Professional Documents
Culture Documents
Laboratory Observation of Spalling Process Induced by Tangential Stress Concentration in Hard Rock Tunnel
Laboratory Observation of Spalling Process Induced by Tangential Stress Concentration in Hard Rock Tunnel
Abstract: A series of laboratory experiments were carried out to study the tangential stress concentration–induced spalling failure process of
rock in an arch-shaped tunnel model under plane-strain loading condition by considering different geological conditions, i.e., a tunnel with
and without a fracture nearby. Different fracture properties, i.e., open fracture and fracture with different infilling materials around the tunnel,
were designed to study their influence on the spalling behavior. Digital image correlation (DIC) and acoustic emission (AE) were combined to
track the spalling initiation and formation, which is characterized by the deformation field and energy release rate evolution. Spalling at two
sidewalls of the tunnel without a fracture under concentrated tangential stress, and formation of various thin slabs and detachment from the
free boundary due to the initiation and propagation of dilatant cracks were well reflected. Asymmetrical spalling failure was observed due to
the existence of the fracture around the tunnel. It was found that the open fracture makes the spalling more concentrated and leads to larger
failure extent and more intensity than the tunnel with a filled fracture. DOI: 10.1061/(ASCE)GM.1943-5622.0001620. © 2020 American
Society of Civil Engineers.
Author keywords: Rock spalling; Fracture; Digital image correlation; Acoustic emission; Deeply buried tunnel.
rocks. Structural model tests have been carried out on rectangular axial load was added to the specimen through a rigid loading platen
prismatic specimens with a central circular hole by many research- on the top of the specimen. The boundary condition of specimens in
ers, and the brittle failure process is simulated and reproduced the plane-strain frame is schematically illustrated in Fig. 1(b).
(Cheon et al. 2011; Haimson 2007; Hu et al. 2018). As mentioned,
discontinuities adjacent to tunnels may cause negligible effects on
the failure and damage of surrounding rocks. It was attempted to Specimen Preparation
establish relations between the induced damage zone and disconti- Red sandstone with good homogeneity, which comes from Sichuan
nuities through model tests (Huang et al. 2013; Jeon et al. 2004; province, was used in this study. It shows typical brittle failure
Song et al. 2018) and numerical analyses (Barton and Shen 2017; characteristics under uniaxial compression test. The uniaxial com-
Manouchehrian and Cai 2018; Sagong et al. 2011). These results pressive strength (UCS) of sandstone is about 95 MPa. The physi-
indicate that the stability of the surrounding rocks is easily affected cal and mechanical parameters of the sandstone are presented
by discontinuities by increasing the damage zones around the in Table 1, which shows that the sandstone belongs to hard rock.
tunnel and causing uniform stress distributions. Even though sub- Rectangular specimens with a length of 150 mm, a width of 30 mm,
stantial works on brittle failure characteristics of surrounding rocks and height of 150 mm were cut from massive intact rock and proc-
have been carried out, the focus is mainly on the failure around a essed carefully to meet the dimension and orthogonality require-
homogenous tunnel or tunnel with fractures of single contact stiff- ments. To investigate the influence of an adjacent fracture on the
ness. In fact, different geological conditions (e.g., with and without failure modes of surrounding rock around the arch-shaped tunnel,
fractures) have been proven to be important factors influencing the tunnel models with and without fracture were prepared and
cracking behaviors (Miao et al. 2018; Shen et al. 1995). Further- tested for comparison. Key dimensions of the tunnel models are
more, the whole spalling failure process of rocks, including crack presented in Fig. 2. Fig. 2(a) is the arch-shaped tunnel model
initiation, propagation, and coalescence, has been given little atten- without any adjacent fracture and the key dimensions are given.
tion in previous studies. Fig. 2(b) shows the dimensions for the tunnel model with a nearby
After excavation, the radial stress around the tunnel will be re- fracture. The fracture with a length of 60 mm, an aperture of 2 mm,
leased and the tangential stress will increase. The tangential stress and inclination of 45° was located 5 mm from the tunnel boundary.
concentration is one of the main factors to induce spalling failure of Then the cubic specimen was cut precisely according to predeter-
rock mass. Therefore, a series of laboratory experiments were de- mined geometry by high-pressure water-jet cutting machine.
signed to study the spalling process induced by tangential stress Tunnel models with a nearby open fracture and a filled fracture
concentration in an arch-shaped tunnel model under plane-strain were prepared to study the effect of fracture properties on the
loading condition by considering different geological conditions, failure of surrounding rocks. In this study, three types of fillers
i.e., a tunnel with and without a fracture nearby. During testing, with varied properties, i.e., cement, gypsum, and epoxy resin, were
digital image correlation (DIC) and acoustic emission (AE) were chosen, and their mechanical parameters are given in Table 1. As
used together to track the spalling failure process and cracking the common grouting materials, cement and gypsum were chosen
behaviors of surrounding rocks around the tunnel. The rock spall- for infilling, and the mechanical parameters for them are close.
ing failure extent and intensity as well as the mechanism caused by
Resin was also chosen because of its high strength, stiffness,
different geological conditions were analyzed.
and adhesion, which could fit seamlessly with the rock matrix.
Cement slurry and gypsum slurry were respectively prepared for
cement filler and gypsum filler. The water–gypsum ratio of the gyp-
Experiment Procedure sum filler and the water–cement ratio of the cement filler were both
equal to 1∶2. In addition, transparent epoxy resin was also prepared
Plane-Strain Loading Frame for infilling. To ensure the quality of infilling, some steps were
In this paper, the tunnel is simplified as a plane-strain model. There- performed. First, the bottom of the fracture was sealed to prevent
fore, a loading instrument was designed and operated for tunnel the slurry from leakage. Second, the filler was injected into the
instability analysis under plane-strain conditions (Fig. 1). It was fracture using a grouting device. After injecting for 10 minutes,
achieved by using a stiff box to force all deformations occurring the filler in the fracture was crushed and replenished repeatedly
around tunnels or openings, which is consistent with the stress con- to increase the compactness of the filler. After infilling for 24 h,
dition of deeply buried tunnels in the field. The design of this de- suitable water (25°C) was poured onto the filled fracture to ob-
vice was inspired by the surface instability apparatus (Kao et al. serve whether or not bubbles were generated. If bubbles occurred,
2015) and plane-strain loading frame (Huang et al. 2013). The load- indicating that nonnegligible voids existed in the filler, refilling
ing instrument included bottom platen, the front wall, the rear wall, was required to ensure the tightness and homogeneity of the infill-
left sidewall, right sidewall, and back wall, all made by annealed ing materials. Before testing, a random gray intensity distribution
steel plate (24 mm thick) and polished on the surface contacting the was made by spraying black paint and white paint onto the speci-
specimen [Fig. 1(a)]. Highly transparent plexiglass (24 mm thick) men surface for carrying out digital image correlation analysis.
Fig. 1. (a) Plane-strain loading frame for tunnel model tests; and (b) boundary conditions applied to the specimen.
Testing Process hand, pressing the specimen against the right end of the loading
device. The 10 locking bolts on the front of the loading device were
First, a transparent friction reducing agent was uniformly applied to
tightened, squeezing the plexiglass and specimen against the rear
six faces of the specimen to reduce the friction between the speci-
wall. A wrench was used to alternately tighten all the locking bolts
men and loading device. The excess lubricant on the specimen ob- at the left wall and front wall to limit the specimen deformation in
servation surface was wiped to avoid blurring the highly transparent the x-direction and y-direction on the boundary, respectively.
plexiglass and affecting observation. Then a specimen was put into Fig. 3 shows the rock mechanic testing machine, loading device,
the loading instrument. The left locking bolts were fastened by acoustic emission systems, and digital image acquisition system.
Fig. 2. Key dimensions of tunnel models: (a) arch-shaped tunnel without fracture; and (b) arch-shaped tunnel with a nearby fracture.
Fig. 3. Arrangement of the experimental system and monitoring systems: (a) loading system, AE monitoring system, and DIC acquisition system;
(b) optical observation region; and (c) locations of AE sensors.
with a resolution of 3,376 × 2,704 pixels, two white light sources, sidewalls of the tunnel. The final failure pattern can well reflect
and a computer with image acquisition software. Details about that real brittle failure occurred in deep underground engineering
the system can be found in previous studies (Pan et al. 2009a). The [Figs. 4(c and d)].
deformation and failure around the tunnel were recorded by the
CCD camera and image acquisition software through the observa-
tion window and plexiglass [Fig. 3(b)]. A sequence of high- Influence of Nearby Fault on the Spalling Behavior
resolution images was recorded using image acquisition software In the construction of underground openings, it is inevitable to
for determining displacement field and strain field at any given encounter various geological discontinuities, whose properties,
moment by image correlation algorithm. These images can also such as stiffness and contact state, may change the stress state of
be used to track the visible failure range and location of surround- the tunnel and in turn affect the spalling failure behavior. In this
ing rocks. The displacement and strain field can be calculated by study, four different fracture (or fault) properties, i.e., open fracture,
using GOM Correlate software version 2018. AE was also adapted cement-filled fracture, gypsum-filled fracture, and resin-filled frac-
to detect damage to the specimens. The technique has been proven ture, are considered to simulate the geological structures that may
to be powerful for tracking the rock damage and fracture by exist in the engineering.
detecting high-frequency elastic waves emitted from defects (Ishida Fig. 5 presents the failure process of rocks surrounding an
et al. 2017; Zheng and Tannant 2016; Mao et al. 2018; Xiao et al. arch-shaped tunnel with an open fracture nearby. As shown in
2019). The experimental setup included eight sensors, eight pream- Fig. 5(a), the axial stress–displacement curve is presented, and
plifiers, and a digital AE data acquisition system. Sensors were the AE energy release rate during the test process is also obtained.
all fixed on the back of specimens to record AE activities. Pream- The failure characteristics of the arch-shaped tunnel with an open
plification of AE signals was provided by preamplifiers with a fracture at typical stress levels are presented in Fig. 5(b). At Point
gain set to 40 dB, and their locations are presented in Fig. 3(c). A, no spalling or slabbing is observed in surrounding rocks around
The threshold level for signal recording was set to 45 dB to avoid the tunnel, and there is no obvious AE energy released from the
background noise. specimen. With the increase of stress, thin slivers detach from side-
walls, indicating the spalling initiation. At this moment, systematic
AE signals are detected. At Point B, a macro tensile crack subpar-
Results allel to the free boundary occurs at the right sidewall. The corre-
sponding AE energy release rate achieves a larger peak. As axial
load increases, at Point C, more tensile cracks subparallel to the
Characteristics of Rock Spalling Process in an
free face are found at the right sidewall, and various thin slabs tend
Arch-Shaped Tunnel
to detach from the right sidewall. Meanwhile, small rock pieces
Fig. 4 presents the spalling failure process of intact rocks surround- burst out from the free boundary beneath the fracture accompanied
ing the arch-shaped tunnel without a fracture nearby. In Fig. 4(a), by violent voice. The loud failure can also be demonstrated from
the axial stress–time curve with the AE energy release rate evolu- the AE energy release rate, where substantial energy is released at
tion is presented. AE is a transient elastic wave generated from this stress level. At Point D, an antiwing tensile crack initiates from
the sudden release of strain energy caused by microcracks or other the upper tip of the fracture. The failure depth at the right sidewall
irreversible changes in the material. From the AE energy release becomes much larger, and numerous macro tensile cracks form
rate, the failure intensity caused by spalling can be evaluated. and develop rapidly, characterized by an unstable cracking pro-
The axial stress–time curve shown in Fig. 4(a) is generally linear cess. After Point D, an antiwing shear crack initiates from the
except for the initial nonlinear segment caused by compression lower tip of the fracture. Some flakes and slivers are completely
deformation of the opening. The failure characteristics of the detached from each other and fall into the tunnel from the right
arch-shaped tunnel at typical points (or loading level) marked in sidewall. Due to the existence of the open fracture, collapse occurs
Fig. 4(a) are shown in Fig. 4(b). At Point A, no spalling is observed at the arch roof of the tunnel (around Point E).
in surrounding rocks around the tunnel, and the AE energy release In real situations, the open fracture is seldom found. However,
rate is negligible. When the load increases from Point A to Point B, a fracture or fault with weak infilling materials may be revealed
some AE events with small energy are detected, which is related to near underground openings after excavation (Duan et al. 2017).
the microcracks’ initiation in the specimen. From DIC observation, Unlike the tunnel without a fracture, spalling occurs mainly at the
at Point B, thin slabs’ spalling occurs near the dome at the right right sidewall under concentrated tangential stress, and various
sidewall, and a tensile fracture initiated from the bottom of the left thin slabs are formed and detached from each other. However, no
sidewall is also observed. Larger AE energy release rate begins to obvious damage and spalling is observed at the left sidewall. Fur-
occur systematically due to the beginning of macrocracks forma- thermore, compared with a tunnel without a fracture, the spalling
tion. As axial load increases, the spalling or slabbing behavior failure of a tunnel with an open fracture is more severe.
becomes more prominent due to the increase of tangential stress. Fig. 6 presents the failure process of rocks surrounding an arch-
At Point C, macro tensile cracks, initiated from the bottom of shaped tunnel with a cement-filled fracture nearby. As shown in
Fig. 4. Failure process of rocks surrounding the arch-shaped tunnel: (a) AE energy release rate together with the applied stress–displacement curve;
and (b) failure characteristics at typical points. Typical spalling occurring in underground engineering in China: (c) Jinping II Hydropower Station;
and (d) Baihetan underground powerhouse.
Fig. 5. Failure process of rocks surrounding the arch-shaped tunnel with an open fracture: (a) AE energy release rate together with the applied stress–
displacement curve; and (b) failure characteristics at typical points.
Fig. 6(a), the axial stress–displacement curve is presented, and the than that of the tunnel model with an open fracture shown in
AE energy release response during the test process is also shown. Fig. 5(a). The failure characteristics of the arch-shaped tunnel with
The variation of AE energy release rate with axial stress shows a a nearby cement-filled fracture at these typical points are shown in
rising trend in general. However, its magnitude is much smaller Fig. 6(b). No spalling or slabbing is observed in surrounding rocks
(a)
Thin slabs
Spalling
Thin slabs
V-shaped
notch
Detached
slabs
Detached
slabs
New cracks Tensile
initiation cracks
(b)
Fig. 6. Failure process of rocks surrounding the arch-shaped tunnel with a cement-filled fracture: (a) AE energy release rate together with the applied
stress–displacement curve; and (b) failure characteristic at typical points.
around the tunnel at Point A. With the increase of the loading, thin under the concentration of the tangential stress, and a thin slab at
slabs’ spalling begin to occur at the right sidewall, and there are still the free boundary is found, which indicates the beginning of the
no macro dilation cracks appearing at the left sidewall (Point B). At spalling failure at the left sidewall. With the increase of the axial
Point C, obvious spalling and slabbing occurs at the right sidewall load, the spalling failure is developing and advancing toward
Fig. 7. Failure modes of (a) the tunnel with a cement-filled fracture; (b) the tunnel with a gypsum-filled fracture; and (c) the tunnel with a resin-filled
fracture.
deeper areas. At Points D and E, the failure depth becomes greater Spalling Extent and Intensity
at the two sidewalls and a V-shaped notch forms at the right side-
From deformation by DIC and energy release rate by AE, the fail-
wall of the tunnel. During loading, no antiwing cracks initiate from ure extent (or scope) and intensity can be roughly estimated.
two fracture tips, which is different from the tunnel with an open As presented, the spalling or slabbing occurs at two sidewalls
fracture shown in Fig. 5. of a tunnel. The horizontal deformation can directly reflect the
Fig. 7 shows the failure mode of rocks surrounding the arch- failure behavior and extent. The failure zone in this study is con-
shaped tunnel with a filled fracture. In Fig. 7(a), some subparallel sidered as the macroscopic failure area formed by macro tensile
tensile cracks and incompletely detached thin slabs presented at cracks.
the right sidewall. Various thin rock slabs and flakes that detached The maximum depth of the failure zone is defined as the failure
from the free boundary fell into the floor of the tunnel. Unlike depth in this study. Horizontal sampling lines are set on the two
failure occurring in the arch-shaped tunnel with an open fracture, sidewalls crossing the section with maximum failure depth. The
obvious failure occurs beneath the cement-filled fracture at the left horizontal displacement on the sampling line is also obtained at
sidewall. different moments to present the deformation evolution of sur-
The failure process and characteristics for a tunnel model with a rounding rocks. The relative horizontal displacement ΔX is defined
gypsum-filled fracture are similar to that for a tunnel model with by removing the assumed homogeneous deformation of the interior
a cement-filled fracture [Fig. 7(b)]. The failure characteristics for (i.e., X ¼ 28 mm).
the tunnel with a resin-filled fracture are a bit different. As shown in Fig. 8 presents the deformation evolution of surrounding rocks
Fig. 7(c), the failure that occurs beneath the fracture is insignificant for the tunnel model without a fracture nearby. The location of
compared with that occurring in the tunnel with a cement-filled maximum failure depth is determined by moving the vertical line
fracture and a gypsum-filled fracture. The thin slabs are also ob- tangent to the macrofailure profile, and the horizontal sampling
served, and the failure depths at the two sidewalls are comparable. lines at the two sidewalls are shown in Fig. 8(a). The relative hori-
Moreover, the failure process and AE characteristics for the tunnel zontal displacement on sampling lines at different moments are
with a resin-filled fracture are similar to that for a tunnel without presented in Figs. 8(b and c). On the abscissa axis, X ¼ 0 mm
fracture, indicating that the resin-filled fracture has little influence refers to the free surface, and X ¼ −28 mm corresponds to the
on the failure of surrounding rocks. deepest surrounding rock in the observation range. The evolution
Fig. 8. Relative horizontal displacement evolution of rocks surrounding the arch-shaped tunnel: (a) the location of horizontal sampling lines;
(b) the left sidewall; and (c) the right sidewall.
Fig. 10. Relative horizontal displacement evolution of rocks surrounding the arch-shaped tunnel with a cement-filled fracture: (a) the location of the
horizontal cross sections; (b) the left sidewall; and (c) the right sidewall.
X ¼ −2.5 mm, characterized by the displacement jump. When the tunnel. As shown in Figs. 11(c and d), the failure depth and failure
applied stress reaches 65 MPa, the displacement jump extends to areas are nearly consistent at the two sidewalls.
the depth at X ¼ −4.5 mm. The failure depth reaches 8.5 mm at the
right sidewall of the arch-shaped tunnel with an open fracture as
the load increases to 68 MPa, which is about 83% of peak load. Discussion
After that, the increase of the load will lead to the collapse of
the specimen. The spalling process of the rock mass in deeply buried rock engi-
As depicted in Fig. 10, the horizontal displacement evolution at neering may involve crack initiation, propagation, and interaction,
two sampling lines are shown for the arch-shaped tunnel with a which were reproduced and simulated in this study. To understand
cement-filled fracture. The failure evolution is similar to that pre- the brittle failure characteristics of surrounding rocks in deeply
viously mentioned. As the load increases to 80 MPa (about 83% of buried tunnels induced by different geological conditions, an arch-
peak load), the failure depth is about 6.5 mm at the right sidewall shaped tunnel, tunnels with an open fracture, and fracture with
and 2.5 mm at the left sidewall, which is related to the existence of different infilling materials were tested in the laboratory under
the nearby cement-filled fracture. plane-strain loading conditions. Even though open fracture occurs
The failure area is calculated through obtaining the failure depth rarely in deep buried underground engineering, it can be regarded
at various sections covering the entire failure zone. Schematic il- as the extreme case for weak structures that the normal stiffness and
lustrations of the failure profile, depth, areas, and accumulated en- shear stiffness are very small. Due to the existence of open fracture,
ergy release for the four tunnel models at about 83% peak stress stress was concentrated on the right sidewall of the tunnel. As a
are shown in Fig. 11. In Fig. 11(a), for the arch-shaped tunnel with result, violent spalling at the right sidewall and then a large volume
an open fracture, due to the existence of open fracture around the of collapse at the top of the tunnel occurred. Its failure extent and
tunnel, violent spalling or slabbing occurs at the right side wall. It is intensity were greater than the other cases. The cement filler in the
found that the total energy release at 83% peak stress is much larger fracture can provide some resistance for fracture normal deforma-
than the other three cases. In Fig. 11(b), spalling or slabbing ap- tion and tangential deformation, so the cement-filled fracture can
pears at two sidewalls for the arch-shaped tunnel with a cement- somewhat coordinate deformation of rock nearby and transfer some
filled fracture. Even though the cement filler in the fracture can normal and tangential stress. The cement-filled fracture can also be
transfer a certain amount of stress and release the stress concentra- regarded as a structural plane with medium fracture stiffness. Its
tion at the right sidewall, the failure extent at the right sidewall is normal stiffness and shear stiffness depend on the strength ratio
much larger than the left sidewall. Because the mechanical proper- and stiffness ratio of cement filler and sandstone. Due to the exist-
ties of resin are closer to sandstone, the stress can be transferred by ence of cement filler in the fracture, spalling failure also developed
the filled fault well. As a result, the failure depth, area, and AE at the left sidewall. However, the spalling failure extent was smaller
energy release are close to the case without a fault around the than that at the right sidewall. With further increase of fracture
Conclusions
Fig. 12. Stress distribution in the y-direction for the four tunnel modes (σ ¼ 40 MPa): (a) the tunnel with an open fracture; (b) the tunnel with
a cement-filled fracture; (c) the tunnel with a resin-filled fracture; (d) the tunnel without fracture; and (e) the tangential stress along the sampling
lines for the four tunnel models.
For the tunnel model without fracture or with fracture filled by the AE energy release rate was much larger than that of the other
stronger materials (e.g., resin), the energy release rate was approx- two cases. The spalling failure developed in an unstable and violent
imately at a relatively fixed value after spalling initiation, indicating manner.
that the spalling process developed in a steady manner with the As mentioned, after excavation, the radial stress was released
increase of axial load. However, AE activities for the tunnel model and the tangential stress was concentrated. Although the loading
with an open fracture were much more active, and the magnitude of condition of the study cannot simulate the actual stress path in
Pan, P.-Z., X.-T. Feng, and J. A. Hudson. 2009b. “Study of failure and scale
Zhang, C., X. T. Feng, H. Zhou, S. Qiu, and W. Wu. 2012. “Case
effects in rocks under uniaxial compression using 3D cellular
histories of four extremely intense rockbursts in deep tunnels.” Rock
automata.” Int. J. Rock Mech. Min. Sci. 46 (4): 674–685. https://doi
Mech. Rock Eng. 45 (3): 275–288. https://doi.org/10.1007/s00603
.org/10.1016/j.ijrmms.2008.11.001.
-011-0218-6.
Read, R. S. 2004. “20 years of excavation response studies at AECL’s
Underground Research Laboratory.” Int. J. Rock Mech. Min. Sci. Zhang, C., X. T. Feng, H. Zhou, S. Qiu, and W. Wu. 2013. “Rockmass
41: 1251–1275. https://doi.org/10.1016/j.ijrmms.2004.09.012. damage development following two extremely intense rockbursts in
Sagong, M., D. Park, J. Yoo, and J. S. Lee. 2011. “Experimental and deep tunnels at Jinping II hydropower station, southwestern China.”
numerical analyses of an opening in a jointed rock mass under biaxial Bull. Eng. Geol. Environ. 72 (2): 237–247. https://doi.org/10.1007
compression.” Int. J. Rock Mech. Min. Sci. 48 (7): 1055–1067. https:// /s10064-013-0470-y.
doi.org/10.1016/j.ijrmms.2011.09.001. Zheng, W., and D. Tannant. 2016. “Frac sand crushing characteristics and
Shen, B., and N. Barton. 1997. “The disturbed zone around tunnels in morphology changes under high compressive stress and implications
jointed rock Masses.” Int. J. Rock Mech. Min. Sci. 34 (1): 117–125. for sand pack permeability.” Can. Geotech. J. 53 (9): 1412–1423.
https://doi.org/10.1016/S1365-1609(97)80037-8. https://doi.org/10.1139/cgj-2016-0045.
Shen, B., O. Stephansson, H. H. Einstein, and B. Ghahreman. 1995. Zhou, H., F. Meng, C. Zhang, D. Hu, F. Yang, and J. Lu. 2015. “Analysis of
“Coalescence of fractures under shear stresses in experiments.” J. Geo- rockburst mechanisms induced by structural planes in deep tunnels.”
phys. Res. Solid Earth 100 (B4): 5975–5990. https://doi.org/10.1029 Bull. Eng. Geol. Environ. 74 (4): 1435–1451. https://doi.org/10.1007
/95JB00040. /s10064-014-0696-3.