanical (HM) behaviour.
Finite element formulations describing joint behaviour started
in the pioneering contribution of Goodman et al. (1968). Since
then, published formulations have steadily improved the capabil-
ities of the joint models. In particular, attention is given here to
the coupled hydraulic and mechanical behaviour of joints. Recent
contributions were published by Guiducci et al. (2002) and Segura
(2008).
The HM behaviour under varying normal stress has been exten-
sively studied. The experimental results obtained by Hans (2002)
show that transmissivity decreased as normal stress increased. This
 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213 201

decrease is due to the reduction of the void space between the joint
walls, the increase of the contact area and the changes in tortuosity.
When the compression stress on the discontinuity is released, an
irreversible behaviour can be observed.
When a shear stress is applied, before the peak conditions
are attained, transmissivity initially decreases. However when
peak conditions are met, transmissivity increases substantially
(approximately two orders of magnitude). The increment of
transmissivity is directly related to joint dilatancy (Lee and Cho,
2002). Even if dilatancy increases continuously with relative shear
displacements, joint permeability may reach a constant value. This
is a consequence of the gouge material generated by the breakage
of asperities. The roughness degradation depends on the strength
of asperities, the applied normal load and the shear stiffness.
Olsson and Barton (2001) described this behaviour and proposed
a model to consider these phenomena. Indraratna et al. (2003)
reported an analytical and experimental study of the two-phase
flow through rock joints.
Taking these results as a starting point, a finite element formu-
lation for the coupled thermo-hydro-mechanical (THM) behaviour
of joint elements has been developed. It considers a two-phase
(air and water) flow and vapour diffusivity through joints. A
further motivation for this work was related to the conditions
found in nuclear waste disposal designs. Bentonite barriers, ini-
tially unsaturated, exhibit strong suctions at early phases. The Fig. 1. Joint element with double nodes. (a) Stress state at the mid-plane of the joint
heat generation imposed by nuclear canisters results in a drying element. (b) Relative displacement defined at mid-plane.
of the engineered barrier that is also subjected to inflow from
the host rock. In the long term, the gas generated in the waste
2. A coupled thermo-hydro-mechanical formulation of
may escape through interfaces and rock joints, a phenomenon
joints
which depends on gas generation rates. The set of conditions
outlined imply that artificial joints (those existing between
2.1. Mechanical formulation
engineered barriers and excavated rock surface, for instance)
and natural rock joints may be exposed to partially saturated
The mechanical formulation of the joint element is defined by
conditions.
the relationship between stress and relative displacements of the
Finally, joints above an existing water level or exposed to ambi-
joint element mid-plane (Fig. 1). Then, the mid-plane relative dis-
ent conditions are involved in slope stability and excavations.
placements are interpolated using the nodal displacements and the
It was then natural to attempt a generalized formulation of joint
shape functions:
behaviour for partially saturated conditions. This is achieved by
providing a separate consideration to water and air transfer. In
 
un
addition, since heat transfer is also involved in some applications w mp = = rN ump [ −I 4 I 4 ]uj (1)
(notably nuclear waste disposal), an energy balance was added to us
mp
the field equations.
The effect of suction on the mechanical behaviour of rock joints where un and us are the normal and tangential relative displace-
has not been reported in the literature to the authors’ knowl- ments of the element’s mid-plane (see Fig. 1), respectively; r is the
edge. Since the prevailing suction has a very significant effect on rotation matrix that transforms the relative displacements in the
rock strength (Oldecop and Alonso, 2001), it was anticipated that local orthogonal coordinate system into the global coordinate sys-
rock joint behaviour would be also significantly affected. This was tem; N ump is a matrix of shape functions; I4 is a identity matrix of
the motivation for the performance of a laboratory testing pro- 4th order; and uj is the vector of nodal displacements.The stress
gram concentrated on the mechanical behaviour of rock joints tensor of the mid-plane is calculated as
subjected to direct shear under suction control. Suction was con-  
trolled by a vapour equilibrium technique (Fredlund and Rahardjo, 
1993; Romero, 2001). Artificially prepared joints of Lilla clay-  mp = = Dw mp (2)

stone were tested. Joint roughness of varying intensity was created mp
by carving the surfaces in contact in such a manner that rock
where  is the tangential stress at mid-plane;   is the normal
ridges of different tip angles were formed. These angles varied
effective stress; D is the stiffness matrix, which relates relative dis-
between 0◦ (smooth joint) and 45◦ (very rough joint profile). The
placements, defined by vector w mp , to stress state (see Fig. 1); and
geometric profiles of the two surfaces in contact were initially posi-
 mp is the net effective stress at the mid-plane of the element and
tioned in a “matching” situation. Several tests were performed
it is defined as
for different values of suctions (200, 100, and 20 MPa) and for
different values of vertical stresses (30, 60, and 150 kPa). From  mp =  mp − max{pgmp , plmp } (3)
the analysis of test results, a constitutive law was proposed. It
takes into account the effect of suction on the strength parame- where  mp is the total mean stress, pgmp is the gas pressure, and
ters and the degradation of rock joints. The performance of the plmp is the liquid pressure in the mid-plane of the element.
model was checked against the recorded shear stress-relative Note that the mechanical response is defined in terms of a net
displacements. stress (excess of total stress over air pressure) when the joint is
202 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213

The fluxes at mid-plane are calculated by

w a dl

[j l ]mp = [lw qlt dl + ilt
w dl] + [ w q a + iw a]
0 l ll ll 0
(5)
w a dl
[j g ]mp = [gw qgt dl + igt
w dl] + [ w q a + iw a]
g gl gl
0 0

where qlt , qgt , qll and qgl are the advective (liquid or gas) transver-
sal and longitudinal fluxes at the element boundaries, respectively;
w , iw , iw , iw are the nonadvective (liquid or gas), transver-
and ilt gt ll gl
sal, and longitudinal fluxes at the element boundaries (Fig. 2a),
respectively. The first term of Eq. (5) corresponds to the transversal
fluxes at mid-plane of the joint (calculated by means of the pres-
sure drop between surfaces, pmp1 = p3 − p1 and pmp2 = p4 − p2 )
(Fig. 2b). The second term corresponds to the longitudinal fluxes
at mid-plane calculated considering the average pressure in nodes
(pmp1 = p3 + p1 /2 and pmp2 = p4 + p2 /2; Fig. 2c).

2.2.2. Air mass balance equation

The air mass balance equation considers the dry air and the air
dissolved in the water phase. Its expression is

∂(la Sl + ga Sg ) da
dl + [j l ]mp + [j g ]mp = f a
a a
adl + (la Sl + ga Sg ) (6)
∂t dt
where la is the mass of air dissolved in liquid phase, ga is the mass
a
of gas phase (dry air), [j l ]mp is the air dissolved fluxes at mid-plane,
a
[j g ]mp is the gas flux at mid-planes, and fa is an external supply of
air.

2.2.3. Internal energy balance for the element

The internal energy balance for the element is expressed by

∂(El l Sl + Eg g Sg ) da
Fig. 2. (a) Schematic view of the mass balance of joint element; (b) transversal
adl + (El l Sl + Eg g Sg ) dl
∂t dt
fluxes; (c) longitudinal fluxes.
+ [ic ]mp + [jEl ]mp + [jEg ]mp = f E (7)

not saturated. Once saturated, the definition adopted for effective
stress results in Terzaghi’s principle.
2.2. Mass and energy balance equation
The two-phase flow through a single joint is analyzed by formu-
lating the water, air and energy balance equations at the mid-plane
of the element. The fluxes at mid-plane are calculated by interpo-
lating the leak-off at the element boundaries (see Fig. 2).
2.2.1. Water mass balance equation
The water mass balance equation for a differential volume of
joint is
∂(lw Sl + gw Sg ) da  
adl + (lw Sl + gw Sg ) dl + [jlw ]mp + [jgw ]mp = f w
∂t dt
(4)
where lw is the mass of water in liquid phase, gw is the mass of
gas in liquid phase (vapour), a is the opening of the joint element,
dl is the discrete length of the joint element, Sl is the liquid degree
w
of saturation, Sg is the gas degree of saturation, [j l ]mp is the liquid
w
flux at mid-plane, [j g ]mp is the vapour flux at the mid-plane, and fw
is an external supply of water. The first term of Eq. (4) considers the
storage change of mass at constant volume, and the second term is
the storage change caused by changes of joint opening.
The energies of the liquid and gas phases are calculated by

El l = (Elw ωlw + Ela ωla )l = Elw lw + Ela la
(8)
Eg g = (Egw ωgw + Ega ωga )g = Egw gw + Ega ga
where Elw and Ela are the internal energies of water and/or air in
liquid phase per unit mass of water and/or air, respectively; ωlw and
ωla are the masses of water and/or air in liquid phase, respectively;
Egw and Ega are the internal energies of water and/or air in gas phase
per unit mass of water and/or air, respectively; ωgw and ωga are the
masses of water and/or air in gas phase, respectively; l and g are
the liquid and gas densities, respectively; El and Eg are the internal
energies of liquid and mass per unit mass, respectively; [ic ]mp is
the heat flux at the mid-plane of the joint element; and [jEl ]mp and
[jEg ]mp are the advective energy fluxes in the joint mid-plane for
liquid and gas, respectively.
The conduction of heat [ic ]mp at mid-plane of joint is calculated
by
a dl
[ic ]mp = [ict dl]0 + [icl a]0 (9)
where ict is the transversal heat flux, and icl is the longitudinal heat
flux at the element boundaries.
The energy fluxes [jE1 ]mp and [jEg ]mp are calculated considering
the advective fluxes:
[jEl ]mp = [j l ]mp Elw + [j l ]mp Ela
w a
(10)
[jEg ]mp = [j g ]mp Egw + [j g ]mp Ega
w a
(11)
 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213 203

The weighted residual method is applied to obtain the discrete

form of equations. Finally, Eqs. (2), (4), (6), (7) are solved simulta-
neously. The vector of unknowns for each node includes the normal
and shear relative displacements (un , us ), the gas and liquid pres-
sures (Pg , Pl ) and the temperature (T).

2.3. Constitutive models

The mechanical response of the joint was modelled by means

of nonlinear elasto-viscoplastic formulation. The viscoplastic
approach provides numerical advantages (there is no need to use
return algorithm in particular).
Darcy’s law describes the advective flow for longitudinal direc-
tions. A flow proportional to the pressure drop is used in transversal
direction. The non-advective fluxes (vapour diffusivity) were mod-
elled by Fick’s law. The longitudinal permeability and the air entry
pressure of the joint depend on its opening. Finally, the heat con-
duction through the joint is calculated by Fourier’s law.

2.3.1. Mechanical model based on elasto-viscoplastic formulation

The elastic formulation proposed describes the elastic normal
stiffness by means of a nonlinear law which depends on the joint
opening (Gens et al., 1990).
The viscoplastic formulation (Perzyna, 1963; Zienkiewicz and Fig. 3. Elastic constitutive law of the joint element. Normal stiffness depends on
joint opening.
Cormeau, 1974) allows the treatment of a non-associated plasticity
and a softening behaviour of joints subjected to shear displace-
ments.
where c is the effective cohesion and tanϕ is the tangent of internal
Total displacements w are calculated by adding reversible elas-
friction angle. Note that cohesion and friction angle are defined for
tic displacements, we , and viscoplastic displacements wvp , which
the asymptote of the hyperbolic yield function.
are zero when stresses are below a threshold value (the yield sur-
Variation of these parameters results in a family of yield surfaces
face):
(Fig. 4a).
w = w e + w vp (12)

Normal and shear (relative) displacements are represented by 2.3.1.3. Softening law. The strain-softening of the joint subjected to
a two-element vector [un , us ] in the two-dimensional (2D) case: shear stress is modelled by means of the degradation of the strength
parameters. The degradation of parameters c and tanϕ depends
w T = [un , us ] (13) linearly on viscoplastic relative shear displacements. This is based
on the slip weakening model introduced by Palmer and Rice (1973).
2.3.1.1. Elastic behaviour. The elastic behaviour of the joint relates In this way the cohesion decays from the initial value c0 to zero and
stresses (  and ) to displacements (un and us ) through the nor- the tangent of friction angle decays from the peak (intact material)
mal (Kn ) and tangential stiffness (Ks ), respectively. Normal stiffness to the residual value as a function of a critical viscoplastic shear
depends on the opening of the joint, as indicated in Fig. 3 and in displacements (u* ). Two different values of u* are used to define
the following equations: the decrease of cohesion (u* c ) and friction angle (u* tanϕ ) (see Fig. 4b
     and c). The mathematical expressions are
un 1/Kn 0 
= (14)
 vp

us
us 0 1/Ks  c  = c0 1− (17)
u∗c
m
Kn = (15)
a − amin where c is the effective cohesion which corresponds to the vis-
vp
coplastic shear displacement us , c0 is the initial value of the
where m is a parameter of the model; a is the opening of the element ∗
effective cohesion, and uc is the critical value of shear displacement
and amin is the minimum opening of the element (at this opening,
for which the value of c is zero. Also,
the element is closed; see Fig. 3).
vp
us
tan ϕ = tan ϕ0 − (tan ϕ0 − tan ϕres

) (18)
2.3.1.2. Viscoplastic behaviour. The viscoplastic behaviour of the u∗ϕ
joint was developed taking into account the formulations proposed
by Gens et al. (1990) and Carol et al. (1997) for rock joints. According where tanϕ corresponds to viscoplastic shear displacement us ,
vp
to these theories, it is necessary to define a yield surface, a plastic tan ϕ0 is the tangent of the peak friction angle, tan ϕres
 is the tangent
potential and a softening law. of the internal friction effective residual angle, and u∗ϕ is the critical
Viscoplastic displacements occur when the stress state of the value of shear displacement when the value of tanϕ is equal to
joint reaches a yielding condition. This condition depends on a pre-  .
tanϕres
viously defined yield surface. In this study a hyperbolic yield surface
(Fig. 4) based on Gens et al. (1990) was adopted:
2.3.1.4. Viscoplastic displacements. If F < 0, the stress state of the
F ≡  2 − (c  −   tan ϕ )
2
(16) joint element is inside the elastic region; if F ≥ 0, the displacements
204 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213

of the joint element have a viscoplastic component. The viscoplastic

displacements are calculated by

dẇ
vp F ∂G
= <
> (19)
dt F0 ∂
where G is a plastic potential, and is a viscosity parameter. In
order to ensure that there is no viscoplastic flow inside the yield
locus, the following consistency conditions should be met:
F ⎫
<
≥ 0 (if F < 0) ⎪
⎬
F0
F ⎪
(20)
<
≥
(F) (if F ≥ 0) ⎭
F0
where F0 can be any convenient value of F to render the above
expressions non-dimensional. In this study F0 = 1.
vp
The normal and shear viscoplastic displacement rates, un and
vp
us , are given by a power law:
⎫
vp
u̇n = F N
∂G ⎪
⎬
∂
(21)
vp ∂G ⎪
⎭
u̇s = F N
∂
where N is the exponent of the power law.

2.3.1.5. Plastic potential surface and dilatancy. The associated rule

allows the calculation of displacements directions. The derivative
of G with respect to stresses includes the parameters fdil and fcdil ,
which take into account the dilatant behaviour of the joint under
shear stresses (López, 1999):
∂G   T
= [2 tan ϕ c  −   tan ϕ fdil fcdil , 2] (22)
∂
where fdil accounts for the decrease of dilatancy with the level of
the normal stress acting on the joint, and fcdil defines the degrada-
tion of the joint surfaces due to shear displacements. The following
expressions describe these effects:
     
     
f␴dil = 1− exp −ˇd (23)
qu qu

c
fcdil = (24)
c0

where qu is the compression strength of the material for which

dilatancy vanishes, ˇd is a model parameter, and c is the cohesion
vp
value for the viscoplastic shear displacement us (Fig. 4b).

2.3.2. Hydraulic model

The transversal advective flux flow through the joint is cal-
culated by means of a transversal intrinsic permeability and the
pressure drop between joint surfaces (Segura, 2008). Furthermore,
the longitudinal advective flow is calculated using a longitudinal
intrinsic permeability and a generalized Darcy’s law. Therefore,
it is necessary to define the longitudinal and transversal intrinsic
permeabilities of joints. Likewise, in the case of joints under unsat-
urated conditions, the water retention curve should be specified.

2.3.2.1. Advective fluxes. The transversal flux is calculated as

Fig. 4. (a) Evolution of the failure surface due to softening of the strength parame- klt krlt
ters. (b) Softening law for cohesion. (c) Softening and law for tanϕ . qlt = pmp (25)
l
where klt is the transversal intrinsic permeability for the liquid,
krlt is the transversal relative permeability for the liquid, l is

the dynamic viscosity of the liquids and pmp is the pressure drop
between the two surfaces of the joint element.
 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213 205

The generalized Darcy’s law for the longitudinal flow reads: Also, P is scaled with surface tension, , if temperature effects
  are considered:
qll = −
kll krll ∂pmp
− g (26) 
l ∂l k10 
P = P0  (33)
k 0
1
where kll is the longitudinal intrinsic permeability for the liquid,
krll is the longitudinal relative permeability for the liquid, g is the
gravity vector, and pmp is the liquid pressure in the mid-plane. 2.3.2.5. Relative permeability. The relative permeability is calcu-
lated by
2.3.2.2. Non-advective fluxes (vapour diffusivity). The nonadvective krl = AS1n (34)
flux (vapour diffusivity) is calculated by means of Fick’s law:
where A = 1.0 and n = 3 in cases analyzed here.
∗
iw
g = − g Sg Dg I ∇ ωg
w w
(27)

where * is the tortuosity; Dgw is the molecular diffusion coeffi- 2.3.2.6. Thermal model. The heat conduction is given by Fourier’s
cient, which depends on temperature and gas pressure; and I is the law:
identity matrix.
ic = −∇ T (35)

2.3.2.3. Intrinsic permeability. The longitudinal fluid flow has been where  is the thermal conductivity, and T is the temperature
analyzed as a laminar flow between two smooth and parallel plates gradient.
separated a given hydraulic opening (e). Based on this hypothesis, The thermal conductivity is made dependant on the degree of
the longitudinal hydraulic conductivity of the joint is calculated by saturation of the joint as follows:
means of cubic law:  
 = sat Sl + dry (1 − Sl ) (36)
g e3
Kl = (28)
 12 where sat is the thermal conductivity of the water saturated joint,
and dry is the thermal conductivity of the dry joint.
where  is the fluid density, g is the gravity, and  is the fluid
viscosity.
Then the equation of intrinsic permeability is given by 3. Discretization of equations of stress equilibrium, mass
and energy balance
e2
kll = (29)
12 The discrete form of stress equilibrium relations can be directly
established for the joint element. Then, the integration to average
The hydraulic opening (e) of joints will be related to its geomet-
the residuals provides
rical aperture (a) and to the roughness of joint surfaces (JRC) by
 
means of the law proposed by Barton et al. (1985). Substituting in −I 4
Barton’s expression (Eq. (29)), the longitudinal intrinsic permeabil- N ump r T DrN ump [ −I 4 I 4 ]uj dl
ity can be expressed as I4 lmp

 2  
1 a2 −I 4 p 1
kll = (30) + N ump r T mN mp [ I 2 I 2 ]P j dl − b = 0 (37)
12 JRC 2.5 I4 lmp
2

The transversal intrinsic permeability klt is considered equal to where b is the vector of the external body forces, m is the unit
the value for the continuum media. vector, and Pj is the vector of nodal fluid pressures.
In order to describe the numerical treatment of mass and energy
2.3.2.4. Water retention curve. The degree of saturation of joints balance equations, the water mass balance equation is used as an
is calculated using the standard retention curve proposed by van example. Only terms describing water vapour transfer will be con-
Genuchten Th (1980): sidered. For the remaining mass and energy balance equations, the
 −∗ treatment is identical (see also Olivella et al., 1995).
  1−
1
∗ The weighted residual method is applied to obtain the discrete
Sl = 1+ (31) form of balance equations. The discrete forms of the terms of the
P
equations are given as follows:

where  = Pg − Pl is the current suction, where Pg and Pl are the gas

(1) Storage changes of mass or energy at constant joint opening are
and liquid pressures, respectively; * is a model parameter; and P
written as
is the air entry pressure necessary to desaturate the joints.
   
The air entry pressure of a joint depends on the hydraulic open- I2 w
1 p ∂(l Sl + gw Sgw )
ing as suggested by Olivella and Alonso (2008). If the Laplace N mp adl
2 I2 ∂t
expression for P is combined with Eq. (29), the equation for the lmp
air entry pressure for a given opening is obtained in terms of the    
intrinsic permeability for the reference state (k10 ) and the current 1 I2 p (lw Sl + gw Sgw )
= N mp a dl
value (k1 ): 2 t
I2 lmp

k10 (lw Sl + gw Sgw ) al
P = P0  (32) = [I 4 ] (38)
k1 t 4
206 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213

Fig. 5. (a) Joint specimens of Hwangdeung granite. (b) Discretized geometry and
boundary conditions used for the hydro-mechanical simulation of the test per-
formed by Lee and Cho (2002).

(2) Storage change induced by changes of joint opening can be

written as

   
1 I2 pT du
N mp (lw Sl + gw Sg )mT rN ump [ −I 4 I4 ] dl
2 I2 lmp
dt
   
1 I2 pT u
= N mp (lw Sl + gw Sg )mT rN ump dl [ −I 4 I4 ]
2 I2 lmp
t
(39)

(3) For advective fluxes, the nodal liquid pressures Plj are differ-
entiated to provide the mid-plane values of the joint pressure
drop:
Fig. 6. Comparison between experimental results obtained by Lee and Cho (2002),
and results from numerical simulations. (a) Shear stress-shear displacement curves;
p (b) normal displacement vs. shear displacement, and (c) intrinsic permeability vs.
pmp = N mp [ I 2 −I 2 ]P lj (40)
shear displacement.
 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213 207

(4) The discretized expression for the transversal advective flux is Table 1
Hydro-mechanical parameters of granite matrix and joint used in the numerical
   
model.
I2 pT Mechanical parameters Symbol Units Value
N mp lw qlt dl
−I 2 lmp Young’s modulus E MPa 54,100
     
Poisson’s ratio  0.29
I2 w Porosity n0 % 49.0
pT krlt l p
= N mp klt N mp dl [I2 − I2 ]P lj (41) Hydraulic parameters
−I 2 lmp
l
Intrinsic permeability k m2 1 × 10–16

(5) The liquid pressure at the mid-plane is calculated by averaging

the nodal liquid pressures:

p 1
pmp = N mp [ I 2 I 2 ]P lj (42)
2
(6) The discretized expression for the longitudinal advective flux
is
     
1 I2 pT
∂N mp kll krll
p
∂pmp
− l g dl
2 I2 lmp
∂l l ∂l
     
1 I2 ∂N mp
pT
krll lw ∂N mp
p
1
= kll dl [ I2 I 2 ] P lj
2 I2 lmp
∂l l ∂l 2
     
1 I2 ∂N mp
pT
krll lw l
+ kll gdl (43)
2 I2 lmp
∂l l
Fig. 7. Scheme of the direct shear device.

The discretized equations for the non-advective fluxes and heat

conduction are analogous to the equations for advective fluxes 4.1. Geometry and material parameters of the model
given above.
The tests described above were modelled using the coupled
hydro-mechanical formulation described before, implemented
4. Numerical simulation of a hydraulic shear test on rough
into the finite element program Code Bright (DIT-UPC, 2000). The
granite surfaces
model is 120 mm high and 110 mm wide (Fig. 5b). The rock matrix
was discretized using 200 quadrilateral continuum elements hav-
The hydraulic shear tests selected to check the capabilities of
ing 4 nodes, and the joint was discretized by means of 10 joint
the model were performed on granitic rock from Korea (Lee and
elements. The normal stress is applied at the boundary AB, while
Cho, 2002). An intact rock block was sawed to obtain samples with
shear displacements are applied at boundaries AC and BD. Bound-
a length of 160 mm and equal values of width and height (120 mm)
aries EG and FH are horizontally fixed and GH is vertically fixed. The
(Fig. 5a). The fracture surfaces were created by means of a tensile
water injection pressure (Pl ) on the joint was applied at boundary
fracture exerted by a splitter. The fracture opening was measured
CE, while at boundary DF a drainage boundary condition was con-
using a 3D (three-dimensional) laser profilometer. The mean value
sidered. The pressure at boundary CE was increased when the shear
of the opening was 0.65 mm.
displacement increased 1 mm, as done in the real test. The joint is
Shear hydraulic tests were performed maintaining constant
considered to be saturated (Sl = 1) during the test.
normal stresses of 1, 2 and 3 MPa. The tangential displacement was
A linear elastic constitutive law was used to simulate the
applied at a rate of 0.05–0.08 mm/s. The hydraulic pressure applied
mechanical behaviour of the rock matrix and the intrinsic
to the joint varied from 4.91 kPa to 19.64 kPa. For each stage of shear
displacement of about 1 mm, hydraulic pressure was kept constant.
When the fluid flow reached a steady state, the mean flow rate
was calculated recording the amount of outflow measured for a
period of 2 min. These measurements were also used to calculate
the permeability of the joint.
The shear behaviour of the rock joint is shown in Fig. 6a and b.
The results obtained are characterized by a peak shear strength and
a pronounced dilation that greatly affected the hydraulic behaviour
of the rough fractures. Dilatancy increases rapidly before shear
stress reaches its peak value. Then, dilatancy increases at a lower
rate during the shear stress drop to reach residual values.
The permeability changes with respect to the increments of
shear displacements are plotted in Fig. 6c. The fracture permeabil-
ity changes slightly during the initial stage of shear loading. But, as
dilation occurs close to the peak strength, permeability increases
dramatically, about 2 orders of magnitude. When shear displace-
ments reach 7 mm, permeability becomes constant. Fig. 8. Example of carved rock joint.
208 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213

Fig. 9. Comparison of shear stress vs. shear displacements and normal displacements vs. shear displacements for experimental tests and simulation results (˛a = 15◦ ).

permeability was considered constant during the test. The param-
eters adopted for the granite matrix are summarized in Table 1.
The mechanical behaviour for the rock joint was modelled
using the elasto-viscoplastic constitutive laws described before.
The longitudinal permeability changes during the test according
to the joint opening. The parameters for rock joint are indicated in
Table 2.
4.2. Numerical results against test data
The results obtained from the simulations are compared with
the tests results in Fig. 6. The mechanical behaviour of the joint
is closely reproduced by the model. The numerical formulation is
able to reproduce the increment of peak shear stress with nor-
mal stresses. Also, it is possible to capture how the shear strength
 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213 209

Table 2
Hydro-mechanical parameters of the joint model used in calculations.

Mechanical parameters Initial normal stiffness Tangential stiffness (Ks Initial cohesion (c 0 Initial friction angle (ϕ0 Residual friction angle
parameter m (MPa) (MPa/m)) (MPa)) (◦ )) (ϕres (◦ ))

Value 90 1500 0.02 47 37◦

Mechanical parameters Initial opening Minimum opening Viscosity parameter Stress power Critical displacement for cohesion
a0 (mm) amin (mm) (s−1 ) N uc * (mm)

Value 0.65 0.065 10−4 2.0 15.0

Mechanical parameters Critical displacement for tanϕ Uniaxial compressive strength Model parameter Joint roughness coefficient
(mm) qu (MPa) ˇd JRC

15.0 151 40 2.70

Hydraulic parameters Hydraulic opening Longitudinal intrinsic permeability Transversal intrinsic permeability
e (mm) kl (m2 ) kt (m2 )

Value 0.035 10−8 10−16

decreases with displacements. Fig. 6b also compares the measured roughnesses. Fig. 8 shows one of the rock joints tested. To obtain
and calculated dilatancy of the joint. the topographical data of rock fracture surfaces, 2D laser-scanning
The evolution of the intrinsic permeability of the joints is also profiles of both sides of joints were measured before and after the
simulated. Even though the permeability in the model increases shear test.
continuously with dilatancy, the permeability measured in the test Prior to shearing, each sample was equilibrated at the required
for different normal stresses became constant and was indepen- suction. Samples were placed in a desiccator with a solution, whose
dent of normal stress. This is mainly caused by the gouge materials concentration is known, at a constant temperature of 20 ◦ C. The
generated from the degradation of asperities during shearing. This equilibrium was considered complete when there was no mea-
phenomenon is not considered in the model (Fig. 6c). surable change in the weight of samples. Samples weights were
measured and the total suction was measured on small samples
of rock using a dew-point psychrometer. A small sample of rock
5. Direct shear tests on rock joints with suction control was placed in the desiccator together with the joints samples and
its suction was used as a standard average value. It was assumed
A suction-controlled direct shear cell using the principle of axis that the suction measured with the psychrometer is the suction of
translation (Escario and Saez, 1986) was modified and updated to the joint. Samples reached the equalization stage after a period of
perform the tests reported here (Fig. 7). fifteen days.
The air overpressure chamber was connected to a vapour circu- Once equilibrated at a given suction, samples were transferred
lation system with the objective of controlling the relative humidity to the shear cell and the vapour control system was put into oper-
(RH) of the sample. Sensors were installed to measure the tem- ation. When the RH measured by the sensor reached the constant
perature and RH of the air within the chamber. A modern data
acquisition system was attached to the cell and a digital program
was developed, using LabVIEW programming language, for data
acquisition.
The rock tested (Lilla claystone) is a sulphate-bearing argilla-
ceous rock located in the Lower Ebro Basin, in the northeast of
Spain. The characterization of the rock was taken from previ-
ous works performed by Garcia-Castellanos et al. (2003), Berdugo
(2007), Tarragó (2005), and Pineda et al. (2010). These sulphated
rocks formed during the Tertiary Period range from Early Eocene
to Late Miocene in age. The clay matrix has a low plasticity. The
porosity varies from 0.09 to 0.11. The Young’s modulus E0 varies
from 26.5 GPa to 28.5 GPa and the shear stiffness G0 varies from
11 GPa to 12.5 GPa.
The testing methodology consists in: (1) preparing the samples
by carving joints with different geometric angles; (2) measuring
the profile of the joint wall surface; (3) applying a wetting or a dry-
ing cycle on the samples using vapour equilibrium technique; (4)
performing the direct shear test with suction control; and (5) mea-
suring the profile of the joints surface after the test. The samples
were extracted from a borehole core of Lilla claystone drilled from
the floor of Lilla tunnel. The core had 110 mm in diameter and a
length of 1 m. The core was cut into pieces with a nominal length
of 50 mm. Then, these pieces were drilled and cut in a machine to
obtain samples 50 mm in diameter and 12 mm in height. Finally,
the joints were carved with a diamond drill in order to create regu-
lar geometric asperities having “opening” angles of 0◦ , 5◦ , 15◦ , 30◦ ,
and 45◦ , respectively. The intention was to test different asperity Fig. 10. Relationship between Wg /Wi and total work.
210 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213
Fig. 11. Peak shear stresses vs. net normal stresses for different values of ˛a . The associated parameters of the yield surface are also given.
value of suction required, the test began by applying the vertical
load. Three different values of air pressure (net normal stress) were
applied: 30, 60, and 150 kPa, respectively. When the vertical dis-
placements induced by the vertical stress remained constant, shear
displacements were applied at a rate of 0.05 mm/min. The shear
test ended when the shear stress reached its residual value. An
example of recorded results for an asperity angle of 15◦ is given in
Fig. 9.
The recorded plots of shear stress versus shear displacement
show that the shear strength of joints depends on the three vari-
ables, namely normal stress, suction, and joint roughness angle.
The effect of the normal stress is well known: the larger the normal
stress, the higher the shear strength.
The value of suction imposed also affects the peak and resid-
ual shear strengths. Increasing suction results in higher values of
peak shear strength. However, the effect of suction on the residual
strength is not seen as clearly as that on the peak strength. Residual
strength depends not only on suction, but also on the degradation
of the roughness of asperities. Degradation of asperities is influ-
enced by suction and by the irregular matching due to defects of
joint construction and the heterogeneity of the rock. This implies
the existence of contact areas with higher or lower strength. These
phenomena result in some heterogeneity of results.
Increasing the asperity roughness is associated with higher
strength. Furthermore, the roughness also affects the strength
softening of the joint. In joints with higher roughness, the residual
strength is reached for smaller displacements. The flat joint
exhibited a ductile behaviour; in these cases, the softening effect
is negligible.
It was generally observed that increasing normal stress resulted
in lower dilatancy. However this behaviour was not always
recorded. This is the case for ˛a = 15◦ . It is believed that this anoma-
lous behaviour is due to some irregularity of matching and probably
a consequence of the heterogeneity of the rock, which influences
the degradation of asperities. The influence of suction on dila-
tancy is also apparent in plots. Joints equilibrated at low suction
( = 20 MPa, RH = 86%) exhibited the lowest dilatancy.
Suction increases strength. It was observed that the sliding of the
joint walls, one over the other, occurs without breakage of asperi-
ties when suction is high. Even if breakage occurs at a given suction,
it is likely that the gouge material equilibrated at high suctions is
capable of rolling on the joint surface. The gouge material at lower
suctions is easier to crush without rolling. These phenomena are
capable of explaining the recorded effect of suction on strength
and dilatancy.
Since the global damage of a rough joint may be a consequence
of the work spent in shearing the joint, there was an interest in
relating a measure of the joint damage to the irreversible (plastic)
work induced by external stress. Joint damage was defined by an
index relating the weight of the gouge material (Wg ) generated in
a sheared joint to the initial weight of the sample tested (Wi ). The
ratio Wg /Wi is plotted in Fig. 10 against the total work (shear plus
volumetric strain). The plot shows that suction is also a controlling
factor not fully accounted for by the work. The trend lines plotted in
 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213 211

Fig. 12. (a) Effective cohesion vs. ˛a (above) and effective cohesion vs. suction (below). (b) Effective tangent of internal friction angle vs. ˛a (above) and effective tangent of
internal friction angle vs. suction (below).

Fig. 10 (dash blue, green and red lines) indicate that the degradation 
where tan ϕ0( is the tangent of the effective initial angle of
,˛a )
of joints increases with the work exerted, in all cases. In addition, internal friction; is the total suction; t0 is the value of tan ϕ0 for
the higher is the suction, for a given value of total work, the lower = 0 and ˛a = 0◦ ; t1 is the slope of tan ϕ0 vs. suction line for ˛a = 0◦ ;
is the joint degradation. d0 and d1 are model parameters which control the increment of
tan ϕ0 with suction for ˛a = 5◦ –45◦ .
6. Influence of suction and roughness on yield function Fig. 12a and b shows the fitting of the experimental values of
parameters 
c0( 
and tan ϕ0( with the equations previously proposed.
,˛ ) a ,˛ ) a

The parameters and c0 tan ϕ0 ,

which define the yield surfaces 7. Numerical simulation of direct shear tests of joints of
(Eqs. (17and 18)) were obtained for each test by plotting the max- Lilla claystone
imum shear stress measured against net normal stress (Fig. 11).
Values of c0 and tan ϕ0 were plotted against roughness angle The numerical simulation of the shear stress tests was carried
(˛a ) and suction (Fig. 12a and b). Fig. 12a (above) shows that mini- out with the help of Code Bright using the joint element developed
mum values of cohesion are obtained for ˛a = 0◦ . Cohesion increases and the new mechanical constitutive law proposed before.
up to ˛a = 15◦ , but remains essentially constant for others values of 
The values of c0( 
and tan ϕ0( incorporated the strain-
˛a . Fig. 12a (below) shows a linear increment of c0 with suction. ,˛ ) ,˛ )
a a
softening law. This allows considering the effect of suction and
A mathematical expression is proposed for c0 taking into
roughness angle in the softening of the joint.
account the effect of suction and asperity roughness angle:
The parameters fdil and fcdil , which control the dilatant

c0( ,˛a )
= (c0 + c1 ) + (b0 + b1 )[1 − exp(−b2 tan ˛a )] (44) behaviour of the joint with shear stresses, were modified through

the following expressions:
where c0( ,˛a )
is the effective initial cohesion; is the total suction;
  1
 

c0 is the cohesion for = 0 and ˛a = 0◦ ; c1 is the slope of c0 vs. suction fdil = d tan ˛a 1 − exp −ˇd (46)
qu qu
line for ˛a = 0◦ ; b0 is the average value of c0 for ˛a = 15◦ –45◦ ; b1 is
a parameter of the model that controls the increment of cohesion c( ,˛a )
with  for ˛a = 5◦ –45◦ ; and b2 is a parameter of the model. fcdil =  (47)
c0( ,˛a )
Fig. 12b shows that tan ϕ0 increases with ˛a and suction. The √
increment with respect to ˛a is considered to be dependent on where d and ˇd are model parameters, and the term tan ˛a con-
tan˛a , and the increment is made linearly dependent on suction. siders the influence of roughness on dilatancy.
The equation proposed for tan ϕ0 is Then, the amount of dilatancy depends on the level of the normal
 stress, on the roughness of the joint surfaces and on the degradation
tan ϕ0( = (t0 + t1 ) + (d0 + d1 ) tan ˛a (45)
,˛a ) of the interface surface, which varies with suction.
212 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213
Table 3
Material parameters.
Rock matrix Rock joints
Young’s Poisson’s Initial normal Tangential Initial
modulus ratio () stiffness stiffness (Ks friction
E (MPa) parameter (MPa/m)) angle (ϕ0
(m (MPa)) (◦ ))
27,000 0.29 100 500 35◦
Fig. 13. Finite element mesh geometry used to numerically simulate the experi-
mental results.
The geometry of model is shown in Fig. 13. The rock was
assumed to be an elastic material and the joint was modelled
as a viscoplastic joint element. The joint is discretized using 10
elements (in red). The rate of displacements used in the test
(0.05 mm/min) is applied on boundaries AC and BD. Boundaries EG
and FH are horizontally fixed and boundary GH is vertically fixed.
The net normal stresses used in the test (30, 60, and 150 kPa) are
applied on boundary AB. The initial liquid pressures are −20, −100,
and −200 MPa which are the values of the applied suction.
All the simulations were performed with the same parameters,
except that the critical values of shear displacements u∗c and u∗ϕ
were changed according to the strength softening of shear stress
and dilatancy of the joints. The parameters are listed in Table 3. The
predictions of the numerical analysis for an asperity angle of 15◦
are plotted in Fig. 9.
8. Conclusions
A coupled THM formulation for a joint element was proposed
and implemented in the finite element program Code Bright.
A mechanical constitutive law considering the elastic and plastic
displacements of the joint is adopted to describe the stress-
displacement behaviour of the joint. In the elastic law, the normal
stiffness depends on the evolution of the joint element open-
ing. Plastic behaviour is defined by a hyperbolic yield surface
and softening is based on a slip weakening model. The equations
theoretically developed were transformed into a viscoplastic for-
mulation.
Darcy’s law was adopted for the longitudinal hydraulic consti-
tutive law. However, the transversal flux is calculated proportional
to pressure drop between joint surfaces (Segura, 2008). A reten-
tion curve with an air pressure entry dependant on joint aperture
(Olivella and Alonso, 2008) is adopted to calculate the degree of sat-
uration of the joint. The vapour diffusivity is calculated by Fick’s law
and the heat conduction through the joint is obtained by Fourier’s
law.
A numerical simulation of rough rock joints under shear stress
subjected to forced flow along the joint was carried out to validate
the numerical tool. The comparison between test and numerical
Residual Initial Minimum Viscosity Uniaxial Model Model
friction opening (a0 opening ( (s−1 )) compressive parameter parameter
angle (ϕres (mm)) (amin strength (qu (MPa)) (d )
(◦ )) (mm)) (N)
8◦ 0.1 0.01 10−2 2.0 20 0.3
results was positive and it was concluded that the formulation is
able to reproduce the main characteristics of coupled mechanical-
flow joint behaviour. Shear stress softening and dilatancy with
shear displacements as well as the increments of permeability with
displacement was well captured.
The influence of suction on joint behaviour was also experimen-
tally investigated. It is believed that this is an important issue in
applications. No reference of this effect, which was found to be
very significant in the rock tested, was found in the literature.
An available direct shear device was successfully modified to
test rock joints under controlled RH of the specimens. Modifica-
tions included the addition of a vapour circulation system and the
improvement of the acquisition data incorporating an analogue
data acquisition device.
The carving process adopted to build different asperity angles
allowed exploring the roughness effects on the shear strength and
the dilatancy of joints.
The shear test results showed the marked dependency of peak
shear stress and dilatancy on suction and roughness. The shear
strength and dilatancy decrease when suction decreases. However,
the dependency of residual strength with suction was not so clear.
Comparing the shear strength and dilatancy recorded for different
roughnesses, it was noted that greater roughness implied greater
shear strength as expected. It was also observed that a rougher
asperity results in smaller values of displacements necessary to
reach residual strength. In other words, rougher joints are more
brittle. This brittle behaviour induces higher damage on the joints
surfaces, and this damage extends to the whole joint surfaces. A
consequence of this phenomenon is that rougher surfaces exhibit
a lower dilatancy.
It was also obtained that the degradation of joints increases with
the applied work in all cases. However, the additional effect of suc-
tion should be considered an independent contribution. The higher
is the suction, for a given value of total work, the lower is the joint
degradation.
New mathematical expressions for the strength parameters, the
initial effective cohesion (c0 ), and the initial effective tangent of
internal friction angle (tanϕ0 ) of the asymptote of the hyperbolic
yield surface are proposed. These expressions consider the effects
of suction and asperity roughness on strength parameters. Also,
the dilatancy parameters were modified taking into account suc-
tion and geometry of joints. Both modifications were introduced
in the constitutive law of the interface element implemented in
Code Bright.
The numerical simulation performed reproduces in a satisfac-
tory manner the experiments run on rock joints of Lilla claystone.
References
Barton N, Bandis S, Bakhtar K. Strength, deformation and conductivity coupling of
rock joints. International Journal of Rock Mechanics and Mining Science and
Geomechanical Abstracts 1985;22(3):121–40.
Berdugo IR. Tunnelling in sulphate-bearing rocks expansive phenomena. Barcelona:
Department of Geotechnical Engineering and Geosciences; 2007 [PhD thesis,
UPC].
 E.E. Alonso et al. / Journal of Rock Mechanics and Geotechnical Engineering 5 (2013) 200–213
Carol I, Prat P, López CM. A normal/shear cracking model. Application to discrete
crack analysis. ASCE Journal of Engineering Mechanics 1997;123(8):765–73.
DIT-UP. C. Code Bright: a 3D program for thermo-hydro-mechanical analysis in
geological media. In: User’s guide. Barcelona: Centro Internacional de Métodos
Numéricos en Ingeniería (CIMNE); 2000.
Escario V, Saez J. The shear strength of partly saturated soils. Géotechnique
1986;36(3):453–6.
Fredlund DG, Rahardjo H. Soil mechanics for unsaturated soils. New York: Wiley
and Sons, Inc; 1993.
Garcia-Castellanos D, Vergés J, Gaspar-Escribano J, Cloetingh S. Interplay between
tectonics, climate, and fluvial transport during the Cenozonic evolution of the
Ebro Basin (NE Iberia). Journal of Geophysical Research 2003;108 (B7): ETG 8-
1-8-18.
Gens A, Carol I, Alonso EE. A constitutive model for rock joints; formulation and
numerical implementation. Computers and Geotechnics 1990;9(1–2):3–20.
Goodman RE, Taylor RL, Brekke TL. A model for the mechanics of jointed rock.
ASCE Journal of the Soil Mechanics and Foundations Division 1968;94 (SM3):
637–59.
Guiducci C, Pellegrino A, Radu JP, Collin F, Charlier R. Numerical modelling of
hydro-mechanical fracture behaviour. In: Pande, Pietruszczak, editors. Numer-
ical models in geomechanis-NUMOG VIII. Lisse: Swets and Zeitlinger; 2002. p.
293–9.
Hans J. Etude expérimental et modélisation numérique multiéchelle du compor-
tamente hydromécanique de répliques de joints rocheux. Université Joseph
Fourier: Grenoble; 2002 [Thése de Doctorant].
Indraratna B, Ranjith PG, Price JR, Gale W. Two-phase (air and water) flow through
rock joints: analytical and experimental study. Journal of Geotechnical and
Geoenvironmental Engineering 2003;129(10):918–28.
Jennings JE, Robertson AM. The stability of slopes cut into natural rock. In: Confer-
ence on Soil Mechanics and Foundation Engineering; 1969. p. 585–90.
Lee HS, Cho TF. Hydraulic characteristics of rough fractures in linear flow under nor-
mal and shear load. Rock Mechanics and Rock Engineering 2002;35(4):229–318.
López CM. Análisis microestructural de la fractura del hormigón utilizando ele-
mentos finitos tipo junta. Aplicación a diferentes hormigones. Barcelona:
Department of Geotechnical Engineering and Geosciences; 1999 [PhD thesis,
UPC].
Oldecop L, Alonso EE. A model for rockfill compressibility. Géotechnique
2001;51(2):127–39.
Olivella S, Gens A, Carrera J, Alonso EE. Numerical formulation for a simulator
(CODE BRIGHT) for the coupled analysis of saline media. Engineering Compu-
tations 1995;13(7):87–112.
Olivella S, Alonso EE. Gas flow through clay barriers. Géotechnique
2008;58(3):157–76.
Olsson R, Barton N. An improved model for hydromechanical coupling during shear-
ing of rock joints. International Journal of Rock Mechanics and Mining Sciences
2001;38(3):317–29.
213
Palmer AC, Rice JR. The growth of slip surfaces in the progressive failure of
over-consolidated clay. Proceedings of the Royal Society of London, Series A,
Mathematical and Physical Sciences 1973;332:527–48.
Perzyna P. The constitutive equations for rate sensitive plastic materials. Quarterly
of Applied Mathematics 1963;20:321–32.
Pineda JA, De Gracia M, Romero E. Degradation of partially saturated argillaceous
rocks: influence on the stability of geotechnical structures. In: Buzzi, Fityus,
Sheng, editors. 4th Asia-Pacific conference on unsaturated soils. Taylor and
Francis Group; 2010. p. 449–54.
Romero EE. Controlled suction techniques. In: Schnaid G, editor. Proc 4 Simposio
Brasileiro de Sols Nao Saturados; 2001. p. 535–42.
Segura JM. Coupled HM analysis using zero-thickness interface elements with
double nodes. Barcelona: Department of Geotechnical Engineering and Geo-
sciences; 2008 [PhD Thesis, UPC].
Tarragó D. Degradación mecánica de argilitas sulfatadas y su efecto sobre la
expansividad. Barcelona: Universitat Politécnica de Catalunya; 2005 [BSc dis-
sertation].
van Genuchten Th M. A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils. Soil Science Society of America Journal
1980;49(9):892–8.
Zienkiewicz OC, Cormeau IC. Visco-plasticity-plasticity and creep in elastic solids-a
unified numerical solution approach. International Journal for Numerical Meth-
ods in Engineering 1974;8(4):821–45.
E.E. Alonso, born in 1947, got his degree in Civil Engineer-
ing (Ingeniero de Caminos, Canales y Puertos) in Madrid in
June 1969. He obtained a PhD in Northwestern University
in 1973. At present, he is a Professor of Geotechnical Engi-
neering at the UPC in Barcelona. He is the author of more
than 300 papers published in Proceedings of Conferences
and learned journals. Professional activities include foun-
dation problems, deep excavations, nuclear power plants,
slope stability, breakwaters, earth dams, tunnelling and
underground waste disposal. Awards: Thomas Telford
medal (Institution of Civil Engineers, London – ICE) in
1994 and 2007; J. Torán Prize in 1995; N. Monturiol medal
(Barcelona) in 2000; Second Coulomb lecturer (French
Committee on Soil Mechanics and Geotechnical Engineering) in 2003, Eleventh
Buchanan lecturer (Texas A&M, USA) in 2003, Tenth Sowers lecturer (Georgia Insti-
tute of Technology, USA) in 2007, Crampton Prize (ICE) in 2006, Geotechnical
Research Medal (ICE) in 2009 and 2010. He is a member of the Royal Academy
of Engineering of Spain since 1995 and member of the Royal Academy of Sciences
and Art of Barcelona since 2007. He is co-author of the books “Geomechanics of
Failures” and “Geomechanics of Failures, Advanced Topics”, published by Springer
in 2010.