You are on page 1of 11

Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 191 (2017) 20 – 30

Symposium of the International Society for Rock Mechanics

Geomechanical Model Tests and Infrared Detection of Rock


Responses for Tunnels Excavated in Sedimentary Rocks
Weili Gonga,b*, Manchao Hea,b, Hong Yanb, Lifeng Lia,b,c, Xiaodong Xua,b
a
State Key Laboratory for Geomechanics & Deep Underground Engineering, Beijing 100083, China
b
School of Mechanics & Civil Engineering, China University of Mining & Technology Beijing, Beijing 100083, China
c
School of Architecture & Civil Engineering, Guizhou University of Engineering Science, Guizhou, Bijie 551700, China

Abstract

Rock mass behavior is controlled mainly by weak planes in sedimentary rocks. Geomechanical model tests were conducted for
simulating tunnel excavations in horizontal, the 45q and vertical inclined rock strata. Infrared thermography was employed for
detecting rock responses during the excavations. Infrared temperature (IRT) curve obtained by averaging the matrices of
the infrared sequence can be viewed as temporal observation of the overall energy release from the rock under excavation.
The IRT curve characterizes the horizontal and vertical strata as linear behavior and the 45q inclined strata as piecewise linear
behavior over the full-face excavation and the three strata as plastic-like behavior over the staged excavation, respectively.
The constitutive heterogeneity can be represented by the Weibull modus obtained by curve-fitting to the Weibull model using
the probability distribution of the IRT temporal observations. The vertical strata has the smallest Weibull modulus values
corresponding to the higher heterogeneity compared with the else two strata models. The structural response of the rock under
excavation was characterized by the IRT distribution of the infrared image. IRT distribution of the horizontal strata evolved from
scattering distribution to localized high-temperature zone around the face. In contrast, IRT distribution for the 45q and vertical
strata distributed as belt-like IRT parallel to the weak surface; indicating the frictional sliding damage mechanism. Most intense
friction was observed in the excavation in the 45q inclined rock strata.
©
© 2017
2017TheTheAuthors. Published
Authors. by Elsevier
Published Ltd. This
by Elsevier Ltd. is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of EUROCK 2017.
Peer-review under responsibility of the organizing committee of EUROCK 2017
Keywords: Geomechanical model test; Sedimentary rocks; Tunnel excavation; Infrared detection; Excavation damaged zone

* Corresponding author. Tel.: +8610-15910785289; fax: +8610-010-62339820.


E-mail address: gwl@cumtb.edu.cn

1877-7058 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of EUROCK 2017
doi:10.1016/j.proeng.2017.05.149
Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30 21

1. Introduction

Sedimentary rocks cover the majority of earth’s surface and are frequently encountered in underground mining.
In the sedimentary rocks, two main sources of discontinuities are beddings and joints. The beddings can be assumed
continuous over areas greater than that of any designed excavation and joints, however, are typically constrained
between beddings [1]. Both beddings and joints are surfaces of relatively low shear and negligible tensile strength.
Under the condition of stratified rock masses stretching across the roadway section, the engineering geological
behaviour during tunnel development and operation is mainly controlled by the characteristics of the stratification
planes [2]. Thus, existence of the discontinuities may exert a significant impact on the stability of the surrounding
rock masses.
Excavation damaged zone (EDZ) is closely related geological disasters such as rock bursts, bumps and water
inrush, etc. Better understanding of rock behaviour during the excavation, therefore, has been an important issue for
design of the underground opening and mitigation of the potential geological disasters. Excavation induced damage
has been investigated extensively with different methods including the field experiments [3], numerical study [4],
and laboratory experiments [5–7]. Geomechanical model tests with judicious choice of the model materials may
yield some important information on the mechanisms of the EDZ which are not available from numerical models or
field experiments [8].
One of the critical issues in geomechanical model tests is the use of the monitoring techniques which were
developed purposefully for acquiring the “real” information of the rock behaviour in varied regimes. Strain gauge is
widely used because it can measure the stressed rocks consistent with the theory of rock mechanics but is limited to
the elastic phase. Alternatively, the use of remote sensing techniques is getting more popular. Digital imaging has
been applied for determining the displacement and strain [8, 9]. The photogrammetry requires supplementary
lighting while three-dimensional laser scanning (Lidar) acts as its own source of “illumination”, and can be used in
in active tunnelling environment under dusty, damp, and dark conditions and collected very accurate, high resolution
3-dimensional images of its surroundings [10].
The advantages of non-contact optical vision lie in their ability to represent the structural change by realistic and
practical surface models or geometrical features such as cracks and fractures. However, it is hard to get a definite
relationship with respect to the stress redistribution in large-deformation problems [1]. Infrared thermography is
another remote sensing technique which produces thermal image by detecting electromagnetic waves within infrared
band [11]. Infrared image represents rock response based on the thermal-mechanical coupling effect and does not
require supplementary lighting as well. When processed with proper algorithms, thermal image will not only be able
to detect geometrical features such as crack propagation, but also the static and dynamic friction [12] which could
hardly be observed by the conventional optical visualization techniques. The outstanding feature of
the thermography is that the image matrix represents the infrared temperature field on the surface in view induced
by energy release of the straining rocks.
IR thermography has been widely used in detecting damage in the deformed materials such as composite [13],
carbon fiber reinforced polymers [14], metals [11, 15], concrete [16], and rocks [17–20]. In recent decades, infrared
thermography has been employed to detect the EDZ in the large-scale geomechanical model tests for simulation of
tunnel excavations at China University of Mining and Technology Beijing (CUMTB). Compared to detection of
small-scale laboratory specimens with infrared thermography, major difficulty for detecting the large-scale
geomechanical models is the small signal to noise ratio (SNR) of the raw image due to the environmental and
instrumental noises and lower spatial resolution defined by imaging area per pixel. Hence enhanced imaging
processing algorithms were developed to analyse the thermal images [12].
To date, the reported experimental investigations at CUMTB involved the excavation in the 0q [21], 45q [22], 60q
[23] and 90q [24] inclined stratified rocks [21–24]. However, these experimental investigations were just reported
separately and a comprehensive analysis is lacking. At same time, heterogeneity existed in the constitutive
behaviour of the differently inclined stratified rocks under excavation has not been discussed. The objective of this
research is to deepen our understanding on the mechanisms for EDZ from excavation in differently inclined
sedimentary rocks. It was achieved by carrying out an comparative study on the thermal-infrared precursors
including the energy release rate represented by the averaged infrared temperature (IRT), structural features
22 Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30

manifested in the infrared image and Weibull modulus obtained by fitting the thermal temperature time series from
the experimental results [21–22, 24].

2. Experimental

2.1. Geomechanical model

The field case simulated by these geomechanical model tests are the main haulageway excavated in Qishan
underground coal mine in Xuzhou coal mining district, located in Jiangshu province, eastern China. The roadeay
was at a depth of 1000 m below the ground surface. Main exposure rock types are sandy mud rock, mudstone and
sandy rock. The coal seam is inclined at different inclination angles from 0q–90q with respect to the horizontal. Thus
the 0q, 45q and 90q inclined strata model can be used as a miniature of the geological formations of the simulated
field case. Fig. 1 shows schematically the three geomechanical models including 0q inclined strata model (Fig. 1a),
the 45q inclined strata model (Fig. 1b) and 90q inclined strata model (Fig. 1c), respectively. Each of the models was
constructed with nine strata consisting of one sandstone stratum, four mudstone strata and four coal seam strata,
indicated by digits 1-9 respectively. Their thickness from No. 1–9 strata are: 440 mm. 140, 120, 250, 150, 60, 140,
60 and 240 mm, respectively.

Fig. 1. Schematic of large-scale geomechanical models; (a) 0q inclined stratified rocks; (b) the 45q inclined stratified rocks,
and (c) 90q inclined stratified rocks.

2.2. Dimensional analysis and model materials

The constructed models have a 1.6 m u 1.6 m plane dimension and 0.4 m thickness. The excavation zones were
located in the No. 4 stratum (coal seam) centered on the model with a 250 u 200 mm sectional dimension.
The model is able to simulate a real excavation problem with rock mass dimension of 19.2 u 19.2 m and tunnel face
dimension of 3 u 2.4 m, by employing the geometrical scaling factor ߙ௟ ൌ ͳʹ. The force scaling factor was
determined as ߙఙ ൌ ͺ and body force scaling factor ߙఊ ൌ ߙఙ Τߙ௟ ൌ ͺΤͳʹ ൌ ͲǤ͸͹. The following stress boundary
condition was adopted, i.e., vertical stress ߪଵ was at 1 MPa while lateral stress ߪଶ at 0.3 MPa, corresponding to
a lateral pressure coefficient ߣ ൌ ߪଵ Τߪଶ ൌ ͲǤ͵. The loading was applied by the testing machine YDM-C developed
by Manchao He. Mechanical parameters of the model rock materials was also presented in Table 1. Three real rock
properties were simulated by using gypsum with three water-gypsum ratios, among them 0.8:1 for sandstone, 1:1 for
mudstone and 1.2:1 for coal respectively. Detailed introduction could be found in [21–22, 24].

Table 1. Real and artificial rock properties.


Rock type Unit weight UCS Tensile strength Young’s modulus Internal friction angle Poisson’s ratio
(kN/m3) (MPa) (MPa) (GPa) δqε
Real Artificial Real Artificial Real Artificial Real Artificial Real Artificial Real Artificial
rock rock rock rock rock rock rock rock rock rock rock rock
Sandstone 27 15 64 8 5.83 0.72 26 3.22 34 32 0.15 0.13
Mudstone 26 11 44 5 5.59 0.69 21 2.62 36 33 0.13 0.12
Coal seam 14 8 26 3 0.90 0.11 5 0.61 40 33 0.36 0.32
Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30 23

2.3. Rock strata fabrication

Fig. 2 shows the principle for producing rock strata. The rock strata are constructed with a large number of small
prismatic plates, the so-called “elementary slab”. The elementary slab is made by model materials reported
previously and casted in a mold. All the elemental slabs were fabricated with a same surface dimension 400 u 400 m
and three thicknesses of 1, 2 and 3 cm, respectively [21]. A large number of the elementary slabs with the same
model-rock property were used to ensemble a rock layer of the same property and a certain number of the rock
layers with the same rock property were used to construct a rock stratum. As seen in the figure, the mudstone
stratum and coal seam stratum are assembled by placing the elementary slabs in layers with perfectly mating
interfaces. A weak surface is formed between the two different model rock strata. This weak surface is
homogeneous along its striking at macroscopic scale. At the same time, the strata also have many asperities
constituted by many small gaps (secondary joints) between the parallel-placed elemental slabs in the rock layer.
Three classes of the model-rock strata were constructed, i.e. sandstone, mudstone and coal seam which were t used
for construction of the geomechanical models.

Fig. 2. Model rock strata construction.

2.4. Infrared detection

Fig. 3 shows schematically the scenario where the operator performs the excavation with the hammer and chisel,
starting from the back side of the model and went through to the front face. At the same time, the infrared camera
detects the temperature rise at the front face. Infrared images were acquired by an infrared thermography TVS-
8100MKII which was cooled and operated in a passive mode (no extra heat sources used). It works at wave length
of 3.64.6 Pm with measuring temperature range of -40 to 300 qC; minimum detection temperature difference of
0.025 qC; a field of view of 13.6q u 18.2q/25 mm; spatial resolution of 2 mrad; on-line display resolution 240 u 320
pixels. The raw thermogram were stored in the computer as digital image of 120 u 160 pixels for off-line
processing. The IR camera was fixed to a photographic tripod and placed in the front side of the geological model at
a distance of 1333 mm so as to have an imaging area of 400 u 367 mm indicated by a 400 u 300 mm red-colored
frame of plastic tape glued to the model front face as shown in Fig. 3. Before the imaging, the emissivity was set to
0.92 for the simulated model rocks. The image acquisition frequency was set as one frame every four seconds. All
the instruments were placed in the same room with the geomechanical models 24 hours prior to the tests, so that
the detected infrared temperatures show the temperature variation due to the excavation impact.

Fig. 3. Infrared detection of the tunnel excavation; (a) schematic illustration of the excavation without support using drill and blast method
simulated by an operator using a hammer and a chisel and (b) photo of the testing site of the 90q inclined strata models.
24 Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30

2.5. Excavation plan


The tunneling work is to excavate a tunnel-like cubical space by 250 u 200 u 400 mm in the stratum #4
(coalseam). The excavation volume was divided into seven sub-spaces, termed rock block (RB), along the strike of
the #4 stratum. The RBs were numbered #1#7 corresponding to the roadway excavation sequence (see Fig. 3a).
Roadway development in the three geological models is schematically shown in Fig. 5. The excavation was
designed as two phases, i.e. phase 1 (full-face excavation): tunneling on the #1 RB and the adjacent rocks until
a small passage is cut through, and phase 2 (staged excavation): removing one RB at each tunneling stage. Four
panels in Fig. 4 show the excavation plan. Fig. 4a shows the full-face excavation at a sectional view for three models
(also see Fig. 3). Figs. 4b, c and d show the staged excavation for the 0q, 45q, and 90q strata models respectively.

Fig. 4. Excavation plan; (a) full-face excavation for the three models (sectional view); (b) staged excavation for 0q inclined strata (at front view);
(c) staged excavation for 60q inclined strata (front view) and (d) staged excavation for 90q inclined strata (front view).

For the full-face excavation, the excavated volume was referred to as “footage” and for the staged excavation,
the term “excavation stage” denotes the removing each of the RBs, and total of seven excavation stages were
performed. The capical letters E0E6 stand for the excavation steps in the full-face excavation and P0P7 stand for
the excavation stages in the staged excavation. E0 denotes the onset of the excavation, E1E5 denote the face
advancement from step1 to step 5, and E6 denotes the time instant when a small passage was about to cut through
(destruction of the #1 RB for the first time). P0 denotes the time instant at which a small passage on the #1 RB was
cut through. P0P7 correspond to the removal of the RBs during each of the excavation stage.

3. Data reduction

3.1. Excavation plan

When using the thermography in the passive mode, although no requirement for the physical calibration of
the temperature increment, however, finding a reference point mathematically is needed for characterization of
the detected temperature variation. The mathematical calibration can be realized by the image subtraction algorithm.
While the object under detection is subjected to the external loading, the interested features of the infrared sequence
are the temperature increment relative to the initial state. Taking the first frame of infrared sequence as a benchmark,
acquired when the object was at the initial state, subtraction of the first frame from the following images obtains
the temperature increment relative to the initial state of the object. It can be expressed mathematically that,
Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30 25

݂መ௞ ሺ‫ݔ‬ǡ ‫ݕ‬ሻ ൌ ݂௞ ሺ‫ݔ‬ǡ ‫ݕ‬ሻ െ ݂଴ ሺ‫ݔ‬ǡ ‫ݕ‬ሻ (1)

where,݂௞ ሺšǡ ‫ݕ‬ሻ represents the image matrix of kth frame in the sequence; ݂መ௞ (x,y) the incremental temperature field
(also image matrix) at the ݇௧௛ instant, and ݂଴ ሺšǡ ‫ݕ‬ሻ the first frame of the thermal sequence taken at the initial state of
the loading; the subscript k is an integer served as frame index; ‫ ݔ‬ൌ ͳǡ ʹǡ ǥ ǡ ‫ ܯ‬and ‫ ݕ‬ൌ ͳǡ ʹǡ ǥ ǡ ܰ are the pixel
coordinates, and ‫ ܯ‬ൌ ͳ͸Ͳ and ܰ ൌ ͳʹͲ are the maximum pixel number respectively for the image matrix.

3.2. Image processing

Structural changes of the straining rocks could be observed from the infrared sequence when they are processed
with robust and efficient image processing algorithms. The tasks for processing thermal image acquired in the large-
scale geomechanical model tests generally involve the removal of different types of noises and the enhancement of
the low-contrast image in order to extract the interested features such as edges, borders, and high and low
temperature regions. The image processing algorithms used in this research are summarized in the following.
x For removal of the environmental radiation noise, the image subtraction expressed in Eq. (1) should be employed.
The image subtraction can also be used as the temperature calibration procedure as introduced above;
x For eliminating the salt-and-pepper noise induced by the electronic current in the measurement instruments,
median filter should be used;
x For reduction the additive-periodical noise which may come from the rotating parts in the cooling system
embedded in the infrared camera, Gaussian high-pass filters (GHPF) in the frequency domain could be utilized;
x When detecting a large-scale object with infrared camera working in the passive mode, the raw thermal image
will have a small dynamic scope. As a result, the images should be enhanced in order to represent the rock
response clearly. In our research, the morphological enhancement filter, ‫ܨܫܯ‬, proved to be very effective which
was developed by Gong et al. (2015) [12].
Detailed discussion on the image process algorithms for the treatment of the low-contrast and noisy thermal
images can be found in the references [1, 12].

3.3. Energy release rate

Apart from the image analysis, the infrared sequence can be transferred into observations of energy release at
a specific instnat of time. In fact, matrix for an thermal image is the infrared temperature (IRT) filed created by
the thermal-mechanical coupling effect, representing the energy release from the stressed rocks. Mean value of
the matrix is accordingly the averaged IRT field, representing the overall energy release at the time instant.
Computing the statistical mean on the infrared sequence yields a time series which is observations of the energy
release.
Statistical mean value of the matrix of a thermal image, ൏ ‫ ܴܶܫ‬൐, can be written as,

ଵ ଵ
൏ ‫ ܴܶܫ‬൐ൌ ெ ே σெ ே
௫ୀଵ σ௬ୀଵ ݂ሺ‫ݔ‬ǡ ‫ݕ‬ሻ (2)

Compute statistical mean over the processed infrared sequence using Eq. (2), time series, ൏ ‫ ܴܶܫ‬൐௞ ,
݇ ൌ Ͳǡͳǡʹǡ ǥ, can be generated, where k corresponds to the sampling period of time, i.e. 4 second. As introduced in
section 3.1, the first frame, ݇ ൌ Ͳ, of the sequence was taken when the physical model was at initial state and
performing the image subtraction using Eq. (1) makes the time series ൏ ‫ ܴܶܫ‬൐௞ being the IRT variations relative to
the initial loading state. In the following, to make the decription simple, we use IRT stands for ൏ ‫ ܴܶܫ‬൐ in
the related context.
26 Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30

4. Comparison

4.1. Energy release precursors

Fig. 5 shows the energy release rate against time with respect to the face development for the three models.
The capical letters E0E6 and P0P7 correspond to the excavation plans in the full face excavation and staged
excavation respectively (see Fig. 4). It is seen from Fig. 5a (horizontal strata model) that IRT increased linearly with
the face development over the full face excavation; it attained the global maximum energy release by a temperature
increment of 31.25 qC at E6. Cutting through a small passage on #1 RB (state P0) caused an abrupt drop of
the energy release by 31.25 30.86 = 0.39 qC. After that the IRT evolves in the plastic manner fluctuating with
multiple small peaks corresponding to removing the RBs. The energy release level much lower than that in the full-
face excavation. It demonstrates that less energy was consumed in removing the RBs and expanding the tunnel
space.

Fig. 5. Energy release rate < IRT> against time with respect to the face development; (a) horizontal strata model; (b) the 45q inclined strata model
and, (c) vertical strata model.

IRT of the 45q inclined strata model (Fig. 5b) has a piecewise linear increase against face development during
the full face excavation. The global maximum energy release was found at E2 by 25.83 qC, much smaller than that
of the horizontal strata. At steps E2, E4 and E6, sharp drops of the IRT were found and the maximum value
corresponding to E2 equals to 25.83  25.13 = 0.7 qC, much larger than the IRT drop in the horizontal strata. These
sharp drops were caused by the relative sliding along the weak surfaces. It revealed the fact that the steep-inclined
rock strata are prone to slide and hard to store the elastic energy under loading. At the cutting through step E6, IRT
dropped by 25.49  25.21 = 0.28 qC, much smaller than that of the horizontal strata showing a drop of 0.39 qC,
indicating less energy being consumed over the full-face excavation in the 45q inclined strata than that in horizontal
strata as a result of the energy dissipation in the frictional sliding. Over the staged excavation, IRT oscillates with
high amplitudes. The local peaks for the stages P1–P6 increased as the excavation stage advances. The intense energy
release was due primarily to the excavation-induced static friction of the strata caused by removing the RBs.
The time-marching scheme of the IRT fully revealed the unstable nature for the steep-angled strata under
excavation.
IRT of the vertical strata model (Fig. 5c) increased monotonically against the face development, similar to
the IRT curve for the horizontal strata. The global maximum IRT was attained at E6 with 31.24 qC, higher than those
for the 0q and the 45q strata models. The IRT had a sharp drop at P0 by 31.42 30.99 = 0.43 qC, larger than that of
the horizontal strata with 0.39 qC. These differences of the IRT evolution pattern was attributed to the stronger static
friction between the vertically inclined layers under the excavation impact. During the staged excavation, IRT
increased sharply with multiple peaks with higher amplitude than those for the horizontal and the 45q inclined strata
models, demonstrating the fact that vertical strata is the most unstable geological structure during the staged
excavation.
Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30 27

4.2. Constitutive heterogeneity

Time series, ൏ ‫ ܴܶܫ‬൐௞ , ݇ ൌ Ͳǡͳǡʹǡ ǥ, can be viewed as a stochastic process sampled in the IRT ensemble space.
Rock behavior can also be characterized by means of p.d.d. (probability density distribution) or c.d.f. (cumulative
distribution function) of the ൏ ‫ ܴܶܫ‬൐ data sets. Among the probability distributions, Weibull model has been widely
used in rock mechanics for developing the stochastic constitutive equations used in numerical simulation of the rock
mechanics studies [25].
The p.d.d. for the Weibull model is given by,

݂ሺ‫ݔ‬ሻ ൌ ሺ݉Τߟ ሻሺ‫ ݔ‬Τߟሻ௠ିଵ ݁‫݌ݔ‬ሾെሺ‫ݔ‬Ȁߟሻ௠ ሿ (3)

where, K is the scale factor; m is the homogeneity index (also known as Weibull modulus). The parameter m
characterizes degree of the homogeneity for rock masses. An infinitely high m value corresponds to a homogeneous
structure with a uniform strength, whereas a heterogeneous structure with a broad distribution of local strength is
associated with a relatively small m value [26]. Integration of Eq. (3) yields c. d. f. of the Weibull distribution,

‫ܨ‬ሺ‫ݔ‬ሻ ൌ ͳ െ ݁‫݌ݔ‬ሾെሺ‫ݔ‬Ȁߟሻ௠ ሿ (4)

For evaluating the constitutive heterogeneity exhibited by the three geomechanical models under excavation, IRT
curve fitting was performed for obtaining the Weibull modulus from the thermal time series, by using the c.d.f.
Weibull model given by Eq. (4). It was done based on the assumption that the IRT time series data follow a Weibull
distribution. Fig. 6 shows the curve fitting results for the three strata models; the rectangular solid blue square stands
for the data sets ൏ ‫ ܴܶܫ‬൐௞ and the red line for the fitted curves. It is seen that all the IRT data follow the c. d. f.
Weibull model very well.

Fig. 6. Weibull model-based IRT curve fitting results for the three geomechanical models: (a) horizontal strata; (b) 45q inclined strata,
and (c) vertical strata; it was done based on the assumption that the IRT time series data follow a Weibull distribution.

The upper panel for Fig. 6a is the curve-fitting result for the full-face excavation, and the lower panel for Fig. 6a
is the fitting result for the staged excavation. Likewise, Fig. 6b and 6c are the curve-fitting relusts for tow phase
excavations in the 45q inclined strata and vertical strata models respectively. By observing the Weibull modulus
presented in Fig. 6, the constitutive heterogeneous precursors can be understood as:
x horizontal strata: having the largest Weibull modulus values, i.e., 2.9868 for full-face excavation and 2.4933 for
staged excavation, exhibited a relatively weak heterogeneity behavior;
28 Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30

x the 45q inclined strata: having a moderate Weibull modulus values, i.e. 1.5606 for the full-face excavaiton and
2.1073 for the staged excavation, exhibited a moderate heterogeneity;
x vertical strata: having the smallest Weibull modulus values, i.e. 1.3767 for full-face excavation and 1.5061 for
staged excavation, corresponding to the most heterogeneity as compared with the horizontal and the 45q strata
models.

4.3. Image precursors for full-face excavation

Comprehension and interpretation of an infrared image is based on such image features as colors, borders and
edges that segment different zones in the image, as well as the temperature distribution scale [12]. Pseudo-colors
represent the temperature levels; hot (positive) colors stand for high-level temperatures and cool (negative) colors
for the low-level temperatures. High-level temperature indicates high stress level due to friction, shear or stress
concentration, and low-level temperature indicates low stress level due to tensile cracking, stress release or
unloading. Borders or edges that separate hot and cool colored zones reflect different modes of the rock behavior
and scale of the high temperature zones corresponds to the scale of the EDZ. Image precursors for characterizing the
EDZ will be extracted by analyzing the infrared sequence processed using the algorithms presented in section 3.2.
Fig. 7 shows the infrared sequences corresponding to E0 – E6 for the full-face excavation. Fig. 7a is for
horizontal strata, Fig. 7b for the 45q inclined strata and Fig. 7c for vertical strata geomechanical models respectively.
By analyzing these infrared sequences, the following image features could be regarded as the image puecursors, i.e.
(a) at onset of the excavation corresponding to E0, images for the three strata exhibit a random-scattering IRT
distribution indicating the intact state of the model rocks; (b) heterogeneous IRT distribution was observed in
the subsequent excavation steps E2–E6, but the configurations of the EDZ are different:
x for horizontal strata model, IRT evolved from the scattered distribution to the localized high-temperature zone
(i.e. EDZ) distributed around the face, behaving like isotropic materials;
x for the 45q inclined and vertical strata models, belt-like IRT parallel to the weak surface was observed; indicating
the frictional sliding damage mechamisms;
x for the 45q strata, most intense static friction took place at the excavation step E5 and E6; for the vertical strata,
intense frictions were observed over the entire excavation steps E1–E6.

Fig. 7. Infrared sequences corresponding to E0 – E6 in the full-face excavations; (a) horizontal strata model; (b) the 45q inclined strata
and (c) vertical strata.

4.4. Image precursors for staged excavation

Fig. 12 shows the infrared sequences corresponding to P1–P7 for the staged excavation. Fig. 7a is for horizontal
strata, Fig. 7b for the 45q inclined strata and Fig. 7c for vertical strata geomechanical models respectively.
Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30 29

Following image features could be regarded as the image precursors, i.e. (a) for horizontal strata, EDZ was localized
around the face, closely associated with removing RBs, appearing on overlying rock and floor (P2, P3, P4), the side
walls (P4, P5, P6, P7); (b) for the 45q strata, EDZ was hybrid, including localized damage around the face plus
frictional sliding represented by the IRT belts over the entire excavation stages, and (c) for vertical strata, during
P2–P4, the frictions along the weak surfaces were appreciable by the distinct IRT belts; while over P5–P7, EDZ was
localized around face without frictional sliding being observed. Difference of the EDZ among the three strata may
be attributed to the extent of the excavation influence area. Apparently, the excavation in the 45q strata has
the largest influenced zone.

Fig. 8. Infrared sequences corresponding to P1–P7 for the staged excavation: (a) horizontal strata model; (b) the 45q inclined strata model,
and (c) vertical strata model.

5. Conclusions

IRT time series, obtained by averaging matrices of the infrared sequence, can be viewed as time observations of
the energy release characterizing overall rock response to the excavation. During the full-face excavation, IRT of
the horizontal and vertical strata increased linearly with the global maximum energy release rate higher than that of
the 45q strata. IRT of the 45q strata has a piecewise increase with several major energy release induced by
the frictional sliding along the weak surface. At the transfer point from the full-face excavation to staged excavation,
the vertical strata has larger IRT drop than those of the horizontal and 45q strata, indicating the fact that the frictional
sliding occurred in the vertical strata was more intense than that in the other two strata models. Over the staged
excavation, the time-marching scheme of the IRT for the three strata models had a plastic manner with multiple local
peaks. Among them, the vertical strata had higher energy release level.
Constitutive heterogeneity precursor could be represented by the Weibull modulus which can be obtained by
curve fitting of probability distribution of the time observation ൏ ‫ ܴܶܫ‬൐௞ to Weibull model. By comparing
the resultant Weibull moduli, constitutive heterogeneous precursors could be understood as horizontal strata model
has the largest Weibull modulus values, i.e., 2.9868 for full-face excavation and 2.4933 for staged excavation,
exhibited a weak heterogeneity behavior; the 45q inclined strata model has a moderate Weibull modulus values, i.e.
1.5606 for the full-face excavaiton and 2.1073 for the staged excavation, exhibited a moderate heterogeneity; and
vertical strata has the largest Weibull modulus values, i.e. 1.3767 for full-face excavation and 1.5061 for staged
excavation, exhibited the most heterogeneity behavior.
Image precursors for characterizing the EDZ were extracted from IRT distribution pattern in the processed
infrared sequence. Over the full-face excavation, IRT distribution for horizontal strata evolved from scattered
distribution to localized high-temperature zone (i.e. EDZ) distributed around the face, behaving like isotropic
materials; IRT distribution for the 45q and vertical strata models distributed as belt-like IRT parallel to the weak
surface; indicating the frictional sliding damage mechamism. The most intense friction was observed in the 45q
30 Weili Gong et al. / Procedia Engineering 191 (2017) 20 – 30

strata. During the staged excavation, for horizontal strata, EDZ was localized around the face, closely associated
with removing each of the rock blocks; for the 45q and vertical strata, EDZ was hybrid, including localized damage
around the face plus frictional sliding represented by the IRT belts. Likewise, in the 45q strata, the most intense
frictional EDZ was also observed.

Acknowledgement
Financial support from the Special Funds for the National Natural Science Foundation of China under Grant
No.51574248 are gratefully acknowledged. Heartfelt thanks should also be given to the anonymous reviewer for
the valuable comments and modification suggestions on the improvement of this article.

References
[1] W.L. Gong et al., Thermal image and spectral characterization of roadway failure process in geologically 45q inclined rocks, Tunnel. Underg.
Space Technol. 49 (2015) 156–173.
[2] P. Fortsakis, K. Nikas, V. Marinos, P. Marinos, Anisotropic behavior of stratified rock masses in tunneling. Eng. Geol., 141-142 (2012) 74–83.
[3] R.S. Read, 20 years of excavation response studies at AECL’s Underground Research Laboratory, Int. J. Rock Mech. Min. Sci. 41(8) (2004)
1251–1275.
[4] T.G. Sitharam, L.G. Madhavi, Simulation of excavations in jointed rock masses using a practical equivalent continuum approach, Int. J. Rock
Mech. Min. Sci. 39 (2002) 517–525.
[5] M.A. Meguid, O. Saada, M.A. Nunes, J. Mattar, Physical modeling of tunnels in soft ground: A review. Tunnel. Underg. Space Technol. 23
(2008) 185–198.
[6] S. Fekete, M. Diederichs, M. Lato, Geotechnical and operational applications for 3-dimentional laser scanning in drill and blast tunnels,
Tunnel. Underg. Space Technol. 25 (2010) 614–628.
[7] S.C. Li, H.L. Liu, L.P. Li, Q.Q. Zhang, K. Wang, Large scale three-dimensional seepage analysis model test and numerical simulation
research on undersea tunnel, App. Ocean Res. 5 (2016) 129–142.
[8] W.S. Zhu, Y. Li, S.C. Li, S.G. Wang, Q.B. Zhang, Quasi-three-dimensional physical model tests on a cavern complex under high in-situ
stresses, Int. J. Rock Mech. Min. Sci. 48 (2011)199–209.
[9] Y.J. Lee, R.H. Bassett, Application of a photogrammetric technique to a model tunnel, Tunnel. Underg. Space Technol. 21(1) (2006) 79–65.
[10] S. Fekete, M. Diederichs, M. Lato, Geotechnical and operational applications for 3-dimensional laser scanning in drill and blast tunnels.
Tunnel. Underg. Space Technol. 25 (2010) 614–628.
[11] M.P. Long, Infrared thermographic scanning of fatigue in metals, Nuclear Eng. Design 158 (1995) 363–376.
[12] W.L. Gong, Y.Y. Peng, X.M. Sun, M.C. He, S.Y. Zhao, H.Q. Chen, T. Xie, Enhancement of low-contrast thermograms for detecting
the stressed tunnel in horizontally stratified rocks, Int. J. Rock Mech. Min. Sci. 74 (2015) 69–80.
[13] M. Connolly, D. Copley, Thermographic inspection of composite material, Materials Evaluation 48(12) (1990) 1461–1463.
[14] R. Steinberger, T.I.V. Leitão, E. Ladstätter, G Pinter, Infrared thermographic techniques for non-destructive damage characterization of
carbon fiber reinforced polymers during tensile fatigue testing, Int. J. Fatig. 28 (2006) 1340–1347.
[15] M.L. Pastor, X. Balandraud, M. Grédiac, J.L. Robert, Applying infrared thermography to study the heating of 2024-T3 aluminum specimens
under fatigue loading, Infrared Physics & Technology, 51 (2008) 505–515.
[16] M.P. Luong, Infrared thermovision of damage processes in concrete and rock, Engineering Fracture Mechanics 35 (1990) 127–135.
[17] B.T. Brady, G.A. Rowell, Laboratory investigation for the electrodynamics of rock fracture, Nature. 321 (1986) 488–492.
[18] M.P. Luong, Introducing infrared thermography in soil dynamics, Infrared Phys. Technol. 49 (2007) 306–311.
[19] E. Grinzato, S. Marinetti, P.G. Bison, M. Concas, S. Fais, Comparison of ultrasonic velocity and IR thermography for the characterization of
stones, Infrared Phys. Technol. 46 (2004) 63–68.
[20] L.X. Wu, S.J. Liu, Y.H. Wu, C.Y. Wang, Precursors for rock fracturing and failure-Part I, IRR image abnormalities, Int J Rock Mech Min
Sci. 43 (2006) 473–482.
[21] M.C. He, W.L. Gong, H.M. Zhai, H.P. Zhang, Physical modeling of deep ground excavation in geologically horizontally strata based on
infrared thermography. Tunnel.Underg. Space Technol. 25 (2010) 366–376.
[22] M.C. He, X.N. Jia, W.L. Gong, L. Faramarzi, Physical modeling of an underground roadway excavation vertically stratified rock using
infrared thermography, Int. J. Rock Mech. Min. Sci. 47 (2010) 1212–1221.
[23] M.C. He, Physical modeling of an underground roadway excavation in geologically 45 inclined rock using infrared thermography, Eng. Geol.
121(3-4) (2011) 165–176.
[24] W.L. Gong, J. Wang, Y.X. Gong, P.Y. Guo, Thermography analysis of a roadway excavation experiment in 60th inclined stratified rocks, Int.
J. Rock Mech. Sci. 60 (2013) 134–147.
[25] S.C. Yuan, J.P. Harrison, A review of the state of the art in modeling progressive mechanical breakdown and associated fluid flow in intact
heterogeneous rocks, Int. J. Rock Mech. Min. Sci. 43 (2006) 1001-1022.
[26] C.A. Tang, P.K. Kaiser, 1998. Numerical simulation of cumulative damage and seismic energy release during brittle rock failure-Partĉ:
Fundamentals. Int. J. Rock Mech. Min. Sci. 35(2) (1998) 113–121.

You might also like