You are on page 1of 28

INTERNATIONAL JOURNAL FOR NUMERICAL AND ANALYTICAL METHODS IN GEOMECHANICS

Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888


Published online 19 December 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/nag.2341

Numerical analysis of the geotechnical behaviour of energy piles

AIice Di Donna*,† and Lyesse Laloui


1
Laboratory for Soil Mechanics EPFL-ENAC-IIC-LMS, Swiss Federal Institute of Technology, EPFL CH-1015
Lausanne, Switzerland

SUMMARY
Energy geostructures are rapidly gaining acceptance around the world; they represent a renewable and clean
source of energy that can be used for the heating and cooling of buildings and for de-icing of infrastructures.
This technology couples the structural role of geostructures with the energy supply, using the principle of
shallow geothermal energy. The geothermal energy exploitation represents an additional thermal loading,
seasonally cyclic, which is imposed on the soil and the structure itself. Because the primary role of the piles
is the stability of the superstructure, this aspect needs to be ensured even in the presence of the additional
thermal load. The goal of this paper is to numerically investigate the behaviour of energy pile foundations
during heating–cooling cycles. For this purpose, the finite element method is used to simulate both a single
and a group of energy piles. The piles are subjected to a constant mechanical load and a seasonally cyclic
thermal load over several years, imposed in terms of injected–extracted thermal power. The soil and the
pile–soil interface behaviours are reproduced using a thermoelastic-thermoplastic constitutive model. The
thermal-induced stresses inside the piles and the additional displacements of the foundations are discussed.
The group model is used to investigate the interactions between the piles during thermo-mechanical loading.
The presented results are specific to the studied cases but lead to the conclusion that both the thermal-
induced displacements and stresses, despite being acceptable under normal working conditions, deserve to
be taken into account in the geotechnical design of energy piles. Copyright © 2014 John Wiley & Sons, Ltd.

Received 21 July 2014; Revised 27 October 2014; Accepted 29 October 2014

KEY WORDS: energy piles; soil thermo-plasticity; thermal cyclic load; long-term effects; finite element
modelling; pile group effects

1. INTRODUCTION

The use of energy geostructures is increasing in Europe and around the world because they represent a
renewable and clean source of energy that can be used for heating and cooling of buildings,
infrastructures, stations and so on. [1]. This technology couples the structural role of geostructures
with the energy supply, using the principle of shallow geothermal energy. Polyethylene pipes are
embedded into the concrete structures, and a heat-carrying fluid circulates through them and
exchanges heat with the ground. The pipes are then connected to a heat pump system, which
circulates the fluid in the heating–cooling plant of the building. This system allows the heat to be
extracted from the ground during winter to satisfy the heating needs of the buildings and injected
into the ground during summer, to meet air conditioning requirements. The advantage of this
technology is that it incorporates the geothermal equipment inside geostructures that are already in
place for the stability of the infrastructure, reducing the initial costs of construction with respect to
other geothermal installations. Since the beginning of 1980s, geothermal energy has been
increasingly obtained from foundation elements [2]: at first from base slabs, then from piles (1984),

*Correspondence to: Alice Di Donna, Laboratory for Soil Mechanics EPFL-ENAC-IIC-LMS, Swiss Federal Institute of
Technology, EPFL, Station 18, CH-1015 Lausanne, Switzerland.

E-mail: alicedidonna@teknemaprogetti.it

Copyright © 2014 John Wiley & Sons, Ltd.


862 A. DI DONNA AND L. LALOUI

diaphragm walls (1996) and tunnels. Energy geostructures have been increasingly constructed in
Europe, particularly in Switzerland [3], Austria [4], Germany [5] and England, but also in Australia
[6], Japan [7], China, Scotland [8], the Netherlands [9] and U.S.A. [10]. Because they represent a
new engineering technology and are becoming more and more common, there is a need for
improved scientific knowledge of their behaviour. Several efforts have been devoted to the
investigation and optimization of the energy performance of such structures [11]. With regard to
their mechanical behaviour and geotechnical design, research has been devoted to the in situ
characterization [12–14], numerical analysis [15, 16] and the development of design tools [17]. A
recent state-of-the-art article on the subject can be found in [18]. The goal of the present paper is to
numerically investigate the behaviour of energy piles, focusing on their long-term response to a
seasonally cyclic thermal loading. After a brief state of the knowledge on the subject, the
mathematical formulation employed in this paper to simulate the problem is introduced. Then,
the constitutive model used to reproduce the thermo-mechanical behaviour of the soil is
described, with particular attention given to the thermal cycling aspect. Then two cases are
considered: the first concerns a single energy pile (axisymmetric problem), while the second a
group of energy piles (three-dimensional problem). For both cases, the results are discussed,
focusing mainly on the effects of temperature change on different aspects involved in the
geotechnical design of deep foundations.

2. STATE OF KNOWLEDGE

The most common approaches employed to investigate the thermal deformation of energy piles, the
thermal-induced down drag and axial stresses and the interaction between the pile and the soil are as
follows: (i) physical modelling and (ii) numerical simulations. The former category includes in situ
real scale testing [2, 12–14], model piles at the laboratory scale [19] and centrifuge experiments
[20]. All of them have been fundamental to understand the physical mechanisms of energy piles. On
the other hand, advanced numerical techniques have been developed and validated based on the
acquired experimental knowledge, with the purpose to use them as prediction tools to investigate
further configurations that cannot be tested physically or would be too expensive. Among the
advantages of the numerical techniques, there is the possibility to study the behaviour of entire deep
foundations including large number of piles, the thermal interactions between them, the response
under extreme thermal loading and the long-term performance. In what follows, the main concepts
describing the physical behaviour of energy piles acquired through physical testing are listed,
followed by a brief overview on the numerical results already available in the literature.

2.1. Physical behaviour of energy piles


The thermo-mechanical behaviour of energy piles can be described through the definition of the
following key concepts [12, 14].
2.1.1. Free thermal deformation of a pile. An energy pile is a structural element, usually made of
concrete, which behaves thermo-elastically. Its free response when subjected to a temperature
change can be assimilated to that of a bar, which dilates when heated and contracts when cooled, in
a reversible manner. Its free axial thermal deformation εth
f can be computed according to the theory
of thermo-elasticity as follows:

β′p
f ¼
εth ΔT (1)
3

where β′p is the volumetric axial thermal expansion coefficient of the construction material (i.e. concrete)
and ΔT the temperature variation.
2.1.2. Observed and blocked pile thermal deformation. Piles are not completely free to deform when
heated or cooled, because the surrounding soil, the eventual stiff layer at the tip (end-bearing piles) and

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 863

the superstructure represent a partial constraint. The observed pile axial thermal deformation εth
o is thus
smaller than the free one or eventually null if the pile is completely blocked. Calling εb the blocked
th

axial deformation, the observed one is as follows:

o ¼ εf  εb
εth th th
(2)

2.1.3. Thermal-induced stress. The portion of thermal axial deformation that is prevented generates
additional stresses inside the pile, which are compressive during heating (prevented expansion) and
tensile during cooling (prevented contraction). According to the theory of thermo-elasticity, the
additional thermal-induced axial stresses σth developed inside the pile is as follows:

σth ¼ Ep εth
b (3)

where Ep is the Young’s modulus of the pile.


2.1.4. Degree of freedom of a pile. The degree of freedom nf of the pile is an indicator of the entity of
observed axial deformation (and related additional axial displacements of the superstructure) and of the
thermal-induced axial stresses inside the pile. It is defined as follows:

εth
nf ¼ o
(4)
εth
f

and is equal to 1 when the pile is completely free to deform axially and to 0 when it is completely
blocked. In the first case, the thermal-induced axial displacements are the maximum, and the
thermal-induced axial stresses are null, while in the second case, the opposite occurs.
2.1.5. Null point. The deformation is a local measurement that, in the case of a free bar subjected to a
thermal loading, is constant along the whole length (Equation (1)). The axial displacement of a pile
subjected to a temperature variation at a given point along its shaft can be computed as the
integration of the pile axial deformation over the length, with respect to a fixed point, called null
point. This point is characterized by zero thermal-induced axial displacement. In the case of no or
symmetric constraints, the null point corresponds to the mid-length of the pile. In all other cases, its
positions depend on the external constraints.
2.1.6. Thermal mobilization of the interface shear stress. During axial compressive mechanical load,
the shear stress at the interface is mobilized upwards as the pile moves downwards. During heating, the
portion of pile above the null point moves upwards, mobilizing the interface shear stress in the opposite
direction with respect to the mechanical load, while the portion of pile below the null point will move
downwards, mobilizing the interface shear stress in the same direction as the mechanical load. The
opposite occurs during cooling.
2.1.7. Orders of magnitude. The more critical configuration in terms of additional movements of the
superstructure is the completely free pile with null point at the tip. In this case, there will be no induced
thermal axial stress, while the displacement at the pile head will be the maximum. To have an order of
magnitude,
 it will be about 10mm for a temperature variation of 30 °C and a 30-m length pile made of
concrete β′p ¼ 3:6105 °C1 ; that is, 0.36 mm/°C. Conversely, the most critical condition in terms of
thermal stresses is the one of a completely blocked pile (no thermal-induced axial displacement).
Considering the same thermal loading, the same geometry and a Young’s modulus for the pile of
30 GPa, the maximum additional stress will be about 10 MPa; that is, 333 kPa/°C. Real conditions
are between these two extreme ones. From the available in situ tests [14], it results that the
maximum thermally induced stress is around 100 to 300 kPa/°C and the head displacement around
0.1 to 0.16 mm/°C.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
864 A. DI DONNA AND L. LALOUI

2.2. Numerical methods for the analysis of energy piles


Knowing the physical response of piles when subjected to a thermo-mechanical loading is the first step
towards their proper geotechnical design. This allowed the establishment of the first design
recommendations. Indeed, the currently employed approach is to dimension energy piles assuming
as maximum thermal displacement the one that corresponds to the free thermal condition (Equation
(1)) and as maximum thermal-induced stress the one that corresponds to the fully constrained
condition [21]. This clearly results to be on the side of safety. The step forward into a more precise
and cost-effective geotechnical design of energy piles is represented by the development of
numerical tools able to reproduce the response of such foundations to thermo-mechanical loading
accounting for the specific constraints and boundary conditions of each specific case. Mainly, two
numerical approaches were employed in the past to investigate the behaviour of energy piles: (i) the
methods based on the load-transfer curves theory adapted to thermal loading and (ii) the finite
element method (FEM). The second one is chosen to run the analyses presented in this paper. In the
framework of energy pile foundations, the FEM was firstly employed and validated by Laloui et al.
[15] who compared the numerical results with the experimentally observed behaviour of a real scale
instrumented pile. They employed a fully coupled thermo-hydro-mechanical formulation, and the
experimental results were globally well reproduced. Later on, Silvani et al. [22] went on with
further simulations on an energy pile group, with the purpose to investigate the effects of extreme
temperatures and piles interactions. In terms of long-term predictions of the energy piles behaviour,
the main aspect that needs to be considered is the cyclic nature of the thermal loading.
Suryatriyastuti et al. [23] investigated this aspect numerically for piles installed in sand, by
introducing the temperature variations in terms of corresponding induced deformations and
assuming a cyclic interface constitutive law (degradation). They studied the combined effects of
mechanical and thermal loading on both a single pile and a raft deep foundation. They showed that
cyclic temperature variations in thermo-active piles affect the long-term geotechnical response in
terms of (i) alteration of the head reaction force and head settlement and (ii) alteration of the
mobilized shaft and tip stress. The magnitude of the additional stress produced results to be strongly
related to the entity of the mechanical load. A high mechanical load is an advantage in the first
thermal cycle but becomes risky in the long-term operation of thermal cycles. In what follows,
further analyses are presented, with the main goal of studying the effects of the thermo-mechanical
behaviour of soils and pile–soil interface on the geotechnical performance of energy piles.

3. COUPLED THERMO-HYDRO-MECHANICAL FORMULATION

The considered problem is represented by a concrete piled foundation subjected to a mechanical


vertical load and able to exchange heat with the surrounding ground. Both the concrete and the soil
are considered to be porous materials composed by a solid and a fluid phase. The whole medium is
considered to be fully saturated by water. Hence, three main aspects are involved in the problem:
mechanical, thermal and hydraulic. These three aspects are coupled because the solid volume
variations are affected by the temperature gradient, the heat exchanged depends on the presence of
water flow, the water density varies with thermal loading and the mechanical response of the
materials depends on the pore water pressure (effective stress concept). Hence, a fully coupled
thermo-hydro-mechanical formulation is needed to correctly analyse the problem. The FEM
software LAGAMINE is used for the simulations [24–27]. In this software, the equilibrium and
balance equations, as well as the water and heat diffusion laws, are expressed in the moving current
configuration through a Lagrangian-updated formulation. Details about the definition of tensors and
invariants considered are provided in Appendix A. The equilibrium equation is as follows:

 
div σij þ ρgi ¼ 0 (5)

where div denotes the divergence, σij is the total stress tensor, gi the gravity vector and ρ the bulk
density of the material, which includes the density of water ρw and of solid particles ρs, through the

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 865

porosity n, as ρ = nρw + (1  n)ρs. The Terzaghi formulation for the effective stress (hydro-mechanical
coupling) is introduced, so that the equilibrium equation becomes
 
div σ ′ij þ ∇pw þ ½npw þ ð1  nÞρs gi ¼ 0 (6)

where ∇ represents the gradient, pw the pore water pressure and σ′ij the effective stress tensor that can be
written in incremental form by introducing the constitutive law (see the succeeding texts). The mass
conservation equation reads as follows:
 
1 1   
∂t pw n þ ð1  nÞ þ ∂t T nβ′w þ ð1  nÞβ′s þ div vw;i ¼ 0 (7)
Kw Ks

where ∂t denotes the time derivative, K1w and K1s denote the water and solid skeleton compressibility, T
denotes the temperature, β′w and β′s denote the volumetric thermal expansion coefficients of water and of
the solid skeleton and vw,i denotes the relative velocity of water with respect to the solid. The first two
terms of Equation (7) represent the internal mass variation of the two phases (water and solid) induced
by the changes in the pore water pressure and temperature, respectively (thermo-hydraulic coupling).
The third term represents the exchange of water between the reference volume and the outside. In
the Lagrangian-updated formulation implemented in the code, the mass conservation equation is
verified at each step in the deformed configuration so that the solid mass exchange is null. The
relative velocity of the water with respect to the solid skeleton can be expressed by Darcy’s law:

k
vw;i ¼  ∇ðpw þ ρw gzÞ (8)
ρw g

where k is the hydraulic conductivity (measured in m/s) and z the vertical coordinate. The hydraulic
conductivity k is expressed in terms of the intrinsic permeability kint (measured in m2), which is
temperature independent, as
kint ρw g
k¼ (9)
μw

where μw is the water dynamic viscosity. The hydraulic conductivity’s thermal dependence (thermo-
hydraulic coupling) is represented by the dependency of μw and ρw on temperature, as follows [28]:

μw ¼ 0:6612ðT  229Þ1:562 (10)

 
1
ρw ¼ ρw0 1 þ ∂t pw  β′w ∂t T (11)
Kw

where μw is in pascals per second, T is the absolute temperature in kelvin and ρw0 is the water density
at the reference temperature and pressure. The energy conservation equation reads as follows:
 
ρ^c ∂t T  div λth ∇T þ ρw cp;w vw;i ∇T ¼ 0 (12)

where ρĉ = nρwcp,w + (1  n)ρscp,s is the soil specific heat (including the specific heat for water cp,w and
solid cp,s) and λth ¼ nλth
w þ ð1  nÞλs is the soil thermal conductivity (which includes water λw and
th th

s thermal conductivities). In Equation (12), the first term represents the heat stored in the
solid λth
medium, the second one the heat transferred by conduction (Fourier’s law) and the third one the
heat transferred by convection (thermo-hydraulic coupling). The thermal conductivity and capacity
of the two phases are considered to be temperature independent.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
866 A. DI DONNA AND L. LALOUI

3.1. Constitutive models


In the numerical simulations presented hereafter, the energy piles are made of concrete and behave
thermo-elastically. Concerning the soil, the case of a clay is considered, and it is assumed to behave
according to the Advanced Constitutive Model for Environmental Geomechanics, with temperature
effects included (ACMEG-T model), which is presented hereafter.
3.1.1. Thermal cyclic ACMEG-T model. The model considered in this paper was proposed by Laloui
and Francois [29], and it is known as ACMEG-T. It is a thermoelastic-thermoplastic model belonging
to the Cam Clay family, and thus, it is developed in the framework of the theory of elasto-plasticity and
critical state [30]. The isothermal part of the ACMEG-T model is based on the work of Hujeux [31].
Various successive improvements were made to extend it to non-isothermal monotonic conditions
[32–34]. The extension to the thermal cyclic part is presented in Di Donna and Laloui [35]. The
nonlinear elastic response is described by the following equations for the bulk, Ks and shear, Gs,
moduli:
 ′  ′
p ρ
Ks ¼ Kref ′ and Gs ¼ Gref ′ (13)
pref ρref

where p’ is the mean effective stress, and Kref and Gref are the two moduli at the reference mean
effective stress p’ref. The model is based on the multi-mechanism [31] and bounding surface [36]
theories. According to the former, the yield surface is the combination of two surfaces, or in other
words, the model includes two plastic mechanisms, one isotropic and one deviatoric, which are
coupled through the concept of pre-consolidation pressure (Figure 1). In this paper, for both
mechanisms, an associated flow rule is considered. The isotropic yield limit is as follows:

εp;iso
f iso ¼ p′  p′c riso ¼ 0 with riso ¼ reliso þ v
(14)
c þ εp;iso
v

where p’c is the pre-consolidation pressure, riso the degree of mobilization of plasticity of the isotropic
mechanism, reliso its initial value, c a material parameter and εp;iso
v the volumetric plastic deformation
developed by the activation of the isotropic mechanism. The deviatoric yield limit is as follows:
  
′ dp′ εpd
f dev ¼ q  Mp 1  bIn rdev ¼ 0 rdev ¼ reldev þ (15)
p′c a þ εpd

where q is the deviatoric stress invariant, b and d are two material parameters, rdev is the degree of
mobilization of plasticity for the deviatoric mechanism, reldev its initial value, a a material parameter

Figure 1. Combined effect of strain hardening and thermal softening on the size of the elastic domain at (a)
different temperatures and (b) different amounts of developed plastic volumetric strain.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 867

and εpd the deviatoric plastic deformation. The coefficient M depends on the Lode angle ϑ to account for
the effect of the stress path direction in the π plane, perpendicular to the stress space diagonal [37]. The
model adopts the formulation proposed by Van Eekelen [38], in which the M coefficient is defined as
follows:
pffiffiffiffi
M¼3 3aL ð1 þ bL  sinð3ϑÞÞnL (16)

where aL, bL and nL are three material parameters. The first and second of these parameters depend on
the friction angles in compression and extension, φc and φe, respectively, so that by defining
   
1 2sinφc 1 2sinφe
rc ¼ pffiffiffi and re ¼ pffiffiffi (17)
3 3  sinφc 3 3 þ sinφe
they read as follows:

 1=
rc nL
1
re rc
bL ¼   and aL ¼ with aL > 0; bL nL > 0; 1 < bL < 1 (18)
rc 1=nL ð1 þ bL ÞnL
þ1
re

The value of nL must be assumed to ensure the convexity condition [39]. This dependency was
introduced for the purpose of this work, and details about its implementation and validation are
provided in Appendix B. In this model, the dependency of the pre-consolidation pressure p’c on the
volumetric plastic deformation (strain hardening, horizontal plane in Figure 1a) is described
according to the critical state theory, and its evolution with temperature (thermal softening,
horizontal plane in Figure 1b) is introduced according to the equation proposed by Laloui and
Cekerevac [40]:
 
p T
p′c ¼ p′co eβεv 1  γT In (19)
T0

where p’c0 is the pre-consolidation pressure


 at the initial temperature T0, and before the generation of
any volumetric plastic deformation εpv ¼ 0 , β is the plastic rigidity index and γT is a material
parameter that defines the shape of the isotropic yield function with respect to temperature. The
coupling between thermal softening and strain hardening is represented in Figure 1: for a certain
amount of developed volumetric plastic deformation, the size of the elastic domain decreases with
increasing temperature (thermal softening, Figure 1b), while for a constant temperature, the size of
the elastic domain increases for increasing volumetric plastic strain (strain hardening, Figure 1a).
The global response of the soil under a thermo-mechanical load is a combination of the two effects.
Purely isotropic loading causes only volumetric plastic deformation, while a pure deviatoric loading
causes both deviatoric and volumetric plastic deformations. The flow rules for the volumetric and
the deviatoric plastic strains are as follows:

∂f iso ∂f dev
dεpv ¼ λpiso ′
þ λpdev (20)
∂p ∂p′

∂f dev
dεpd ¼ λpdev (21)
∂q

where λpiso and λpdev are the plastic multipliers for the isotropic and the deviatoric mechanisms,
respectively. Considering the two plastic mechanisms, the consistency equation for this model is as
follows:

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
868 A. DI DONNA AND L. LALOUI

!
∂f l ∂f l ∂f l ∂f l ∂kk p ∂f l
Cij dεj þ β′s;j dT  λp1 ′ þ dT þ λ ¼0 (22)
∂σ ′i ∂σ j ∂T ∂kk ∂dεpi 1 ∂σ ′i

where fl represents the vector containing the two yield surfaces fiso and fdev and λp1 the vector containing
the two plastic multipliers λpiso and λpdev .
The thermal cyclic component was introduced in the ACMEG-T model by the activation of a new
nesting surface at the beginning of each new thermal cycle, according to the theory proposed by Mroz
et al. [41]. From a mathematical viewpoint, riso is re-initialized at each temperature reduction:

p′
riso ¼ rcyc
iso þ (23)
p′c

where rcyc
iso is a material parameter that depends on the tendency of the material to accumulate plastic
deformation during thermal cycles. When heating restarts, riso remains constant as long as the re-
initialized yield surface is reached. Therefore, during re-heating, plasticity starts again at the
temperature T e that verifies the following equation:

" ! # 
′ p e
T p′
f iso ¼ p  p′co eβεv 1  γT In rcyc
iso þ ′ βεp ¼0 (24)
T0 pc0 e v

Once this temperature l is reached, if heating continues, the degree of mobilization of plasticity for
the isotropic mechanism evolves according to

p′ εp;iso;cyc
riso ¼ rcyc
iso þ þ v
(25)
p′c;cool c þ εp;iso;cyc
v

where p′c;cool is the pre-consolidation pressure at the end of the previous cooling phase and εp;iso;cyc
v the
volumetric plastic deformation induced by the isotropic mechanism starting from the last re-heating.
The temperature T, e at which plasticity occurs, increases at each cycle. Consequently, after a certain
number of cycles, it coincides with the maximum temperature imposed, and no further thermal
plastic deformation is developed starting from the next cycle. This reproduces the thermal cyclic
accommodation phenomenon observed experimentally [35].

4. NUMERICAL MODEL OF A SINGLE ENERGY PILE

The geometry and boundary conditions of the energy pile model are represented in Figure 2. The
simulations are run in axisymmetric conditions (half pile), with the axis of symmetry on the left side
of the mesh. The model is made of 1460 quadrilateral elements (including 80 for thermal power
injection) with eight nodes and four integration points and two linear elements with three nodes and
two integration points, for the application of the mechanical load at the pile head. Each node has 4
degrees of freedom: two for the displacements in the two directions, one for pore water pressure and
one for temperature. The total number of nodes is 40 307. The mesh is finer near the zone of
interest, that is, around the pile, and becomes coarser going towards the boundaries. Mesh
sensitivity analyses have shown that the chosen discretization is adequate. The pile has a diameter D
of 0.8 m and a length H of 20 m. The temperature is fixed at the bottom and on the right side, which
are considered to be far enough to not influence the computation, as well as on the top, in order to
take into account the regulated temperature of the building. The pore water pressure is fixed at the
top, bottom and right side of the mesh, while the axis of symmetry is considered impervious, so that
no water flow is allowed through it. The initial temperature of the soil and the piles is 11 °C, and the

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 869

Figure 2. (a) Geometry and boundary conditions of the single energy pile model and (b) zoom on the points
of interest.

initial pore water pressure is assumed hydrostatic with the water table at the surface level. Both the soil
and the concrete are considered as porous materials, and the parameters used for the heat and water
flow problem are given in Table I. The constitutive parameters of the concrete (thermoelastic) are
given in Table II; those of the clay (ACMEG-T model), calibrated on the basis of laboratory testing
on a natural clay [35], are given in Table III. The clay is normally consolidated, has solid particles
unit weight of 27 kN/m3 and earth pressure coefficient at rest K0 of 0.6, computed through the
formulation proposed by Jaky [42] and knowing that the soil has a friction angle of 24°. The pile–
soil interface is modelled by a thin layer of elements, which behave according to the ACMEG-T
model, and the constitutive parameters are the same as those used for the surrounding soil.
Figure 2b is a zoom of the pile area and indicates the points of interest that will be considered in the
analysis of the results. Besides the model initialization, the stress-temperature path includes two
steps: (i) the application of the mechanical load at the pile head and (ii) the application of a
seasonally cyclic thermal load under constant mechanical load. The applied mechanical load is
750 kN. The thermal load is applied in terms of injected and extracted thermal power in the piles,
according to the law presented in Figure 3. During the summer period (first 6 months), heat injection
is increased from zero to the maximum value of 115 W/m of pile length during 1 month, then kept
constant during 3 months, then decreased during 1 month and kept constant for another month.
Similarly, during the winter period, heat extraction is increased from zero to the maximum value of
115 W/m of pile length during 1 month, then kept constant during 3 months, increased during
1 month and kept constant for another month. The same cyclic thermal loading is repeated for
10 years with the purpose to study the long-term response.

Table I. Parameters for heat and water flow.


Soil Concrete
7
k (m/s) 10 109
n () 0.39 0.12
β’w (°C1) 2·104 2·104
λth
w (W/m°C) 0.6 0.6
cpw (J/Kg°C) 4186 4186
λth
s (W/m°C) 2.4 1.7
cps (J/Kg°C) 930 930

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
870 A. DI DONNA AND L. LALOUI

Table II. Mechanical properties for the pile.


Ep (GPa) 35.0
νp () 0.25
γp (kN/m3) 24.0
β′p (°C1) 3.6 ∙ 105

Table III. Constitutive parameters for the clay (Advanced Constitutive Model for
Environmental Geomechanics, with temperature effects included).
Elastic parameters
Reference bulk modulus Kref (MPa) 110.3
Reference shear modulus Gref (MPa) 66.0
Elastic exponent ne () 1
Thermal expansion coefficient β’s (°C1) 1.8 · 105

Plastic parameters

Material parameters a () 0.004


b () 1
c () 0.01
d () 2
Friction angle in compression φ’c (°) 24
Friction angle in extension φ’e (°) 24
Plastic compressibility β () 67
Plastic radius for the isotropic mechanism reliso () 0.01
Plastic radius for the deviatoric mechanism reldev () 0.01
Van Eekelen exponent nL () 0.229
Thermal effect on the isotropic yield limit γT () 0.25
Degree of plasticity under cyclic conditions rcyc
iso () 0.2

Figure 3. Thermal cyclic load law applied during the simulations.

4.1. Initialization and mechanical loading step


After the initialization of the model, the state of soil stress results to be geostatic with water table at the
surface level. The distribution of total vertical stress inside the pile after the initialization (weight),
when half of the load has been applied (375 kN, i.e. 750 kPa) and at the end of the mechanical
loading phase (750 kN, i.e. 1500 kPa) is reported in Figure 4a. It results that the largest part of the
mechanical load is supported by shaft friction and only a small portion is transmitted at the pile tip
(about 100 kPa). The mechanical-induced settlements for the same three stages are presented in

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 871

Figure 4. Application of the mechanical load at the pile head: (a) total stress transmitted inside the concrete
and (b) induced settlement at the pile head.

Figure 4b along the pile length. The final settlement at the pile head is slightly lower than 5 mm, while
the tip settles of about 4 mm. The difference between the two is related to the pile deformation, and it is
consistent with the pile material properties. The main part of the simulation is represented by the
application of a seasonally cyclic thermal loading under constant mechanical loading. In what
follows, the discussion focuses on the thermal field imposed by the pile on the surrounding soil.
This allows the quantification of the applied thermal load and to link it to the induced effects
described in the subsequent part of the paper. The deformation of the soil under cyclic thermal
loading and the consequent displacements of the foundation are then discussed. Then, the state of
stress in the soil, in the piles and at the interface between them is investigated.

4.2. Thermal field


The injection and extraction of power from energy piles generate a heat storage volume around them,
and the dimension of this volume depends on the amount of power injected or extracted, the soil
permeability, the presence of water flow and the ground thermal properties. In the studied case,
there is no imposed water flow, and the soil has a relatively low permeability, so that the ground
represents a heat storage medium. The heat is stored during summer and extracted during winter,
and this preserves the thermal equilibrium of the ground. The evolution of temperature at 10 m
depth in the pile (point A in Figure 2b), at the pile–soil interface (point B in Figure 2b) and at a
distance of 0.8 m (point C in Figure 2b), 1.6 m (point D in Figure 2b) and 2.4 m (point E in
Figure 2b) from the pile shaft is plotted in Figure 5a for the first 2 years of the simulation. This
period is representative also of the temperature variations of the following years. Whereas,
according to the norms [43], the temperature at the interface does not go below zero, the one inside
the pile reaches negative values (point A in Figure 2b). The maximum temperature variation at the
pile–soil interface (point B in Figure 2b) is of about 24 °C, between a maximum of 25 °C during

Figure 5. Thermal field (a) around and (b) below the pile.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
872 A. DI DONNA AND L. LALOUI

Figure 6. Pile temperature.

summer and a minimum of 1 °C during winter. The maximum and minimum temperatures reached
inside the pile are 25 and 2 °C, respectively. The observed temperature variations at the pile’s tip
(point F in Figure 2b) and 0.5, 1.2 and 2.8 m below it (points G, H and I in Figure 2b) are plotted in
Figure 5b. The thermal variation becomes negligible from about 23 m depth; that is, 3 m below the
tip. It is interesting to notice that the thermal load imposed to the soil below the pile tip is less
significant than the one imposed laterally. Figure 6 shows the temperature profiles inside the pile for
the heating and cooling phases of the first, fifth and seventh years. It can be observed that the
temperature is almost uniform along the pile length, except at the extremities, where it must satisfy
the equilibrium with the soil below and the regulated temperature above.

4.3. Soil strain and pile displacement


The thermoelastic-thermoplastic response of the soil around the energy pile plays a role as responsible
of additional displacements of the superstructure during thermal cycles. The total volumetric strain at
the points B, C, D and E (Figure 2b) is represented in Figure 7 (positive is compression). The first
heating phase results in a plastic deformation (irreversible compression) that is more evident for
point B that is subjected to the most important temperature variation, less significant but still present
for points C and D and negligible for point E that is the extremity of the thermal field and thus
almost not subjected to thermal loading. Only the soil close to the pile (point B) shows the cyclic
accommodation effect because for the others the thermal loading is not high enough to activate it.
The cooling branches of the curves of the points C, D and E have a different slope with respect to
that of B, because the response of point B is affected by the presence of the pile (pile–soil
interaction), while the other points respond according to the soil’s thermal expansion coefficient. In

Figure 7. Evolution of volumetric total strain for different points of the soil domain.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 873

Figure 8. Settlements in the soil at different depths at a distance of (a) 2.4 and (b) 0.8 m from the pile shaft.

the view of the geotechnical design of energy piles, the interest in studying the soil deformation in the
surroundings on the foundation lies in the need to verify that the displacements of the superstructure
remain inside admissible limits to guarantee the required comfort and performance. With this in
mind, the displacements induced by the thermo-mechanical loading of the modelled energy pile are
studied here, starting from those of the soil at the extremity of the thermal field up to those of the
pile itself. The vertical displacements (positive downwards) in the soil at 5, 10 and 15 m depths at a
distance of 2.4 and 0.8 m from the pile shaft are presented in Figure 8a and b, respectively.
Comparing the two figures, it is possible to see that the responses are qualitatively similar but more
marked for the points closer to the pile shaft (Figure 8b), which are subjected to a higher
temperature variation. With respect to the mechanical-induced displacements, the first heating phase
induces a thermoplastic collapse of the soil, and this is reflected by the fact that the average
displacement during the subsequent thermal cycles is higher than the mechanical one. After the first
thermal cycle, the points at 15 m depth show no further significant thermal-induced displacement.
For the points at 10 and 5 m depth, the displacement is due to the accumulated thermal deformation
of the lower portion of soil, so it is higher than for the points at 15 m depth. However, starting from
the second cycle, the behaviour becomes thermoelastic (reversible). The difference between the
mechanical-induced displacement and the average one during thermal cycles is in the order of 1 mm,
while the amplitude of the elastic displacement is between 1 mm (at 2.4 m from the shaft) and 2 mm
(at 0.8 mm from the shaft). The displacements of the pile at 5, 10 and 15 m depth are plotted in
Figure 9a as function of temperature. In accordance with the pile thermal-induced axial deformation,
the point situated in the upper part (5 m depth) moves upwards during heating and downwards
during cooling, while the point in the lower part (15 m depth) does the opposite. The displacement
of the point at mid-length (10 m depth) during the first heating phase is almost zero. This can lead to
the conclusion that the null point is not far from this depth. After the first cycle, the behaviour
changes, and the slope of the stabilized branches reflects the thermoelastic response of the model.
From the second cycle, the point situated at 5 m depth has a reversible upwards (heating) and

Figure 9. Pile vertical displacements (a) with temperature at different depths and (b) along the pile’s shaft.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
874 A. DI DONNA AND L. LALOUI

Figure 10. Pile head settlement during thermal cycles.

downwards (cooling) displacement of about 3 mm, the point at 10 m depth moves in the same sense but
of about 1.5 mm and the point at 15 m depth moves in the opposite sense (downwards during heating
and upwards during cooling) and definitely less with respect to the two previous ones, of about 0.4 mm.
This indicates that, from the second cycle, the null point is close to 15 m depth. Besides this behaviour
mainly governed by the thermoelastic response of pile, the three points accumulate an additional
settlement during thermal cycles that is due to the thermal collapse of the surrounding soil.
Coherently with what has been presented previously, almost all the additional settlements are
accumulated during the first cycle, and only little is added during the following thermal cycles. The
displacements of the pile along its length are plotted in Figure 9b for the mechanical loading stage
and after the heating and cooling phases of the first, fifth and seventh years. This plot confirms that
at long term the response becomes reversible: the curves related to the fifth year are coincident with
those related to the seventh year. Coherently with what has been discussed earlier, the null point of
the first heating phase (indicated as NP-H1) is situated slightly deeper than 10 m depth. Comparing
the curve at the end of the first heating phase with the one related to the end of the first cooling
phase, it is possible to identify the position of the null point during this stage: according to the
observation deduced from Figure 9a, it is deeper than the previous one, at around 15 m depth
(NP-C1). For the subsequent cycles, its position remains fairly constant. The displacements at the pile
head, which result from the presented complex behaviour, are plotted in Figure 10 as function of time.
The pile has an average settlement during the thermal cycles, which is about 1 mm higher than the
initial one induced by the mechanical loading. The effect of the subsequent thermal cycles on the
thermoplastic response of the soil (accommodation) is subtle. The reversible upwards (heating) and
downwards (cooling) displacements of the pile’s head during thermal cycles is about 4 mm.

4.4. Stress state in the soil and at the pile–soil interface


It is known that heating a soil might induce an excess pore water pressure because the thermal
expansion coefficient of water is higher than the one of the solid skeleton. Therefore, heating
induces a decrease of the effective stresses. Depending on the soil hydraulic conductivity and
heating rate, the excess pore water pressure may be dissipated in a relatively short period. In the
simulated case, the heating rate and the soil hydraulic conductivity are such that the process occurs
in drained conditions, so that the pore water pressure increases/decreases during heating/cooling, and
the consequent decreases/increases of effective stress are not significant (about 1 kPa). However, if
this is true for the soil surrounding the pile, the behaviour of the soil at the pile–soil interface is
slightly different and deserves to be analysed separately. The evolution of the horizontal effective
stress at the pile–soil interface at 10 m depth (point B) is presented in Figure 11, together with the
evolution of pore water pressure. It shows that, although the pore water pressure is unaffected by
temperature, the horizontal effective stress varies cyclically because of the interaction with the pile.
The horizontal effective stress increases during heating and decreases during cooling of about 5 kPa

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 875

Figure 11. (a) Evolution of the state of stress at the pile–soil interface (10 m depth, point B) and (b) constant
normal stiffness conditions at the pile–soil interface.

because of the radial thermal expansion of the pile, in accordance with the constant normal stiffness
(CNS) approach [44]. Figure 11b presents the evolution of the normal (i.e. horizontal) effective
stress to the pile–soil interface at 5, 10 and 15 m depth against the normal displacement of the same
points. The slope of these curves represents the far field stiffness K, which varies between 10 and
100 kPa/mm, in accordance with the theoretical value that can be computed as follows [45]:

2G0
K¼ (26)
R

for the pile radius R of 0.4 m and a soil shear modulus at small deformation G0 varying with depth from
2 to 20 MPa. The highest value of K is found for the deepest point, which has a higher shear modulus.
Figure 12 shows the response of the pile–soil interface at 5, 10 and 15 m depth in terms of mobilized
shear stress during mechanical loading and thermal cycles. The mechanical loading is responsible for
the mobilization of the largest part of the interface shear stress (ts). Nevertheless, the thermal cycles
result in additional displacements and consequent additional mobilization of interface shear stress,
both in the same direction and in the opposite one with respect to the mechanical loading. The
thermal-induced mobilization of the interface shear stress is more important for the points that are
further from the null point (5 and 10 m depth) and less significant for the point closer to the null
point (15 m depth). Among the three studied points, the one at 5 m depth is the more stressed as it is
subjected to a thermal cyclic displacement of about 2.5 mm in amplitude. The point at 15 m depth is
the less stressed, as it is the closest to the null point, and its additional displacements during thermal
cycles are negligible. The average thermal-mobilized shear stress at the pile–soil interface is about
1 kPa/°C, consistent with available in situ tests measurements [14]. Figure 12b shows the stress
paths of the same three points in the Mohr plane, together with the Mohr–Coulomb failure envelope

Figure 12. Pile–soil interface response: (a) shear stress and (b) stress path.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
876 A. DI DONNA AND L. LALOUI

corresponding to the imposed interface friction angle of 24°. Also in this picture, it is clear that the
CNS condition is properly reproduced.

4.5. Thermal-induced stress in the pile


According to what was discussed in the first part of this paper, the magnitude and distribution of the
additional thermal-induced axial stresses inside the pile depend on the type and entity of restraints
[14]. Figure 13 shows the total vertical stress resulting from the finite element (FE) simulation inside
the pile after the heating and cooling phases of the first year, compared with the one induced by the
mechanical loading. The total stress at the pile head after thermal loading corresponds to the one
applied as the pile head is free to deform thermally (no restraint). During heating, the pile dilates
with respect to the null point that is, during the first heating phase, situated at about mid-pile length
(Figure 9b). The additional thermally induced displacement measured at the pile head (δv,o) is
1.5 mm (the minus indicates that it is upwards), which corresponds to an observed average
deformation (expansion) equal to

δv;o
εo ¼ H ¼ 0:00015 (27)
=2

If the pile was completely free to thermally expand about the same null point, its free thermal axial
deformation would have been as follows:

β′p
εf ¼  ΔT ¼ 0:00017 (28)
3

with ΔT = 14 °C and β′p ¼ 3:6105 °C1 . The difference between the free and the observed axial
deformations represents the portion of deformation that is prevented by the presence of the
surrounding soil and induces additional axial compression stresses inside the pile of
 
f  εo ¼ 7000 KPa
σth ¼ Ep εth th
(29)

Starting from the end of the heating phase, during the subsequent cooling, the pile contracts about a
point that is situated at 15 m depth (Figure 9b). It is subjected to a temperature variation of 27 °C,
from +25 °C to 2 °C. The observed head displacement during this phase (variation with respect to
the end of heating) is 4.2 mm, which corresponds to an observed axial deformation of 0.00028
(considering the null point at 15 m depth). The free thermal axial deformation of the pile would have
been 0.000324, which corresponds to a blocked axial deformation of 0.000044 and a consequent
thermal-induced stress (tension) variation of 1540 kPa. These values are consistent with the one

Figure 13. Stresses transmitted inside the piles.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 877

numerically computed (Figure 13). In the case study, the reduction in compression axial stress during
cooling results in a tensile stress (negative) in the lower part of the pile. However, it has to be
mentioned that it depends on the entity of the mechanical load and on the boundary conditions. The
thermal-induced total stresses inside the pile remain constant cycle after cycle, after each heating
and after each cooling phase. Accordingly, the pile could be designed accounting for the maximum
stresses (negative and positive) presented in Figure 13. Finally, it is worth mentioning that from a
structural point of view, the additional thermal compression stresses do not represent a significant
risk for the stability of the considered foundation. Indeed, assuming a concrete with a 28-day
compressive strength fcu of 25 MPa, the pile is able to carry both the mechanical and the thermal
loads applied (maximum of 2.5 MPa). Conversely, deep foundations are usually not designed to
support tension, so a possible issue could be represented by the eventual tensile stresses at the pile
tip, which have to be specifically taken into account. In any case, the achievement of the maximum
acceptable stress for concrete (compression or tension) results in a stress redistribution and
consequently in a displacement problem.

4.6. Geotechnical bearing capacity


A number of factors need to be addressed when dealing with the geotechnical bearing capacity of
energy piles, particularly what follows: (i) the thermally induced change in the state of stress at the
pile–soil interface (CNS), (ii) the eventual thermal effect on the soil behaviour and soil properties
(ACMEG-T), (iii) the possible freezing of the soil–pile interface during winter and (iv) the way in
which the lateral and tip capacities are mobilized during thermo-mechanical loading. The forces
mobilized along the lateral interface Ts and at the base of the piles Tb during the heating and
cooling cycles have been computed by integrating the stresses at the interface between the pile and
the soil, ts and tb, as

H 2π
Ts ¼ ∫0 ∫0 ts R dωdz (30)

R 2π
Tb ¼ ∫0 ∫0 tb r dωdr (31)

The mobilization of these two components during the application of the mechanical loading, as well
as their sum (i.e. the total mobilized capacity), is represented in Figure 14a. The tip component starts
from an initial value of about 200 kN (i.e. 400 kPa), which corresponds to the pile weight (Figure 4a),
and it increases slightly during the application of the mechanical loading up to 250 kN, thus supporting
about 50 kN of the externally applied load. The largest part of the mechanical load (700 kN) is
supported by shaft friction. The evolution of Ts and Tb during the heating and cooling cycles is
presented in Figure 14b and c, respectively. Their analysis is interesting to gain inside the thermo-
mechanical behaviour of the energy piles.
The first important observation that can be carried out is that, in the absence of a head reaction at the
top of the pile (slab), the sum of the shaft and tip-mobilized resistance must remain constant during
thermal cycles and equal to the sum of the pile’s weight WP and external load P, in order to verify
the pile’s equilibrium equation:

P þ WP ¼ TS þ TB (32)

When the pile is heated, it expands axially about the null point, so that the mobilized tip resistance
increases because the pile’s basis moves downwards. If it increases, the mobilized lateral component
must decrease in order to keep the sum constant. In elastic conditions, the null point must be
situated in the lower part of the pile, so that during heating the upper part that moves upwards and
thus mobilizes the interface friction in the opposite direction with respect to the mechanical load
(reduction of Ts) is bigger than the lower one that moves downwards and thus mobilizes the

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
878 A. DI DONNA AND L. LALOUI

Figure 14. Mobilization of (a) lateral, tip and total resistances during mechanical loading and (b) lateral, (c)
tip and (d) total resistances during thermal cycles.

interface friction in the same direction than the mechanical loading (increase of Ts). In non-elastic
conditions, the position of the null point might vary depending on the transfer mechanisms. In the
presented simulations, Tb increases during heating and decreases during cooling (Figure 14c), and Ts
does the opposite (Figure 14b), so that their sum Ttot is constant during thermal cycles (Figure 14d).
If a head constraint is present (slab), it generates a reaction TH and the equilibrium equation becomes

P þ WP þ TH ¼ TS þ TB (33)

These conditions are more representative of most of real cases, as piles are usually designed to work
in groups. The thermo-mechanical analysis of a group of energy piles is more complex, and it
represents the focus of the next paragraph.

5. NUMERICAL MODEL OF A GROUP OF ENERGY PILES

With the purpose to study the effect of a head restraint and the possible interactions between energy
piles in an entire deep foundation, a three-dimensional FE model was developed. It consists in a
rectangular group of 7 × 3 energy piles equally spaced by 7 m between each other. The geometry,
dimensions and mechanical boundary conditions applied to the model are presented in Figure 15.
Figure 16 shows the plan view of the model with the labels used to identify the piles and the
indication of the soil profile analysed in what follows. Thanks to the symmetry of the case study;
only a quarter of the foundation is modelled. This includes three entire external piles (P6, P7 and
P8), four half piles on the planes of symmetry (P2, P3, P4 and P5) and a quarter of the central pile
(P1). The model includes 9145 hexahedral elements (including 264 for thermal power injection and
extraction) with eight nodes and eight integration points and 33 surface elements with four nodes
and four integration points for the application of the mechanical load at the head of each pile. The

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 879

Figure 15. Geometry of the three-dimensional model of the group of energy piles.

Figure 16. Plan view of the three-dimensional model with the labels used to identify the piles (P) and the soil
profiles (S) analysed.

total number of nodes is 10 350. The temperature and the pore water pressure are fixed at the bottom,
on the right side and on the back of the model, while the two planes of symmetry are considered to be
impervious, and no water flow is allowed through them. As for the single pile study, all the piles have a
diameter of 0.8 m and a length of 20 m, but in this case, a concrete slab was also considered, having
thickness of 0.5 m, dimension of 44 × 16 m2 (a quarter being 22 × 8 m2) and same mechanical,
thermal and hydraulic properties as the piles (Tables I and II). For the soil, the same constitutive
model and material parameters used for the single pile simulation have been also used in this case.
The thermo-mechanical load applied on each pile is the same as that imposed for the single pile case.

5.1. Initialization and mechanical loading step


The model is initialized in the same way described for the axisymmetric simulation of the single pile.
The responses of the eight piles after the application of the mechanical load, in terms of transmitted
total vertical stress inside the concrete and vertical displacements along the length, are reported in
Figure 17a and b, respectively. The curve indicated as ‘initial’ in Figure 17a refers to the initial state
of stress inside the pile (the same for all of them) and corresponds to its weight. The presence of the
concrete slab results in a redistribution of the applied loads: the central piles settle slightly more

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
880 A. DI DONNA AND L. LALOUI

Figure 17. (a) Total vertical stress and (b) vertical displacement in the piles after application of the mechan-
ical load.

than the external ones, and the slab transfers part of their load to the external piles that consequently are
more charged. It is commonly known that a rigid slab results in a quite uniform settlement of the
foundation and a redistribution of load towards the external piles. Conversely, the presence of a
flexible slab leads to more significant differential displacements between the external and central
areas of the foundation (bending of the slab). In such a case, the settlements of the central piles are
higher with respect to those of the external ones, and the loads are concentrated on them [46].
Reality is likely to be between these two extreme configurations, and this is the case for the
simulated foundation. The average settlement of the foundation is about 1 cm, and the differential
settlement between the central and the external area of the foundation is less than 2 mm, which is
definitely admissible with regard to the current norms [47].

5.2. Thermal field


The evolution of temperature at 10 m depth for a point inside pile P1 (it is the same for the others), at
the interface between the pile and the soil and in the soil in correspondence of the profiles S1 and S2
indicated in Figure 16, is plotted in Figure 18 for the first years, which are representative also for the
subsequent cycles. The maximum temperature reached inside the pile is 25 °C, while the minimum one
is 2 °C. The pile–soil interface is subjected to a temperature variation of 17 °C, from a maximum of
21 °C and a minimum of 4 °C. The soil at the maximum distance from the piles inside the foundation
domain (profiles S1 and S2) is also subjected to a temperature variation, which is of about 4 °C,
reaching a maximum temperature of 15 °C. In Figure 18, the transient heat diffusion problem is

Figure 18. Evolution of temperature in the pile P1 (same for the other piles), at the interface between the soil
and pile P1 (same for the other piles) and in the soil profiles S1 and S2 (Figure 16).

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 881

Figure 19. Evolution of vertical displacements during thermal cycles at the piles head for (a) central piles
and (b) external piles.

shown by the fact that the maximum temperature in the soil profiles S1 and S2 is reached later with
respect to the moment at which the temperature is the maximum inside the pile, where the heat power
is injected.

5.3. Piles displacements


The vertical displacements at the top of the eight piles are plotted in Figure 19 (positive downwards).
The central piles (P1, P2 and P3) start from the same mechanical-induced settlement of roughly 1 cm,
and their response during heating and cooling is definitely similar (Figure 19a). As for the case of the
single pile, the effect of the thermoplastic response of the soil corresponds to 1 mm additional
irreversible settlement, but in this case, it results to be negligible with respect to the mechanical one.
The vertical reversible displacement due to the thermoelastic deformation of the pile (upwards
during heating and downwards during cooling) has total amplitude of 4 mm, as for the case of single
pile. As it has been already presented in Figure 17b, the external piles (P4, P5, P6, P7 and P8) have
a differential displacement, although small, between them and with respect to the central ones
(Figure 19b). However, the thermal loading does not enhance it, and for these piles, the same
conclusions drawn for the central ones can be made. In order to have a global vision of the thermal-
induced movements of the foundation, Figure 20 presents the vertical displacements of the eight
piles, along with their depth, due to the mechanical loading and after the first heating and cooling
phases. Piles P2 and P3 show the same behaviour, so their response is plotted only once
(Figure 20b), as well as piles P6 and P7 (Figure 20e). Comparing the different plots, it is possible to
see that the null point related to the first heating phase is between 12 and 14 m depth for all the
piles, while the one related to the subsequent cooling phase is between 18 m depth and the pile’s tip.
It is worth mentioning that in the case study the same thermal loading is applied to all the piles
contemporaneously. A differential thermal load between the piles or the absence of it for one or
more piles would result in differential displacements definitely more significant than those shown here.

5.4. Thermal-induced stresses in the piles


To complete the geotechnical analysis of the energy pile foundation considered, the thermal-induced
stresses inside the piles are studied here. Figure 21 reports the total vertical stress inside the piles
due to the mechanical loading and after the first heating and cooling phases. As for the case of
vertical displacements, the behaviour of piles P2 and P3, as well as that of piles P6 and P7, is
comparable, and thus, they are presented only once (Figure 21b and e). According to the
expectations, the increase in temperature results in an increase of compressive stress inside all the
piles. The subsequent decrease in temperature results in a reduction of compressive stress, which
can eventually lead to tension. Despite all the piles have the same thermal expansion coefficient and
are subjected to the same temperature variation (i.e. they have the same free thermal deformation)
and show roughly the same (observed) thermal deformation, they react in a significantly different
way in terms of thermal-induced stress. The central piles (P1, P2 and P3) are subjected to an

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
882 A. DI DONNA AND L. LALOUI

Figure 20. Vertical displacements due to mechanical loading, first heating and first cooling phases for (a)
pile P1, (b) piles P2 and P3, (c) pile P4, (d) pile P5, (e) piles P6 and P7 and (f) pile P8 (Figure 16).

increase of compressive stress during heating, which is largely higher than the one undergone by the
external piles (in particular by piles P5, P6, P7 and P8). Pile P1 has the highest increase in
compressive stress due to heating of 1500 kPa, and pile P5 has the lowest one of roughly 200 kPa.
Similarly, cooling results in a strong reduction of compression stress, eventually turning into
tension, for the central piles and in a limited reduction for the external piles. The extreme cases are
once again the pile P1, for which cooling induces a zero or negative (tension) vertical total stress for
almost the entire length, and the pile P5 that is subjected, after cooling, to a compressive stress that
is roughly 300 kPa lower than the one at the end of heating or 100 kPa lower than the one at the end
of the mechanical stage. The theory developed for the case of a single pile, according to which the
freer is the pile, the higher is the thermal-induced observed deformation and the lowest is the
thermal-induced stress, cannot be directly applied to the case of a group of energy piles. This is
attributed to the presence of the slab and to the consequent redistribution of stresses from the
external piles to the central ones that result to be consequently more charged. As for the case of
mechanical loading, the response is likely to be sensitive to the slab stiffness: the more rigid the
slab, the more the response will be governed by it, and the less rigid the slab, the more the response

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 883

Figure 21. Total axial stress induced by mechanical loading, heating and cooling in (a) pile P1, (b) piles P2
and P3, (c) pile P4, (d) pile P5, (e) piles P6 and P7 and (f) pile P8 (Figure 16).

will be governed by the single pile behaviour. The group effects are in this sense an important aspect
that has to be taken into account in the design of an energy pile foundation. As for the case of the single
pile, it is worth mentioning that the applied mechanical load at the piles head is low with respect to
their bearing capacity, and consequently, the cooling-induced reduction in compression stress results
in tensile stress easier than a similar case with a higher mechanical load. Nevertheless, the
simulations show that the problem of the eventual development of tensile stress inside the piles after
the cooling phase might be enhanced by the presence of a rigid slab with respect to the case of an
energy pile considered alone.

6. CONCLUSIONS

In situ tests on thermo-active piled foundations, as well as laboratory and centrifuge experiments on
scaled model piles, have permitted to lay the basis for the comprehension of the physical behaviour
of energy piles and develop a theoretical framework to represent it. When subjected to a temperature
variation, a pile deforms thermally but a portion of its free thermal deformation is prevented and

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
884 A. DI DONNA AND L. LALOUI

generates additional stresses in the pile that are compressive during heating and tensile during cooling.
Moreover, the interface shear stress, which is mobilized upwards during axial compressive mechanical
loading, is mobilized both upwards and downwards during thermal loading depending on the pile
deformation and position of the null point. However, despite the undeniable potentiality of physical
modelling in the understanding of the thermo-mechanical behaviour of energy pile foundations, it is
clear that the numerical approaches are often preferable if not essential. They permit to investigate a
large variety of configurations, geometries, materials, thermo-mechanical loading and boundary
conditions in a controlled and relatively easy and cost-effective manner. In the engineering practice,
when in situ testing is not always possible, they represent a powerful design tool.
Two FE models were developed and presented in this paper: one devoted to study the behaviour of a
single energy pile, and the second devoted to the investigation of a group of energy piles. Ten years of
thermal cyclic loading was applied to study the long-term performance. From the single pile
simulation, it appeared that the thermoelastic-thermoplastic soil behaviour results in an additional
settlement of the foundation, which is nevertheless admissible (1 mm) and entirely developed during
the first year. During the subsequent thermal cycles, the pile reacts thermo-elastically, showing an
upwards head displacement during heating and a downwards head displacement during cooling,
with amplitude of about 4 mm. The additional heating-induced compressive stress is significant
although admissible with respect to the concrete strength (about 700 kPa). The cooling-induced
reduction in compression stress leads to a zone of tensile stress at the pile tip.
The results of the more complex three-dimensional simulation of an energy pile group are
interesting to understand the interactions between the piles and analyse the response of the entire
foundation when subjected to mechanical loading and thermal cyclic solicitations. From the first
mechanical phase, it was possible to conclude that the presence of the modelled slab results in a
medium rigid foundation, showing a slight inflection (settlement of 1 to 2 mm higher in the central
zone) and a redistribution of vertical loads towards the external piles. This fairly rigid response,
mainly governed by the concrete slab, is also reflected during the subsequent thermal cycles. The
thermal-induced displacement is fairly uniform over the whole foundation, and the additional
settlement induced by the thermoplastic behaviour of the soil shown by the single pile case is subtle
for the group of piles. The thermal-induced stresses inside the piles are redistributed towards the
central ones. As for the case of the single pile, whereas the heating-induced compressive stress,
although significant, is admissible with respect to the concrete strength (maximum of 1500 kPa in
the central pile), the cooling-induced reduction of compressive stress turns into tension inside the
central pile. This aspect needs to be specifically considered in the design practice. The admissibility
of the maximum compressive stress during heating with respect to the concrete strength, as well as
the development of tensile stress during cooling, depends also on the entity of the mechanical-
induced stress. In any case, the achievement of the maximum acceptable stress for concrete
(compression or tension) results in a displacement problem. The presented results are specific to the
cases studied but lead to the conclusion that both the thermal-induced displacements and stress
deserve to be taken into account in the geotechnical design of energy piles.

APPENDIX A: DEFINITION OF TENSORS AND INVARIANTS

The effective stress tensor σ′ij is defined as follows:

σ′ij ¼ σij þ pw δij

where σij is the total stress tensor, pw is the pore water pressure and δij is the Kronecker’s delta. The
deviatoric stress q is defined as follows:
rffiffiffiqffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3  2 1
q¼ tr sij with sij ¼ σij  δij σkk
2 3
where sij is the deviatoric stress tensor and tr(∙) stands for the trace of the tensor. The mean effective

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 885

stress p’ is defined as follows:


σ′ kk
p′ ¼
3

The Lode angle ϑ is defined as follows:

 pffiffiffi 
3 3IIIs 1 1
sinð3ϑÞ ¼  with IIs ¼ sij sij and IIIs ¼ sij sjk ski
2 II3s 2 3

where IIs and IIIs are the second and third invariants of the deviatoric stress tensor, respectively. The
strain tensor εij is defined as follows:

pffiffiffiqffiffiffiffiffiffiffiffiffiffiffiffiffiffi
εv   6  2
εij ¼ δij þ eij with εv ¼ tr εij and εd ¼ tr eij
3 3

where εv is the volumetric strain, eij is the deviatoric strain tensor and εd is the deviatoric strain.

APPENDIX B: VALIDATION OF THE IMPLEMENTATION OF THE VAN EEKELEN


SURFACE

The dependency of the yield limit on the Lode’s angle for the ACMEG-T model was implemented in
the FE code of the LAGAMINE software, according to the Van Eekelen formulation, as presented
through Equations (16), ( 17) and ( 18). From a mathematical viewpoint, this implied the introduction
of a new term in the expressions of the derivatives of the deviatoric yield limit as follows:

∂f dev ∂f dev ∂p′ ∂f dev ∂q ∂f dev ∂ sinð3ϑ Þ



¼ ′
þ ′
þ (34)
∂σ ij ∂p ′ ∂σ ij ∂q ∂σ ij ∂ sin ð3ϑ Þ ∂σ ′ij

where the first two terms of this equation were already included in the previous version of the model,
while the third one had to be newly introduced. The derivatives of fdev with respect to the sine of three
times the Lode’s angle is as follows:

 
∂f dev pffiffiffi dp′
¼ 3 3aL bL nL ð1 þ bL sinð3ϑ ÞÞnL 1 p′ 1  b log ′ rdev (35)
∂ sinð3ϑ Þ pc

Figure 22. Geometry, initial and boundary conditions of the model used for the validation of the implemen-
tation of the Van Eekelen surface.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
886 A. DI DONNA AND L. LALOUI

Table IV. B1Advanced Constitutive Model for Environmental Geomechanics, with temperature effects
included parameters used for the validation.
Elastic parameters

Reference bulk modulus Kref (MPa) 83.3


Reference shear modulus Gref (MPa) 38.5
Elastic exponent ne () 1

Plastic parameters

Material parameters a () 0.003


b () 0.8
c () 0.02
d () 2
Friction angle in compression φ’c (°) 26
Friction angle in extension φ’e (°) 26
Plastic compressibility β () 10
Dilatancy parameter α () 1.1
Plastic radius for the isotropic mechanism reliso () 0.01
Plastic radius for the deviatoric mechanism reldev () 0.01
Van Eekelen exponent nL () 0.229

while the derivative of the sine of three times the Lode’s angle with respect to the effective stress tensor
is as follows:
pffiffiffi  
∂ sinð3ϑ Þ 3 3 2 2 3IIIs sij
¼  3 sik skj  IIs δij  (36)
∂σ ′ij 2IIs 3 IIs 2IIs

In order to validate the numerical implementation, a series of tests on a quadrilateral element with
eight nodes and four integration points was carried out. The geometry, initial and boundary conditions
used are represented in Figure 22. The imposed axial displacement was 20 cm; that is, 20% of the
element height. Three different stress paths were considered: that is, (i) triaxial compression (TX-C),
(ii) triaxial extension (TX-E) and (iii) axial compression in plane strain conditions. For the first two
cases, axisymmetric conditions were imposed. The material parameters are those presented in Table IV.
The results for the two triaxial stress paths are presented in Figure 23, in terms of deviatoric stress mo-
bilized and volumetric strain. According to the expectations, the result does not change under triaxial
compression, while, under triaxial extension, plasticity is achieved at a lower deviatoric stress when the
Van Eekelen surface is used. The third stress path considered is particularly of interest for the numer-
ous geotechnical problems that are usually analysed under plane strain conditions. This is the case, for
instance, of shallow foundations, slope stability problems or tunnels. The results for this stress path are
presented in Figure 24, in terms of vertical effective stress evolution and volumetric deformation. It is
possible to see that, when the Van Eekelen surface is used, the material undergoes plasticity at a lower

Figure 23. Validation of the implementation of the Van Eekelen surface under triaxial compression (TX-C)
and extension tests (TX-E): (a) deviatoric stress and (b) volumetric deformation.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
NUMERICAL ANALYSIS OF THE GEOTECHNICAL BEHAVIOUR OF ENERGY PILES 887

Figure 24. Validation of the implementation of the Van Eekelen surface during axial compression in plane
strain conditions: (a) vertical effective stress and (b) volumetric deformation.

vertical effective stress with respect to the case where the circular deviatoric surface is assumed. For
both the triaxial extension (Lode’s angle, θ = 30°) and the plane strain (Lode’s angle, θ ≈ 0°) stress
paths, for a certain imposed axial deformation, the material is closer to the constant volume condition
when the Van Eekelen surface is used.

ACKNOWLEDGEMENTS
This research project was funded by the Swiss Federal Office of Energy (contract Nb. 154’426). Frederic
Collin (University of Liege) is acknowledged for his help with the numerical implementation.

REFERENCES
1. Laloui L, Di Donna A. Energy Geostructures: Innovation in Underground Engineering, Vol. 304. ISTE Ltd and John
Wiley & Sons Inc., 2013.
2. Brandl H. Energy foundations and other thermo-active ground structures. Geotechnique 2006; 56(2):81–122.
3. Pahud D. A case study: the Dock Midfield of zurich Airport. In Energy Geostructures: Innovation in Underground
Engineering, Laloui L, Di Donna A (eds). 2013; ISTE Ltd and John Wiley & Sons Inc. 281–295.
4. Adam D, Markiewicz R. Energy from earth-coupled structures, foundations, tunnels and sewers. Géotechnique 2009;
59(3):229–236.
5. Kipry H, Bockelmann F, Plesser S. Evaluation and optimization of UTES systems of energy efficient office buildings
(WKSP). In 11th international conference on thermal energy storage2009: Stockholm, EFFSTOCK.
6. De Moel M et al. Technological advances and applications of geothermal energy pile foundations and their feasibility
in Australia. Renewable and Sustainable Energy Reviews 2010; 14(9):2683–2696.
7. Hamada Y et al. Field performance of an energy pile system for space heating. Energy and Buildings 2007; 39(5):
517–524.
8. Lennon DJ, Watt E, Suckling TP. Energy piles in Scotland. Deep Foundations on Bored and Auger Piles, Proceed-
ings, 2009; p. 349–355.
9. Riederer P, et al. Conception de foundations geothermiques, CSTB, Département Energie Santé Environnement,
Pole Energie Renouvelables, 2007
10. Henderson HI, Carlson SW, Walberger AC. North America monitoring of a hotel with room size GSHPs. In IEA
room size heat pump Conference, 1998; Niagara falls, Canada.
11. Fromentin A et al. Pieux échangeurs: conception et règles de pré—dimensionnement. Revue Française de Génie
Civil 1999; 3(6):387–421.
12. Laloui L, Moreni M, Vulliet L. Behavior of a dual-purpose pile as foundation and heat exchanger. Canadian Geo-
technical Journal 2003; 40(2):388–402.
13. Bourne-Webb PJ et al. Energy pile test at Lambeth College, London: geotechnical and thermodynamic aspects of
pile response to heat cycles. Geotechnique 2009; 59(3):237–248.
14. Amatya BL et al. Thermo-mechanical behaviour of energy piles. Geotechnique 2012; 62(6):503–519.
15. Laloui L, Nuth M, Vulliet L. Experimental and numerical investigations of the behaviour of a heat exchanger pile.
International Journal for Numerical and Analytical Methods in Geomechanics 2006; 30(8):763–781.
16. Dupray F, Laloui L, Kazabgba A. Understanding the thermo-hydro-mechanical behaviour of seasonal heat storage in
an energy pile foundation. Computers and Geotechnics 2014; 55(1):67–77.
17. Knellwolf C, Peron H, Laloui L. Geotechnical analysis of heat exchanger piles. Journal of Geotechnical and
Geoenvironmental Engineering 2011; 137(10):890–902.
18. Laloui L, Di Donna A. Understanding the behaviour of energy geo-structures. ICE Proceedings of the Institution of
Civil Engineers 2011; 164(4):184–191.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag
888 A. DI DONNA AND L. LALOUI

19. Tang AM, et al. Behaviour of heat-exchanger piles from physical modelling. In Energy Geostructures: Innovation in
Underground Engineering, Laloui L, Di Donna A (eds). ISTE Ltd and John Wiley & Sons Inc. 2013; 304.
20. McCartney JS. Centrifuge modelling of energy foundations. In Energy Geostructures: Innovation in Underground
Engineering, Laloui L, Di Donna A (eds). ISTE Ltd and John Wiley & Sons Inc. 2013; 99–114.
21. GSHP. Thermal pile system: design, installation and materials standards, 2012.
22. Silvani C, et al. Understanding the thermo-mechanical response of heat exchanger piles. In First International Sym-
posium on Computational Geomechanics (ComGeo I)2009: Juan-les-Pins (France). p. 589–596.
23. Suryatriyastuti ME, et al. Numerical analysis of the bearing capacity of thermoactive piles under cyclic axial loading.
In Energy Geostructures: Innovation in Underground Engineering, Laloui L, Di Donna A (eds). ISTE Ltd and John
Wiley & Sons Inc. 2013; 304.
24. Charlier R. Approche unifiée de quelques problèmes non linéaires de mécanique des milieux continus par la méthode
des éléments finis: grandes déformations des métaux et des sols, contact unilatéral de solides, conduction thermique
et écoulements en milieu poreux. Université de Liège, Faculté des sciences appliquées, 1987.
25. Collin F. Couplages thermo-hydro-mécaniques dans les sols et les roches tendres partiellement saturés. Université de
Liège, Faculté des sciences appliquées, 2003.
26. Charlier R et al. Multi-physical processes in geomechanics. An introduction to constitutive modeling and coupling
aspects. European Journal of Environmental and Civil Engineering 2009; 13(7/8):803–830.
27. Charlier R, Laloui L, Collin F. Numerical modelling of coupled poromechanics processes. European Journal of Civil
Engineering 2006; 10(6/7):669–702.
28. Thomas HR, King SD. A nonlinear, 2-dimensional, potential-based analysis of coupled heat and mass-transfer in a
porous-medium. International Journal for Numerical Methods in Engineering 1994; 37(21):3707–3722.
29. Laloui L, Francois B. ACMEG-T: soil thermoplasticity model. Journal of Engineering Mechanics-ASCE 2009; 135(9):
932–944.
30. Schofield AN, Wroth CP. Critical State Soil Mechanics. McGraw-Hill, 1968.
31. Hujeux JC. Calcul numérique de problèmes de consolidation elastoplastique. Ecole Centrale, Paris, 1979.
32. Laloui L. Modélisation du comportement thermo-hydro-mécanique des milieux poreux anélastique. Ecole Centrale
de Paris, 1993.
33. Modaressi H, Laloui L. A thermo-viscoplastic constitutive model for clays. International Journal for Numerical and
Analytical Methods in Geomechanics 1997; 21(5):313–335.
34. Laloui L, Cekerevac C. Non-isothermal plasticity model for cyclic behaviour of soils. International Journal for Nu-
merical and Analytical Methods in Geomechanics 2008; 32(5):437–460.
35. Di Donna A, Laloui L. Response of soil subjected to thermal cyclic loading: experimental and constitutive study. En-
gineering Geology 2014. Submitted.
36. Dafalias YF. Materials with vanishing elastic region. International Journal for Numerical and Analytical Methods in
Geomechanics 1980; 4(4):389–389.
37. Potts DM, Zdravković L. Finite Element Analysis in Geotechnical Engineering: Theory. Thomas Telford Limited,
1999.
38. Van Eekelen HAM. Isotropic yield surfaces in three dimensions for use in soil mechanics. International Journal for
Numerical and Analytical Methods in Geomechanics 1980; 4(1):89–101.
39. Barnichon JD. Finite Element Modelling in Structural and Petroleum Geology. Universite de Liege, Faculte des Sci-
ences appliquees, 1998.
40. Laloui L, Cekerevac C. Thermo-plasticity of clays: an isotropic yield mechanism. Computers and Geotechnics 2003;
30(8):649–660.
41. Mroz Z, Norris VA, Zienkiewicz OC. An anisotropic, critical state model for soils subjected to cyclic loading.
Géotechnique 1981; 31(4):451–469.
42. Jaky J. The coefficient of earth pressure at rest. Journal of the Society of Hungarian Architects and Engineers 1944;
22:355–358.
43. SIA. Utilisation de la chaleur du sol par des ouvrages de fondation et de soutènement en béton. Guide pour la con-
ception, la realization et la maintenance. 2005.
44. Lehane BM et al. Mechanisms of shaft friction in sand from instrumented pile tests. Journal of Geotechnical
Engineering-Asce 1993; 119(1):19–35.
45. Wernick E. Skin friction of cylindrical anchors in non-cohesive soils. In Symposium on soil reinforcing and
stabilising techniques1978: Sydney, Australia.
46. Lancellotta R, Calavera J, Ruiz JC. Fondazioni. McGraw-Hill Companies, 1999.
47. EN-1997. Eurocode 7: Geotechnical Design. British Standards Institution: London, 2004.

Copyright © 2014 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2015; 39:861–888
DOI: 10.1002/nag

You might also like