Rock Mechanics and Rock Engineering (2024) 57:4367–4388

https://doi.org/10.1007/s00603-023-03536-4

ORIGINAL PAPER

Three‑dimensional Modeling of Cracking

with Thermo‑hydromechanical Process by Considering Rock
Heterogeneity
Zhan Yu1,2 · Jianfu Shao2,3 · Yue Sun2 · Minh‑ngoc Vu4 · Carlos Plua4 · Gilles Armand4

Received: 2 March 2023 / Accepted: 23 August 2023 / Published online: 8 September 2023
© The Author(s), under exclusive licence to Springer-Verlag GmbH Austria, part of Springer Nature 2023

Abstract
This study is part of numerical simulations performed on an in-situ heating test conducted by the French National Radioac-
tive Waste Management Agency (Andra) at the Meuse/Haute-Marne Underground Research Laboratory (URL) to study the
thermo-hydromechanical behavior of the host Callovo-Oxfordian COx claystone in quasi real conditions, through the interna-
tional research project DECOVALEX. We present a numerical study of damage and cracking process in saturated claystone
subjected to thermo-hydromechanical coupling by considering material heterogeneity distribution. For this purpose, a mac-
roscopic elastic model is first determined by using two steps of homogenization by taking into account the effects of porosity
and mineral inclusions. This model is implemented into a finite element code devoted to solving thermo-hydromechanical
coupling problems. The nucleation and propagation of cracks are described by using an extended phase-field method, con-
sidering the effects of temperature and fluid pressure on the evolution of phase-field. The proposed model is applied to the
numerical analysis of cracking process due to excavation and heating around a group of boreholes (CRQ). The numerical
results of the 3D simulation are compared with in-situ measurements of temperature and pore pressure distribution. The
excavation damage zone and heating fracture is reproduced and analysed according to the structure of the heating position
and the heterogeneity of the rock.

Highlights

• A phase-field model is developed with thermo-hydromechanical processes;

• Both tensile and shear cracks are taken into account;
• Crack nucleation is emphasized by spatial heterogeneity of material properties;
• Cracking processes due to thermal-hydraulic interaction are analyzed in three-dimensional conditions.

Keywords Hydro-thermal cracking · Material anisotropy · Heterogeneous rocks · Claystone · Phase-field method ·
Radioactive waste

List of symbols
𝛼b Scalar of thermal dilation coefficient in drained
* Jianfu Shao condition for isotropic material
[email protected]
𝛼f Thermal dilation coefficient of fluid
1
Key Laboratory of Ministry of Education on Safe Mining 𝛼m Differential thermal dilation of saturated porous
of Deep Metal Mines, College of Resources and Civil medium
Engineering, Northeastern University, Shenyang 110819, 𝝈 Stress tensor
China
I Second order unit tensor
2
University of Lille, CNRS, EC Lille, LaMcube, UMR9013, 𝜖 e Elastic strain tensor
59000 Lille, France
Γ Set of cracks
3
Institut Universitaire de France (IUF), Paris, France 𝛾 Crack density
4
Andra, Chatenay Malabry, France 𝜙 Porosity

13
Vol.:(0123456789)

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4368 Z. Yu et al.

𝜌0f Volumetric density of fluid ht Degradation variable for tensile crack

𝜃 Variation of temperature k Small positive value to avoid calculating error
AΓ Crack surface ld Length scale parameter for width of smeared
b Biot coefficient for isotropic material cracks
C𝜎b Volumetric specific heat for constant stress under mi Homogeneity index for mineral inclusions
drained condition mp Homogeneity index for pores
ds Damage variable of shear crack
dt Damage variable of tensile crack
E Total energy 1 Introduction
Ec Fracture surface energy
Ee Elastic strain energy of damaged material Clayey rocks are considered in France as a potential geo-
Kb Bulk modulus of drained material logical barrier for underground disposal of radioactive
Kf Bulk modulus of fluid waste. In this context, it is necessary to investigate thermo-
Km Bulk modulus of solid matrix hydromechanical (THM) properties of such rocks and other
Kreuss Reuss equivalent bulk modulus materials involved in the storage facilities. For this purpose,
M Scalar coefficient of Biot modulus under the coordination of the French National Agency for
m Fluid mass change per unit initial volume radioactive waste management (Andra), a series of in-situ
p Fluid pressure or pore pressure experiments have been and are being carried out at the Bure
s Entropy Underground Research Laboratory (MHM URL). Detailed
T Temperature presentations of some experiments can be found in Armand
wc Energy density per unit volume requested to create et al. (2014), Seyedi et al. (2021), Conil et al. (2020). The
crack main objective of those experiments is to investigate thermo-
we Elastic strain energy of cracked material hydromechanical responses of the host Callovo-Oxfordian
𝛽i Mean value of random variable for mineral inclu- (COx) claystone and other components during excavation,
sions volumetric fraction heating, and gas injection. The results obtained from these
𝛽p Mean value of random variable for porosity experiments are used for the validation of numerical models
ℂhom Macroscopic elastic tensor of heterogeneous rocks developed for the analysis and design of the future geologi-
ℂin Elastic tensor of inclusions cal storage project known as Cigeo.
ℂmp Effective elastic tensor of porous matrix In order to foster international collaboration in research
ℂm Elastic tensor of solid matrix on the geological disposal of radioactive waste, an active and
𝕀 Fourth order unit tensor long-standing international project called DECOVALEX has
ℙi Fourth order Hill tensor for spherical inclusions been established. Various topics and benchmark modeling
ℙp Fourth order Hill tensor for ellipsoidal pores studies have been investigated and performed (Tsang et al.
ℙ±𝜎 Operators for spectral decomposition of stress 2009; Birkholzer et al. 2018, 2019).
tensor In the phase of DECOVALEX-2023, the task A1 has been
Ωi Volume of inclusions proposed to improve the numerical models’ ability to pre-
Ωm Volume of solid clay matrix dict the mechanisms and processes of fracture initiation and
Ωp Volume of pores growth in clayey rocks due to heating induced fluid pres-
𝜎1t Major Terzaghi effective principal stress sure variation. In the task, the step 1.2 is to reproduce the
𝜎3t Minor Terzaghi effective principal stress THM responses of an in-situ heating experiment performed
𝜎t Average value of uniaxial tensile strength by Andra at the MHM URL in June 2019, called “Repre-
𝜑 Friction angle sentative THM Behavior of a high-level waste (HLW) Cell"
𝜉i Probability density for inclusions (CRQ). This experiment involved two heating phases, last-
𝜉p Probability density for pores ing two and one months respectively, separated by a cooling
c Cohesion phase. It is an accelerated heating scenario with the aim
fi Random distribution of mineral inclusions volu- of replicating an effective stress evolution path which is
metric fraction expected to close to that in the area around a HLW cell in the
fp Random distribution of porosity Cigeo project in a short period of time. The rock is quickly
G Shear modulus heated until the pore fluid pressure reaches a critical value
gsc Toughness parameter for shear crack for the possible creation of fractures.
gtc Toughness parameter for tensile crack Therefore, for the feasibility study of geological disposal
hs Degradation variable for shear crack of nuclear waste, the nucleation and evolution of cracks
is an essential issue. Such cracks can be initiated during

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4369

gallery excavation, temperature increase due to waste heat investigated. The numerical results are compared with experi-
production, and gas pressure rise. They are also expected to mental data obtained from in-situ experiments.
progressively close due to self-sealing and healing process.
During the last decades, different types of numerical meth- 2 Phase‑Field Method
ods have been developed for modeling cracking process. for Thermo‑hydromechanical Problems
Without giving an exhaustive review, we can mention for
instance enriched finite element methods (Oliver 1996) and In this work, a saturated porous medium is assumed to occupy
extended finite element methods (XFEM) (Moes et al. 1999, the volume Ω with the external boundary 𝜕Ω. Three physical
2010; Bernard et al. 2012). In those methods, it is needed to fields are considered to describe the thermo-hydromechanical
define specific criteria to describe the onset and propagation behavior of this material: (i) In the mechanical field, it has the
of cracks. Other types of methods, including various discrete body force fb in Ω, the surface force tN on the part of external
element methods, have also been developed and can be used boundary 𝜕Ωf and the prescribed displacement u on the com-
for modeling cracking processes. However, the description plementary part external boundary 𝜕Ωu; (ii) In the pore fluid
of progressive transition from diffuse deformation to local- pressure field, it has the fluid flux 𝜔 on the external boundary
ized cracking is still a pending issue. The analysis of three- 𝜕Ω𝜔 and to the prescribed fluid pressure p on the external
dimensional multiple cracks is also a hard task. boundary 𝜕Ωp; (iii) In the temperature field, it has the heat flux
Based on the variational principle for fracture mechan- q on the external boundary 𝜕Ωq and the prescribed temperature
ics (Francfort and Marigo 1998), the so-called phase-field change 𝜃 on the external boundary 𝜕Ω𝜃. In additional, in this
method has been developed (Bourdin et al. 2000; Miehe et al. THM problem the process of onset and propagation of crack
2010a). By the approximation of sharp cracks by a regularized is taken into account by using phase-field method.
crack density function and the minimization of a total energy Since it has been seen many times in the literature, it is the
functional, the phase-field method is particularly efficient to author choice to keep the writing as light and simple as pos-
deal with the progressive nucleation of localized cracks under sible, emphasizing only the several strong assumptions made
complex three dimensional problems. It is applied to mod- and their consequences, for the sake of the reading.
eling dynamic brittle fracture (Borden et al. 2012), extended
to multi-physics problems (Miehe et al. 2015), and plastic 2.1 Variational Framework
materials (Choo and Sun 2018; Fang et al. 2019), etc.
In many previous studies, the accent was put on tensile Based on the spirit of Griffith theory, (Francfort and Marigo
cracks. For rocks subjected to compressive stresses as in 1998) proposed a variation approach for quasi-static brittle
most rock engineering problems, shear cracks represent an fracture. The nucleation and propagation of quasi-static crack-
essential mechanism. Some extensions have been proposed ing process are considered as the minimisation of energy func-
(Zhang et al. 2017; Bryant and Sun 2018), generally using tion that comprises stored energy and energy necessary for
a specific decomposition of energy functional into a tensile crack formation. This variational framework is suitable to
part and a shear part. However, some critical issues should extended to consider the crack process on the mentioned satu-
still be addressed. Under compressive stresses, complex rated porous material subjected to THM loads. In this way, the
mixed cracking modes can be observed (Evans et al. 1990; total energy functional is developed as:
Wong and Baud 2012). The propagation of frictional shear
E(𝜺e , m, 𝜃, Γ) = Ee (𝜺e , m, 𝜃, Γ) + Ec (Γ) (1)
cracks is not only effected by deviatoric stress but also by
confining pressure. Different types of extensions have been where Ec (Γ) is the fracture surface energy and Ee (𝜺e , m, 𝜃, Γ)
proposed considering frictional contact of cracks (Fei and denotes the elastic strain energy of cracked material, which
Choo 2020a, b; Ulloa et al. 2022; You et al. 2021). Another is a function of elastic strain tensor 𝜺e, fluid mass change per
interesting approach is to introduce two independent phase unit initial volume m and variation of temperature 𝜃 , as well
fields, respectively representing the tensile and shear cracks as the set of crack Γ.
(Yu et al. 2021b; Fei and Choo 2021).
In the present study, the concept of two phase fields devel- 2.1.1 Phase Field Model
oped in our previous work (Yu et al. 2021b) is adopted for mod-
eling tensile and shear cracks. It is further extended to three- Following the regularization strategy proposed by Bourdin
dimensional coupled thermo-hydromechanical problems. In et al. (2000), the crack can be represented by the scalar phase
particular, the influence of fluid pressure due to heating process field variable d, which has the range from 0 (intact material)
on the initiation and propagation of cracks is taken into account. to 1 (fully broken). Furthermore, in order to easily deal with
Moreover, the effect of material heterogeneity due to spatial mixed-mode cracks of rock-like material, a so-called double
variability of mineralogical composition of clayey rocks is also phase field method (Yu et al. 2023b) is considered to introduce

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4370 Z. Yu et al.

tensile damage dt and shear damage ds. In this way, the approx- For the sake of readability, the mentioned symbols are
imated crack surface area can be represented as: explained in nomenclature. In this study, though the ini-
tial anisotropy of elastic properties of COx claystone is

∫Ω
AΓ ≈ AtΓ (dt ) + AsΓ (ds ) = {𝛾 t (dt , ∇dt ) taken into account, for the sake of simplicity, the tensors of
(2) Biot coefficients, thermal dilation coefficients and drained
+ 𝛾 s (ds , ∇ds )}dΩ thermo-elastic coupling coefficients are taken as isotropic
ones. In order to deal with anisotropic elastic materials,
where total surface area of sharp crack AΓ is approximated
the so-called Reuss equivalent bulk modulus is here used:
by the volumetric integration of a crack density function for
Kb = kReuss (dt , ds ). The Biot modulus is accordingly given by
tensile crack 𝛾 t (dt , ∇dt ) and shear crack 𝛾 s (ds , ∇ds ). Among
1∕M = (b − 𝜙)∕Km + 𝜙∕Kf . The differential thermal coef-
three commonly employed phase-field models proposed by
ficient is given by 𝛼m = (b − 𝜙)𝛼b + 𝜙𝛼f . By using these
Bourdin et al. (2000), Miehe et al. (2010b), Wu (2017), we
constitutive relations, the elastic free energy of an intact
adopt the formulation presented by Miehe et al. (2010b),
material can be determined as:
which can be expressed as follows:
{ } 1 1 (p − p0 )
2
1 C𝜎 2
b
1 1 𝛼 2 w0e (𝜺e , p, 𝜃) = 𝝈 b ∶ 𝜺e + − 𝜃 (9)
(3)
𝛼 𝛼 𝛼 𝛼 𝛼
𝛾 (d , ∇d ) = (d ) + ld ∇d .∇d ; 𝛼 = t, s 2 2 M 2 T0
2 ld
with Biot effective stress tensor 𝜎b:
where ld denotes a length scale parameter which is in rela-
tion with the width of smeared cracks. The crack density 𝝈 b = (𝝈 − 𝝈 0 ) + B(p − p0 ) (10)
functions, denoted as 𝛾 𝛼 (d𝛼 , ∇d𝛼 ), are dependent on both the
damage variables d𝛼 and their gradients ∇d𝛼 . This non-local Given that the effective stress plays an important role in the
formulation enables the regularization of damage localiza- onset and propagation of cracking process for saturated rock-
tion issues. With the help of the approximation work, Eq. like material, we decompose the Biot effective stress into a
(1) can be rewritten as: tensile part and a compressive part:
{ b+
𝝈 = ℙ+𝜎 ∶ 𝝈 b
∫Ω
E(𝜺e , m, 𝜃, dt , ds ) = we (𝜺e , m, 𝜃, dt , ds )dΩ (11)
𝝈 b− = ℙ−𝜎 ∶ 𝝈 b
(4)
∫Ω
+ wc (dt , ds , ∇dt , ∇ds )dΩ with the operators for spectral decomposition of stress ten-
sor ℙ±𝜎 proposed in Lubarda et al. (1994), Murakami (2012).
with elastic strain energy of cracked material we , and the Assuming the extension behavior is affected by tensile crack
energy density per unit volume requested to create the crack and the compression behavior is mainly influenced by shear
wc: crack, the elastic free energy of the damaged material can
be expressed as:
wc (dt , ds , ∇dt , ∇ds ) = gtc 𝛾 t (dt , ∇dt ) + gsc 𝛾 s (ds , ∇ds ) (5)
we (𝜺e , p, 𝜃, dt , ds ) = ht (dt )w0+
e
+ hs (ds )w0−
e

1 (p − p0 )
2
1 C𝜎 2
b
(12)
2.1.2 Elastic Free Energy + − 𝜃
2 M 2 T0
For an intact saturated porous medium, the constitutive rela-
with
tions due to thermo-poroelastic theory (Coussy 2004; Cheng
{
2016) can be expressed as: w0+ = 12 𝝈 b+ ∶ 𝜺e
(13)
e

𝝈 − 𝝈 0 = ℂ0b ∶ 𝜺e − bI(p − p0 ) − 3Kb 𝛼b I𝜃 (6) w0−

e
= 12 𝝈 b− ∶ 𝜺e

( ) The decomposition of Biot effective stress tensor is due

m to the operators ℙ±𝜎 (see the detail of old study (Yu et al.
p − p0 = M −bI ∶ 𝜺e + 0 + 3𝛼m M𝜃 (7)
𝜌f 2021a)). A common form h𝛼 (d𝛼 ) = (1 − k)(1 − d𝛼 )2 + k
with 𝛼 = t, s is used to represent the degradation of failure
process. In this study, k = 1 × 10−9 is applied to avoid the
s − s0 = s0m m + 𝛼b I ∶ (𝝈 − 𝝈 0 ) calculating error after the material is fully broken ( d𝛼 = 1).
C𝜎b (8)
− (3𝛼m − b𝛼b I)(p − p0 ) + 𝜃
T0

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4371

2.1.3 Evolution of Crack Fields crack respectively, the irreversibility of cracking process is

taken into account.
Given that all the terms of total energy functional are obtained
and introduced, the governing equations for tensile crack dt
and shear crack ds can be determined by solving the minimiza- 3 Numerical Implementation in Finite
tion problem of Eq. (4): Element Method
⎧ 𝜕w
⎪ − e − g𝛼 𝛿d𝛼 𝛾 𝛼 = 0 , ḋ𝛼 > 0 , in Ω The THM problem considering tensile and shear crack can
⎪ 𝜕d𝛼 c
be solved by using the framework of finite element method.
⎪ According to Eqs. (6), (7), (8) and (15), the weak forms for
⎨ − 𝛼 − g𝛼c 𝛿d𝛼 𝛾 𝛼 ≤ 0 , ḋ𝛼 = 0 , in Ω
⎪ 𝜕we
(14) five unknown fields are expressed as:
⎪ 𝜕d
⎪
∫Ω ∫Ω
⎪ 𝜕𝛾 .n = 0 , on 𝛿Ω 𝛿𝜺 ∶ ℂb (dt , ds ) ∶ 𝜺dV − 𝛿𝜺 ∶ (𝛿p)IdV
⎪ 𝜕𝛁d𝛼 (19)
⎩
∫Ω ∫Ωf
− 𝛿𝜺 ∶ (3𝛼b Kb (dt , ds )T)IdV = t.𝛿udS
More precisely, the evolution of two crack fields are
described by the following equations:
� � k(dt ) k(dt )
∫Ω ∫
∇p.∇(𝛿p)dV = 𝛿p.∇p̄ndS
⎧ −h�t (dt )w0e+ − gtc 1 dt − ld Δdt = 0 , ḋ t > 0 𝜇 Ω𝜔 𝜇
⎪ ld
(15)
∫Ω M 𝜕t
⎨ 1 𝜕p
� � − 𝛿pdV
⎪ −h� (ds )w0 − gs 1 ds − l Δds = 0 , ḋ s > 0
⎩ s e− c ld d
𝜕𝜀
∫Ω 𝜕t ∫Ω m 𝜕t
𝜕𝜃
− b kk 𝛿pdV + 3𝛼 𝛿pdV
where the two elastic energy of intact material w0+ e and we
0−

are represented in Eq. (13). It is reasonable to define the ten- (20)

sile crack evolution is driven by the positive part of elastic
∫Ω ∫Ωq ∫Ω
𝜕𝜃
𝜆∇T.∇(𝛿𝜃)dV = 𝜆𝛿T∇T.̄ndS − C𝜀b 𝛿𝜃dV
energy w0+e . On the other hand, given that the pure compres-
𝜕t

sion behavior can not contribute to create the shear crack, 𝜕𝜀kk
∫Ω
− (3𝛼b Kb (dt , ds )T0 ) 𝛿𝜃dV
only a part of negative part of elastic energy w0−
e can be seen
𝜕t
(21)
∫Ω
𝜕p
as the driving quantity for compressive-shear crack (see (Yu + (3𝛼m T0 ) 𝛿𝜃dV
𝜕t
et al. 2023b) for detail):
k(dt )
∫Ω 𝜇
+ ∇p.(Cp ∇𝜃)𝛿𝜃dV
1 ⟨𝜎1 ⟩− − ⟨𝜎3 ⟩− ⟨𝜎1 ⟩− + ⟨𝜎3 ⟩−
t t t t
ws− = ⟨ + tan𝜑 − c⟩2+ (16)
2G 2cos𝜑 2

∫Ω
{(gtc ∕ld + 2Ht )dt 𝛿dt + gtc ∇dt ⋅ ∇(𝜕dt )}dV
with the bracket ⟨.⟩± such as:
(22)
⟨a⟩+ = a, ⟨a⟩− = 0, a ≥ 0
�
∫Ω
= 2Ht 𝜕dt dV
⟨a⟩+ = 0, ⟨a⟩− = a, a < 0 (17)

with cohesion c and friction angle 𝜑 of material. The major

∫Ω
and minor Terzaghi effective principal stress 𝜎1t and 𝜎3t are {(gsc ∕ld + 2Hs− )ds 𝛿ds + gsc ∇ds ⋅ ∇(𝜕ds )}dV
considered according to the study (DeBuhan and Dormieux (23)
∫Ω
1996; Lydzba and Shao 2002). By substituting the physi- = 2Hs− 𝜕ds dV
cally-based driving energy into (15), w0e− can be replaced by
ws−. Finally, based on the concept of Miehe et al. (2010a), Among these relations, the effect of damage is mainly real-
two energy history functionals are defined as follows: ized on mechanical field and pore pressure field according
{ t to the elastic stiffness of undamaged material:
H (t) = max[w0e+ (𝜏)]𝜏∈[0,t] { }
(18)
Hs− (t) = max[ws− (𝜏)]𝜏∈[0,t] ℂb (dt , ds ) = ht (dt )ℙ+𝜎 + hs (ds )ℙ−𝜎 ∶ ℂ0b (24)

Given that the history functionals H t and H−s are used instead and the permeability tensor of cracked porous medium:
of w0e+ and ws− in the evolution equation of tensile and shear
k(dt , ds ) = k0 exp(𝜂k max(dt , ds )) (25)

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4372 Z. Yu et al.

with the elastic stiffness of undamaged material ℂ0b , evolu- Ωp

tion rate of permeability 𝜂k and the initial permeability ten- fp = (26)
Ω p + Ωm
sor of intact porous medium k0 . The drained bulk modulus
Kb (dt , ds ) can be calculated according to obtained ℂb (dt , ds ).
The damage effect for thermal parameters of the material is Ωi
fi = (27)
ignored in this study. Ω i + Ωp + Ωm
In the framework of the finite element method, fol-
lowing the previous work, the THM coupling problem Ω is the volume of representative volume element, while
formulation (Yu et al. 2021b) and the formulation of the Ωp , Ωi and Ωm are respectively the volumes occupied by
tensile and shear crack problem (Yu et al. 2023b) can be pores, inclusions and solid clay matrix. To account for the
solved. Considering the physical characteristic of the rock inclusion-matrix morphology, we use the standard Mori-
material, the temperature field is solved by an explicit Tanaka scheme for each homogenization step (Mori and
decoupled algorithm, as it is not sensitive to the effects of Tanaka 1973). In the first step, we determine the effective
the mechanical field, the pore pressure field and the dam- elastic tensor of the porous matrix (ℂmp) by considering the
age fields. In order to save the computation memory and effect of porosity. One gets:
time, the fixed stress iteration (Kim et al. 2011) is used to ℂmp = (1 − fp )ℂm ∶ [(1 − fp )𝕀 + fp (𝕀 − ℙp ∶ ℂm )−1 ]−1 (28)
solve the hydromechanical coupling. And the so-called
Alternate Minimization (AM) solver proposed in Bourdin ℙp (ℂm ) is the fourth order Hill tensor for ellipsoidal pores
et al. (2000), Bourdin et al. (2008) is used to couple the (Giraud et al. 2007; Zhao et al. 2022), ℂm is the elastic tensor
displacement field and phase fields because of its good of solid matrix, and 𝕀 is the fourth order unit tensor. At the
robustness performance. mesoscopic step of homogenization, the macroscopic elastic
tensor of heterogeneous rocks is determined after including
the influence of mineral inclusions:

4 Consideration of Material Heterogeneity ℂhom = ℂmp + [fi (ℂin − ℂmp ) ∶ 𝔻in ] ∶ [𝕀 + fi (ℂin − 𝕀)]−1
(29)
In rock-like materials, cracks often start due to variations in with
the material, such as at interfaces between stiff inclusions
and a weak matrix, or in weak zones with high porosity. 𝔻in = [𝕀 + ℙi ∶ (ℂin − ℂmp )]−1 (30)
Previous studies have used a numerical trick of placing ℙi (ℂmp ) is the Hill tensor for spherical inclusions (Giraud
weak elements in the material to facilitate crack initiation et al. 2007; Zhao et al. 2022). It is worth noting that, we
and propagation. However, in this work, we use a micro- combine different families of mineral particles into a single
mechanics based approach to more accurately reflect the equivalent inclusion phase, and represent its elastic stiffness
effect of material heterogeneity on mechanical properties. tensor as ℂin.
Specifically, we use an analytical homogenization method Since the spatial variability of macroscopic elastic
to determine macroscopic elastic properties as explicit func- properties in COx claystone is attributed to the non-uni-
tions of porosity and inclusion volume fraction at the micro- form distribution of porosity and inclusion concentration,
scopic scale. in this study, we assume a random distribution of poros-
For this study, we focus on the Callovo-Oxfordian (COx) ity ( fp ) and volumetric fraction of mineral inclusions ( fi )
claystone, which has been widely studied in France for its using the standard Weibull distribution function:
potential use in storing radioactive waste. Previous research
(Robinet et al. 2012) has identified two main scales of inter- ( )m𝛼 −1 [ ( ) m𝛼 ]
m𝛼 f𝛼 f
est for this material: at the mesoscopic scale, mineral parti- 𝜉𝛼 = exp − 𝛼 , 𝛼 = p, i (31)
𝛽𝛼 𝛽𝛼 𝛽𝛼
cles like quartz, calcite, and pyrite are randomly distributed
in a clay matrix; at the microscopic scale, the majority of where f𝛼 is the volume fraction of pores (𝛼 = p) and mineral
pores are found inside the clay matrix. inclusions (𝛼 = i ). 𝛽𝛼 is the scale parameter of the distribu-
In order to determine the macroscopic elastic properties, tion which represents the mean value of random variable. m𝛼
two steps of linear homogenization is performed consider- is the homogeneity index of the material.
ing two micro-structure parameters: the porosity of the clay In this study, we have adopted an elastic-damage model.
matrix fp and the volume fraction of mineral inclusions fi , Plastic deformation is neglected. Moreover, there exist
one gets: theoretical homogenization models for predicting mac-
roscopic elastic properties with mineral compositions
and porosity. Therefore, it seems to be a good choice

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4373

Fig. 1  Partial view of Andra URL (Seyedi et al. 2017a) (left) and view of CRQ heating test (right)

to consider effect of material heterogeneity on elastic heating phase was only heated by the remaining nine heater
properties. devices.

5.2 Input Data of Modeling

5 Numerical Simulation
As shown in Fig. 3 (left), the studied 3D domain consists of
In this section, the proposed THM modeling, which takes a cube with dimensions of 50 m × 50 m × 50 m in order to
into account the double phase field method, is applied to represent the far field. On one surface of the studied cube,
simulate the results of the in-situ CRQ test performed at half of the GCS drift with 5.2 m diameter is located. Ten
Andra’s Underground Research Laboratory (URL) (Seyedi heater boreholes with a diameter of 0.172 m and a length of
et al. 2017a). The simulation focuses on capturing the 20 m are placed on the wall of the GCS drift. The detailed
thermo-hydromechanical responses and failure behavior coordinates of these ten heater boreholes are shown in the
of COx claystone during excavation and heating-cooling appendix. For simplicity, the four boreholes with sensors
processes. are not included. In this way, 883,232 tetrahedral elements
with 148,807 nodes are used to mesh the studied domain,
5.1 General Presentation of CRQ Test as shown in Fig. 3 (right). Since the rock around the heaters
is expected to be damaged, the finest mesh is defined for
Figure 1 provides a partial view of the Andra URL, which this heater part, which is 10-20 m deep from the GCS drift
is situated at a depth of -490 m. The study focuses on the wall. This is balanced choice between a fine description of
GCS gallery located within the COx claystone layer. The damaged zones and computation costs such as computing
CRQ (Representative THM Behavior of a HLW Cell) in-situ memory and time used.
experiment was performed in the GCS gallery, as shown in Based on the study of Armand et al. (2014), Seyedi et al.
Fig. 1 (right). The experiment involved drilling ten 20 m (2017b), the initial stress state of the studied domain is
long heater boreholes (CRQ1701–CRQ1710) horizontally defined as follows: (i) the minimum principal stress 𝜎h is −
from the GCS drift wall, with the heaters installed in the 12.4 MPa (parallel to the heater boreholes); (ii) the maxi-
last 10 m. In addition, sixteen sensors were positioned in mum principal stress 𝜎H is −16.1 MPa (parallel to the GCS
four additional boreholes (CRQ1720–CRQ1723) to moni- gallery); and (iii) the vertical principal stress 𝜎v is −12.7
tor temperature and pore pressure. The aim of the experi- MPa (perpendicular to the heater boreholes and GCS gal-
ment was to examine the cracking process resulting from lery). The initial temperature and pore pressure of the whole
the THM behavior of the rock, with the anticipated fracture domain are assumed to be uniform, with T0 = 22◦ C and p0
location being between the two central boreholes and 15 m =4.7 MPa, which are the values at a depth of – 490 m from
from the GCS drift wall, which is at the mid-plane of the the ground.
heater devices. The boundary conditions are defined as shown in Table 1.
Two heating phases were performed using the mentioned The lateral faces are considered to have symmetrical con-
heaters, and the heat power applied is shown in Fig. 2. It ditions for all THM fields. The top and bottom faces are
is worth noting that the heater devised in CRQ1704 was permeable to heat flux and fluid flux. After excavation of
damaged after the first heating phase. Therefore, the second the GCS gallery, the drift wall is not fixed for displacement

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4374 Z. Yu et al.

Fig. 2  Timeline of the CRQ

experiment with evolution of
the heat power history of each
heater device: (Left) first heat-
ing phase and (Right) second
heating phase (provided by task
specification)

Fig. 3  Geometry and mesh of CRQ simulation

Table 1  Boundary conditions of Boundary Thermal Hydraulic Mechanical

the CRQ simulation
Lateral faces Free surface Free surface Zero normal displacement
Top face 22◦ C 4.7 MPa −12.7 MPa
Bottom face 22◦ C 4.7 MPa Zero normal displacement
GCS drift wall T(t) 0.1 MPa Free surface
Borehole wall (before heating) Free surface 0.1 MPa Free surface
Borehole wall (during heating) Free surface Free surface Free surface

but is fixed for pore pressure at 0.1 MPa. The temperature of strength, thermal, and hydraulic properties. For simplicity,
the gallery is applied based on the values measured during only the initial anisotropy of elastic properties, permeabil-
the experiment. The boundary conditions for the boreholes ity, and thermal conductivity is considered, while the other
is changed while the heating phase begins. The surface of parameters are assumed to be isotropic. Following the
the drift changes from permeable to impermeable due to the previous experiment (Braun 2019) and simulation work
installation of the resin. (Yu et al. 2023a) for a laboratory-scale heating test on
According to previous studies (Armand et al. 2017a), the same rock, material parameters are provided as shown
the COx claystone is considered to be a transversely iso- in Table 2. It is important to highlight that the measured
tropic material with regards to elastic modulus, failure parameters in the experiment can be influenced by several

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4375

Table 2  Material parameters Parameters Unit Notation Value

for COx claystone from Braun
(2019) Young’s modulus (parallel to bedding) 109 Pa E∥ 6.0
Young’s modulus (perpendicular to bedding) 10 Pa
9 E⟂ 4.0
Poisson ratio (parallel to bedding) – 𝜇∥∥ 0.3
Poisson ratio (perpendicular to bedding) – 𝜇⟂∥ 0.3
Shear modulus (perpendicular to bedding) 109 Pa G⟂∥ 1.7
Density of solid grains kg/m3 𝜌s 2770
Biot coefficient – b 0.8
Porosity – 𝜙 0.18
Initial permeability (parallel to bedding) 10−20 m2 k∥0 4.0
Initial permeability (perpendicular to bedding) 10−20 m2 k⟂0 1.33
Equivalent thermal conductivity (parallel to bedding) W/m/K 𝜆∥ 2.1
Equivalent thermal conductivity (perpendicular to bedding) W/m/K 𝜆⟂ 1.28
Heat capacity of porous medium J/kg/K Cp 1050
Volumetric thermal expansion of solid grains 10−5 1/K 𝛼s 4.5

factors, including potential variable discrepancies and Table 3  Parameter of phase field model for CRQ simulation
inherent variability among individual rock samples. The Parameters Value
material parameters employed in this simulation have been
carefully calibrated, taking into account the distinctions Material toughness gtc=4500
between laboratory-scale and in-situ tests. This calibra- N/m; gsc
=500 N/m
tion falls within an appropriate range, informed by the
Friction angle 𝜑=15◦
accumulated knowledge of COx claystone from ANDRA’s
Cohesion c=0.1 MPa
experience.
Scale length ld=0.18 m
The regular parameters of water are used in this work.
Permeability variation 𝜂k=40
For simplicity, the following parameters are considered as
constant: bulk modulus of water Kf = 2.2 GPa, density of
water 𝜌f = 1000 kg/m 3 , heat capacity of water Cf = 4180
J/kg/K. The water dynamic viscosity 𝜇f (Pa ⋅ s) is due to parameters (Choo and Sun 2018; Wu et al. 2019). For the
temperature ( ◦ C ), it has: evolution of tensile damage, one gets:
√
𝜇f (T) = 4.2844 × 10−5 + (0.157(T + 64.993)2 − 91.296)−1 Eeq
(32) 𝜎t = gtc (34)
3ld
And the volumetric thermal expansion of water (1/◦ C ) is
also dependent on temperature (◦ C): with the average value of uniaxial tensile strength for COx
claystone 𝜎t = 3 MPa, and the equivalent isotropic elastic
𝛼f (T) = −6T 4 + 1660T 3 − 197796T 2 + 16862446T − 64319951 modulus Eeq = (E∥ × E∥ × E⟂ )1∕3 . Following the work of
(33) Miehe et al. (2010a), a scale length is defined according to
It is worth noting that, for the reason of comparison between the element size h used in the area predicted to formation
different teams which worked on this simulation of in-situ of damage: ld = 2h . On the other hand, according to Eq.
test, Andra proposed that these two parameters are the most (15), the evolution of shear damage zone is due to three
sensitive ones to temperature variation. parameters: toughness coefficient of shear damage gsc , fric-
Furthermore, since the phase field method is applied tion angle 𝜑 and cohesion c. It is noticed that although the
to describe the cracking process in this study, the param- physical significance of the friction and cohesion is the same
eters defined in this numerical method should be chosen as that widely adopted, their values in the context of phase-
carefully as shown in Table 3. Although scale length ld field model are generally different with those used in clas-
is a defined parameter which controls the width of local- sical Mohr-Coulomb criterion. Their value are chosen so
ized damage, it also serves as a material property within that one obtains reasonable values of uniaxial compression
phase-field modeling. This is due to the fact that the peak strength and tensile strength. The detailed explanation of the
stress of a material is determined by the scale length in choice of these parameters for COx claystone is presented
conjunction with the toughness parameter and elasticity in the previous study (Yu et al. 2021a). Moreover, the scale

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4376 Z. Yu et al.

6 CRQ1720-1723. The two peaks in the figure correspond

to the end of the first heating phase and the second heat-
5 ing phase. By comparing the experimental and numerical
results, it is clear that the tendencies are well-reproduced and
4
the peak values are almost identical. The peak values of the
3 second heating are higher than the peak values of the first
heating, except for the points on CRQ1722. This is because
2 the heater CRQ1704 was damaged during the second heating
phase, and CRQ1722 is the closest borehole to the damaged
1
heater. Nevertheless, it should be noted that reproducing the
0 temperature evolution in the in-situ test scale is relatively
0 5 10 15 20 straightforward due to the fact that the heat transfer process
in the rock-like material remains unaffected by initial frac-
ture or crack nucleation.
Fig. 4  Pore pressure evolution around the GCS drift at 2018/7/31, Figure 6 illustrates the temperature distributions at the
comparing the experimental result and numerical result end of the first heating phase and the end of the second
heating phase. The temperature concentration area plays
length ld , which controls the width of localized damage, is an important role to study the thermal damage. The middle
determined by the element size. In this study, the value used plane of the heating section is expected to exhibit the high-
is twice the element size of the predicted damaged area. The est temperatures. Hence, this plane is selected to analyze the
permeability variation coefficient 𝜂k , defined in Eq. (25), is THM coupling behavior resulting from heating. After the
identified with the help of experimental work (de La Vais- completion of the first heating phase, it is evident that the
siere et al. 2015), and it has worked well in previous numeri- region with the highest temperature is concentrated around
cal studies (Yu et al. 2021a) that considered 2D problems the two middle boreholes, namely CRQ1709 and CRQ1710.
for the same rock. However, at the end of the second heating phase, a shift in
Only the excavation damage zone (EDZ) around the GCS the position of the zone with the highest temperature can be
drift is considered. The EDZ is located in the area with a observed. This change is attributed to the damage sustained
depth of 2.6 m from the GCS drift wall. The permeability by the heater in CRQ1704 (the top center borehole), caus-
is increased by 10 times the initial value in this zone. In ing the zone of elevated temperature to relocate towards the
this way, after the excavation of the GCS gallery, there is a lower of the two central boreholes, as depicted in the iso-
gradient of pore pressure close to the drift wall due to the thermal map.
seepage phenomenon. Comparing the pore pressure meas- The onset of damage is mainly caused by the operation
ured by the sensor around the GCS gallery on 2018/7/31, as of excavation and heating during the CRQ test. As shown
shown in Fig. 4, the numerical results show good agreement in Fig. 7, the damage due to the excavation of boreholes is
with the effect of the EDZ. It is indeed worth noting that the reproduced as the shear damage. This damage creates what
main feature of the pore pressure distribution that needs to is known as a borehole damaged zone (BDZ), a phenom-
be reproduced is influenced by the EDZ of the GCS gallery. enon commonly observed in in-situ experiments with COx
However, it’s important to acknowledge that the modeling claystone. The maximum value of the shear damage variable
of GCS excavation in this study has certain limitations. For ds is about 0.23, which indicates that only micro-cracking
example, the mesh around the GCS may not be fine enough, occurs around the boreholes, and not macro-cracking. These
and the time-dependent behavior during the excavation pro- micro-cracks lead to increased permeability, which is a con-
cess is not taken into account. sequence of pores being connected by fractures, as described
The simulation is applied from the excavation of the GCS in Eq. (25).
gallery to the end of the second cooling phase. The detailed We visualize in Fig. 8 the region with higher permeability
dates of the main operations are shown in the appendix. For due to the excavation, with elements having permeability
better readability, in the following numerical presentations, values 10 times higher than the initial permeability being
the dates are replaced by days, where day t = 0 is defined as highlighted. It can be seen that these elements within the
the day before the excavation of the GCS gallery. BDZ eventually form a tube that connects with the excava-
tion damaged zone (EDZ) of the GCS gallery. As mentioned
5.3 Simulation Results previously, the borehole walls are defined as impermeable
due to the installation of the resin. As a result, water can eas-
Figure 5 shows the variation of temperature over time ily flow to the GCS gallery through the connected high-per-
at the sixteen sensor points located in four boreholes: meability zones. This phenomenon is verified in the in-situ

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4377

100 100
Exp. CRQ1720_01 Exp. CRQ1721_01
Exp. CRQ1720_02 Exp. CRQ1721_02
Exp. CRQ1720_03 Exp. CRQ1721_03
Exp. CRQ1721_04
80 Exp. CRQ1720_04
Exp. CRQ1720_05
80 Exp. CRQ1721_05
Num. CRQ1720_01 Num. CRQ1721_01
Num. CRQ1720_02 Num. CRQ1721_02
Num. CRQ1720_03 Num. CRQ1721_03
60 Num. CRQ1720_04 60 Num. CRQ1721_04
Num. CRQ1720_05 Num. CRQ1721_05

40 40

20 20
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

100 100
Exp. CRQ1722_01 Exp. CRQ1723_01
Exp. CRQ1722_02 Exp. CRQ1723_02
Exp. CRQ1722_03 Exp. CRQ1723_03
80 Exp. CRQ1722_04
80 Exp. CRQ1723_04
Exp. CRQ1722_05 Exp. CRQ1723_05
Num. CRQ1722_01 Num. CRQ1723_01
Num. CRQ1722_02 Num. CRQ1723_02
Num. CRQ1722_03 Num. CRQ1723_03
60 Num. CRQ1722_04 60 Num. CRQ1723_04
Num. CRQ1722_05 Num. CRQ1723_05

40 40

20 20
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

Fig. 5  Evolution of temperature at the sensor points of CRQ 1720-1723, comparing the experimental results and numerical results

Fig. 6  Distribution of temperature (◦ C ) at the end of the first heating phase (top) and the end of second heating phase (bottom)

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4378 Z. Yu et al.

Fig. 7  Distribution of the shear

damage according to the exca-
vation of heater boreholes

Fig. 8  Distribution of the elements with greater than 10 times value higher than initial permeability k0

test according to the observation of the seepage around the are presented in the right figure with different colored icons.
boreholes on the GCS drift wall. Among them, the sensors CRQ1720 and 1721 are located
By incorporating the seepage phenomenon, the numeri- inside the square surrounded by ten heaters, while CRQ1722
cal simulation results for pore pressure is depicted in Fig. 9. and 1723 are located outside of it. It results in the blue and
The pore pressure has a reduction because of the borehole orange dotted curves of the pore pressure having higher peak
excavation at the beginning of the simulation. After the values than the green and red ones. (The colors of the curves
holding phase, the first heating phase begins at 273rd day. and icons are the same.) This feature is well reproduced in
The simulation results closely align with the experimental the numerical results. However, the simulation results have
findings for both the heating and cooling phases, with one the limitation. Some features, such as: (i) the green dotted
exception. The data from sensors 01 and 05, which are situ- curve being slightly higher than the red one; (ii) the green
ated far from the heating section, exhibit some disparities. and red dotted curves having a slower reduction than the
Regarding the reduction observed in the measured data from blue and orange ones during the two cooling phases, are not
sensor 05 of CRQ1720 before the end of the first heating able to be reproduced in the simulation. The model used in
phase, it is likely attributable to a technical issue during the this work has its limitation. To improve the simulation and
experiment. As for the 01 sensors, located at the deepest address these issues, considering the elasto-plastic behavior
region approximately 5 ms below the end of the boreholes, of the material and incorporating an open-close crack func-
the measured pore pressure shows minimal impact from the tion in the numerical model could be beneficial. However,
two heating phases. This discrepancy, particularly evident the reason about these differences between experimental and
in the sensors of CRQ1720, could indicate the presence of numerical results may be complex. The evolution of pore
an initial fracture zone between the BDZ and sensor 01 of pressure in the in-situ test can indeed be influenced by vari-
CRQ1720. ous factors, including the leakage of resin or the presence of
Of note, sensors 03 of four boreholes are located in the initial fractures in the rock. These factors can introduce com-
mid-plane of the heating area, which is considered the most plexities and uncertainties in the behavior of pore pressure.
important region to study failure behavior. As such, the To gain a deeper understanding of the damage effects
evolution of pore pressure at the sensor 03 locations is pre- observed during the in-situ heating test, we present an addi-
sented together in Fig. 10. The location of four sensor 03 tional set of numerical results depicted in Fig. 11. These

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4379

25 Exp. CRQ1720_01
25
Exp. CRQ1721_01
Exp. CRQ1720_02 Exp. CRQ1721_02
20 Exp. CRQ1720_03
Exp. CRQ1720_04
20 Exp. CRQ1721_03
Exp. CRQ1721_04
Exp. CRQ1720_05 Exp. CRQ1721_05
15 Num. CRQ1720_01 15 Num. CRQ1721_01
Num. CRQ1720_02 Num. CRQ1721_02
Num. CRQ1720_03 Num. CRQ1721_03
10 Num. CRQ1720_04 10 Num. CRQ1721_04
Num. CRQ1720_05 Num. CRQ1721_05

5 5

0 0

-5 -5
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

25 25
Exp. CRQ1722_01 Exp. CRQ1723_01
Exp. CRQ1722_02 Exp. CRQ1723_02
20 Exp. CRQ1722_03 20 Exp. CRQ1723_03
Exp. CRQ1722_04 Exp. CRQ1723_04
Exp. CRQ1722_05 Exp. CRQ1723_05
15 Num. CRQ1722_01 15 Num. CRQ1723_01
Num. CRQ1722_02 Num. CRQ1723_02
Num. CRQ1722_03 Num. CRQ1723_03
10 Num. CRQ1722_04 10 Num. CRQ1723_04
Num. CRQ1722_05 Num. CRQ1723_05

5 5

0 0

-5 -5
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

Fig. 9  Evolution of pore pressure at the sensor points of CRQ 1720-1723, comparing the experimental results and numerical results

Fig. 10  Evolution of pore pres- 25

sure at the sensor points of sen- Exp. CRQ1720_03
Exp. CRQ1721_03
sor 03, comparing the experi-
mental results and numerical 20 Exp. CRQ1722_03
Exp. CRQ1723_03
results with considering damage Num. CRQ1720_03
effect 15 Num. CRQ1721_03
Num. CRQ1722_03
Num. CRQ1723_03
10

-5
0 100 200 300 400 500 600 700

results are obtained through a thermo-poro-elastic model be attributed to the absence of the BDZ, where pore water
simulation that does not take into account the influence of is trapped due to the low permeability of COx claystone.
damage. Similar to Fig. 10, the numerical results of pore Consequently, the pore pressure in the numerical simulation
pressure is compared with the experiment data of the sensors increases to an unrealistic value during the heating phases.
03. While the trends of the curves are effectively reproduced, Furthermore, Fig. 12 displays the distribution of pore
the peak value in the numerical results is notably higher than pressure at the two ends of the heating phase, taking into
that observed in the experimental data. This discrepancy can account the damage effect. In contrast, Fig. 13 represents

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4380 Z. Yu et al.

Fig. 11  Evolution of pore pres- 40

sure at the sensor points of sen- Exp. CRQ1720_03
Exp. CRQ1721_03
sor 03, comparing the experi-
Exp. CRQ1722_03
mental results and numerical 30 Exp. CRQ1723_03
results without considering Num. CRQ1720_03
damage effect Num. CRQ1721_03
Num. CRQ1722_03
20 Num. CRQ1723_03

10

0 100 200 300 400 500 600 700

Fig. 12  Distribution of pore pressure (Pa) at the end of the first heating phase (top) and the end of second heating phase (bottom), simulation
with considering damage effect

the distribution of pore pressure in a simulation where the presence of the BDZ leads to a larger and more uniform
damage effect is not considered, allowing for a meaning- overpressure zone. As a consequence, the distribution of
ful comparison between the two scenarios. Similar to the pore pressure, which is influenced by the presence of the
temperature results, the pore pressure induced by the sec- BDZ, can directly impact the formation of thermal damage.
ond heating phase is higher compared to the pore pressure After successfully reproducing the variation of pore pres-
induced by the first heating phase. Additionally, due to the sure, the thermal damage evolution is now investigated. It
failure of the heater, the overpressure zone that was previ- is noticed that it is very difficult to quantitatively measure
ously observed around the three top boreholes is no longer intensity of thermal damage in in-situ experiment. Only dis-
present. In Fig. 12, it is evident that the presence of the BDZ tributions of micro-cracks are characterized through drilled
results in a reduction of pore pressure around the boreholes. samples. Therefore, the comparison between modeling and
On the other hand, unlike the overpressure zone is more experiment is mainly focused on the extent of cracked or
concentrated in the case without damage consideration, the damaged zone. Figure 14 shows the distribution of tensile

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4381

Fig. 13  Distribution of pore pressure (Pa) at the end of the first heating phase (top) and the end of second heating phase (bottom), simulation
without considering damage effect

damage due to the two heating phases. The distribution of claystone, it should be considered in the simulation of CRQ
shear damage is not shown here, since the shear damage is test. According to the previous studies (Armand et al. 2017b;
almost not changed during the heating. The maximum value Zhao et al. 2022), both the solid clay matrix and COx clay-
of tensile damage indicates that there is almost no hydro- stone are assumed to exhibit transversely isotropic elastic
thermal damage at the end of the first heating. A few dam- behavior, while the equivalent inclusion is assumed to be
age appears due to the second heating, but no macro crack isotropic. The parameters used in this work for homogeniza-
is predicted to see. The damaged zone is mainly located tion and random distribution are presented in detail in
around the lower hole of the two center boreholes, because Table 4 based on the previous studies (Farhat et al. 2017;
the highest pore pressure can be found in this area. Zhao et al. 2022). It is worth noting that the elastic proper-
In this subsection, a simulation of the CRQ test is con- ties of constituent phase are identified by using an inverse
ducted by incorporating the consideration of damage. Both optimization procedure by using the Eqs. (28) and (29).
excavation-induced damage and heating-induced damage are After taking typical values for calcite and quartz inclusions,
reproduced using the proposed double phase field method. the elastic parameters of clay matrix are calculated from the
In addition, the effect of the BDZ on reducing the pore pres- measured macroscopic values for a given set of porosity and
sure in the over-pressure zone around the boreholes is taken mineral composition. With these values, the homogenized
into account. After successfully reproducing the peak value elastic properties can be obtained: E∥hom = 6.08 GPa,
of the pore pressure, the simulation results can be used to E⟂hom = 4.42 GPa, 𝜈∥hom = 0.28, 𝜈∥⟂hom
= 0.33 and Ghom = 2.7
∥⟂
predict the hydro-thermal fracture that occurs at the end of GPa. These macroscopic elastic properties are suitable com-
the simulation. Overall, this approach provides a convincing pare with the measured parameters of COx claystone from
and effective result for predicting and studying the thermo- Armand et al. (2017b).
hydromechanical behavior of claystone in the context of the Based on previous simulation results that considered a
CRQ test. uniform rock, it was observed that the critical failure behav-
ior occurred in the vicinity of the heated part of boreholes.
5.3.1 Study of Material Heterogeneity Consequently, we have limited our study of material het-
erogeneity to the area with a refined mesh, as illustrated
As introduced before, given that material heterogeneity plays in Fig. 3. In this way, Fig. 15 illustrates a representative
an important role during the cracking process of COx

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4382 Z. Yu et al.

Fig. 14  Distribution of the tensile damage according to the first heating phase (left) and second heating phase (right)

Table 4  Parameter of material heterogeneity for CRQ simulation example of the random distribution of porosity and volu-
Parameters Value
metric fraction of inclusions. Since the Weibull distribution
function is widely used in many situations, it is also adopted
Young’s modulus (solid matrix) E∥m = 3.6 GPa, E⟂m = 2.4 GPa here. The involved parameters are chosen so that the varia-
Poisson’s ratio (solid matrix) 𝜈∥m = 0.3, 𝜈∥⟂
m
= 0.3 tion of porosity and inclusion fraction correspond to typical
Shear modulus (solid matrix) Gm∥⟂
= 1.02 GPa variability of these parameters observed on COx claystone.
Young’s modulus (inclusion) Ei = 98 GPa Since we anticipate that the simulation outcomes will
Poisson’s ratio (inclusion) 𝜈i = 0.15 vary depending on the random distribution of pores and
Inclusion volume fraction 𝛽i = 0.4 inclusions, we selected three simulations (denotes Num.
Pore volume fraction 𝛽p = 0.16 1, 2 and 3) with different sets of random distribution for
Homogeneity index mi = mp = 1.5 analysis. At first, Fig. 16 shows the distribution of shear
damage induced to excavation by using three different

Fig. 15  Distribution of pores volume fraction (top) and inclusion volume fraction (bottom)

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4383

Fig. 16  Distribution of shear damage induced by excavation, simulated by using different random distribution of pore and inclusion: Num. 1, 2
and 3

sets of random distribution. The maximum value of ds dif- pore pressure within the BDZ in the simulation that con-
fers among the three simulations due to variations in the siders heterogeneity. This significant reduction improves
homogenized elastic properties resulting from differences the evolution of pore pressure during the cooling phases,
in porosity and inclusion volume fraction. However, the bringing it closer to the experimental data. The reason for
form of BDZ remains unchanged compared to the simu- this reduction in the BDZ can be attributed to the presence
lation results considering a uniform rock. Similarly, this of a larger BDZ due to the weak regions obtained through
BDZ leads to the mentioned seepage phenomenon, which the random distribution of pores.
can reduce the pore pressure at the study points as shown Figure 19 shows the distribution of tensile damage at
in Fig. 17. Even though the three simulations used differ- the end of the second heating phase. It is worth noting that
ent random distributions, the resulting numerical results the tensile damage induced by the first heating phase is not
are identical as depicted in Figure (a–c). For instance, shown since the maximum value of dt is less than 0.1. It
the three curves of CRQ1720-03 are exactly coincided as is reasonable to observe that the maximum values of dt ,
shown in Figure (d). which are 0.43, 0.79, and 0.83, respectively, are greater
Additionally, upon comparing the numerical results to the compared to the uniform simulation ( dt = 0.3) due to the
previous subsection, it is evident that the values are reduced, presence of a weak region. The element with this maxi-
particularly during the cooling phases and at the end of the mum value is located between the two central boreholes
second heating phase. To investigate this difference, a com- as expected. Since there are no elements with high damage
parison of the distribution between the two simulations is variables outside of this area in all three simulations, it
provided in Fig. 18. At the initial stage, the distribution of appears that the results are consistent.
pore pressure does not show any non-uniform characteris- By considering the material heterogeneity, a more real-
tics. Material heterogeneity is only taken into account for the istic representation of the rock behavior is achieved, and
stiffness matrix in the mechanical field, while major param- the simulation better captures the behavior of pore pres-
eters in the pore pressure field, such as permeability, are not sure in the BDZ, resulting in improved agreement with
considered. In other words, the mechanical heterogeneity is experimental data during the cooling phases. Additionally,
not strong enough to significantly influence the pore pres- the presence of weak regions in the rock, such as pores
sure field. and inclusions, can influence the occurrence of thermal
However, the principal difference between the two sim- fractures.
ulation results lies in the pore pressure within the BDZ. However, it should be noted that the simulation con-
Comparing the simulation with considering heterogeneity sidering material heterogeneity ensures that the overall
to the one without, there is a slightly greater reduction in features of the thermal crack remain unchanged compared

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4384 Z. Yu et al.

25 25
Exp. CRQ1720_03 Exp. CRQ1720_03
Exp. CRQ1721_03 Exp. CRQ1721_03
20 Exp. CRQ1722_03 20 Exp. CRQ1722_03
Exp. CRQ1723_03 Exp. CRQ1723_03
Num. CRQ1720_03 Num. CRQ1720_03
15 Num. CRQ1721_03 15 Num. CRQ1721_03
Num. CRQ1722_03 Num. CRQ1722_03
Num. CRQ1723_03 Num. CRQ1723_03
10 10

5 5

0 0

-5 -5
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

25 25
Exp. CRQ1720_03 Exp. CRQ1720_03
Exp. CRQ1721_03
20 Exp. CRQ1722_03 20
Exp. CRQ1723_03
Num. CRQ1720_03
15 Num. CRQ1721_03 15
Num. CRQ1722_03
Num. CRQ1723_03
10 10

5 5

0 0

-5 -5
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

Fig. 17  Evolution of pore pressure at sensor 03 of CRQ 1720-1723 sensor CRQ 1720-03 for simulations using different distributions of
for simulations using different distributions of pores and inclusions: pores and inclusions: Num. 1, 2 and 3, as well as the simulation result
Num. 1, 2 and 3 (a–c), and comparison of pore pressure evolution at without considering heterogeneity (d)

to the simulation without considering heterogeneity. Spe- that the THM coupling process is significantly affected by
cifically, the thermal damage mainly appears between the the BDZ induced seepage phenomenon. The excavation and
two center boreholes, and a macroscopic crack does not heating induced damage zone are identified as shear and
form (damage variable dt or ds = 1). tensile damage, respectively. Moreover, the study verifies the
model’s consistency with material heterogeneity. It is worth
noticing that in the present study, the mechanical proper-
6 Conclusion ties are identified from typical laboratory tests such as tri-
axial compression tests. The parameters related to fluid flow
This study proposes a THM modeling approach that incor- and heat diffusion are chosen by Andra, for all the teams
porates the phase-field method to account for damage involved in the DECOVALEX project. There is no fitting of
effects. The model is capable of identifying both tensile and parameters with respect to the in-situ experiment.
shear cracks induced by THM coupling behavior. To test the Therefore, the numerical simulation presented here rep-
model’s effectiveness, it is applied to simulate the CRQ test. resents an interesting phase of validation of the proposed
The results show that the model accurately reproduces the numerical model. However, further simulations of other
evolution of temperature and pore pressure at sensor points experiments are still useful for a deep validation of the pro-
and the corresponding damage zone. The study also finds posed model. Moreover, other important features, such as the
variation of critical fracture energy with temperature change,
should also be investigated.

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4385

Fig. 18  Distribution of pore pressure (Pa) at the end of the first and second heating: comparison between the simulation with and without con-
sidering damage effect

Fig. 19  Distribution of tensile damage at end of second heating phase, simulated by using random distribution of pore and inclusion: Num. 1, 2
and 3

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4386 Z. Yu et al.

Table 5  The position of sensor points used in simulation of CRQ test • The drift of GCS gallery: 15th September 2010
Sensor Coordinate-x Coordinate-y Coordinate-z (m)
• The drift of heater boreholes: between 20th and 3030
(m) (m) October 2017 (see Table)
• The first heating phase: between 3rd June and 31st July
CRQ1720-01 24.99 0.91 −0.21
2019
CRQ1720-02 17.99 0.86 −0.27
• The second heating phase: between 13th January and 14th
CRQ1720-03 15.00 0.84 −0.29
February 2020
CRQ1720-04 12.00 0.82 −0.31
CRQ1720-05 8.00 0.79 −0.32
CRQ1721-01 25.39 − 0.86 0.65
CRQ1721-02 18.38 −0.84 0.55 Acknowledgements This work is jointly supported by the French
CRQ1721-03 15.38 − 0.82 0.51 National Agency for radioactive waste management (ANDRA), the
DECOVALEX project, the National Natural Science Foundation of
CRQ1721-04 12.38 −0.81 0.47
China (No. 12202099) and the China Postdoctoral Science Foundation
CRQ1721-05 8.38 − 0.79 0.43 funded project (No. 2023M730525). DECOVALEX is an international
CRQ1722-01 24.92 − 0.57 3.73 research project comprising participants from industry, government
CRQ1722-02 17.90 −0.21 2.75 and academia, focusing on development of understanding, models
and codes in complex coupled problems in sub-surface geological and
CRQ1722-03 14.93 −0.05 2.34
engineering applications. The authors appreciate and thank the DECO-
CRQ1722-04 11.97 0.10 1.94 VALEX-2023 Funding Organizations for their financial and technical
CRQ1722-05 8.01 0.32 1.41 support of the work described in this paper. The statements made in the
CRQ1723-01 24.57 −2.34 0.27 paper are, however, solely those of the authors and do not necessarily
reflect those of the Funding Organizations.
CRQ1723-02 19.56 −2.33 0.21
CRQ1723-03 14.56 −2.31 0.15 Data Availability The data that support the findings of this study are
CRQ1723-04 9.57 −2.29 0.1 available from the corresponding author upon reasonable request.
CRQ1723-05 4.57 −2.27 0.07

Table 6  Dates of the heater boreholes drilling for CRQ test

References
Borehole Drilling date Borehole Drilling date Armand G, Leveau F, Nussbaum C, deLaVaissiere R, Noiret A, Jaeggi
D, Landrein P, Righini C (2014) Geometry and properties of the
CRQ1701 14/09/2018 CRQ1706 03/09/2018 excavation induced fractures at the meuse-haute-marne url drifts.
CRQ1702 13/09/2018 CRQ1707 05/09/2018 Rock Mech Rock Eng 47:21–41
CRQ1703 27/09/2018 CRQ1708 07/09/2018 Armand G, Bumbieler F, Conil N, delaVaissiere R, Bosgiraud JM, Vu
MN (2017) Main outcomes from in situ thermo-hydro-mechanical
CRQ1704 02/10/2018 CRQ1709 11/09/2018
experiments programme to demonstrate feasibility of radioactive
CRQ1705 04/10/2018 CRQ1710 24/09/2018 high-level waste disposal in the Callovo-Oxfordian claystone. J
Rock Mech Geotech Eng 9:415–427
Armand G, Conil N, Talandier J, Seyedi DM (2017) Fundamental
aspects of the hydromechanical behaviour of Callovo-Oxfordian
Appendix: Positions of Sensors and Dates claystone: from experimental studies to model calibration and
of the Heater Boreholes Drilling validation. Comput Geotech 85:277–286
Bernard PE, Moes N, Chevaugeon N (2012) Damage growth mod-
eling using the thick level set (TLS) approach: efficient discretiza-
In order to identify the coordinates of the sensor points, tion for quasi-static loadings. Comput Methods Appl Mech Eng
the origin (0,0,0) of the model domain is located at the 233–236:11–27
mid-distance between the heads of borehole CRQ1720 and Birkholzer JT, Bond AE, Hudson JA, Jing L, Tsang CF, Shao H,
CRQ1721 (See Tables 5 and 6). In this way, the studied 3D Kolditz O (2018) Decovalex-2015: an international collabora-
tion for advancing the understanding and modeling of coupled
domain of cube with a side length of 50 m vary between thermo-hydro-mechanical-chemical (thmc) processes in geologi-
x=− 2.6 m to 47.4 m (parallel to the heater boreholes); cal systems
y=− 25.0 m to 25.0 m (parallel to the GCS gallery) and z Birkholzer JT, Tsang CF, Bond AE, Hudson JA, Jing L, Stephansson
= – 25 m to 25 m (perpendicular to the heater boreholes O (2019) 25 years of Decovalex-scientific advances and lessons
learned from an international research collaboration in coupled
and the GCS gallery). In this defined coordinate system, subsurface processes. Int J Rock Mech Min Sci 122:103995
the position of sensor points are shown as follows: Borden MJ, Verhoosel CV, Scott MA, Hughes TJ, Landis CM (2012)
The detail date of the main operations during CRQ test A phase-field description of dynamic brittle fracture. Comput
is: Methods Appl Mech Eng 217:77–95
Bourdin B, Francfort GA, Marigo JJ (2000) Numerical experiments in
revisited brittle fracture. J Mech Phys Solids 48:797–826

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Three‑dimensional Modeling of Cracking with Thermo‑hydromechanical Process by Considering… 4387

Bourdin B, Francfort GA, Marigo JJ (2008) The variational approach Moes N, Dolbow J, Belytschko T (1999) A finite element method
to fracture. J Elast 91:5–148 for crack growth without remeshing. Int J Numer Methods Eng
Braun P (2019) Thermo-hydro-mechanical behavior of the Callovo- 46:131–150
Oxfordian claystone: Effects of stress paths and temperature Moes N, Stolz C, Chevaugeon N, Bernard PE (2010) A level set based
changes. Ph.D. thesis. Université Paris-Est model for damage growth: the thick level set approach. Int J
Bryant EC, Sun W (2018) A mixed-mode phase field fracture model Numer Methods Eng 86:358–380
in anisotropic rocks with consistent kinematics. Comput Methods Mori T, Tanaka K (1973) Averages stress in matrix and average elas-
Appl Mech Eng 342:561–584 tic energy of materials with misfitting inclusions. Acta Metall
Cheng AHD (2016) Poroelasticity, vol 27. Springer, Heidelberg 21:571–574
Choo J, Sun W (2018) Coupled phase-field and plasticity modeling of Murakami S (2012) Continuum damage mechanics: a continuum
geological materials: from brittle fracture to ductile flow. Comput mechanics approach to the analysis of damage and fracture, vol
Methods Appl Mech Eng 330:1–32 185. Springer Science & Business Media, Heidelberg
Conil N, Manon V, Plua C, Vu NM, Seyedi D, Armand G (2020) In situ Oliver J (1996) Modelling strong discontinuities in solid mechanics via
investigation of the THM behavior of the Callovo-Oxfordian clay- strain softening constitutive equations, part 1: fundamentales. Int
stone. Rock Mech Rock Eng 53:2747–2769 J Numer Methods Eng 39:3575–3600
Coussy O (2004) Poromechanics. Wiley, Hoboken Robinet JC, Sardini P, Coelho D, Parneix JC, Pret D, Sammartino S,
DeBuhan P, Dormieux L (1996) On the validity of the effective stress Boller E, Altmann S (2012) Effects of mineral distribution at
concept for assessing the strength of saturated porous materials: mesoscopic scale on solute diffusion in a clay-rich rock: example
a homogenization approach. J Mech Phys Solids 44:1649–1677 of the Callovo-Oxfordian mudstone (Bure, France). Water Resour
de La Vaissiere R, Armand G, Talandier J (2015) Gas and water flow Res 48:W05554
in an excavation-induced fracture network around an underground Seyedi D et al (2021) Upscaling THM modelling from small-scale to
drift: a case study for a radioactive waste repository in clay rock. full-scale in-situ experiment in the Callovo-Oxfordian claystone.
J Hydrol 521:141–156 Int J Rock Mech Min Sci 144:1365–1609
Evans B, Fredrich JT, Wong TF (1990) The brittle-ductile transition in Seyedi D, Armand G, Conil N, Vitel M, Vu MN (2017) On the thermo-
rocks: recent experimental and theoretical progress. The Brittle- hydro-mechanical pressurization in Callovo-Oxfordian claystone
Ductile Transition in rocks. Geophys Monogr Ser 56:1–20 under thermal loading. Poromechanics VI 2017:754–761
Fang J, Wu C, Li J, Liu Q, Wu C, Sun G, Qing L (2019) Phase field Seyedi DM, Armand G, Noiret A (2017) transverse action-a model
fracture in elasto-plastic solids: variational formulation for multi- benchmark exercise for numerical analysis of the Callovo-Oxford-
surface plasticity and effects of plastic yield surfaces and harden- ian claystone hydromechanical response to excavation operations.
ing. Int J Mech Sci 156:382–396 Comput Geotech 85:287–305
Farhat F, Shen W, Shao J (2017) A micro-mechanics based viscoplastic Tsang CF, Stephansson O, Jing L, Kautsky F (2009) Decovalex project:
model for clayey rocks. Comput Geotech 89:92–102 from 1992 to 2007. Environ Geol 57:1221–1237
Fei F, Choo J (2020) A phase-field method for modeling cracks with Ulloa J, Wambacq J, Alessi R, Samaniego E, Degrande G, François
frictional contact. Int J Numer Methods Eng 121:740–762 S (2022) A micromechanics-based variational phase-field model
Fei F, Choo J (2020) A phase-field model of frictional shear fracture in for fracture in geomaterials with brittle-tensile and compressive-
geologic materials. Comput Methods Appl Mech Eng 369:113265 ductile behavior. J Mech Phys Solids 159:104684
Fei F, Choo J (2021) Double-phase-field formulation for mixed-mode Wong Tf, Baud P (2012) The brittle-ductile transition in porous rock:
fracture in rocks. Comput Methods Appl Mech Eng 376:113655 a review. J Struct Geol 44:25–53
Francfort GA, Marigo JJ (1998) Revisiting brittle fracture as an energy Wu JY (2017) A unified phase-field theory for the mechanics of dam-
minimization problem. J Mech Phys Solids 46:1319–1342 age and quasi-brittle failure. J Mech Phys Solids 103:72–99
Giraud A, Huynh Q, Hoxha D, Kondo D (2007) Application of results Wu JY, Nguyen VP, Thanh Nguyen C, Sutula D, Bordas S, Sinaie
on Eshelby tensor to the determination of effective poroelastic S (2019) Phase field modelling of fracture. Adv Appl Mech
properties of anisotropic rocks-like composites. Int J Solids Struct 53:1–183
44:3756–3772 You T, Waisman H, Zhu QZ (2021) Brittle-ductile failure transition in
Kim J, Tchelepi H, Juanes R (2011) Stability and convergence of geomaterials modeled by a modified phase-field method with a
sequential methods for coupled flow and geomechanics: fixed- varying damage-driving energy coefficient. Int J Plast 136:102836
stress and fixed-strain splits. Comput Methods Appl Mech Eng Yu Z, Shao J, Duveau G, Vu MN, Armand G (2021) Numerical mod-
200:1591–1606 eling of deformation and damage around underground excavation
Lubarda V, Krajcinovic D, Mastilovic S (1994) Damage model for brit- by phase-field method with hydromechanical coupling. Comput
tle elastic solids with unequal tensile and compressive strengths. Geotech 138:104369
Eng Fract Mech 49:681–697 Yu Z, Shao JF, Vu MN, Armand G (2021) Numerical study of thermo-
Lydzba D, Shao JF (2002) Stress equivalence principle for saturated hydro-mechanical responses of in situ heating test with phase-field
porous media. Comptes Rendus Mec 330:297–303 model. Int J Rock Mech Min Sci 138:104542
Miehe C, Hofacker M, Welschinger F (2010) A phase field model for Yu Z, Shao J, Sun Y, Wang M, ngoc Vu M, Plua C (2023) Numerical
rate-independent crack propagation: robust algorithmic implemen- analysis of hydro-thermal fracturing in saturated rocks by consid-
tation based on operator splits. Comput Methods Appl Mech Eng ering material anisotropy and micro-structural heterogeneity. Int
199:2765–2778 J Rock Mech Min Sci 170:105457
Miehe C, Welschinger F, Hofacker M (2010) Thermodynamically Yu Z, Sun Y, Vu MN, Shao JF (2023) Modeling of mixed cracks in
consistent phase-field models of fracture: variational principles rock-like brittle materials under compressive stresses by a double-
and multi-field fe implementations. Int J Numer Methods Eng phase-field method. Rock Mech Rock Eng 56:2779–2792
83:1273–1311 Zhang X, Sloan SW, Vignes C, Sheng D (2017) A modification of the
Miehe C, Hofacker M, Schänzel LM, Aldakheel F (2015) Phase field phase-field model for mixed mode crack propagation in rock-like
modeling of fracture in multi-physics problems. Part II. Cou- materials. Comput Methods Appl Mech Eng 322:123–136
pled brittle-to-ductile failure criteria and crack propagation in
thermo-elastic-plastic solids. Comput Methods Appl Mech Eng
294:486–522

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

4388 Z. Yu et al.

Zhao JJ, Shen WQ, Shao JF, anc Liu ZB, Vu MN (2022) A constitutive Springer Nature or its licensor (e.g. a society or other partner) holds
model for anisotropic clay-rich rocks considering micro-structural exclusive rights to this article under a publishing agreement with the
composition. Int J Rock Mech Min Sci 151:105029 author(s) or other rightsholder(s); author self-archiving of the accepted
manuscript version of this article is solely governed by the terms of
Publisher's Note Springer Nature remains neutral with regard to such publishing agreement and applicable law.
jurisdictional claims in published maps and institutional affiliations.

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at

[email protected]