Engineering Geology 121 (2011) 165176

Contents lists available at ScienceDirect

Engineering Geology
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e n g g e o

Physical modeling of an underground roadway excavation in geologically 45

inclined rock using infrared thermography
Manchao He
State Key Laboratory for Geomechanics & Deep Underground Engineering, Beijing 100083, China
School of Mechanics and Civil Engineering, China University of Mining & Technology, Beijing 100083, China

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

Article history: Physical modeling of excavation of a roadway tunnel in the geologically 45 inclined rock strata by using a
Received 21 April 2010 newly developed physical modeling approach, the so-called Physically Finite Elemental Slab Assemblage
Received in revised form 4 December 2010 (PFESA), together with infrared (IR) thermography, were conducted for visualization and observation of the
Accepted 15 December 2010
progressive development of the excavation damaged zone (EDZ). The excavation was schemed as two phases,
Available online 23 December 2010
i.e. phase 1: the full-face excavation and phase 2: the staged excavation. The experiment was performed under
Keywords:
unbalanced connement found in underground coal mining. The relative temperatures of the excavation zone
Physical modeling were mapped by the IR camera, in real time and over the entire eld, into thermography series which were
Roadway tunneling processed using such procedures as statistics, noise removal and spectral-frequency analysis for image feature
Stratied rock extraction. IRT, dened as the statistical mean of the thermography matrix, was used as a measure of rock
Excavation damage responses, characterizing the full-face excavation and the staged excavation as a multi-linearity and
Infrared thermography nonlinearity processes respectively by its time-marching scheme. The temporal and spatial evolutions of the
Spectral analysis EDZ were described in detail by the denoised thermographies. Directional propagations of the excavation
induced stress wave were characterized by the spectral-frequency analysis of the resultant thermographies.
2010 Published by Elsevier B.V.

1. Introduction
Understanding the mechanism of rock damage in the creation and
operation of an underground cavern in jointed rock masses has been a
topic of research for engineers in various elds, including developing
transportation tunnels, petroleum production drillings, and under-
ground mines, for example (He, 2006). Extensive researches with
varied methods, including numerical, laboratory experiments, and in-
situ tests, have been conducted to investigate the excavation related
problems. In-situ tests involve the investigation of the excavation
responses (Read, 2004); monitoring the displacement of a power-
house cavern during excavation (Li et al., 2008), back analysis of
excavation-induced wall deection (Tang and Kung, 2009), etc. In
recent decades, numerical methods have increasingly been used in
simulating the stress redistribution eld created by the tunneling
impacts (Cai, 2008), the swelling behavior around tunnels (Barla,
2008); acoustic emission in underground excavations (Cai et al.,
2007); failure mechanism of tunnel in jointed rock mass (Jia and Tang,
2008). Although computational techniques have progressed fast,
geotechnical researchers depend heavily on physical modeling tests
with a reduced model have been widely used at laboratory in
investigating the failure mechanism and behavior of tunnel face
This article is published with Editor corrections alone.
Tel.: + 86 10 62331091; fax: + 86 10 51733713.
E-mail address: hemanchao@263.net.
(Chambon and Corte, 1994); the stability of the face and unsupported
span over tunnel excavation in weak rock (Lee and Schubert, 2008);
ground movement and collapse mechanisms induced by tunneling in
clayey soil (Wu and Lee, 2003); progressive failure and scale effect of
trapdoor problem (Tanaka and Sakai, 1993; Park, et al., 1999); tunnel
face reinforcement by bolting with centrifuge model test (Sharma et
al., 2001; Kamata and Masimo, 2003).
Field experiments and in-situ tests are limited both in time and
costs. Numerical modeling allows one to conduct more realistic
analyses that take into account the complexity of the jointed rock
masses except for rock masses containing natural discontinuities of
varying size, strength and orientation. In practice, it is almost
impossible to explore all of the joint systems or to investigate all
their mechanical characteristics and implementing them explicitly in
a theoretical model (Sitharam and Latha, 2002). The physical
modeling tests at laboratory with a small-scale model are also
limited by the fact that in situ stresses are not realistically simulated,
and there is inconsistency of scaling factors for different measured
quantities, e.g. length, inertia force, creep (Meguid et al., 2008). Due
to the increase in mining depth and in urbanization found all over the
world, so many tunnels were and will be built under complex
geological conditions. As a result, a comprehensive and deepened
understanding of the tunneling induced displacements and stresses
and their impact on nearby structures have been becoming
imperative. To achieve these goals, new physical modeling methods
for simulating the excavation related problems in a large-scale

0013-7952/$ see front matter 2010 Published by Elsevier B.V.

doi:10.1016/j.enggeo.2010.12.001
166 M. He / Engineering Geology 121 (2011) 165176
geological model, incorporated with the state-of-the art measuring
techniques that is capable of visualization of the stress eld in real-
time and over the entire eld with a nondestructive manner around
the tunneling face, should be introduced into the laboratory model
test.
The objective of this study is to understand the dynamical rock
responses to the tunnel excavation in the geologically 45 inclined
rock strata using a newly developed physical modeling approach by
the author, i.e. the so called Physically Finite Elemental Slab
Assemblage (PFESA). The PFESA method uses a specimen-sized
rectangular plate, referred to as elemental slab thereafter, for
building the large-scale physical model by making layers with these
elemental slabs at specic dip angles, simulating the geologically
inclined rock strata. The elemental slab, made of articial materials
(also referred to as similarity material), simulate the real rock types
in terms of material and mechanical properties. The PFESA method
incorporated with infrared (IR) thermography has been utilized, at
the State Key Laboratory for Geomechanics and Deep Underground
Engineering at China University of Mining and Technology (CUMTB),
for modeling the creation and operation of the underground roadway
tunnels at depths ranging from 300 m to more than 1000 m below
the ground surface (He et al., 2009). IR thermography, as a
nondestructive and noncontact measuring technique was used to
inspect coated materials (carbon-to-carbon composites) and can
detect disbonds between coating and substrate (Connolly and
Copley, 1990), examine the progressive failure of concrete and rock
specimens (Luong, 2007), describing the dynamical process of
turbulence by its spectral-frequency characteristics (Gong et al.,
2008). During the test, IR camera looks at the surface in view and
maps the relative IR temperature eld and shows a contrast between
solid and void spaces. The resultant IR thermography series were
then utilized in observing and capturing the mechanical and
structural behaviors of the rocks under the roadway tunnel
excavation.
The physical modeling works at CUMTB on the roadway
excavation were conducted under a two-phase excavation proce-
dure, i.e. the full-face excavation and staged excavation, including
excavations in the rock strata with inclinations ranging from 0, 15, 30,
45, 60, 90 respectively. In the physical modeling in the geologically
horizontal strata (He et al., 2010a), IRT, dened as the statistical
mean of the thermography matrix, characterizes the full-face
excavation and staged excavation as a liner process and a non-linear
process respectively; the same relationship between IRT and the
excavation phases was conrmed in the physical modeling in the
geologically vertical strata (He et al., 2010b). The EDZ is distributed
with a localized manner around the tunneling face in the horizontal
strata, while disturbed-belt EDZ is dominant in the vertical strata.
The spectral-frequency analysis of the thermography found the
dependency of the excavation-induced stress wave propagation in
the vertical strata, whereas the directionality of the stress wave in the
horizontal strata is random. These ndings will be compared with the
results obtained in this paper from the excavation of an underground
roadway in the geologically 45 inclined rock strata.
Table 1
Physical and mechanical parameters for the real rocks.
Unit weight Unconned compressive
(kN/m3) strength (MPa)
Sandstone 27 64
Mudstone 26 44
Coalseam 14 26
Analogous ratio 1.9:1.9:1 2.5:1.7:1
(sand:mud:coal)
2. Physical model
This physical modeling test is to simulate excavation of a roadway
tunnel in QI SHAN underground coal mine in Xuzhou coal mining district,
located in Jiangshu province, eastern China. The roadway tunnels in this
underground coal mine are at depths ranging from 300 m to more than
1000 m below the ground surface (note that the depth in the following
context is referred to counting from the ground surface). The main
exposure lithologics are sandy mud rock, mudstone and sandy rock. The
coal seam is inclined at different inclination angles from 0 to 90 with
respect to the horizontal, with black colored, semi-bright coal rocks in the
form of block and granulous. Rock types at depth in underground coal
mines in China can be generalized into three categories, i.e. the
sandstones, mudstones and coal rocks (He et al., 2006). Unconned
uniaxial compressive tests were conducted on the rock specimens
sampled from the surrounding rock masses of the eld case. According to
the in situ survey and laboratory specimen tests, the physical and
mechanical parameters of the real rocks are reported in Table 1, which
are the real rock properties simulated by the articial materials in our
test. From Table 1 one can see that the unconned uniaxial compressive
strength (UCS) of the coal rock is close to the threshold strength 25 MPa
for the soft rocks at depth (He et al., 2002).
Dimensional analysis was undertaken to link the rock reality of the
eld case to the physical model in our test. The dimensions of the
PFESA model were 1600 1600 400 mm, and the roadway excava-
tion zone was designed as a cubic space with a face area of
250 200 mm and tunneling length of 400 mm (see Figure 1a). The
geometric scale factor l, dened as the ratio of the dimension of the
real rockmass to the PFESA model, was determined as l = 12. Thus
the PFESA model can simulate a rectangular-shaped underground
cavern with the width of 3 m and height of 2.4 m and a dimension of
the surrounding rockmass of 19.2 19.2 m (see Figure 1b), which was
the typical conguration and dimension of the underground roadway
found in China. According to the uniaxial compressive strength of the
real rocks and the maximum loading capacity (5 MPa) of the testing
machine, the force scale factor was evaluated as = 8. In addition,
for the sake of calculating the body force (i.e. the unit weight, see
Table 1), we need to dene a body force scale factor , which can be
derived by the relation = /l = 8/12 = 0.67. These force scale
factors were used in the test for fabrication the articial materials of
the PFESA model.
Real rock properties reported in Table 1 were simulated by the
articial materials (analogue materials, actually gypsum and water
used in our test) in terms of the material and mechanical properties.
There are discrepancies for the parameters between the real rocks and
the articial materials, due primarily to the fact that the material and
mechanical quantities for the articial materials are random at a given
ratio of water to gypsum. We choose the optimal result of the
quantities for the articial material among many trials which was
closest to that of the real rocks for the simulated eld case. For
producing three kinds of model rocks corresponding to the simulated
real rock types, three watergypsum ratios were used, i.e. 0.8:1 for
producing sandstone of the model rock, 1:1 for mudstone of the
Tensile strength Young's modulus Internal friction Poisson's
(MPa) (GPa) angle () ratio
5.83 26 34 0.15
5.59 21 36 0.13
0.90 5 40 0.36
6.5:6.2:1 5.2:4.2:1 0.9:0.9:1 0.4:0.4:1
 M. He / Engineering Geology 121 (2011) 165176 167

Fig. 1. The tested PFESA model. (a) Schematic of the model; (b) rock reality of the underground cavern simulated by the PFESA model.

model rock and 1.2:1 for coal of the model rock respectively. The
mechanical and material properties of the model rocks were reported
in Table 2. For evaluation of the analogous degree of the analogue and
real materials, we dened a factor, analogous ratio (AR, AR = sand:
mud:coal), to compute the relative ratios for each of the material and
mechanical parameters for the real rocks and model rocks (see
Tables 1 and 2).
Those three kinds of analogue materials were used to manufacture
the elementary slabs three model rock types in simulating three real
rock types, i.e. sandstone, mudstone and coal rock. All the elementary
slabs were fabricated with the same surface dimension 400 400 mm
and three kinds of thickness of 1, 3 and 3 mm respectively. The
geometrical parameters of the elementary slabs were also reported in
Table 2. A large number of the elementary slabs were assembled in
layers with the specied mechanical properties, simulating the real
rock strata of the sandstone, mudstone and coal seam. The integration
of the layers with elementary slabs of the same type constitutes the
model rock stratum. The physical analogue model was built with
total nine strata, including one sandstone stratum, four mudstone
strata and four coal seam strata. All the strata were inclined at an angle
of 45 with respect to the horizontal, to mimic the geologically 45
inclined real rock strata. These model rock strata, indexed 19 from
the left to the right in the physical model, were assembled in
accordance with the sequence of the geological section of the QI SHAN
coal mine. The constructed physical analogue model in our test is
shown in Fig. 1. Fig. 1a is the schematic drawing, and b is the diagram
of the deep ground roadway simulated by the model. The geological
section, material and structural parameters of the physical analogue
model were reported in Table 3.
3. Excavation scheme and boundary conditions
The roadway excavation zone was located on the stratum 6 (coal
seam) centered on the PFESA model. The tunnel excavation space in
the model was a cubic volume measuring 250 200 400 mm. The
total roadway excavation volume in the physical model was divided
into seven sub-spaces along the strike of the strata as seen in Fig. 2a.
These sub-spaces being excavated were referred to as rock block
(RB) as shown in Fig. 2. The RBs were numbered #1#7
corresponding to the roadway excavation sequence as seen in Fig. 2.
The whole amount of tunneling work planned as two phases: phase 1:
full-face excavation, i.e. tunneling on the #1 RB and the adjacent rocks
until a small passage is cut through; and phase 2: staged excavation,
i.e. removing one RB at each tunneling stage. For the full-face
excavation, the excavated volume of the articial materials from the
roadway tunnel was referred to as footage thereafter. For the staged
excavation, the term excavation stage denotes the removing each of
the RBs, and total of seven excavation stages were performed step by
step in sequence during the phase 2 excavation. The excavation of the
roadway tunnel without support was started at the back side of the
model, and went through to the front face, with a chisel and a hammer
as the tunneling tools, simulating the tunnel excavation by drill and
blast method.
During the excavation, the vertical load and lateral load were
applied on the model (simulated as quasi-2D plane strain state) by
using the YDM-C Geological Disaster Simulation Testing Machine,
developed in our institute, as seen in Fig. 3. The testing machine YDM-
C was equipped with a hydraulic servo system, imposing the
uniformly distributed load on the top and two side boundaries of
the model. The bottom of the model was xed on the basement of the
machine, simulated as a rigid boundary condition during the
excavation. During the excavation, the vertical stress 1 was kept at
a constant value of 1.0 MPa, corresponding to the mining depth
300 m. We chose the lateral pressure coefcient = 0.3, (dened by
= 2/1, 2 is the lateral stress), for reproducing the unbalanced
stress state on boundaries in deep ground, and 2 was also kept
constant at 0.3 MPa during the whole excavation processes as
illustrated in Fig. 4.

Table 2
Mechanical, material and geometric parameters of the articial materials used in the physical model.

Rock types simulated by Dimensions of the Unit weight Unconned compressive Tensile Young's Internal friction Poisson's
the elementary slab elementary slabs (cm) (kN/m3) strength (MPa) strength (MPa) modulus (GPa) angle () ratio

Sandstone 40 40 3 15 8 0.72 3.22 32 0.13

Mudstone 40 40 2 11 5 0.69 2.62 33 0.12
Coal seam 40 40 1 8. 3 0.11 0.61 33 0.32
Sandstone 40 40 3 15 8 0.72 3.22 32 0.13
Mudstone 40 40 2 11 5 0.69 2.62 33 0.25
Analogous ratio 1.9:1.4:1 2.5:1.7:1 6.5:6.3:1 5.3:4.3:1 0.9:1:1 0.4:0.4:1
(sand:mud:coal)
168 M. He / Engineering Geology 121 (2011) 165176

Table 3 system, one can observe and capture the physical and structural
Geological section, material and structural parameters of the physical model from the changes of an object, represented by the infrared radiation temper-
left to the right.
ature changes, in real time and over the entire eld of the viewed
Geological Stratum no. Rock types Layer thickness Layer surface. The relationship between dissipated energy and the tested
sectio n (mm) numbers objects temperature obeys the StefanBoltzmann law:
1 Sandstone 440 14
4
M = T 1
2 Coal-seam 140 10

where M is the radiant exitance, W m 2, the object's emissity,

3 Mudstone 120
4 Coal-seam 250
5 Mudstone 150
6 Coal-seam 60
7 Mudstone 140
8 Coal-seam 60
9 Mudstone 240
4. Infrared detection
Although some contact detection technologies, such as accelera-
tion sensors, strain gauges, stress wave propagation meters and
displacement measurement, have been successfully applied both in
situ and laboratory, but it is not suited for such dynamical processes as
rock impaction because of the intrusive disturbance and damage-
prone problem of the contacting detectors (Shi et al., 2007). The
widely used non-destructive detection technique in rock mechanics
includes acoustic emission (AE) and electromagnetic radiation. It
describes the progressive development failure of the rock indirectly
by the transformation or statistics of the test data sets (Majewska and
Mortimer, 1998; Shiotani, 2006; He et al., 2010c).
Owing to the thermomechanical coupling, infrared thermography
provides a non-destructive, non contact and real-time test to observe
the physical process of material degradation and to detect the
occurrence of intrinsic dissipation without surface contact or in any
way inuencing the actual surface temperature of the tested object
(Long, 1995). It produces heat images directly from the invisible
radiant energy (dissipated energy) emitted from stationary or moving
objects at any distance. By using an infrared thermographic detection
7
0 b b 1, the StefanBoltzmann constant, 5.67 10 8, J m 2 K 4,
17 and T the absolute temperature.
Abnormal rising of the surface radiation temperature were
5 observed on the infrared thermal images in the climate satellites
several days before and after earthquakes. It was recognized that the
4
earthquakes are related to the changes of the stress eld of the earth's
5 crust, and then the change of the stress eld causes electromagnetic
radiation including the infrared radiation which has the thermal effect
3 (Qiang et al., 1990). The fact that infrared radiation energy varies with
the change of stress eld of the loaded rocks was further veried by
8
the subsequent studies: uniaxial and biaxial loading of sedimentary
rocks and igneous rocks (Cui et al., 1993; Zhi et al., 1996); the
structural stress elds of rock and gas burst disasters resulting in local
abnormal high temperature in mine (Wu and Wang, 1998); projectile
impact on rock (Shi et al., 2007); the roadway tunnel under plane
loading (He et al., 2009) and its excavation process (He et al., 2010a,b)
in the physical analogue models in a horizontal and vertical strata
consisting of alternating layers of sandstone, mudstone and coal.
The thermal effect caused temperature increment T is affected by
the combined factors, i.e. thermal-elastic effect, rock fracturing, rock
friction, environmental radiation and heat transfer. As the physical model
in our test is large in size, the heat transfer between the loading platen
and the analogous model is small and the same for the environmental
radiation, the loading process can be considered adiabatic, the heat
transfer and the environmental radiation can be neglected. The IR
temperature increment T, K, can, therefore, be described qualitatively
by the following equation (Wu and Wang, 1998):
T = T1 + T2 + T3 2
terms at the right-hand side of Eq. (2) can be explained in what
follows:
T1 is IR temperature increment due to the thermal-elastic effect, K,
positive or negative, which has a relation for a plane loading state
(Wu et al., 2006a,b):

1
T1 = T1 + 2 3

where T is the physical temperature of rock, (K); is a constant

Fig. 2. Roadway excavation scheme illustrated in the front surface of the excavation
zone on the physical analogue model.
correction factor related to rock surface radiant emittance, rock
thermoelastic fractor, (MPaKV 1); is a transfer factor between
the detected voltage signal and the IR temperature, (KV 1).
According to thermoelastic theory, the change of surface tem-
perature of any solid unit is linearly related with the change in the
sum of its three principal stresses (1 + 2), MPa, (Harwood and
Cummings, 1991). If (1 + 2) N 0. T1 will be positive and vice
versa.
T2 is the temperature decrement due to the expansion of the initial
ssures, joints and newly produced fractures, K, negative. The
production of new fracture needs to consume energy, and the
extension of ssures, joints and fractures also needs to consume
energy. Hence, T2 is always negative. The more the production and
extension, the more the negative effect of T2.
T3, K, is the temperature increment due to the friction heat, K,
positive. The friction heat is produced by the friction action between
 M. He / Engineering Geology 121 (2011) 165176
Fig. 3. Experimental test setup. (a) Photograph of the IR camera and the excavation area in view; (b) the diagram for the testing system.
joints, ssures and grains of deforming rock when subject to the
external load. The heat induced by the friction is always positive,
and monotonically increases with the increasing deformation and
newly produced fractures.
Identical or similar results on thermal effect were achieved by such
infrared thermography based studies as: thermographic inspection of
loaded composite material (Connolly and Copley, 1990; Steinberger et
al., 2006); loaded metals (Luong, 1995; Pastor et al., 2008); and
loaded rocks and rock-like materials (Brady and Rowell, 1986; Luong,
1990; Qiang et al., 1990; Cui et al., 1993; Zhi et al., 1996; Geng et al.,
1998; Wu and Wang, 1998; Wu et al., 2002, 2004, 2006a,b; Grinzato et
al., 2004; Shi et al., 2007). These well-established thermomechanical
coupling knowledge were utilized as references in the analyzing and
interpreting the heat images in this study.
Infrared thermography of TVS-8100MKIImodel was used in our test
for detection of the IR radiation eld on the excavation zone in the
physical model. The infrared camera operates at the wave length of 3.6
4.6 m, measuring temperature range of 40 C to +300 C, and
minimum detection temperature difference of 0.025 C. The IR thermog-
raphy visualized the temperature changes over the surface in view in
real-time with an on-line display resolution of 240 320 pixels, and
stored them as a series of thermography in the computer with an image
matrix of 120 160 pixels for off-line processing. During the test, the
image acquisition frequency was set as one frame every four seconds, and
imaging area is 400 267 mm, covering the roadway cross section and
1.0 MPa
250 mm
1600mm
0.3 MPa 0.3 MPa
200 mm
1 which actually represents: (a) IRT variations relative to the
Excavation zone
1600 mm
2 procedures can be found in (He et al., 2010a).
o
Fig. 4. Boundary stress condition applied on the physical analogue model during the
roadway excavation processes.
169
the nearby structures. The experimental setup is shown in Fig. 3. Fig. 3a is
the photograph of the experiment setup and b is the schematic. For 24 h
earlier prior to the testing, all the instruments were placed in the same
room with the PFESA model, so that the detected IR temperatures are the
temperature variation due to the excavation impact. The 400 400 mm
mark frame of red colored plastic thin membrane was attached with glue
to the model front face centrosymmetric to the roadway cross section,
indicating infrared imaging area, as seen in Figs. 1a and 3a. The infrared
camera and video camera were placed at the model's front face while the
excavating operations started from the rear face.
5. IR data processing
The thermography matrix data sets is the infrared radiation
temperature (IRT) distribution on the viewed surface, which is essentially
the rock responses as a result of the thermomechanical coupling nature of
the engineering materials. Let f(i,j) (i=1,2,,M;j=1,2,N) denotes the
IR matrix data at an instant time, where M and N are the pixel number
(M=120, N=160). We used the IR matrix data sets for characterizing the
structural and mechanical responses by the following image processing
procedures.
(1) Computing the statistical mean from the IR matrix data sets by
1 1 N M
bfij N = f i; j 1
M N i=1 j=1
where b fij N represents the averaged IRT distribution which is
actually the density of the infrared emittance at an instant.
Thus we got a time series of b fij N k, k = 0, 1, 2,, where k
corresponds to the sampling time. The rst frame of the IR
image was taken when the model was at the intact state, which
was used as a benchmark reference for calibration of the
temperature changes of the following IR therographies. We
used a normalized b fij N k as a measure in characterizing rock
behaviors over the roadway excavation. For the sake of
simplicity, the normalized b fij N k is also referred to as IRT,
model's intact state, and (b) the energy release level relative to
the maximum mean value of IRT. Detailed computation
(2) The IR image series, representing the temporal and spatial
evolution of the structural and mechanical rock behaviors, can be
reconstructed by the IR matrix data sets. But misunderstanding of
170 M. He / Engineering Geology 121 (2011) 165176

Fig. 5. Illustration of the Fourier analysis of the IR image; the image matrix was resampled along the horizontal (x-) and vertical (y-) axes passing through the center (80, 60) of the
image; (a) thermography (a) is with respect to E1; (b) the vertical spectrum obtained by 120 points DFT and (c) the horizontal spectrum obtained by 160 points DFT.

the IR image would be inevitable without the appropriate
processing of the data sets, because the obtained thermographies
were vague in contrast due primarily to the presence of the noises.
Contrast sharpening and feature extraction from the noise
contained IR thermographies are our expertise. In this study, we
used the so-called multiple denoising algorithm (He et al., 2010a)
to process the IR data sets and reconstruct the IR images. Very good
results were achieved by using the proposed algorithms, leading to
an enhanced understanding of the rock response. Detailed
introduction of the above mentioned algorithms was reported in
literatures (Gong et al., 2008; He et al., 2010a).
(3) As the excavation advances, the dynamic loading and unloading
conditions exist, which generate stress wave. The excavation
induced stress wave will travel in the rock layers and reect at the
interfaces of the rock boundaries. Stress wave propagation can be
described in the frequency domain by Fourier transformation of
the IR images. We resample the two 1-dimentional (1-D) data
sets, i.e. xi, i= 1, 2, , 160 and yj, j= 1, 2, , 160, along the vertical
and horizontal axes right across that center of the IR image and
performed 1-D Fourier transformation upon them, which were
referred to as the horizontal spectrum and vertical spectrum
respectively. Fig. 5 illustrates the diagram of the sampling and
performing Fourier transformation from the thermography
collected at the instant E1 in the phase 1 excavation. The 1-D
Fourier transformation was achieved by the DFT (Discrete Fourier
Transform) algorithm (Pinsdy, 2003):
M over the full-face excavation and uctuation during the staged
F u = DFT xi  = xi exp2iui = M
i=0
and
h i N
F v = DFT yj = xi exp2juj = N
j=0
where u (Hz) is the discrete horizontal frequency variable, and v
(Hz) is the discrete vertical variable (u=1,2,,M,v=1,2,,N); F
(u) and F(v) are the horizontal and vertical spectra respectively.
In our analysis of the thermographies, the amplitudes of the
horizontal and vertical spectra, i.e. | F(u)| and | F(v)| , were used to
characterize the directional propagation of the excavating
induced stress waves along the horizontal and vertical respec-
tively. In the spectral-frequency analysis, we asses propagating
stress waves according to the level of the amplitude, and the
bandwidth or band in which the high level amplitude
distributes. The detailed introduction of performing a spectral-
frequency analysis on the IR thermographies can be found in the
literatures (Gong et al., 2008; He et al., 2010a).
6. Results and discussions
6.1. IRT characterization of the excavation processes
IRT, the intensity of the IR radiation in essence, was used as a
measure in characterizing the rock behavior during the excavation
processes. Fig. 6a shows the time-marching scheme of IRT over the two
roadway excavation processes, and b illustrates the tunneling face
advancement of the excavation steps over the full-face (phase 1)
excavation, and c depicts the total of seven stages during the staged
(phase 2) excavation. From Fig. 6a we can see that the IRT curve has
two different evolution patterns, i.e. piecewise linear increase with
jump discontinuity points (referred to as multi-linearity thereafter)
2 excavation. We use E0 E6 and P0 P8, referred to as points of
interests (POI) standing for the excavation steps and stages in the two
excavation phases respectively, marked them on the IRT prole, as
seem in Fig. 6a. POIs of E0 E6 corresponds to the full-face excavation,
3
E0 is the intact state of the physical, E1 E5 standing for the excavation
advancement from step1 to step 5, and E6 representing the instant
when a small passage was about to cut through (destruction of the #1
RB for the rst time). During the full-face excavation, the accumulation
 M. He / Engineering Geology 121 (2011) 165176 171

Fig. 6. Evolution of the IRT with respect to the excavation advancement. (a) Prole of the measured IRT against the excavation advancement; (b) diagram for the full-face (phase 1)
excavation process at a sectional view and (c) diagram for the staged (phase 2) excavation process at a front view.

of the elastic energy caused the linear increase in IRT and the
excavation induced rock sliding induced the jump discontinuity points
like POI E2 and E4. This evolution pattern of IRT in this test is different
from that in the physical modeling of the roadway tunnel excavation in
the horizontal strata, as presented in literature (He et al., 2010a),
where the IRT has no jump discontinuity point during the full-face
excavation for the horizontal strata. The IRT time-marching scheme
depicted in Fig. 6a validates the fact that for the jointed rock masses
with steep dip angles, the excavation impact, although the tunnel face
advancing shallow, will induce faulting damage between the rock
layers.
It was noted that the IRT prole attains its global maximal point at
POI E2 during the very early excavations, and reaches its local maximal
points at POIs E4 and E6. It indicates that the steep-angled rock layers
are prone to slide along their faulted bedding planes and hard to store
the elastic energy inside caused by the excavation impact. The time-
marching scheme of the IRT prole is divided by POI E6 i.e. before E6
the IRT has a multi-linearity evolution pattern and after E6 the IRT was
change to uctuate. As a result, E6 is the critical point dividing the
time-marching scheme of the IRT prole in the two excavation
processes. Immediately after E6 (the critical point), a sharp drop was
seen on the IRT curve at POI P0, indicating the faulting-damage by
cutting through a small passage in the #1 RB. During the staged
excavation, removal of one RB at each of the excavation stage caused a
quasi-periodic loading and unloading impact on the strata resulting in
the consequential uctuation of the IRT prole. According to the time-
marching scheme of the IRT curve, the rock behavior during the full-
face excavation was in a multi-linearity manner, and during the
staged excavation in a plastic manner, from an energy dissipation
point of view within the IR radiation regime. At the point E6, the
energy accumulation reached the maximum level, when the operator
did his best to dill the #1 RB out of the model for the rst time. After
E6, the expansion of the roadway tunnel in space caused plastic rock
damage in the surrounding rocks, the tunneling operation needs
lesser energy than that of the full-face excavation. E6 is also the point
parting the stored elastic energy and dissipated plastic energy, which
corresponds to the partition of the averaged deformation, i.e. the
averaged elastic strain and plastic strain during the stress redistribu-
tion processes inside the physical analogue model.
6.2. Thermography characterization and spectral analysis
The enhanced understanding of excavation responses was achieved
by characterization of the rock behavior spatially using the denoised
thermographies and the directional stress wave propagation by the
spectral-frequency analysis of the thermography. Fig. 7 shows the
172 M. He / Engineering Geology 121 (2011) 165176
 M. He / Engineering Geology 121 (2011) 165176 173

Fig. 7. IR characterization of the full-face excavation process. (a) Diagrams for the full-face excavation process at a sectional view; (b) IR images after denoising; (c) the horizontal
spectrum |F(u, 0)|; (d) the vertical spectrum |F(0, v)|.

gure sets for the full-face excavation, and the gures in each row are in
accordance to each of POI E0 E6 respectively, the gures in the
columns from the left to the right are: (a) tunneling diagram of the
excavation status at a sectional view; (b) denoised thermographies; (c)
the horizontal spectrum |F(u, 0)|; and (d) the vertical spectrum |F(0, v)|.
Rock responses over the full-face excavation, depicted in Fig. 7, can
be understood as:
(1) Diagram E0 shows the intact state of the model. E1 and E2 show
the face advancement at 1/5 and 2/5 footages, there was no
sliding between the rock layers as the face went forward
shallow away from the back surface of the model. E3 E5
illustrate the face advancing deep into the model and rock layer
sliding occurred. E6 depicts the instant when the #1 RB was
destructed (the critical point, corresponding to the global
maximal value in the IRT curve).
(2) Thermography E0: the random-scattering IRT distribution indi-
cates the intact state of the model. Thermographies E1 E4: the
EDZ exhibits the strip-formed and stratum-paralleled IRT during
the early excavations when the face advanced shallow; but IRT
distribution looks random, indicating the damage mechanism
that the faulted beddings is dominant and damage propagates
with a relatively small scale. Thermographes: E5 E6: the EDZ are
still strip-formed and stratum-paralleled but the IRT distributes in
a localized conguration, representing the damage localization
caused by the face advanced deep. Especially at E6, the IRT
localized around the center of the excavation zone indicating the
destruction of #1 RB.
(3) From the directional spectra | F(u)| and | F(v)|, we can evaluate
the directional EDZ development at a specic instant. POI E0:
both the horizontal and vertical spectra were distributed with
narrow bands and low amplitude at the same level represent-
ing the intact state of the rock mass. POIs E1 E3: the
directional spectra have much higher amplitude in the vertical
direction and the bandwidth become wider gradually, indicat-
ing the fact that when the face advanced shallow the nearby
rock mass had undergone more damage along the vertical
direction. POIs E4 E6: the amplitude for the directional spectra
has almost the same amplitude denoting that the EDZ
developed in the horizontal and vertical directions at the
same scale.
For deepening our understanding in the EDZ with respect to the
inclined rock layers, it is worthwhile to make comparison between the
ndings about the excavation in the horizontal strata (He et al.,
2010a) and this study. Fig. 8 shows the gure sets for POIs E1 and E6 in

Fig. 8. IR characterization of the full-face excavation process for POIs E1 and E6 (the critical point) in the geologically horizontal strata (He et al., 2010a): (a) IR images after denoising;
(b) the horizontal spectrum |F(u, 0)|; (c) the vertical spectrum |F(0, v)|.
174 M. He / Engineering Geology 121 (2011) 165176

the full-face excavation in the horizontal strata. From Fig. 8 we see
that for POI E1, the EDZ is scattered and for E6, the EDZ is localized in a
plastic form around the destructed #1 RB. Whereas we can see from
Fig. 7 (45 inclined strata, as seen in Fig. 2) that for E1 the EDZ is in a
faulting bedding form and for E6, the EDZ is also localized but with
faulting damage along the same inclination with the strata. As for the
directional Fourier spectra, no difference could be observed at the
excavation status E1 for the horizontal strata, while the amplitude is
much higher in the vertical spectrum at E1 for the 45 inclined strata;
at the excavation status E6, the amplitude of the directional spectra
almost have the same level in the horizontal and vertical directions
both for the horizontal and the 45 inclined strata, but the spectra for

Fig. 9. IR characterization of the staged excavation process. (a) Diagrams for the staged excavation process at a front view; (b) denosied thermographies; (c) horizontal spectrum |F
(u, 0)|; (d) vertical spectrum |F(0, v)|.
 M. He / Engineering Geology 121 (2011) 165176
Fig. 9 (continued).
the horizontal strata have wider bandwidths, indicating the much
more stronger rock damage had taken place at this moment. In
comparison with excavation in the 45 and 90 inclined strata, the
damage pattern manifested in the IR image has some similarities, i.e.
faulting bedding in the full-face excavation and localized with faulting
damage in the staged excavation; while the Fourier spectra is
different. The detailed description of excavation in a physical analogue
model of vertical strata can be found in the publication by (He et al.,
2010b).
Fig. 9 shows the gure sets for the staged excavation, and the
gures in each row correspond to each POI from P0 P7. In Fig. 9, the
gures in the columns from the left to the right are: (a) the tunneling
diagram in the front face of the excavation zone; (b) the denoised
thermographies; (c) the horizontal spectrum |F(u, 0)|; and (d) the
vertical spectrum |F(0, v)|. Rock responses over the staged excavation,
depicted in Fig. 9, can be understood as:
(1) Tunneling diagrams: the front view of the diagram shows the
excavated space in every stage illustrates the dynamical effects
by removing each of the RBs. P0 shows the instant when a small
passage was perforated in the #1 RB. P1 P7 illustrate the
status in every excavation stage.
(2) Thermography P0: a small region of the IRT with deep blue-
colored elsewhere on the image indicates the signicant IRT
drop induced by the unloading impact as a result of cutting
through a small passage on the model for the rst time, also
shown in the IRT curve in Fig. 6a. From thermographies P1 P7
one can see the progressive development of EDZ around
tunneling face. These EDZ were featured with damage localiza-
175
tion in the nearby structures and faulted bedding damage along
the same inclination as the rock strata. The scale of the localized
damage was increased while the faulted bedding damage were
getting fainter in contrast, indicating the fact that during the
staged excavation, the faulting damage was getting lesser and
the plastic damage around the face getting larger in scale as the
roadway tunnel excavation was getting to be nished.
(3) The directional spectra|F(u, 0)|and |F(0, v)|: from the gure
sets in the last two columns in Fig. 9 we can see that from
POIs P1 P7, the horizontal and the vertical spectra have
almost the same bandwidths and amplitude, indicating that
fact that the excavation induced stress wave propagates at
the same level in the horizontal and vertical direction; the
bandwidths for the two directional spectra varied from
narrow to wider and nally to narrower with the excavation
approaching to an end, indicating the excavating damage
level undergone inside the surrounding rock masses over the
staged excavation.
The signicant difference for the EDZ over the staged excavation
between the horizontal strata and this study lies in the spatial
congurations. That is, for the horizontal strata, the EDZ was in the
plastic form and distributed around the tunneling face (for detail
seeing literature (He et al., 2010a)); for the 45 inclined rock strata
(see Figure 2), however, the faulted bedding damage was seen over
the whole staged excavation processes, which was similar to the
damage patterns of vertically inclined strata (He et al., 2010b). This
observation revealed the dependency of the EDZ upon the inclinations
for the jointed rockmasses.
176 M. He / Engineering Geology 121 (2011) 165176
7. Conclusions
IRT was used as a measure in characterizing the rock behavior
during the excavation processes. Over the full-face excavation, the IRT
curve has a piecewise linear increasing pattern between the jump
discontinuity points which was caused by the sliding effect at a
macroscopic scale; and during the staged excavation, the IRT curve
has a uctuating pattern due to the loading and unloading impacts by
removing the rock blocks. As a result, the IRT characterizes the full-
face excavation and the staged excavation as multi-linearity and
nonlinearity processes respectively, in an energy dissipation perspec-
tive within the IR radiation regime.
IR thermography series characterize the dynamical process of the
EDZ development by their spatial IRT distributions over the roadway
excavation processes. During the full-face excavation, the EDZ is in the
form disturbed belt parallel to the rock strata indicating faulted
bedding dominant damage mechanism. During the staged excavation,
the EDZ is in the form of localization around the tunneling face with
the faulted bedding damage seen over the staged excavation process.
The directional propagation of the excavation induced stress wave
was characterized by horizontal and vertical Fourier spectra obtained
by performing DFT on the data sets by resampling on the IR image
matrix along the horizontal and vertical symmetry axes respectively.
More stress waves with higher level amplitude was found in the early
excavations during the full-face excavation; and the stress wave
components were found almost getting to be uniform as the tunneling
face advancing deep and over the whole staged excavation process.
Acknowledgements
Financial support from the Special Funds for the Major State Basic
Research Project under Grant No. 2006CB202200 and the Innovative
Team Development Project of the state Educational Ministry of China
under Grant No. IRT0656 are gratefully acknowledged. Last but not
the least, I would like to express my sincerest gratitude to Giovanni B.
Grosta, Editor of this journal, Mia Schouten, Journal Manager, and the
anonymous reviewers for their valuable comments and modication
suggestions on the improvement of this article.
References
Barla, M., 2008. Numerical simulation of the swelling behaviour around tunnels based
on special triaxial tests. Tunn. Undergr. Space Technol. 23, 618628.
Brady, B.T., Rowell, G.A., 1986. Laboratory investigation for the electrodynamics of rock
fracture. Nature 321, 488492.
Cai, M., 2008. Inuence of stress path on tunnel excavation responsenumerical tool
selection and modeling strategy. Tunn. Undergr. Space Technol. 23, 618628.
Cai, M., Kaiser, P.K., Morioka, H., Minani, M., Maejima, T., Tasaka, Y., Kurose, H., 2007.
FLAC/PFC coupled numerical simulation of AE in large-scale underground
excavations. Int. J. Rock Mech. Min. Sci. 44, 550564.
Chambon, P., Corte, J.F., 1994. Shallow tunnels in cohesionless soil: stability of tunnel
face. J. Geotechnol. Eng. 120 (7), 11481165.
Connolly, M., Copley, D., 1990. Thermographic inspection of composite material. Mater.
Eval. 48 (12), 14611463.
Cui, C., Deng, M., Geng, N., 1993. Study on the features of spectrum radiation of rocks
under different load. Chin. Sci. Bull. 38 (6), 538541.
Geng, N.G., Yu, P., Deng, M.D., et al., 1998. The simulated experimental studies on cause
of thermal infrared precursor of earthquake. Earthquake 18, 8388.
Gong, W.L., Zhao, H.Y., An, L.Q., Mao, L.T., 2008. Temporal and spatial analysis of infrared
images from water jet in frequency domain based on DFT. J. Beijing Univ. Aeronaut.
Astronaut. 34 (6), 690694.
Grinzato, E., Marinetti, S., Bison, P.G., Concas, M., Fais, S., 2004. Comparison of ultrasonic
velocity and IR thermography for the characterization of stones. Infrared Phys.
Tech. 46, 6368.
Harwood, N., Cummings, W.M., 1991. Thermoelastic Stress Analysis. IOP Publishing
Ltd., Bristol.
He, M.C., 2006. Rock mechanics and hazard control in deep mining engineering in
China. Rock Mechanics in Underground Construction. Rock Mechanics in
Underground Construction, Proceedings. ISRM International Symposium 2006
4th Asian Rock Mechanics Symposium, pp. 1431.
He, M.C., Lu, X.J., Jing, H.H., 2002. Characters of surrounding rock mass in deep
engineering and its non-linear dynamic-mechanical design concept. Chin. J. Rock
Mech. Eng. 21 (8), 12151224.
He, M.C., Yong, X.J., Sun, X.M., 2006. Chinese coal mine soft rock-study on characteristics
of the clay minerals. Coal Industry Press, Beijing.
He, M.C., Gong, W.L., Li, D.J., Zhai, H.M., 2009. Physical modeling of failure process of the
excavation in horizontal strata based on IR thermography. Min. Sci. Tech. 19 (6),
689698.
He, M.C., Gong, W.L., Zhai, H.M., Zhang, H.P., 2010a. Physical modeling of deep ground
excavation in geologically horizontal strata based on infrared thermography. Tunn.
Undergr. Space Technol. 25, 366376.
He, M.C., Jia, X.N., Gong, W.L., Faramarzi, L., 2010b. Physical modeling of an
underground roadway excavation vertically stratied rock using infrared ther-
mography. Int. J. Rock Mech. Min. Sci. 10.1016/j.ijrmms.2010.06.020
He, M.C., Miao, J.L., Feng, J.L., 2010c. Rock burst process of limestone and its acoustic
emission characteristic under true-triaxial unloading conditions. Int. J. Rock Mech.
Min. Sci. 47, 286298.
Jia, P., Tang, C.A., 2008. Numerical study on failure mechanism of tunnel in jointed rock
mass. Tunn. Undergr. Space Technol. 23, 500507.
Kamata, G., Masimo, H., 2003. Centrifuge model test of tunnel face reinforcement by
bolting. Tunn. Undergr. Space Technol. 18 (2), 205.
Lee, Y.Z., Schubert, W., 2008. Determination of the length for tunnel excavation in weak
rock. Tunn. Undergr. Space Technol. 23, 221231.
Li, S.J., Yu, H., Liu, Y.X., Wu, F.J., 2008. Results from in situ monitoring of displacement,
bolt load, and disturbed zone of a power house cavern during excavation process.
Int. J. Rock Mech. Min. Sci. 45, 15191525.
Long, M.P., 1995. Infrared thermographic scanning of fatigue in metals. Nucl. Eng. Des.
158, 363376.
Luong, M.P., 1990. Infrared thermovision of damage processes in concrete and rock.
Eng. Fract. Mech. 35 (1/2/3), 291310.
Luong, M.P., 2007. Introducing infrared thermography in soil dynamics. Infrared Phys.
Technol. 49, 306311.
Majewska, Z., Mortimer, Z.M., 1998. Fractal description of acoustic emission produced
in systems: coal-gas and gas-water. J. Acoust. Emiss. 16, 14.
Meguid, M.A., Saada, O., Nunes, M.A., Mattar, J., 2008. Physical modeling of tunnels in
soft ground: a review. Tunn. Undergr. Space Technol. 23, 185198.
Park, S.H., Adachi, T., Kimura, M., Kishida, K., 1999. Trap door test using aluminum
blocks. Proceedings of the 29th Symposium of Rock Mechanics. J. S. C. E, pp.
106111.
Pastor, M.L., Balandraud, X., Grdiac, M., Robert, J.L., 2008. Applying infrared
thermography to study the heating of 2024-T3 aluminium specimens under
fatigue loading. Infrared Phys. Tech. 51, 505515.
Pinsdy, M.A., 2003. Introduction to Fourier analysis and Wavelets. Machine Press,
Beijing, China.
Qiang, Z., Xu, X., Ning, C., 1990. Abnormal infrared thermal of satellite: Forewarning of
earthquake. Chin. Sci. Bull. 35 (17), 13241327.
Read, R.S., 2004. 20 years of excavation response studies at AECL's Underground
Research Laboratory. Int. J. Rock Mech. Min. Sci. 41, 12511275.
Sharma, J.S., Bolton, M.D., Boyle, R.E., 2001. A new technique for simulation of tunnel
excavation in a centrifuge. Geotech. Test. J. 24 (4), 343349.
Shi, W.Z., Wu, Y.H., Wu, L.X., 2007. Quantitative analysis of the projectile impact on rock
using infrared thermography. Int. J. Impact Eng. 34, 9901002.
Shiotani, T., 2006. Evaluation of long-term stability for rock slope by means of acoustic
emission technique. NDT & E Int. 39, 217228.
Sitharam, T.G., Latha, G.M., 2002. Simulation of excavations in jointed rock masses using
a practical equivalent continuum approach. Int. J. Rock Mech. Min. Sci. 39, 517525.
Steinberger, R., Valadas, L.T.I., Ladsttter, E., Pinter, G., 2006. Infrared thermographic
techniques for non-destructive damage characterization of carbon ber reinforced
polymers during tensile fatigue testing. Int. J. Fatigue 28, 13401347.
Tanaka, T., Sakai, T., 1993. Progressive failure and scale effect of trapdoor problem with
granular materials. Soils Found. 33 (1), 1122.
Tang, Y.G., Kung, G.T.C., 2009. Application of nonlinear optimization technique to back
analyses of deep excavation. Compt. Geotech. 36, 276290.
Wu, B.R., Lee, C.J., 2003. Ground movement and collapse mechanisms induced by
tunneling in clayey soil. Int. J. Phys. Model. Geotech. 3 (4), 1327.
Wu, L.X., Wang, J.Z., 1998. Technical note: Infrared radiation features of coal and rocks
under loading. Int. J. Rock Mech. Min. Sci. 35 (7), 969976.
Wu, L.X., Liu, S.J., Wu, Y.H., Wu, H.P., 2002. Technical Note: changes in infrared radiation
with rock deformation. Int. J. Rock Mech. Min. Sci. 39, 825831.
Wu, L.X., Wu, Y.H., Liu, S.J., Li, G.H., Li, Y.Q., 2004. Technical Note: infrared radiation of
rock impacted at low velocity. Int. J. Rock Mech. Min. Sci. 41, 321327.
Wu, L.X., Liu, S.J., Wu, Y.H., Wang, C.Y., 2006a. Precursors for rock fracturing and failure-
Part I: IRR image abnormalities. Int. J. Rock Mech. Min. Sci. 43, 473482.
Wu, L.X., Liu, S.J., Wu, Y.H., Wang, C.Y., 2006b. Precursors for rock fracturing and failure-
Part II: IRR T-Curve abnormalities. Int. J. Rock Mech. Min. Sci. 43, 483493.
Zhi, Y., Cui, C., Zhang, J., 1996. Application of infrared imaging system to the basic
remote sensing experiment on rock mechanics. Remote Sens. Environ. Chin. 11 (3),
161167.