J. Cent. South Univ.

(2021) 28: 2905－2918

DOI: https://doi.org/10.1007/s11771-021-4815-6

Coupled thermo-hydro-mechanical process in buffer material and

self-healing effects with joints

YANG Gao-sheng(杨高升)1, 2*, LIU Yue-miao(刘月妙)3, GAO Yu-feng(高玉峰)3,

LI Jian(李舰)1, 2, CAI Guo-qing(蔡国庆)1, 2

1. Key Laboratory of Urban Underground Engineering of Ministry of Education, Beijing Jiaotong University,
Beijing 100044, China;
2. School of Civil Engineering, Beijing Jiaotong University, Beijing 100044, China;
3. Beijing Research Institute of Uranium Geology, Beijing 100029, China
© Central South University Press and Springer-Verlag GmbH Germany, part of Springer Nature 2021

Abstract: Within the multi-barrier system for high-level waste disposal, the technological gap formed by combined
buffer material block becomes the weak part of buffer layer. In this paper, Gaomiaozi bentonite buffer material with
technological gap was studied, the heat transfer induced by liquid water flow and water vapor was embedded into the
energy conservation equation. Based on the Barcelona basic model, the coupled thermo-hydro-mechanical model of
unsaturated bentonite was established by analyzing the swelling process of bentonite block and the compression process
of joint material. The China-Mock-up test was adopted to compare the numerical calculation results with the test results
so as to verify the rationality of the proposed model. On this basis, the effect of joint self-healing on dry density, thermal
conductivity and permeability coefficient of buffer material was further analyzed. The results show that, with bentonite
hydrating and swelling, the joint material gradually increases in dry density, and exhibits comparatively uniform
hydraulic and thermal conductivity properties as compacted bentonite block. As a result, the buffer material gradually
shifts to homogenization due to the coordinated deformation.

Key words: buffer material; thermo-hydro-mechanical coupling; joints; self-healing effect

Cite this article as: YANG Gao-sheng, LIU Yue-miao, GAO Yu-feng, LI Jian, CAI Guo-qing. Coupled thermo-hydro-
mechanical process in buffer material and self-healing effects with joints [J]. Journal of Central South University, 2021,
28(9): 2905−2918. DOI: https://doi.org/10.1007/s11771-021-4815-6.

barrier, chemical barrier, conduction of radioactive

1 Introduction waste decay heat, closing disposal pit and so on. It
is a key link to prevent radionuclide migration and
Densely compacted bentonite possesses low ensure the safety and stability of the repository [4−
permeability, good thermal conductivity and high 6]. However, the joints between waste canister and
adsorption, and it is among the best available buffer bentonite block, between surrounding rock and
materials [1 − 3]. The densely compacted bentonite bentonite block, and between bentonite blocks are
block is used as a buffer material around the waste inevitable in the construction process. Obviously,
canister and the natural barriers, which plays an the existence of joints reduces the ability to
important role in engineering barrier, hydraulic dissipate radionuclide decay heat, shortens the time
Foundation item: Projects(52078031, U2034204) supported by the National Natural Science Foundation of China
Received date: 2020-10-23; Accepted date: 2021-01-29
Corresponding author: YANG Gao-sheng, PhD; Tel: +86-17812077092; E-mail: [email protected]; ORCID: https://orcid.org/0000-
0002-4440-7094
2906 J. Cent. South Univ. (2021) 28: 2905－2918

for groundwater seepage to reach the waste tank, permeability test on soil samples using three kinds
increases the possibility of radioactive waste of joint material including bentonite powder,
leakage, and weakens the barrier function of buffer bentonite slurry and micro bentonite. The test
structure [7, 8]. Although bentonite has a strong self- results show that the hydraulic conductivity
healing ability, the joint interface is the potential coefficient is (2.51 − 4.94) ×10−10 cm/s for soil
hydraulic defect and weak strength part of buffer samples with different sealing treatment, and the
material, and its impact on the overall performance joint material with 70% particle rate has the best
of high-level radioactive waste repository buffer healing effect. It is pointed out that the dry density
layer remains an urgent problem. of the joint material goes up gradually with
So far, the research on the impact of bentonite hydration and swelling, and the hydraulic
technological gaps on the performance of buffer and thermal conductivity properties of joint material
material in the world is still in its infancy. POPP gradually tend to the soil sample, and the joint zone
et al [9] found that the existence of joint causes the gradually heals and disappears. CHEN et al [13]
decrease of overall sealing on the interface of the reviewed and summarized the research results in the
buffer material. Despite the bentonite having strong world on the water mechanical coupling
self-healing ability, the existence of gas at the joint performance of technological gap on buffer
and the infiltration of high-pressure underground material. It is pointed out that the influence of joint
water may open the seepage channel again. on the overall water force coupling characteristics
GUERRA et al [10] carried out an hydro- of buffer material is analyzed mainly through
mechanical (HM) coupling experimental study on physical test and overall average method. This
the particle/powder mixture. The results show that method cannot express the real water force
after 800 d hydration, the samples are not fully characteristic index of the joint, and cannot reveal
saturated, and the axial and radial swelling pressures the hydraulic transmission characteristic of the
are anisotropic; the bentonite particles in the interface effectively, thus making it difficult to
mixture generate a large swelling force after analyze the relevant mechanism reasonably.
hydration, which is conducive to the self-healing of In summary, the current research is mainly
the mixture pores. CHEN et al [11] carried out focused on the experimental analyses for the
permeability tests on densely compacted Gaomiaozi performance of buffer material with technological
(GMZ) bentonite samples with different initial dry gaps. However, the multi-field coupling process is
density and joint width through the self-developed very complex and must be evaluated for a long time
multifunctional swell and permeability apparatus, for high-level radioactive waste repository
and conducted mercury porosity test on saturated engineering barrier system [14, 15]. Therefore, it is
bentonite samples after the completion of necessary to establish the relevant theoretical model
permeability test. The test results show that the to simulate healing process. Some scholars have
swelling force decreases with the increase of the analyzed the moisture transfer and heat conduction,
joint width, and the permeability coefficient and also the swelling properties of buffer material
increases with the increase of the joint width. The under different boundary conditions [8, 14].
influence mechanism of joints on the permeability However, these researches focused more on the
of densely compacted GMZ bentonite is also numerical study for the whole bentonite block than
studied from the microscopic view. Due to the the buffer material with joints. In this paper, GMZ
existence of the joints, the densely compacted bentonite buffer material with technological gap
bentonite blocks have apparent spaces for swelling was studied, and the moisture transport mechanisms
deformation, which causes the decrease of small induced by heating, mechanical model and
pores (100−300 nm), and the increase of large pores boundary conditions were discussed. Then, a
(1000 − 3000 nm) in the buffer material during the coupled thermos-hydro-mechanical (THM) model
joint healing process. The increase of large pore for unsaturated buffer material is established. In this
sizes indicates that the permeability coefficient is model, the swelling property of bentonite block and
increased. ZHANG et al [12] conducted a the compressive process of joint material were
J. Cent. South Univ. (2021) 28: 2905－2918 2907

considered. On this basis, the effect of joint healing According to the heat conduction test results of
was further analyzed on dry density, thermal bentonite [20], it is found that the thermal
conductivity and permeability coefficient of buffer conductivity coefficient (λk) of bentonite is
material. The research results may provide a approximately linear with its dry density (ρd) and
reference for the safety assessment of the barrier saturation (Sω), which can be expressed as:
system of high-level radioactive waste repository. λ k ( ρ d ,S ω ) = λ k0 + λ k1 ρ d + λ k2 S ω (4)

2 A coupled THM model where λk0, λk1 and λk2 are the parameters obtained by
fitting the relevant test results.
2.1 Energy conservation equation
A coupled THM model should not only 2.2 Moisture transfer equation
consider the heat conduction and convection process The governing equation for the flow of liquid
of water vapor and liquid water but also focus on water and water vapor in unsaturated soil is given
evaporation process of liquid water. Therefore, the by the following mass conservation equation [21]:
energy conservation equation of unsaturated soil ∂θ
= -∇ ( q l + q v ) + Q (5)
should be expressed as [16−19]: ∂t
∂T ∂θ v
C + Ll ρl = ∇ ( λ k ∇T ) - q l ∇ (C l T ) - ρv
∂t ∂t θ = θl + θ (6)
ρl v
q v ∇ (C vT ) - L l ρ l ∇q v (1)
where θ is the total volumetric water content; θl is
where ∇ is a Hamiltonian operator; C, Cl and Cv are the volumetric liquid water content; ρv is the vapor
the volumetric heat capacities (J/(m3·℃)) of the soil, density; Q is the source item.
liquid water and vapor water, respectively; T is soil The flux density of liquid water is described by
temperature (° C); λk is the thermal conductivity Richards equation:
(W/(m∙°C)); ρl is the density of liquid water (kg/m3);
θv is the volumetric vapor content; ql and qv are the q l = -K l ∇ ( s - ρ l gz ) (7)
flux densities of liquid water and water vapor, where s is suction, when the soil is unsaturated, the
respectively; Ll is the latent heat of water absolute value of s is equal to the matric suction; g
vaporization (Ll=2.4×106 J/kg). is the gravitational acceleration; z is the height; Kl is
The first term on the left-hand side of Eq. (1) hydraulic conductivity for liquid water, which is
represents the variations in the energy content, and described by van Genuchten model as [22]:
the second represents the variations in the latent
K l = K s S e0.5 [ 1 - (1 - S e1 m )m ]
2

heat of the vapor phases. The respective terms on (8)

the right-hand side represent soil heat transport by where Ks is the saturated hydraulic conductivity; m
conduction, convection by liquid water and water is the parameter related to soil particle size-grade
vapor, and the transfer of latent heat by diffusion of distribution [23]; Se is the effective saturation, it can
water vapor. As the volumetric heat capacities of be expressed as S e = (θ l - θ r )/ (θ s - θ r ), θl, θr and θs
water vapor are too small, the influence of water are the actual liquid water content, saturated and
vapor on soil heat capacities can be neglected [19]. residual water content, respectively.
The soil and liquid water volumetric heat According to the previous test results of
capacities can be defined as: bentonite [7, 24 − 26], it is found that the saturated
C = (1 - n ) ρ s c s + nS ω ρ l c l (2) hydraulic conductivity of bentonite has a linear
relationship with its dry density index at the same
C l = nS ω ρ l c l (3)
temperature:
where n is the medium porosity; ρs is the density of
lg K s = -k 1 ρ d - k 0 (9)
soil matrix (kg/m3); cs and cl (J/(kg∙°C)) are specific
heat capacity of the soil matrix and liquid water, Therefore, the soil hydraulic conductivity can
respectively. be expressed as:
2908 J. Cent. South Univ. (2021) 28: 2905－2918

S e0.5 [ 1 - (1 - S e1 m )m ]
()
-k 1 ρ d - k 0 2 β
K l = 10 (10) ρd
s ( S ω , T, ρ d ) = s r ( S ω )σ (T ) (21)
ρ0
where k0 and k1 are the experimental fitting
parameters. The relationship between suction (sr) and
The flux density of water vapor is described by saturation (Sω) is described by van Genuchten (VG)
Fick􀆳s law as [27]: model as [22]:
q v = -K v ∇p v
( )
(11) b
1 b

sr ( Sω ) = Sω 1- b
-1 ρ l g/a (22)
where Kv is the hydraulic conductivity for water
vapor; pv is the water partial pressure, which can be where σ(T) is the surface tension coefficient of
expressed as the product of saturated vapor pressure water.
(pvs) and relative humidity (HR), namely: According to Refs. [28, 29], σ(T) can be
p v = p vs H R (12) expressed as:
σ (T ) = σ 0 - σ 1 (T - T r ) (23)
p vs = 0.611 × 10 exp 17.27
T
T + 237
-3
( ) (13)
where ρ0 is the initial dry density; σ0 and σ1 are
fitting parameters; Tr is the reference temperature
where HR is related to temperature and suction in
(20 ° C); sr is the suction at the reference
unsaturated soil.
temperature; and a, b and β are the experimental
é sM l ù fitting parameters [7].
H R = exp ê ú (14)
ë Rρ l (T + 273.15) û
2.3 Mechanical model
where Ml is the molecular weight of water The Barcelona basic model (BBM) [30] can
(0.018 kg/mol); R is the universal gas constant R= describe the swelling and collapsibility of
8.314 J/(mol∙°C). unsaturated soil by considering the influence of
Hydraulic conductivity for water vapor (Kv) suction change on the mechanical behavior of
can be expressed as: unsaturated soil. Under the action of groundwater,
Da Ml g the buffer material mainly causes the swelling of
Kv = ρ vs H R (15)
ρl R (T + 273.15) bentonite block and compression deformation of
joint material. Therefore, the BBM can be used to
where Da is the vapor diffusivity, and ρvs is the
calculate the mechanical behavior of buffer material
saturated vapor density.
[4, 31, 32].
D a = τθ v D 0 (16) 2.3.1 Compression process of joint
Although the BBM model has been approved
D 0 = 2.12 × 10-5 [ (T + 273.15) /273.15 ]2 (17) by many scholars, it lacks relevant test data to
obtain the parameters of unsaturated bentonite.
τ = θ v7 3
θ s2 (18)
According to the hydration process of bentonite
As a function of temperature and saturated block and joint material, the calculation model is
vapor density, vapor density (ρv) is expressed as: reasonably simplified in this paper. For this, their
possible stress paths are analyzed: 1) The outer joint
ρ v = ρ vs H R (19)
material is first saturated under the action of water,

( )
6014.79 the swelling force produced by the joint material is
ρ vs = exp 31.37 - - 7.92 × 10-3T ( 103T ) ignored in hydration process, and it is believed that
T
neither the bentonite block nor the joint deforms.
(20)
The bentonite block then starts to swell in
The matric suction (s) of unsaturated soil is hydration, and produces a large swelling force. The
related to saturation, temperature and dry density, joint material is compressed to point B3 with the
which can be expressed as: compression index λ(0). 2) The bentonite block is
J. Cent. South Univ. (2021) 28: 2905－2918 2909

first hydration and produces large swelling force. corresponding to the preconsolidation pressure of
The inner joint material is extruded to the yield the joint material. Let p'=0 in Eq. (25), v2=v02 is the
surface along the rebound curve, and then it is initial specific volume of the joint material.
compressed to point B4 with the compression index According to Eq. (25), it can be obtained that: v p2 =
λ(s). The inner joint material is saturated and v 02 - λ (0 ). When p' is the preconsolidation pressure
collapsible, and finally reaches point B3. As shown of the joint material, v2 =vp2, which can be obtained
in Figure 1, the specific volume v=1+e, where e is as:

( )
the void ratio, and p' is the effective stress. In
v 0 - v 02 + λ (0 )
practice, the hydration process of bentonite block p 2 = P 0 exp (26)
λ (0 )
and joint material occurs at the same time, so the
actual stress path is between stress path 1 and stress In this paper, the elliptical yield surface of
path 2. When the buffer material is fully saturated, modified Cam-Clay model is used to calculate the
the final point is independent of the stress path, so yield of joint material. The yield function, hardening
the change of compression index λ(s) is ignored; law and flow rule of the modified Cam-Clay model
therefor, λ(0) is taken to calculate the joint material are as follows [33]:
in the numerical calculation. The stress − strain q2
relationship of joint material during compression F= + p (p - pc) = 0 (27)
M2
can be expressed as:
1+ e

( )
ṗ c = p ε̇ p (28)
p′ λ-κ c v
v 2 = v 0 - λ (0 )ln , v 2 ≤ v p2 (24)
P0 ∂F
dε p = dφ (29)
∂σ
where p is the mean stress; q is the deviator stress;
M is the slope of the critical line; pc is the pre-
consolidation pressure, which can be used as the
hardening parameter of the modified Cam-Clay
model; e is the void ratio; λ and κ are compression
index and elastic compression index of soil
respectively; dϕ is a nonnegative plastic multiplier.
2.3.2 Swelling process of block
Swelling force is defined as the pressure
Figure 1 Stress path of bentonite block and joint material required to keep the volume of bentonite unchanged
during wetting after water absorption and swelling. Based on the
previous research results [25, 34, 35], there are three
It is considered that the soil has linear elastic test methods to determine the swelling force,
deformation when v2>vp2. Considering the continuity namely, free swelling compression test method,
of numerical calculation, the elastic modulus is the loading swelling compression test method and
value when v2=vp2. constant volume test method. Actually, the buffer
p′ - p 2 material is filled in the closed space between the
v 2 = v p2 - λ (0 ) , v 2 > v p2 (25)
p2 waste tank and the surrounding rock, and the stress
path of the swelling force is similar to that of the
where P0 and v0 are the reference stress and the
constant volume test method. YE et al [36]
reference specific volume, respectively, and the
measured the relationship between the maximum
coordinates of any point on the compression curve
swelling force (Ps0) and dry density (ρd) of GMZ
can be taken; λ(0) is the plastic slope in the
bentonite with initial water content of 11.14% by
saturated condition; v2 is the specific volume of the
constant volume test method:
joint material; p2 is the preconsolidation pressure of
joint material; vp2 is the specific volume P s0 = 0.0194exp (7.4197ρ d ) (30)
2910 J. Cent. South Univ. (2021) 28: 2905－2918

The relation between Bishop stress (σs) caused where p1 is the preconsolidation pressure of
by suction change and suction (s) and saturation (Sω) bentonite block.
is taken as σ s = P s0 S ω s s 0, so the swelling force (Ps) It is considered that the specific volume vp1=v01
of bentonite can be expressed as: is corresponding to the preconsolidation pressure of
the block, and no unloading rebound deformation
P s = P s0 (1 - S ω s/s 0 ) (31)
occurs after the compaction of the bentonite block.
Under low confining pressure, the deformation According to Eq. (24), the preconsolidation pressure
of bentonite block caused by suction change is p1 of bentonite block can be obtained:

( )
elastic deformation [30, 37], which can be v 0 - v 01
expressed as: p 1 = P 0 exp (36)
λ (0 )
ds
dv 1 = κ s (32) 2.3.3 Thermal expansion
( s + q0 )
The soil skeleton expands as a result of
where v1 is the specific volume of bentonite block; increasing temperature, and the relationship
κ s is the elastic compression index caused by between thermal expansion strain (ε T) and
suction; s is suction; q0 the experimental fitting temperature (T) of clay minerals can be expressed:
parameters.
ε T = α (T - T 0 ) (37)
The deformation of bentonite block caused by
suction change under free swelling state can be where α is the thermal expansion coefficient of clay
obtained by: minerals, which is directly proportional to the dry

( )
density of soil, and can be expressed as α =
s0 + q0
v 1 = v 01 + κ s ln (33) 10−5×ρd [39].
s + q0

where v01 and s0 are the initial specific volume and 3 Model validation and analysis
initial suction of bentonite block, respectively.
3.1 Physical model and boundary conditions
The swelling of bentonite block decreases with
The physical model adopted by this paper is
the increase of external pressure (P) [35, 36, 38].
established according to the China-Mock-up test
Substitute Eq. (31) into Eq. (33), the stress − strain
bench [7, 8]. It mainly includes a central electric
relationship in the swelling process of bentonite
heater which represents the waste tank (Φ300 mm×
block under external pressure can be obtained:
1600 mm) the electric heater is composed of
v 1 = v 01 + κ s ln { s0 + q0
}
s 0 [1 - ( P s - P ) /P s0 ] /S ω + q 0
(34)
temperature adjustable heater strip (Φ60 mm×
800 mm) and high thermal conductivity metal
(Φ300 mm×1600 mm); the heat conducting metal is
In practice, the hydration process of bentonite
surrounded by densely compacted GMZ bentonite
block is a gradual infiltration process from the
block with dry density of 1.70 g/cm3, steel tank
outside to the inside. The soil on the inner side is
(Φ900 mm×2200 mm, steel thickness of 25 mm);
deformed by the swelling force generated by the
the gaps between the steel tank and the bentonite
outer soil in this process. The preconsolidation
blocks, and that between the thermal conductive
pressure of densely compacted bentonite block is
metal and the bentonite blocks are filled with
very high. According to Eq. (30), the
crushed particles made of GMZ bentonite soil with
preconsolidation pressure of the block is greater
a dry density of 1.30 g/cm3 (the thicknesses are 15
than the compressive stress of the inner soil during
and 50 mm, respectively) [7, 8]. Due to the
this process. Therefore, the linear elastic
axisymmetry of the China-Mock-up bench, the two-
deformation of the inner soil with elastic
dimensional model is used for numerical
compression index κ under the compressive stress is:
calculation. The finite element software COMSOL
p′ is used, which includes 585 domain elements, 405
v 1 = v 01 - κ , v 1 ≤ v 01 (35)
p1 edge elements and 48 vertex elements. The specific
J. Cent. South Univ. (2021) 28: 2905－2918 2911

Table 1 Parameters for hydro-thermal-mechanical model

Parameter Value Parameter Value
ρl/(kg·m−3) 1000 β 2
−3
ρs/(kg·m ) 2660 σ0 1
−1 −1
cl/(J·(kg·°C) ) 4200 σ1/°C 0.004
−1
cs/(J·(kg·°C) ) 900 P0/MPa 1.0
w0/% 11.14 v0 1.90
λk0/(W·(m·°C) ) −1
−0.5 λ(0) 0.1333
−1
λk1/(W·(m·°C) ) 0.6 v01 1.588
−1
λk2/(W·(m·°C) ) 1.07 v02 2.455
m 0.22 M 0.558
k0 3.51 κs 0.10
k1 6.0 κ 0.0244
a/m−1 0.009 q0/MPa 5.0
b 1.3386 υ 0.3
Figure 2 Diagram of numerical model and mesh: (a) Size
of numerical model (Unit:mm); (b) Meshes
Table 2 Parameters for metal materials
Parameter Copper Steel
−3
Density/(g·cm ) 8.96 7.85
−1
Thermal conductivity/(W·(m∙°C) ) 400 44.5
−1
Constant pressure heat Capacity/(J·(kg∙°C) ) 385 475
Coefficient of thermal
1.7×10−5 −
expansion
Elastic modulus/GPa 110 −
Poisson ratio 0.35 −

sizes and mesh generation of the physical model are

illustrated in Figure 2. The comparison between the
calculated and tested results (Figures 5−7) indicates Figure 3 Variation of heat and external environment
that the mesh density shown in Figure 2(b) can temperature with time
satisfy the demand of accuracy.
In the numerical calculation, the central instead of applying constant water pressure. The
electric heater is set as the first boundary condition, changes in total injection mass and injection rate
and the heating process includes three stages: the with time are shown in Figure 4. Given the gravity
temperature was maintained at 30 ° C for the first and the low permeability of bentonite, the water
135 d, then gradually increased to 90 ° C from the distribution is not uniform in the vertical direction.
136th day to the 255th d, and remained unchanged It is assumed that the water mass flux is triangular
at 90 ° C for the remaining tests, as shown in along the height.
Figure 3. Although the outer surface of the China-
Mock-up test bench is covered with thermal 3.2 Results
insulation material, it is found that the internal 3.2.1 Temperature
temperature is significantly affected by the external A comparison between the calculated and
environment temperature. Therefore, the external tested results at different monitoring locations is
environment temperature is set as the second shown in Figure 5. Compared with Figures 3 and 5,
boundary condition. In order to avoid damaging the it can be found that the temperature of each
sensor, the hydraulic boundary is realized by monitoring location is affected by the internal heat
injecting a certain amount of water on a daily basis sources and the external environment temperature.
2912 J. Cent. South Univ. (2021) 28: 2905－2918

Figure 4 Variation of total water injection and water

injection rate with time: (a) Total water injection;
(b) Water injection rate

As the internal heat sources temperature gradually

rises to 60 ° C, the temperature of each monitoring
location also gradually increases with the rise of
internal heat sources temperature (Figure 5).
However, the internal heat sources temperature
remains constant at 60 °C from 175 to 209 d, while
the external environment temperature decreases
with time, and the temperature of each monitoring Figure 5 Comparison of calculated and tested temperature
location also declines to a certain extent (Figures 3 result: (a) (350, 100) mm; (b) (300, 1100) mm;
and 5). The numerical simulation also predicts this (c) (350, 1950) mm
result well. After 256 d, when the internal heat
sources temperature remains constant at 90 ° C, the the model (Figure 5(c)), and the calculation results
temperature of each monitoring location is affected of this paper and Ref. [8] are lower than the test
by the external environment temperature, which results. The difference may be due to the lower
changes periodically with time (Figures 3 and 5). It density of water vapor in the buffer material, which
can be seen from the bottom (Figure 5(a)) and carries heat to the upper part of the model. The
middle (Figure 5(b)) of the China-Mock-up China-Mock-up test model is designed as a sealed
experimental model that there is a good consistency cylindrical tank, and the heat carried by the steam
between the test results and the calculation results. accumulates in the top area of the tank body, which
However, there is a great difference between the test causes the temperature to rise. However, the
results and the calculation results in the top area of influence of gravity migration on vapour phase is
J. Cent. South Univ. (2021) 28: 2905－2918 2913

not considered herein or in Ref. [8], which results in hydraulic boundary, the relative humidity of each
a lower temperature value than the test results. monitoring location began to change significantly.
3.2.2 Relative humidity For the monitoring location which is close to the
The calculated distributions of the relative internal heat sources (Figures 6(b) and (c)), the
humidity at different monitoring locations are relative humidity first drops and then gradually
increases. The main influencing factor is that high
shown in Figure 6. In the initial stage of the test, the
temperature causes the water in the locations near
relative humidity of each monitoring location is
the internal heat sources to evaporate into outgoing
basically unchanged before the hydration process.
water vapor, which brings a drop in relative
During this period, the relative humidity calculated
humidity near the internal heat sources. With the
by the two models is highly compatible with the test continuous injection of water from the hydraulic
results. With the gradual injection of water into the boundary to the tank, the relative humidity
gradually rises. In this process, the calculated results
of ZHAO et al [8] are lower than the test results,
and the calculation results in this paper are more
consistent with the test results. The numerical
simulation also well predicts the process, that is, the
relative humidity first drops and then increases
gradually.
3.2.3 Swelling force
A comparison between the calculated and test
results of the swelling force at different monitoring
locations is shown in Figure 7. It can be seen that
the proposed model can approximately predict the
change process of swelling force with time at
different monitoring locations. However, there are
also significant differences between the calculated
results and the test results, especially for the
monitoring locations far away from the hydraulic
boundary (Figures 7(a) and (c)). The contributing
factor is that the model only considers the inner
joint between the bentonite block and the inner
thermal conductive metal and the outer joint
between the block and the steel tank, and the joints
between bentonite blocks are ignored here. The test
monitoring results turn out to be smaller than the
calculation results due to the existence of joints
between blocks. In addition, in the China-Mock-up
test [40], the required pore water reflects the effect
of salt ions. Generally, the existence of salt ions
inhibits the swelling deformation of bentonite [35,
41, 42]. However, the effect of salt ions is neglected
in this numerical calculation, which makes the
calculated results larger than the test results.
However, the relative humidity calculated by ZHAO
Figure 6 Comparison of calculated and tested temperature et al [8] is lower than the test result (Figure 6),
result: (a) (350, 150) mm; (b) (300, 1100) mm; which makes the calculated results of swelling force
(c ) (50, 1900) mm lower than the test results.
2914 J. Cent. South Univ. (2021) 28: 2905－2918

waste repository. The thermal conductivity

determines whether the decay heat of radioactive
waste can be transferred out timely, and the
permeability coefficient affects the time when the
groundwater seepage reaches the waste tank. The
influence of joints healing on buffer material will be
analyzed from these three aspects.

4.1 Effect of self-healing effects on dry density

The variation of dry density of buffer material
with time is shown in Figure 8. When the buffer
material is fully saturated, the dry density of the
inner and outer joint zone increases, but it is still
less than that of the bentonite block. A careful
comparison of the dry density distribution of the
buffer material before and after the self-healing
effect is helpful to understand the self-healing
mechanism of the joints. The initial setting value
of dry density of compacted bentonite block is
1.7 g/cm3. When the buffer material is hydrated, the
dry density of compacted bentonite block reduces to
about 1.63 g/cm3 (Figure 8). On the contrary,
bentonite particles with an average dry density of
1.3 g/cm3 are selected as the inner and outer joint
material before hydration. After hydration, the dry
density of the inner joint material increases to about
1.40 g/cm3, and the dry density of the outer joint
material rises more obviously to 1.46 g/cm3
(Figure 8). Therefore, the dry density of both inner
and outer joint material is increased. That is to say,
with the adjustment of stress in the process of
hydration and swelling of buffer material, the
bentonite block is continuously extruded towards
Figure 7 Comparison of calculated and tested swelling
pressure result: (a) (160, 170) mm; (b) (420, 1000) mm;
(c) (160, 1930) mm

4 Effect of self-healing effects with joints

In the hydration process of buffer material, due

to the existence of technological gaps, the densely
compacted bentonite blocks have apparent space for
swelling deformation, which makes the dry density,
thermal conductivity and permeability of buffer
materials change with time. The dry density directly
affects the swelling deformation of buffer layer, and Figure 8 Distribution of dry density of buffer material
then affects the sealing performance of nuclear with time
J. Cent. South Univ. (2021) 28: 2905－2918 2915

the joint area after hydration and swelling. This and the thermal conductivity of joint material
makes the soil particles in the joint area arrange gradually approaches to that of block with the
more and more densely, the pores are less and less, hydration of buffer material. The change of thermal
and the dry density increases. On the contrary, the conductivity of inner and outer joints reveals that
soil particles of bentonite block arrange more and the thermal conductivity of the outer joint increases
more loosely, the pores become more and more, and faster than that of the inner joint. This is mainly due
the dry density reduces. The buffer material shows a to the gradual saturation of buffer material from the
general tendency for homogenization, and the joint outside to the inside under the action of hydraulic
area gradually tends to heal. boundary. On the whole, the thermal stability of
bentonite block and joint material tends to be
4.2 Effect of self-healing effects on thermal consistent after hydration, which satisfies the
conductivity requirements of International Atomic Energy
Figure 9 shows that how the thermal Agency (IAEA) that the thermal conductivity is
conductivity of buffer material varies with time. The greater than 0.8 W/(m∙°C).
results show that the thermal conductivity of the
inner and outer joint zone rises after the buffer 4.3 Effect of self-healing effects on permeability
material is fully saturated, but it is still slightly less coefficient
The variation of permeability coefficient of
than that of the bentonite block. The thermal
buffer material with time is shown in Figure 10.
conductivity of bentonite block is 1.06 W/(m ∙ ° C)
When the buffer material is fully saturated, the
before hydration. After hydration, the thermal
permeability coefficient of bentonite block
conductivity of bentonite block increases to
increases, and the permeability coefficient of inner
about 1.55 W/(m ∙° C) (Figure 9). The thermal
and outer joint zone declines, but it is still higher
conductivity of the joint material is 0.72 W/(m∙°C)
than that of bentonite block. The permeability
before hydration. After hydration, the thermal
coefficient of bentonite block is 1.94×10−12 cm/s
conductivity of the inner joint material rises to about
before hydration, and after hydration, the
1.41 W/(m∙°C), and the outer joint material rises to
permeability coefficient of bentonite block increases
about 1.45 W/(m ∙ ° C) (Figure 9). Compared with
to about 4.51×10−12 cm/s (Figure 10). On the
the bentonite block, the thermal conductivity of the contrary, the permeability coefficient of inner and
joint material shows greater increase after outer joint material is 4.92×10−10 cm/s before
hydration. The reason is that there is more air in the hydration. After hydration, the permeability
joint material than in the bentonite block before coefficient of the inner joint material is reduced to
hydration, so the thermal conductivity of the joint about 1.26×10−10 cm/s, and that of the outer joint
material is lower than that of the bentonite block, material is reduced to about 6.08×10−11 cm/s
(Figure 10). The permeability coefficient of the
buffer material with joints after hydration is
approximately the same as that of the previous
complete samples [5, 43], which satisfies the low
permeability requirement that the permeability
coefficient of high-level waste engineering barrier is
less than 10−9 cm/s [5]. In other words, bentonite
sealing material is used to seal the technological gap
of the buffer material. After the buffer material is
hydrated, the permeability of the joint containing
buffer material is gradually consistent with that of
the complete bentonite block. Therefore, as long as
the technological gap is properly sealed, the joint
Figure 9 Distribution of thermal conductivity of buffer will not reduce the anti-seepage performance of the
material with time engineering barrier.
2916 J. Cent. South Univ. (2021) 28: 2905－2918

value than the test results.

3) Compared with the variation of temperature,
the hydration process within the buffer material is
more complex. For the monitoring location which is
close to the internal heat sources, the relative
humidity first drops and then gradually increases.
The calculation results are approximately consistent
with the test results, and the numerical simulation
well predicts the process.
4) Overall, the stress variation of buffer
material caused by the swelling of bentonite block is
dominant, while the expansion stress caused by
temperature is insignificant. Under the action of
swelling force, the joint material is gradually
compressed and self-healed. In the hydration
process, the thermal conductivity and anti-seepage
performance of the joint material is gradually
consistent with that of the bentonite blocks. Finally,
the buffer material gradually shifts to
homogenization due to the coordinated deformation.

Contributors
YANG Gao-sheng established the models,
calculated the predicted results and edited the draft
of manuscript. LIU Yue-miao and GAO Yu-feng
provided the measured data, and analyzed the
measured data. LI Jian and CAI Guo-qing edited the
Figure 10 Distribution of permeability coefficient of
draft of manuscript.
buffer material with time: (a) Bentonite block; (b) Joint
material
Conflict of interest
5 Conclusions YANG Gao-sheng, LIU Yue-miao, GAO Yu-
feng, LI Jian and CAI Guo-qing declare that they
1) The heat transfer induced by liquid water have no conflict of interest.
flow and water vapor are embedded into the energy
conservation equation. Based on the BBM, the References
coupled THM model of unsaturated bentonite is
established by analyzing the swelling process of [1] VILLAR M V, GARCIASINERIZ J L, BARCENA I,
LLORET A. State of the bentonite barrier after five years
bentonite block and the compression process of
operation of an in situ test simulating a high level radioactive
joint material. In this model, the swelling property waste repository [J]. Engineering Geology, 2005, 80(3): 175−
of bentonite block and the compressive process of 198. DOI: 10.1016/j.en ggeo.2005.05.001.
joint material are considered, which can be used to [2] SAMPER J, ZHENG L, MONTENEGRO L, FERNÁNDEZ
A M, RIVAS P. Coupled thermo-hydro- chemical models of
simulate the THM coupling of buffer material with
compacted bentonite after FEBEX in situ test [J]. Appl
joints. Geochem, 2008, 23(5): 1186 − 1201. DOI: https: 10.1016/j.
2) The simulation results are in well agreement apgeochem.2007.11.010.
with the test results obtained by the China-Mock-up. [3] BAI Bing, XU Tao, NIE Qing-ke, LI Peng-peng.
Temperature-driven migration of heavy metal Pb2+ along
However, the influence of gravity migration on
with moisture movement in unsaturated soils [J].
vapour phase is not considered in the numerical International Journal of Heat and Mass Transfer, 2020, 153:
calculation, which results in a lower temperature 119573. DOI: https://doi.org/10.1016/j. ijheatmasstransfer.
J. Cent. South Univ. (2021) 28: 2905－2918 2917
2020.119573. International Journal of Heat and Mass Transfer, 2018, 117:
[4] COLLIN F, LI X L, RADU J P, CHARLIER R. Thermo- 517−526. DOI: 10.1016/j.ijheatmasstransfer.2017.10.004.
hydro-mechanical coupling in clay barriers [J]. Engineering [16] BAI Bing, SU Zhong-qin. Thermal responses of saturated
Geology, 2002, 64(2): 179−193. DOI: https://doi.org/10.1016/ silty clay during repeated heating – cooling processes [J].
S0013-7952(01)00124 -7. Transport in Porous Media, 2012, 93(1): 1−11. DOI: 10.1007/
[5] YE Wei-min, CUI Yu-jun, QIAN Li-xin, CHEN Bao. An s11242- 012-9939-6
experimental study of the water transfer through confined [17] JIANG Zhong-ming, HOXHA D, HOMAND F, CHEN Yong-
compacted GMZ-bentonite [J]. Engineering Geology, 2009, gui. Simulation of coupled THM process in surrounding rock
108(3, 4): 169 − 176. DOI: https://doi. org/10.1016/j. mass of nuclear waste repository in argillaceous formation
enggeo.2009.08.003. [J]. Journal of Central South University, 2015, 22(2): 631 −
[6] BAI Bing, ZHANG Jia-xi, LIU Lu-lu, JI Yan-jie. The 637. DOI: 10.1007/s11771-015-2564-0.
deposition characteristics of coupled lead ions and suspended [18] BAI Bing, NIE Qing-ke, ZHANG Yi-ke, WANG Xiao-long,
silicon powders along the migration distance in water HU Wei. Cotransport of heavy metals and SiO2 particles at
seepage [J]. Transport in Porous Media, 2020, 134(3): 707 − different temperatures by seepage [J]. Journal of Hydrology,
724. DOI: https://doi.org/ 10.1007/s11242-020-01464-3. 2021, 597: 125771. DOI: https://doi. org/10.1016/j. jhydrol.
[7] CHEN Liang, LIU Yue-miao, WANG Ju, CAO Sheng-fei, 2020.125771.
XIE Jing-li, MA Li-ke, ZHAO Xing-guang, LI Ya-wei, LIU [19] ZHANG Ming-li, WEN Zhi, XUE Ke, CHEN Liang-zhi, LI
Jian. Investigation of the thermal-hydro-mechanical (THM) De-sheng. A coupled model for liquid water, water vapor and
behavior of GMZ bentonite in the China-Mock-up test [J]. heat transport of saturated–unsaturated soil in cold regions:
Engineering Geology, 2014, 172(8): 57−68. DOI: https://doi. Model formulation and verification [J]. Environmental
org/10.1016/j. enggeo.2014.01.008. Earth Sciences, 2016, 75(8): 701. DOI: 10.1007/s12665-016-
[8] ZHAO Jing-bo, CHEN Liang, COLLIN F, LIU Yue-miao, 5499-3.
WANG Ju. Numerical modeling of coupled thermal-hydro- [20] LIU Yue-miao, CAI Mei-feng, WANG Ju. Thermal
mechanical behavior of GMZ bentonite in the China-Mock- properties of buffer material for high-level radioactive waste
up test [J]. Engineering Geology, 2016, 214(30): 116 − 126. disposal [J]. Chinese Journal of Rock Mechanics and
DOI: https://doi.org/10.1016/j. enggeo.2016.09.015. Engineering, 2007, 26(S2): 3891−3896. (in Chinese)
[9] POPP T, ROLKE C, SALZER K. Hydromechanical [21] RAO Dengyu, BAI Bing. Study of the factors influencing
properties of bentonite-sand block assemblies with interfaces diffusive tortuosity based on porescale SPH simulation of
in engineered barrier systems [J]. Geological Society London granular soil [J]. Transport in Porous Media, 2020, 132(11):
Special Publications, 2014, 415(1): 19 − 33. DOI: https://doi. 333−353. DOI: 10.1007 /s11242-020-01394-0.
org/10.1144/ SP415.1. [22] GENUCHTEN M T V. A closed-form equation for predicting
[10] GUERRA A M, CUI Y, MOKNI N, DELAGE P, the hydraulic conductivity of unsaturated soils [J]. Soil
BORNERT M, AIMEDIEU P, TANG A M, BERNIER F. Science Society of America Journal, 1980, 44(44): 892−898.
Investigation of the hydro-mechanical behaviour of a pellet/ DOI: 10.2136/sssaj1980. 03615995004400050002x.
powder MX80 bentonite mixture using an infiltration column [23] CARSEL R F, PARRISH R S. Developing joint probability
[J]. Engineering Geology, 2018, 243: 18−25. DOI: https://doi. distributions of soil water retention characteristics [J]. Water
org/10.1016/j.enggeo.2018.06.006. Resources Research, 1988, 24(5): 755 − 769. DOI: 10.1029/
[11] CHEN Bao, CHEN Jian-qin, CAO Yong-chao. Influence of WR024i005p00 755.
joint on self-sealing behaviour of highly compacted bentonite [24] LLORET A, VILLAR M V. Advances on the knowledge of
in engineering barrier [J]. Chinese Journal of Rock the thermo-hydro-mechanical behavior of heavily compacted
Mechanics and Engineering, 2012, 31(3): 618 − 624. DOI: FEBEX bentonite [J]. Physics and Chemistry of the Earth,
10.3969 /j.issn.1000-6915.2012.03.021. (in Chinese) 2007, 32(8): 701−715. DOI: 10.1016/j.pce.2006.03.002.
[12] ZHANG Hu-yuan, WANG Xue-wen, LIU Ping, YAN Ming, [25] FERRARI A, SEIPHOORI A, RÜEDI J, LALOUI L. Shot-
PENG Yu. Sealing and healing of compacted bentonite block clay MX-80 bentonite: An assessment of the hydro-
joints in HLW disposal [J]. Chinese Journal of Rock mechanical behavior [J]. Engineering Geology, 2014, 173(2):
Mechanics & Engineering, 2016, 35(S2): 3605 − 3614. DOI: 10−18. DOI: 10.1016/j.engg eo. 2014.01.009.
10.13722/j.cnki.jrme.2015.1691. (in Chinese) [26] CUI Su-li, DU Yan-feng, WANG Xue-pan, HUANG Sen,
[13] CHEN Yong-gui, JIA Ling-yan, YE Wei-ming, CHEN Bao, XIE Wan-li. Influence of temperature on swelling
CUI Yu-jun. Advances in experimental investigation on deformation characteristic of compacted GMZ bentonite-
hydraulic fracturing behavior of bentonite-based materials sand mixtures [J]. Journal of Central South University, 2018,
used for HLW disposal [J]. Environmental Earth Sciences, 25(11): 2819−2830. DOI: 10.1007/s11771-018-3955-9.
2016, 75(9): 787. DOI: 10.1007/s12665-016-5644-z. [27] LIU Fei-fei, MAO Xue-song, ZHANG Jian-xun, WU Qian,
[14] VILLAR M V, SÁNCHEZ M, GENS A. Behaviour of a LI Ying-ying, XU Cheng. Isothermal diffusion of water
bentonite barrier in the laboratory: Experimental results up to vapor in unsaturated soils based on Fick's second law [J].
8 years and numerical simulation [J]. Physics and Chemistry Journal of Central South University, 2020, 27(7): 2017 −
of the Earth, Parts A/B/C, 2008, 33: 476 − 485. DOI: https:// 2031. DOI: 10.1007/ s11771-020-4427-6.
doi.org/10.1016/ j.pce.2008.10.055. [28] BAI Bing, GUO Lan-jie, HAN Song. Pore pressure and
[15] BAI Bing, RAO Deng-yu, XU Tao, CHEN Pei-pei. SPH- consolidation of saturated silty clay induced by progressively
FDM boundary for the analysis of thermal process in heating/cooling [J]. Mechanics of Materials, 2014, 75: 84 −
homogeneous media with a discontinuous interface [J]. 94. DOI: 10.1016/j. mechmat.2014.04.005.
2918 J. Cent. South Univ. (2021) 28: 2905－2918
[29] BAI Bing, RAO Deng-yu, CHANG Tao, GUO Zhi-guang. A 3865. DOI: CNKI:SUN:YSLX.0.2007-S2-039.(in Chinese)
nonlinear attachment-detachment model with adsorption [37] ALONSO E E, VAUNAT J, GENS A. Modelling the
hysteresis for suspension- colloidal transport in porous media mechanical behaviour of expansive clays [J]. Engineering
[J]. Journal of Hydrology, 2019, 578: 124080. DOI: 10.1016/ Geology, 1999, 54(1, 2): 173−183. DOI: 10.1016/S0013-7952
j.jhydrol.2019.124080. (99)00079-4.
[30] ALONSO E E, GENS A, JOSA A. A constitutive model for [38] BAI Bing, LI Tao. Irreversible consolidation problem of a
partially saturated soils [J]. Géotechnique, 1990, 40(3): 405− saturated porothermoelastic spherical body with a spherical
430. DOI: 10.1680/geot.1990.40.3.405. cavity [J]. Applied Mathematical Modelling, 2013, 37(4):
[31] TANG Chao-Sheng, CHENG Qing, LENG Ting, SHI Bin, 1973−1982. DOI: 10.101 6/j.apm.2012.05.003.
ZENG Hao, INYANG H I. Effects of wetting-drying cycles [39] BAI Bing, SHI Xiao-ying. Experimental study on the
and desiccation cracks on mechanical behavior of an consolidation of saturated silty clay subjected to cyclic
unsaturated soil [J]. Catena, 2020, 194: 104721. DOI: thermal loading [J]. Geomechanics and Engineering, 2017, 12
10.1016/j. catena.2020.104721. (4): 707−721. DOI: 10.129 89/gae.2017.12.4.707.
[32] YANG Guang-chang, BAI Bing. Thermo-hydro- mechanical [40] LIU Yue-miao, WANG Ju, CAO Sheng-fei, MA Li-ke, XIE
model for unsaturated clay soils based on granular solid Jing-li, ZHAO Xing-guang, CHEN Liang. A large-scale
hydrodynamics theory [J]. International Journal of THMC experiment of buffer material for geological disposal
Geomechanics, 2019, 19(10): 04019115. DOI: 10.1061/ of high level radioactive waste in China [J]. Rock and Soil
(ASCE)GM.194 3-5622.0001498. Mechanics, 2013, 34(10): 2756 − 2762. DOI: 10.16285/j.
[33] FAN Qing-lai, LUAN Mao-tian, YANG Qing. Numerical rsm.2013.10.043. (in Chinese)
implementation of implicit integration algorithm for [41] YE Wei-ming, ZHU Chun-min, CHEN Yong-gui, CHEN
modified Cam-clay model in ABAQUS [J]. Rock and Soil Bao, CUI Yu-jun, WANG Ju. Influence of salt solutions on
Mechanics, 2008, 29(1): 269 − 273. DOI: 10.16285/j. the swelling behavior of the compacted GMZ01 bentonite
rsm.2008.01.051. (in Chinese) [J]. Environmental Earth Sciences, 2015, 74(1): 793 − 802.
[34] SRIDHARAN A, RAO A S, SIVAPULLAIAH P V. Swelling DOI: 10.1007/s12665-015-4108-1.
pressure of clays [J]. Geotechnical Testing Journal, 1986, [42] NIE Qing-ke, LI You-dong, WANG Guo-hui, BAI Bing.
9(1): 24−33. Physicochemical and microstructural properties of red muds
[35] ZHANG Feng, YE Wei-ming, WANG Qiong, CHEN Yong- under acidic and alkaline conditions [J]. Applied Sciences,
gui, CHEN Bao. An insight into the swelling pressure of 2020, 10(9): 2993. DOI: 10.339 0/app10092993.
GMZ01 bentonite with consideration of salt solution effects [43] YE Wei-ming, CHEN Yong-gui, CHEN Bao, WANG Qiong,
[J]. Engineering Geology, 2019, 251: 190 − 196. DOI: 10. WANG Ju. Advances on the knowledge of the buffer/backfill
1016/j.enggeo.2019.02.016. properties of heavily-compacted GMZ bentonite [J].
[36] YE Wei-min, SCHANZ T, QIAN Li-xin, WANG Ju, Engineering Geology, 2010, 116(1): 12 − 20. DOI: 10.1016/j.
ARIFIN Y. Characteristics of welling pressure of densely enggeo.2010.06.002.
compacted Gaomiaozi bentonite GMZ01 [J]. Chinese Journal (Edited by FANG Jing-hua)
of Rock Mechanics and Engineering, 2007, 26(S2): 3861 −

中文导读
缓冲材料的热-水-力耦合过程及接缝愈合效应

摘要：在高放废物处置库工程屏障系统中，缓冲材料砌块组合堆砌形成的技术接缝成为了缓冲层的薄
摘要：
弱部位。本文研究了技术接缝的高庙子(GMZ)膨润土缓冲材料，将液态水流动和水蒸汽迁移引起的传
热嵌入到能量守恒方程之中，基于 Barcelona 基本模型，通过分析膨润土砌块的膨胀过程及接缝材料的
压缩过程，建立了非饱和膨润土热-水-力耦合模型。然后以 China-Mock-up 试验为例，将数值计算结果
与试验结果进行对比，以验证所提模型的合理性。在此基础上，进一步分析了接缝愈合效应对缓冲材
料的干密度、导热系数和渗透系数的影响。结果表明，随着缓冲材料的水化膨胀，接缝材料的干密度
逐渐增大，导热性能和防渗性能逐渐与膨润土砌块趋于一致，缓冲材料在协调变形的作用下逐渐向均
质化发展。

关键词：
关键词：缓冲材料；热-水-力耦合；施工接缝；愈合效应