Original Article

Proc IMechE Part A:

J Power and Energy
Numerical simulation of coupled 0(0) 1–11
! IMechE 2017

thermal-mechanical fracturing in Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

underground coal gasification DOI: 10.1177/0957650917744699

journals.sagepub.com/home/pia

Tian-Hong Duan1,2,3, Jian-Ming Zhang3, Cliff Mallett3,

Hong-Hang Li4, Li-Wen Huo5 and Kai-Ting Zhao1,2

Abstract
A combination of thermal fracturing and stress-induced fracturing, i.e. coupled thermal-mechanical fracturing, occurs
under the effect of combustion-generated heat and stress in underground coal gasification. Controlling the cracking of
roofs and floors and the precise positioning of the combustion zone in underground coal gasification requires full
knowledge of the characteristics of the coupled thermal-mechanical fracturing of the surrounding rocks. This study
considers the variation in the physical and mechanical parameters of the rock with temperature and rock heterogeneity
in order to derive a mathematical model of coupled thermal-mechanical fracturing. Then, a numerical simulation is
performed, from which the following conclusions can be reached. First, temperature increases expand fractures, which
emerge in the highest temperature area, before extending to lower temperature areas. Second, fracture density is
directly related to temperature, with higher temperature corresponding to greater fracture density. Third, the cracking
rate increases linearly with time in stages. For m ¼ 1, the growth rate of the cracking rate are 55.76%/h, 26.08%/h
correspondingly during the first stage (from 0.8 h to 1.6 h) and the second stage (from 1.6 h to 2 h).While for m ¼ 2, the
growth rates are 18.50%/h, 40.99%/h correspondingly during the first stage (from 0.8 h to 1.3 h) and the second stage
(from 1.3 h to 2 h). Finally, fracture formation speed is slower in more homogeneous conditions when other conditions
are unchanged. A few fractures emerge in the sample after 1.3 h for m ¼ 2; the sample cracking rate is only 14.84%, which
is far lower than for m ¼ 1 at the same time (32.31%). Similarly, the cracking rate reaches 29.94% after 1.6 h for m ¼ 2
(still far lower than that for m ¼ 1 at the same time, which is 46.91%). The average growth rate of cracking rate for m ¼ 1
(27.16%/h) is quicker than that of m ¼ 2(21.77%/h).

Keywords
Underground coal gasification, coupled thermal-mechanical fracturing, numerical simulation, Weibull distribution

Date received: 6 April 2017; accepted: 24 October 2017

pollution from dust and noise. Furthermore, UCG

Introduction
In recent years, the disadvantages affecting the trad-
itional coal mining and processing industry (e.g.
restrictions associated with mining rate and environ-
mental pollution concerns) have become increasingly
obvious.1 These changes have made it urgent for
the coal industry to develop new methods of
coal-resource exploitation and utilisation. The devel-
opment and application of underground coal gasifica-
tion (UCG) technology have emerged as the main
direction of energy acquisition research.2 Compared
with conventional coal mining methods, UCG can
reduce or eliminate the need for underground
miners, thereby improving safety conditions. In add-
ition, UCG eliminates the need for treatment of coal
refuse and tailings at the surface, which reduces
reduces methane emissions and enables deep
resources, which are generally inaccessible to mining
to be utilized.3–8 Although UCG has the potential to
1
Key Laboratory of Deep Coal Resource Mining, School of Mines,
Ministry of Education of China, Beijing, China
2
State Key Laboratory of Coal Resources & Safe Mining, China
University of Mining and Technology, Xuzhou, China
3
International UCG Research Centre, China University of Mining and
Technology, Xuzhou, China
4
Lugou Coal Mine, Zhengzhou Coal Group, Zhengzhou, China
5
New China Trust Co., Ltd, Beijing, China
Corresponding author:
Jian-Ming Zhang, International UCG Research Centre, China University
of Mining and Technology, Xuzhou 221116, China.
Email: jzhang@carbonenergy.com.au
2 Proc IMechE Part A: J Power and Energy 0(0)
create two environmental hazards (ground-water con-
tamination and surface subsidence) if operations are
not optimally managed, both hazards appear avoid-
able through careful site selection and the adoption of
best practices for operations management. Moreover,
while all uses of coal, including UCG, produce carbon
dioxide, there exists a strong synergy between UCG
and carbon capture and sequestration (CCS): the
cavity developed during UCG might be used to
store supercritical carbon dioxide.9
The fundamental principle of UCG is changing
coal into combustible gas using heat and chemical
reactions; the difference between UCG and ordinary
coal gasification is that these processes are performed
underground. Gasification can be realized in the gal-
lery of an underground gasifier through three reaction
zones: the oxidation, reduction, and dry distillation
zones, as shown in Figure 1.10
Unlike traditional mining methods, UCG is imple-
mented in a closed, high-temperature environment;
therefore, delineation of the combustion zone with
monitoring technology is important. In the UCG,
microcracks in the coal and rock extend and form
broken meshes; this behavior is known as the coupled
thermal-mechanical fracturing of coal and rock. In
order to elucidate the mechanism and characteristics
of this phenomenon, many experiments and numer-
ical simulations have been conducted, and the thermal
fracturing characteristics of coal and rock under dif-
ferent temperatures and heating rates have been
researched.
For example, Johnson et al.11 and Wang et al.12
studied the thermal fracturing of US Westerly granite,
finding that coupled thermal-mechanical fracturing of
this rock occurs at 75  C. Further, Mahmutoglu13
demonstrated that the uniaxial compressive strength
and the tensile strength of Carrara marble and
Buchberger sandstone are reduced to approximately
Figure 1. A diagrammatic sketch of the principles of UCG.
50% and 60%, respectively, after heating. Using par-
ticle simulation, Xia et al.14 found that particle dis-
placement and the number of microscopic cracks
increase remarkably when a certain critical tempera-
ture is attained. Similarly, Zhao et al.15 showed that
increases in temperature can cause structural changes
in coal and surrounding rocks. Further, Shen et al.16
reported that multi-region boundary element analysis
can accurately simulate temperature changes and
thermal stresses in rocks. Simpson,17 Booker and
Savvidou,18 Ayotte et al.,19 and Huang20 have all
investigated the impact of heating rate on the thermal
fracturing of igneous rock, finding that a higher heat-
ing rate can significantly increase its thermal fractur-
ing. In addition, Akbarzadeh and Chalaturnyk21 have
found that pore volume, aperture, porosity, and per-
meability increase significantly with temperature in
UCG. Some researchers have also performed numer-
ical analyses of the thermal fracturing characteristics
of rock and coal, obtaining results following the pat-
tern reported above.22–24 Islam et al.25 and Kim et al26
modeled thermal discrete cracking with finite element
analysis.
The studies summarized in the previous paragraph
primarily analyzed the mechanism and characteristics
of the thermal fracturing of coal and rock; these
results indicate that high-temperature UCG induces
strong fracturing. However, few researchers have con-
sidered that most of the characteristics of in situ rocks
vary dramatically with temperature, and that the
characteristics of these in situ rocks can only be deter-
mined from experiments involving simulated in situ
conditions.
Therefore, in this study, we fully consider the in
situ conditions of the examined rock and incorporate
these properties into numerical simulations of coupled
thermal-mechanical fracturing. This approach is the
only method that ensures the accuracy of the
Duan et al.
numerical simulation and yields the coupled thermal-
mechanical fracturing characteristics of in situ rock.
Model and numerical simulation of coupled
thermal-mechanical fracturing of rock
To evaluate the stability of the direct sandstone roof
of the three_2 coal seam in the industrial UCG test
conducted in the Chengjiao coal mine and reduce the
risks of industrial tests, we designed a model experi-
ment at 1:10 scale. A diagram of this scale model
experiment is shown in Figure 2. The sandstone
sample used in this model experiment was mined
from the real direct roof of the three_2 coal seam.
Therefore, the material similitude ratio is 1:1 and
the stress similitude ratio is 1:10. Along with the gas-
ification of the coal below the direct roof, the coal was
burned to ash and a cave was formed. The rectangular
zone of A–B–C–D is subjected to stress and heat.
Therefore, coupled thermal-mechanical fracturing is
inevitable in this zone.
In this section, a mathematical model of the
coupled thermal-mechanical fracturing of rock is
derived and established. Then, a numerical simulation
with relevant parameters determined from a large-
scale rock sample experiment under high temperature
and triaxial stress is implemented. Note that the vari-
ations in physical–mechanical parameters with
Figure 2. Diagram of the scale model experiment.
3
temperature, as well as the heterogeneity of the
rock, are fully considered in the numerical simulation.
Thus, the coupled thermal-mechanical fracture char-
acteristics relevant to the direct sandstone roof of the
three_2 coal seam in the Chengjiao coal mine (oper-
ated by the Shenhuo Corporation in Henan, China)
are obtained.
Rock heterogeneity
In the traditional homogeneous thermo-elastic rock
mechanics model, the deformation and thermal frac-
turing of heated rock are attributed to elasticity-
plasticity; this behavior is modeled using macroscopic
elasticity-plasticity theory. However, this mechanical
model, which is based on classical mechanical theory,
ignores the heterogeneity of the microstructure inside
the rock mass and is not sufficiently exact to express
the complexity that appears throughout rock deform-
ation. Any rock specimen contains a variety of min-
erals; therefore, the different thermal expansion
coefficients of the various mineral particles—as well
as the different thermo-elastic properties of the aniso-
tropic particles with different crystallographic orien-
tations—should be considered. The latter induce
uncoordinated thermal expansion around the particle
boundaries, and all aforementioned factors contribute
to the thermal fracturing mechanism of a rock mass.
To simplify this model for ease of investigation, the
following assumptions and physical mechanics condi-
tions are adopted in this study:
1. At the microscopic level, rock is formed by com-
bining random nonhomogeneous particles. In con-
trast, on the macroscopic level, rock is a
homogeneous isotropic elastic medium containing
many mineral particles.
2. The physical–mechanical parameters of the micro-
scopic elements are the statistical properties of
many mineral particles.
3. Microscopic elements do not have macroscopic
statistical properties because of their small size;
therefore, their characteristics exhibit random
heterogeneity.
Mathematical model of coupled
thermal-mechanical fracturing
Coupled thermal-mechanical fracturing is a common
phenomenon in rock engineering. Rock is composed
of a variety of mineral particles with different thermal
expansion coefficients. However, as a continuum, the
mineral particles inside a rock cannot expand freely
according to their thermal expansion coefficients
under the influence of temperature; therefore, as
shown in Figure 3, thermal stress is generated inside
the rock, and fracturing occurs around the mineral-
particle boundaries. Further, the cracks in the rock
4 Proc IMechE Part A: J Power and Energy 0(0)

Figure 3. Frame diagram of the mechanism of coupled thermal-mechanical fracturing.

expand if the thermal stress exceeds the rock’s where l and G are the Lamé constants, u is the dis-
strength limit. Simultaneously, the mechanical prop- placement, " is the normal strain, and  is the shear
erties of the rock degrade with increased temperature, strain.
facilitating the generation of cracks under stress When equations (4) to (7) are substituted into
action. In return, the change in porosity caused equation (1), the following is obtained
by the development of fractures in the rock can also       
affect the heat transfer of the rock mass. @u1
@ l @x @u2
þ @x @u3
þ @x @u1
þ 2G @x @u1
@ G @x @u2
þ @x
1 2 3 1 2 1
Consequently, these two factors, force and heat, are þ
related. In general, it is easier to generate fractures @x1 @x2
  
under heat; this mechanism can be referred to as @u1 @u3
@ G @x3 þ @x 1 @ ð3KTÞ
coupled thermal-mechanical fracturing. þ þ F1 ¼ ð8Þ
@x3 @x1
In UCG, under high temperatures, coupled ther-
mal-mechanical fracturing of the rocks in the under-
ground gasification stope must occur. Because of the If parameters G, , and K are all considered
complexity of the coupled thermal-mechanical pro- heterogeneous, the model will be very complex and
cess, it is generally modeled by complex differential difficult to solve. In addition,  dominates the
equations, which are coupled via certain variables. thermal fracturing of the rock. Therefore, in this
The main equations for the model established in this mathematical model, only  is considered heteroge-
study are derived as follows. neous, with other parameters assumed to be homo-
The stress equilibrium equations of a rock mass, geneous. Therefore, equation (8) can be simplified as
considering temperature effects,27 are follows

@x1 @x1 x2 @x1 x3 @ ð3KTÞ  

þ þ þ F1 ¼ ð1Þ @ 2 u1 @2 u2 @ 2 u3 @2 u1
@x1 @x2 @x3 @x1 l 2
þ þ þ 2G 2
@ x1 @x1 @x2 @x1 @x3 @ x1
 2 2
  2 
@x2 @x1 x2 @x2 x3 @ ð3KTÞ @ u1 @ u2 @ u1 @2 u3
þ þ þ F2 ¼ ð2Þ þG 2 þ þG 2 þ
@x2 @x1 @x3 @x2 @ x2 @x1 @x2 @ x3 @x1 @x3
@ ð3KTÞ
@x3 @x1 x3 @x2 x3 @ ð3KTÞ þ F1 ¼ ð9Þ
þ þ þ F3 ¼ ð3Þ @x1
@x3 @x1 @x2 @x3
 
where  is the normal stress,  is the shear stress, F is @ 2 u1 @2 u2 @ 2 u3 @2 u1
l 2 þ þ þ 2G 2
the external force,  is the coefficient of linear thermal @ x1 @x1 @x2 @x1 @x3 @ x1
expansion (CLTE), T is the temperature; and K is the  2   2 
@ u1 @ 2 u2 @ u1 @2 u3
volumetric deformation modulus. Further27 þG 2 þ þG 2 þ
@ x2 @x1 @x2 @ x3 @x1 @x3
xi ¼ l" þ 2G"xi ð4Þ @ ð3KTÞ
þ F1 ¼ ð10Þ
@x1
 
@ui @uj
xi xj ¼ Gxi xj ¼ G þ i ¼ 1, 2, 3 j ¼ 1, 2, 3  2 
@xj @xi @ u1 @2 u2 @ 2 u3
ðl þ GÞ 2 þ þ
ð5Þ @ x1 @x1 @x2 @x1 @x3
 2 
@ u1 @2 u1 @2 u1
" ¼ "x1 þ "x2 þ "x3 ð6Þ þG 2 þ 2 þ 2 þ F1
@ x1 @ x2 @ x3
 
@T @
@ui ¼ 3 K þ KT ð11Þ
" xi ¼ i ¼ 1, 2, 3 ð7Þ @x1 @x1
@xi
Duan et al. 5

In the same way

 2 
@ u2 @ 2 u1 @ 2 u3
ðl þ GÞ 2 þ þ
@ x2 @x1 @x2 @x2 @x3
 2 
@ u2 @ 2 u2 @ 2 u2
þG 2 þ 2 þ 2 þ F2
@ x1 @ x2 @ x3
 
@T @
¼ 3 K þ KT ð12Þ
@x2 @x2
 
@ 2 u3 @ 2 u1 @ 2 u2
ðl þ GÞ 2 þ þ Figure 4. Stress–strain curve of the sandstone at a
@ x3 @x1 @x3 @x2 @x3
 2  temperature of 600  C.
@ u3 @ 2 u3 @ 2 u3
þG 2 þ 2 þ 2 þ F3
@ x1 @ x2 @ x3
  Here, S is the shear modulus, sij is the deviatoric
@T @ stress tensor, dsij is the deviatoric stress increment, d"ij
¼ 3 K þ KT ð13Þ
@x3 @x3 is the strain increment, and dl is the nonnegative
scalar proportionality constant.
In general, the stress balance equation of a rock In contrast to the general mathematical model of
mass is expressed as elasticity, two items are incorporated into the rock–
mass balance equation in the mathematical model
X3   developed in this study, namely, the deformation
@" @ 2 ui @T @
ðl þ GÞ þG þ Fi ¼ 3 K þ KT , caused by the temperature gradient and the thermal
@xi j¼1
@x2j @xi @xi
expansion coefficient gradient. These items help yield
i ¼ 1, 2, 3 ð14Þ a more accurate model of the thermal elasticity of a
random heterogeneous rock mass. Thus, in this study,
The sandstone used in this work displays brittle- the heat balance equation of rock is expressed as
ness or complete elasticity at low temperatures.28
 2 
However, as shown in Figure 4, when the @ T @2 T @2 T @T
;200 mm  400 mm sample is heated and compressed kr2 T ¼ k 2
þ 2
þ 2
¼ C ð19Þ
@z @y @x @t
in the 20 MN high-temperature, high-pressure servo-
controlled triaxial rock testing system, it obviously where k is the thermal conductivity of the rock,  is its
changes from brittleness to ductility. In the experi- density, and C is its specific heat capacity.
ments, the temperature was maintained at 600  C The mathematical model of coupled thermal-
and the confining pressure at 5 MPa. Wan showed mechanical fracturing comprises equations (14) to
that higher temperatures aid the brittle–ductile tran- (19), and the relevant characteristics and laws con-
sition,28 and Yu et al.29 reported that greater confin- cerning the coupled thermal-mechanical fracturing
ing pressure plays a positive role in the brittle–ductile of a rock mass can be obtained by solving these
transition. The confining pressure of the real physical equations.
model in our study is 15.2 MPa (greater than 5 MPa)
and the temperature of the corner of the model is
higher than 600  C. Therefore, the elastic–plastic con-
Numerical simulation
stitutive relation is used in this study. Because the scale in the Z direction is much larger
The constitutive relation of a rock mass under than that of the X and Y directions of the scale
changing temperature is determined as follows. For model, and the strain of the Z direction can be neg-
the elastic stage30 lected in comparison with those of the X and Y dir-
ections, the numerical simulation is based on a plane
ii ¼ l" þ 2G"ii  3KT ð15Þ strain model. Assuming that the CLTE obeys a
random distribution (Weibull distribution), the geom-
ij ¼ Gij ð16Þ etry of the plane model is taken as 20 cm  10 cm.
Because the physical–mechanical parameters of the
For the plastic stage31 rock change with temperature, these changes must be
considered as the temperature varies to achieve accur-
1 ate numerical simulation results. In most previous
d"ij ¼ dsij þ dlsij ð17Þ
2S experimental studies, the mechanical parameters of
high-temperature rock were established for nonstress
3d"p or uniaxial stress states; however, the actual rock state
dl ¼ ð18Þ
2i involves triaxial stress. In addition, most samples used
in previous experiments were small, thus, could not
6 Proc IMechE Part A: J Power and Energy 0(0)

effectively reflect the large-scale characteristics of an Table 1. Values of partial parameters at room
in situ rock mass. A large-scale in situ rock mass con- temperature.
tains many original fractures; thus, the larger the
Symbol Unit Value
sample, the more accurate the physical and mechan-
ical parameters. Zhao et al.,32 Wu et al.,33 Feng,34 and RT0 MPa 6.06
Qu35 have all conducted high-temperature, high- ’T0 
46
pressure triaxial pressure tests on rock using the QT0 MPa 13.10
XPS-20MN servo-controlled high-temperature, high-
pressure triaxial rock testing system developed by
Zhao and Wan. The axial and lateral maximum nom- Note that  varies with T, with different behaviors
inal forces of this testing system are 1000 t, and the exhibited at different T ranges, such that
sample size can be set to ;200 mm  400 mm.
8
The variations in parameters with temperature >
>  ¼ 6:66  106  6:30  108 T þ 1:92  106
obtained under high-temperature and high-pressure >
>
<  1010 T2 ðT 5 350  C,
triaxial-pressure testing on sandstone specimens of
;200 mm  400 mm have been reported by Wan,28 >
>  ¼ 7  105  2  107 Tð350  C 5 T 5 450  C,
>
>
who found that :
 ¼ 1:66  106 ðT > 450  C
ð27Þ
k=k0 ¼ 3:0196 T0:2847 ð20Þ

where k0 is the thermal conductivity of the rock at

room temperature. Further, the specific heat capacity The composition and distribution of minerals in
C of the rock is rocks are random; thus,  is taken as a random vari-
able in this paper. Three kinds of random distribution
C ¼ 0:0004T þ 1:0064 ð21Þ (Weibull, uniform, and exponential distributions) are
used to describe the heterogeneity of rocks; however,
and its elastic modulus E is given as the Weibull distribution is the most consistent with
  the heterogeneity of rocks.36,37
T If the probability density function of a random
E ¼ 90:18252 exp  þ 0:37607 ð22Þ
59:83617 variable x is
   m
The Poisson’s ratio  of the rock is given by m x m1 x
pðxÞ ¼ e ð28Þ
894:59068
 ¼ 0:36264 þ pffiffi
868:41092 2 where m and are constants, it is said that x obeys the
 ! ð23Þ Weibull distribution with parameter (m, ), which is
T  309:65286 2 abbreviated as x  (m, ). Then, the distribution
 exp 2
868:41092 function is
Z    m
In addition x
m t m1 t
FðxÞ ¼ e dt
1
RT =RT0 ¼ 0:0008T þ 1 ð24Þ Z    m  m ð29Þ
x
m t m1  t
 x
¼ e dt ¼ 1  e
where RT and RT0 are the uniaxial tensile strengths of 0
the rock at a given T and at 20  C, respectively.
Further In this paper,  is assumed to obey the Weibull
distribution with parameter (m, ). Thus, (m, ) is
’T =’T0 ¼ 0:0008T þ 1 ð25Þ called the ‘‘Weibull distribution parameter of ’’,
with m being the shape parameter, which reflects the
where ’T and ’T0 are the internal friction angles of the dispersion degree of . When m increases, the distri-
rock at a given T and at room temperature, respect- bution density function of  changes from short and
ively. Similarly wide to high and narrow; that is, it becomes more
concentrated and uniform. The smaller the value of
QT =QT0 ¼ 0:0008T þ 1 ð26Þ m, the larger the range of the physical–mechanical
parameters of the rock particles, and the less
where QT and QT0 are the cohesion of the rock at a homogeneous the rock is. In contrast, a larger m
given T and at room temperature, respectively. The corresponds to a smaller fluctuation range in the
initial values of the parameters at room temperature physical–mechanical parameters of the rock par-
in equations (24) to (26) are listed in Table 1. ticles, along with greater uniformity in the rock.
Duan et al.
Therefore, m reflects the heterogeneity of the rock
structure and is called the heterogeneity parameter.
In the numerical simulation, ¼ 9.6  106 K1 was
set at room temperature.
Regarding the other simulation parameters, the
depth H of the direct sandstone roof of the three_2
coal seam in Chengjiao coal mine is 600 m, and the
average density of the upper layers is assumed to be
2.5 kg/m3; therefore, the vertical stress is 15.0 MPa.
The lateral pressure coefficient is l ¼ 140/H þ 0.78
¼ 1.01, and the corresponding confining pressure is
15.2 MPa. Because the thickness of the coal seam is
too small in comparison with its buried depth, the
vertical stress is simplified to be equal. Since the
stress similitude ratio is 1:10, the vertical and confin-
ing stresses are correspondingly set to 1.5 MPa and
1.52 MPa.
When the blowing rate of the gasification agent is
stable and without large geological structures, the
heat release intensity of UCG can be taken as a con-
stant. Although the rectangular physical boundary of
this model is not static, because the exothermic pos-
ition of the oxidation zone moves slowly, the combus-
tion surface also moves slowly. Thus, the physical
boundary of this model can be taken as a slowly
moving boundary which moves along with the com-
bustion surface. The heat transfer of the rock in the
boundary ultimately reaches a balance, and the tem-
perature distribution of the rock in the boundary is
constant. The heat-release intensity of the heat source
is determined by the blowing rate of the gasification
agent. Therefore, when the blowing rate of the gasifi-
cation agent is stable, the heat-release intensity of the
heat source can be taken as a constant. Therefore, the
distribution of the temperatures of the interface
between the coal seam and roof is invariant. An air–
steam mixer is used as the gasification agent in this
paper. If the air proportion in the gasification agent is
72.17%, the maximum temperature of the reduction
zone is 798.35  C.38 Then, there exists one point at
which the temperature is 700  C on the interface
between the coal seam and roof. Because the distance
between the D corner and the gasification injection
pipe is the shortest, the combustion in the collapsed
coal and the coal above the cave on the vertical exten-
sion line of D points is the most violent. Considering
coal is a kind of poor conductor of heat, and for sim-
plifying, the initial temperatures of other units are set
to be 20  C. This is considered as the first boundary
condition of the heat transfer model. Therefore, the
boundary conditions of the stress field in the rock can
be expressed in Figure 5.
Finally, the Drucker–Prager yield criterion is taken
to govern coupled thermal-mechanical fracturing. The
Drucker–Prager yield criterion approximates the
Mohr–Coulomb criterion and modifies the von
Mises yield criterion (containing an additional item).
The yield surface does not change with gradual yield-
ing of the material; therefore, there is no hardening
7
criterion and the plastic behavior is assumed to be
elastic and perfectly plastic. However, the yield
strength increases with increasing confining pressure
(hydrostatic stress). In addition, the Drucker–Prager
criterion incorporates the volume expansion caused
by the yield. Therefore, this criterion is suitable for
granular materials such as concrete, rock, and soil.
The Drucker–Prager yield surface is a curved surface
in the principal stress space, which is expressed as
  pffiffiffiffiffiffiffiffiffiffiffiffi
f ¼ I1 ij þ I2 ðSij Þ þ k ¼ 0 ð30Þ
 
where f is the plastic potential function; I1 ij and
I2 ðSij Þ are the first and second invariants of the
stress tensor, respectively; and and k are the mater-
ial constants, which are functions of the cohesive force
and internal friction angle of the material,
respectively.
Results
Numerical simulations of the coupled thermal-
mechanical fracturing of rock under the random
Weibull distribution were conducted using
COMSOL Multiphysics 5.2a (COMSOL Inc.), and
the coupled thermal-mechanical fractures distribu-
tions, cracking rates, and temperature distributions
were obtained. The geometry model is divided into
400  200 units, with the scale of each unit being
0.5 mm  0.5 mm. The results obtained for m ¼ 1 and
2 are listed in Figures 6 to 11. In these figures, the
cracking rate is defined as the percentage of the total
units that are damaged during rupturing. (Note that
the results for only a portion of the timeframe are
listed because of space limitations.) Further, the
cracking rates as a function of time for different m
values are shown in Figure 12.
From Figures 6 to 8 and Figure 12, it is apparent
that the variety in coupled thermal-mechanical frac-
turing profiles within the rock becomes obvious
within a short period of time for m ¼ l. A small
number of fractures emerge in the sample after 1 h,
and the cracking rate is only 19.67% (Figure 12).
However, after 1.6 h, many fractures have emerged,
and the cracking rate has reached 46.91%. The tem-
perature extends from the upper-left corner, as does
the fracturing behavior. The fracture distribution
density is directly related to temperature, with a
higher temperature corresponding to greater crack
density; therefore, the damaged area expands with
the gradual increase in temperature under the syner-
gistic effect of the stresses. The cracking rate varies
from 2.30% at 0.8 h to 54.31% at approximately
2 h; this behavior is almost linear, as shown in
Figure 12.
The behavior observed in Figures 9 to 11 is similar
to that for Figures 6 to 8. The variety in the coupled
thermal-mechanical fracturing profiles inside the rock
quickly becomes apparent for m ¼ 2. A few fractures
8 Proc IMechE Part A: J Power and Energy 0(0)
Figure 5. Stress field boundary condition.
(a) (b)
Figure 6. Results after 1 h with the Weibull distribution
(m ¼ 1): (a) fracture distribution; (b) temperature distribution.
emerge in the sample after 1.3 h; the sample cracking
rate is only 14.84%, which is far lower than for m ¼ 1
at the same time (32.31%). However, many fractures
have emerged after 1.6 h, and the specimen cracking
rate reaches 29.94% (still far lower than that for m ¼ 1
at the same time, which is 46.91%). As in the previous
case, the temperature extends from the upper-left
(a) (b)
Figure 7. Results after 1.3 h with the Weibull distribution
(m ¼ 1): (a) fracture distribution; (b) temperature distribution.
corner, like the fracturing behavior. The fracture dis-
tribution density is directly related to the temperature;
therefore, the damage area expands with the gradual
increase in temperature. The cracking rate ranges
from 1.80% at 0.8 h to 43.53% at approximately
2 h, which also yields an almost linear plot, as
shown in Figure 12.
Duan et al.
(a) (b)
Figure 8. Results after 1.6 h with the Weibull distribution
(m ¼ 1): (a) fracture distribution; (b) temperature distribution.
(a) (b)
Figure 9. Results after 1.3 h with the Weibull distribution
(m ¼ 2): (a) fracture distribution; (b) temperature distribution.
(a) (b)
Figure 10. Results after 1.6 h with the Weibull distribution
(m ¼ 2): (a) fracture distribution; (b) temperature distribution.
Discussion
The simulation results for m ¼ 1 and 2 are similar,
indicating that fractures extend from the upper-left
corner in accordance with the temperature increase.
Further, the fracture distribution density is directly
related to temperature. These results indicate that
9
(a) (b)
Figure 11. Results after 2 h with the Weibull distribution
(m ¼ 2): (a) fracture distribution; (b) temperature distribution.
Figure 12. Variation in cracking rate with time under  with
the Weibull distribution.
fracture formation and propagation are directly
related to temperature, and that this behavior can
be interpreted as the consequence of thermal stress
and the weakening of the material.
The cracking rate for the m ¼ 2 case after 1.6 h is
29.94%, which is far lower than that obtained at the
same time in the m ¼ 1 case; this difference can be
interpreted as the effect of material homogeneity.
For m ¼ 2, the degree of homogeneity of the rock is
higher, the distribution range of the thermal expan-
sion coefficient is smaller, and the particle expansions
are closer to a number. Therefore, it is more difficult
for thermal stress to emerge in the rock, which yields a
slower fracture growth rate.
The coupled thermal-mechanical fracturing of in
situ rock is an extremely complicated process. Thus,
there are some discrepancies between the numerical
simulation results obtained in this paper and actual
results, which are impractical to measure directly
under in situ rock stress and temperature conditions.
However, this study aids in understanding numerical
simulation of coupled thermal-mechanical fracturing
of rock mass under in situ stress and temperature.
10
Finally, note that acoustic emission events occur
upon coupled thermal-mechanical fracturing of
rock. Acoustic emission monitoring systems are
widely used in mines and can record the energy
levels released upon fracturing in the form of stress
waves, based on acoustic emission generated by the
production, expansion, friction, and energy accumu-
lation of fractures.39–43 In addition, acoustic emission
monitoring is a possible method of determining the
precise location of the combustion area in UCG.
Thus, investigation of coupled thermal-mechanical
fracturing will provide a theoretical basis for deter-
mining the UCG combustion area location.
Therefore, the authors will focus on investigating
the acoustic emission characteristics of coupled ther-
mal-mechanical fracturing in future work.
Conclusions
A mathematical model of the coupled thermal-
mechanical fracturing of rock is established in this
paper; the variation in the physical–mechanical
parameters and heterogeneity of the rock are fully
considered. In addition, numerical simulation of the
coupled thermal-mechanical fracturing of rock is per-
formed using the established model. The following
conclusions are obtained: (1) Fractures appear in the
area with highest temperature before gradually
extending to lower temperature regions. (2) The frac-
ture distribution density is directly related to tempera-
ture, i.e. a higher temperature corresponds to a
greater fracture density. (3) The coupled thermal-
mechanical cracking rate of the rock increases linearly
with time by stages. For m ¼ 1, the growth rate of the
cracking rate is 55.76%/h during the first stage from
0.8 h to 1.6 h. And the growth rate is 26.08%/h during
the second stage from 1.6 h to 2 h. While for m ¼ 2, the
growth rate is 18.50%/h during the first stage from
0.8 h to 1.3 h. And the growth rate is 40.99%/h during
the second stage from 1.3 h to 2 h. (4) Fracture for-
mation is relatively slow under homogenous condi-
tions, if other conditions are unchanged. A few
fractures emerge in the sample after 1.3 h for m ¼ 2,
the sample cracking rate is only 14.84%, which is far
lower than for m ¼ 1 at the same time (32.31%).
Similarly, the cracking rate reaches 29.94% after
1.6 h for m ¼ 2 (still far lower than that for m ¼ 1 at
the same time, which is 46.91%). The average growth
rate of cracking rate for m ¼ 1 (27.16%/h) is quicker
than that of m ¼ 2(21.77%/h).
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of
this article.
Funding
The author(s) disclosed receipt of the following financial
support for the research, authorship, and/or publication
Proc IMechE Part A: J Power and Energy 0(0)
of this article: This study was supported by a project
funded by the Priority Academic Program Development
of Jiangsu Higher Education Institutions and the
Fundamental Research Funds for the Central Universities
(China University of Mining and Technology) under Grant
No. 2017XKQY045.
References
1. Yang LH. Research on seepage combustion method of
underground coal gasification. Research on seepage
combustion method of underground coal gasification.
Xuzhou, Jiangsu: China University of Mining and
Technology Press, 2001.
2. Duan TH, Wang ZT, Liu ZJ, et al. Experimental study
of coal pyrolysis under the simulated high-temperature
and high-stress conditions of underground coal gasifi-
cation. Energy Fuels 2017; 31: 1147–1158.
3. Duan TH, Wang Z, Zhou L, et al. Gas produc-
tion strategy of underground coal gasification based
on multiple gas sources. Sci World J 2014; 2014:
154197–154197.
4. Duan TH, Lu CP and Xiong S. Evaluation method of
the energy conversion efficiency of coal gasification
and related applications. Int J Energy Res 2015; 40:
168–180.
5. Duan TH, Lu CP and Xiong S. Pyrolysis and gasifica-
tion modeling of underground coal gasification and the
optimisation of CO2 as a gasification agent. Fuel 2016;
183: 557–567.
6. Yu L. Academic papers on underground coal gasification.
Beijing: China University of Mining and Technology
Press, 2003.
7. Yang LH, Liang J and Wang JG. Study on the necessity
of underground coal gasification in China. Coal Sci
Technol Mag 1995; 2: 32–34.
8. Yu L. Research and development of underground coal gas-
ification technology. Sci Technol Rev 1995; 2: 54–56.
9. Burton E, Friedmann J and Upadhye R. Best practices
in underground coal gasification. Livermore, CA:
Lawrence Livermore National Laboratory, 2006.
10. Duan TH, Wang ZT, Li NL, et al. Thermodynamic
equilibrium model research and application of under-
ground coal gasification. Energy Educ Sci Technol Part
A: Energy Sci Res 2012; 30: 837–844.
11. Johnson B, Gangi AF and Handin J. Thermal cracking
of rock subject to slow uniform temperature changes.
In: Proceedings of 19th US symposium on rock mech-
anics, Stateline, NV, USA: American Institute of
Mining Engineers, 1978, pp.259–267.
12. Wang HF, Bonner BP, Carlson SR, et al. Thermal stress
cracking in granite. J Geophys Res 1989; 94: 1745–1758.
13. Mahmutoglu Y. Mechanical behaviour of cyclically
heated fine grained rock. Rock Mech-Rock Eng 1998;
31: 169–179.
14. Xia M, Zhao CB and Hobbs BE. Particle simulation of
thermally-induced rock damage with consideration of
temperature-dependent elastic modulus and strength.
Comput Geotech 2014; 55: 461–473.
15. Zhao YS, Qu F and Wan ZJ. Experimental investiga-
tion on correlation between permeability variation and
pore structure during coal pyrolysis. Transp Porous
Med 2010; 82: 401–412.
16. Shen B, Kim HM and Park ES. Multi-region boundary
element analysis for coupled thermal-fracturing
Duan et al.
processes in geomaterials. Rock Mech Rock Eng 2013;
46: 135–151.
17. Simpson C. Deformation of granitic rocks across the
brittle-ductile transition. J Struct Geol 1985; 7: 503–511.
18. Booker JR and Savvidou C. Consolidation around a
point heat source. IJNAMG 1985; 9: 173–184.
19. Ayotte E, Massicotte B and Houde J. Modeling the
thermal stresses at early ages in a concrete monolith.
ACI Mater 1997; 94: 577–587.
20. Huang CX. The three dimensional modeling of thermal
cracks in concrete structure. Mater Struct 1999; 32:
673–678.
21. Hossein A and Richard JC. Structural changes in coal
at elevated temperature pertinent to underground coal
gasification: a review. Int J Coal Geol 2014; 131:
126–146.
22. Yan CZ and Zheng H. A coupled thermo-mechanical
model based on the combined finite-discrete element
method for simulating thermal cracking of rock. Int J
Rock Mech Min Sci 2017; 91: 170–178.
23. Tang SB, Tang CA and Zhu WC. Numerical investiga-
tion on rock failure process induced by thermal stress.
CJRME 2006; 25: 2071–2078.
24. Li LC, Yang TH and Tang CA. Study on coupled ther-
mal-mechanical damage model in rock failure process.
Rock Soil Mech 2006; 27: 1727–1732.
25. Islam M, Sjolind SG and Pramila A. Finite element
analysis of linear thermal expansion coefficients of uni-
directional cracked composites. J Compos Mater 2001;
35: 1762–1776.
26. Kim S, Lee D, Koo J, et al. Evaluation of influence of
cracks on DVC coating using finite element analysis.
Appl Mech Mater 2014; 467: 24–28.
27. Yang GT. An introduction to elastic-plastic mechanics.
2nd ed. Beijing: Tsinghua University Press, 2013.
28. Wan ZJ. Study on thermal-mechanical coupling process
of heterogeneous rock mass and channel’s stability of
underground coal gasification. Xuzhou: China
University of Mining and Technology, 2006.
29. Yu HN, Xu YF and Sun CL. Deformation character-
istics of sandstone under normal temperature and
affecting factors. Rock Soil Mech 1995; 16: 70–77.
30. Kang J. Random media solid-head coupled mathematics
model and numerical tests on hot cracking of rock.
FuXin: LiaoNing Technical University, 2004.
11
31. Zeng XG, Chen HY and Hu YP. Engineering plastic
mechanics. Chengdu, China: Sichuan University, 2013.
32. Zhao YS, Wan ZJ, Zhang Y, et al. Experimental study
of related laws of rock thermal cracking and permeabil-
ity. CJRME 2010; 1970–1976.
33. Wu JW, Zhao YS and Wan ZJ. Experimental study of
acoustic emission characteristics of granite thermal
cracking under middle-high temperature and triaxial
stress. Rock Soil Mech 2009; 30: 3331–3336.
34. Feng ZJ. The theory and experimental research of hydro-
carbon resources extraction from in situ lignite via inject-
ing vapour. Taiyuan: Taiyuan University of Technology,
2012.
35. Qu F. Experimental study on the coal mass pyrolysis and
mechanical characteristics in simulative situ. Xuzhou:
China University of Mining and Technology, 2007.
36. Tang CA. Numerical studies of the influence of micro-
structure on rock failure in uniaxial compression – part
1: effect of heterogeneity. Int J Rock Mech Min Sci
2000; 37: 555–569.
37. Tang CA and Hudson JA. Rock failure mechanisms
explained and illustrated. Boca Raton, FL: Taylor and
Francis, CRC Press, 2010.
38. Duan TH. Study on pyrolysis model experiments and
gasfication parameters of underground coal gasification.
Xuzhou: China University of Mining and Technology,
2014.
39. Dou LM and He XQ. Theory and technique on rockburst
prevention. Xuzhou, Jiangsu: China University of
Mining & Technology Press, 2001.
40. Dou LM and Zhao CG. Mining-induced rockburst dis-
aster prevention. Xuzhou, Jiangsu: China University of
Mining & Technology Press, 2006.
41. Wang HL and Ge MC. Acoustic emission/microseismic
source location analysis for a limestone mine exhibiting
high horizontal stresses. Int J Rock Mech Min Sci 2008;
45: 720–728.
42. Angus FCE, Brian LFD and Arnfinn FP. Energy spec-
tral density characterization of microseismic events in
potash mines. Measurement 2009; 42: 264–268.
43. Xu XF, Dou LM, Lu CP, et al. Frequency spectrum
analysis on micro-seismic signal of rock bursts induced
by dynamic disturbance. J Min Sci Technol 2010; 20:
682–685.