International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

Investigation of the failure mode of shale rocks in biaxial and triaxial

compression tests
Shrey Arora, Brijes Mishra n
Department of Mining Engineering, West Virginia University, Morgantown, WV, USA

art ic l e i nf o a b s t r a c t

Article history: High horizontal in-situ stress and weak sedimentary laminated roof rock can severely affect underground
Received 15 March 2015 coal mine roof stability. These sedimentary rocks possess planes of weakness along the horizontal di-
Received in revised form rection in a mine roof and delaminate easily when acted upon by high lateral stresses. Empirical studies
6 August 2015
have shown that the magnitude of these stresses can be as much as two to three times the local over-
Accepted 11 August 2015
burden stress. The resulting ground control problems (buckling, cutters, etc.) in such conditions are quite
challenging for mining engineers. This paper describes the failure observed in coal mine shale rocks
Keywords: under biaxial and triaxial stress conditions. To carry out the investigation, special platens were fabricated
Coal mines that are capable of applying biaxial compressive stress on a cubic rock specimen when the entire ar-
Cutter roof failure
rangement is used inside a uniaxial compressive loading device. This experimental set-up was further
Laminated shale
modified to apply a pseudo-triaxial compressive stress. Laminated shale specimens tested under biaxial
Biaxial platens
Biaxial tests stress condition showed tensile failure along the laminations at macrolevel. Black shale specimens
Triaxial tests showed extreme brittle failure. Limestone specimens tested under similar conditions failed violently and
it was concluded that failure observed in each rock type tested was unique and not an artifact of the
experimental design. The pseudo-triaxial conditions reduced the influence of laminations on the failure
mode. Laminated shale specimens were found to fail along multiple shear planes.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction The mechanism underlying typical failures in laminated roof rock

According to the Centers for Disease Control and Prevention
(CDC)3 there are 568 active underground coal mines in the United
States. A majority of these mines are located in the eastern part of
the country (Fig. 1(a)). Mines in these regions (like Northern Ap-
palachia, Southern Appalachia and the Illinois Basin) are known to
have been affected by severe ground control problem (Fig. 1(b)). It
is important to note that apart from the high topographic stress
(especially in the Appalachian region), other factors like relatively
high lateral stress and weak roof have been known to immensely
affect the overall stability of underground coal mine openings. A
recent study published by Bajpayee et al.2 analyzed the con-
tributory geologic factors behind roof falls based on well-docu-
mented narratives of mine operators involved with 1825 non-in-
jury roof falls between the years 1999 and 2008. The two major
factors—namely, the presence of discontinuities in the roof rock
and laminated rock surfaces—were determined to be the two
prominent contributors in about 60–65% of the roof falls reported.
n
Correspondence to: Department of Mining Engineering 359E Mineral Resources
Building West Virginia University, Morgantown, WV-26506, USA.
E-mail address: brijes.mishra@mail.wvu.edu (B. Mishra).
has not been researched extensively, and the investigation pre-
sented in this paper is an attempt to move in this direction.
An example of failure in laminated roof rock in underground coal
mines is the “cutter roof”. Hill et al.7 describe the failure as: “a failure
process that initially begins as a fracture plane in the roof rock parallel
to, and located at, the roof-rib intersection. The fracture propagates at
an angle usually steeper than 60° from the horizontal.” A cutter type
roof failure could eventually have disastrous consequences, resulting
in the collapse of the entire roof of an entry or cross-cut13. The entire
roof collapse, starting from the initial occurrence of the cutter failure,
happens in a step-wise manner (Fig. 2), and can occur over a time
period of a few days to a month; and sometimes the intial fracture
remains dormant for the rest of the mine panel life.14 The coal mines
in the Appalachian region in the Unites States have been known to be
severely affected by this ground control problem,1,6,14–16 etc. This
problem is aggravated in the presence of existing high horizontal in-
situ stresses (the maximum horizontal stress in these regions can be
two to three times the overburden stress).4
The use of numerical modeling has been successful in re-
plicating some of the key aspects of cutter roof failure like step-
wise failure, location of failure at the roof rib intersection, and
random spatial distribution of cutters within a mine panel. Con-
tinuum mechanics-based numerical modeling software has been

http://dx.doi.org/10.1016/j.ijrmms.2015.08.014
1365-1609/& 2015 Elsevier Ltd. All rights reserved.
110 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123
Fig. 1. (a) Roof fall rate in U.S. coalfields from 2005–2006.11Note: Percentage in bars indicate the percent of production due to longwall mining method, and (b) location of
active underground coal mining operations, 2012.10
widely used for this purpose,1,9,14,16 etc., with strain softening type
of rock constitute behavior incorporated into some of the recent
modeling attempts.6,16 However, in this project, discontinuum
mechanics-based numerical modeling was deliberately avoided
due to the fact that reliable input parameters (pertaining to dis-
continuity properties) were not readily available.6 The authors are
skeptical about this approach of excluding discontinuum modeling
as a pertinent question related to the influence of mechanical
properties of laminations (cohesion, friction angle, etc.) on the
results of numerical model is raised. Esterhuizen and Bajpayee5
point out, in this regard, that delamination is common in lami-
nated rock, where the rock separates into thin, weak beams which
are then more susceptible to buckling under high lateral stress.
Hence, to develop and analyze numerical models simulating fail-
ure in laminated rock, it is imperative that we fully understand the
failure mechanism in these rocks under varying stress conditions.
The research presented in this article is based on this rationale and
explores the response of laminated shale specimens from im-
mediate roof rock (rock samples collected from the immediate
stratum in the roof of a mine opening) under biaxial and triaxial
stress conditions in a laboratory set-up.
2. Approach
As discussed earlier, the magnitude of in-situ horizontal stress
Fig. 2. Stages of “cutter-roof” failure development (a) initial stage of a cutter, (b) small cutter type roof fall at a corner, and (c) roof profile after a massive fall initiated by
cutters. Photo credits: Murali Gadde and SS Peng.6
in underground coal mines in the eastern United States could be as
high as two to three times the overburden stress, and this fact has
been identified as a key factor responsible for failure in laminated
rock. An analysis of in-situ stress measurements of this region by
Mark10 indicates the presence of a predominant E-NE horizontal
stress orientation, which is also biaxial in nature. In an attempt to
simulate these in-situ stress conditions at laboratory scale, special
platens were designed and fabricated, and these are capable of
applying equal biaxial stress on a two inch cubic specimen. Ad-
ditionally, a pseudo-triaxial stress condition was also simulated
using biaxial platens and industrial c-clamp arrangement. Ideally,
a true-triaxial equipment would have simulated the in-situ con-
ditions in a more realistic fashion by allowing control over in-
dependent triaxial stresses acting on the specimen. However, such
equipment was beyond the financial scope of this research project.
3. Experimental design
3.1. Specimen preparation
Samples of irregular shape and size were collected from un-
derground coal mines in West Virginia and Pennsylvania. These
samples, consisting of black and laminated shale rocks, were re-
trieved from mines affected with weak laminated roof (Fig. 3
(a) and (b)). Laminated roof rocks had laminae less than 1 in. The
 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123 111

Fig. 3. Rock chunks (a, b) and cylindrical specimens (c) from which two-inch cubic specimens (d) were prepared.15

black shale were massive shale with presence of laminations.
Three-inch cylindrical specimens (laminated shale and limestone)
provided by a mine in Illinois basin were also used (Fig. 3(c)).
These rocks were then cut into two-inch cubic specimens (Fig. 3
(d)) using a machine-operated rock saw with a circular cutting
blade, and the surfaces were ground using a combination of a
handheld grinding machine and sandpaper. Water was not used in
any of the specimen preparation steps described above for shale
rock, in an attempt to prevent flaking and weakening of the spe-
cimen. Limestone specimens were also used in order to investigate
whether the experimental results obtained for other rock types
were not just an artifact of the experimental design. These speci-
mens were prepared from the rocks cored at the Illinois mining
site and represent massive type with no laminations. The average
uniaxial compressive strength was approximately 15,000 psi.16
Each specimen was referenced with a letter of the English al-
phabet based on the coal mine seam from which it came (“P” for
Pittsburgh seam, “I” for Illinois no. 6 seam, and “G” for specimens
from No. 2 Gas seam) and a number. As for example, (G6) meant
the 6th specimen from No. 2 Gas seam. Specimens P1–P4 (black
shale), I1–I4 (laminated shale), G1–G7 (laminated shale), and I5–I8
(limestone) were tested under biaxial stress condition. And, spe-
cimens P5–P8 (laminated shale) were tested under pseudo-triaxial
stress condition. (Note: Only 23 specimens could be prepared for
this experimental work. This was largely due to the restrictions
imposed by the mines, logistical issues, and difficulty in preparing
laminated shale specimens. It was recognized that such a low
number of specimens per specific test condition might hinder the
final analysis leading up to making a generalization of failure
mode of each rock type. Hence, each test was carefully monitored
and every minute detail was documented.)
3.2. Test platen design and dimensions
The design of biaxial platens is based on a direct shear device
(Fig. 4(b)) developed by I 1975.8 This device splits a vertical load
(P) into two components: (N) and (T), acting normal and tangential
to a particular plane in a cubic specimen as shown in Fig. 4(a). The

Fig. 4. (a) Free body diagram of the device indicating the forces acting on the specimen (b) the shear device with a rock specimen.8
112 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123
Fig. 5. (a) Fiberboard material was used for uniform distribution of the load on an
imperfect specimen and (b) CAD drawing of the biaxial platen showing the exact
dimensions of the platens.15
combination of ball bearing support on which the bottom platen
sits (allows it move in a horizontal direction during loading) along
with the shortened arm on of its side, allows shearing of the cubic
specimen.
These platens were modified to apply a biaxial compressive
stress on a 2-in. specimen by elongating the small arm (Fig. 5(a))
and making the bottom platen fixed. Made out of mild hardened
steel, the base of the platens is 4.95 in. in length and 2.2 in. in
thickness (Fig. 5(b)). Two of the arms of the platens (“g” and “h”)
are 2 in. in length and completely cover two faces of the specimen.
The remaining arms, “e” and “f”, are slightly shorter in length
(1.8 in.) and prevent contact between the two platens during
loading. The inner surfaces of the platen were ground smooth to
prevent any stress concentrations arising from undulations. In
addition, corrugated fiberboard material was used to ensure uni-
form loading of an imperfect specimen.
To analyze how these platens simulate in-situ biaxial stress
conditions, consider the underground coal mine entry condition
with laminated roof as shown in Fig. 6(a). The roof of the entry is
acted upon by principal horizontal stresses sh1 and sh2 and over-
burden stress V. The platens replicate the case when sh1 ¼ sh2 by
distributing the applied vertical stresses P and R in two equal and
Fig. 6. (a) Stress acting on the roof of an entry (b) stress distribution on specimen
inside UCS testing machine.15
mutually perpendicular stresses N acting normal to the lamina-
tions. The vertical stresses P and R are applied by the bottom
moving and top fixed platens of the uniaxial compression strength
(UCS) testing machine (Fig. 5(a)). The overburden stress “V” is
replicated by confining the exposed surface of the specimen (Fig. 5
(a)) using industrial clamps and this set-up will be described in the
subsequent section.
3.3. Test procedure
The compression-testing machine used for the experimental
procedure described in this section can be programmed to control
the movement of the lower movable platen in two ways, namely,
force and displacement rate control. This allows a user-defined
value to be set for force or displacement rate to be applied by the
machine to the specimen being tested. For the current tests, dis-
placement control mode with a lower strain rate was used
(0.0001 in./s). This assists in minimizing sudden failure of a spe-
cimen and in observing its complete behavior including post-
failure characteristics. These tests were run until a rock specimen
entered its post-failure region and to prevent any contact between
the two biaxial platens.
3.3.1. Post-test state
Post-test analysis of the specimen consisted of visual observa-
tion through photographic images (to get a general idea of failure
mode and mechanism) and analysis of vertical load vs. vertical
displacement relationship through the graphs generated by the
load cell sensors and LVDT's of the compression test rig. For fur-
ther investigation into a typical lamination failure under biaxial
stress condition (discussed in Section 4), two faces of seven la-
minated shale specimens had special markings (Fig. 7(c)) on them
to assist in determining the failure mode (shear or tensile). One
face had two concentric squares (sides with 1.0 and 1.5 in. in
length, respectively) and common diagonals drawn on it using a
white permanent marker (Fig. 7(a)). Any distortion in shape of
these squares, post-test, should be suggestive of shear failure
mode at a macrolevel. In addition, one-half of another face of the
same specimen was sprayed with white enamel paint (Fig. 7(b)) to
highlight the newly developed fracture after the test and thereby
indicating tensile splitting.
3.3.2. Visual recording of the test
Three biaxial compressive tests on laminated shale specimens
with markings were recorded on a digital camera that closely
monitored the lamination failure process. The camera was or-
iented to capture the failure occurring within the lamination and
at the exposed surface of the specimen (Fig. 14). Analysis of spe-
cific frames of these videos, which recorded significant movement
and failure of laminations, is presented in Section 4.
3.3.3. Triaxial test
As discussed earlier the biaxial compression test set-up was
modified to apply a pseudo-triaxial stress condition. The exposed
surfaces of the cubic specimen inside the platens were confined by
pressing square metallic plates (2.00 in.  2.00 in.  0.25 in.) on
them with the help of c-clamps (Fig. 8(b)). The c-clamps are made
out of ultra-force steel and has a maximum load capacity of 12,500
pounds.
3.3.4. Data analysis
As part of the postprocessing of the force vs. displacement data
retrieved from the data logger of the compression test rig (details
of the device can be found in Ref. 15), approximate stress acting on
a face of the specimen in contact with the biaxial platen at peak
load was also calculated. This was done to have a quantitative
 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123
Fig. 7. Markings on specimen to determine (a) shearing of lamination, (b) tensile fracturing under biaxial loading conditions,15 and (c) position of failure mode indicators on
the specimen with respect to the biaxial platens.15
assessment of the strength of the different rock types tested under
biaxial and triaxial stress conditions. If P is the force applied by the
lower movable platen of the compression test rig, N is the force
component acting normal to the face of the specimen (Fig. 6(b)),
and A is the approximate area of the face of the cubic specimen in
contact with the biaxial platens (assumed to be 4 in.2 for the ease
of calculation. Actual area depends on the area of contact between
specimen and platens), then:
N = P cos 45°
Approximate peak stress = N /A
3.4. Test limitations
This section lists the limitations of the biaxial and triaxial
compressive tests performed for this research, along with some of
the assumptions made. (1) The specimens were not instrumented
to obtain strain measurements. (2) Friction between biaxial pla-
tens and platens of the compression test rig, and between biaxial
platens and specimens, was ignored. (c) The c-clamps were
113
tightened to exert confinement, which ensured proper contact
between the specimen and the metal plates for the triaxial test.
However, the value of this confining stress could not be recorded
during the test.
4. Results and discussion
(1) 4.1. Effect of biaxial stress on non-laminated shale
Four non-laminated shale specimens (or black shale with no
(2)
distinct laminations visually observed) were tested first under
biaxial stress condition. The two tests (Fig. 9(e) and (f)) were
manually terminated before the specimen could reach its peak
load to avoid the biaxial platens meeting each other. The common
observation was failure of one or both exposed surfaces with a
large chunk of material failing along this zone. The load bearing
ability of the specimen did not diminish with the failure of the
exposed surfaces. Specimen P3 (Fig. 9(c)) showed sets of localized
failure close to and moving inward from the unconfined zone. The
rest of the specimens showed superficial cracks but no major
114 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123
Fig. 8. (a) Schematic view of experimental set-up with confinement, and (b) test
set-up with biaxial platens and the c-clamp inside the test frame (inset shows the
specification of c-clamp).15
fracture developed within the specimen.
The load vs. displacement graph of all the specimens showed a
non-uniform region from 0 to 0.03 in. axial displacement. This was
caused by the squeezing of the fiberboard material between the
specimen and also adjustment of the platen. Specimen P3 reached
a peak load of 55,000 lbs., beyond which it started exhibiting re-
sidual strength (Fig. 9(g)). Specimen P4 attained a peak load of
35,000 lbs. with the graph (Fig. 9(h)) indicating a strain softening-
type behavior in the post-peak load region.
It was concluded that some of the specimens were extremely
brittle in nature as indicated by the crushing and ejection of pieced
of rock from the exposed surface. It was also observed that failure
was progressive in nature as the load transferred from the un-
confined to the confined zone. It is, however, difficult to generalize
this failure mechanism in black shale cubic specimens based on
these observations, as the number of specimens tested was small.
4.2. Effect of biaxial stress on laminated shale
Eleven laminated shale specimens from mines in Illinois and
West Virginia were tested. The first set of four specimens from the
Illinois no. 6 seam failed in a peculiar way as compared to the
black shale specimens. The laminated layers in the first specimen
failed in a domino effect, with the unconfined zones failing first
(Fig. 10(a)) followed by the apparent separation along the lami-
nations. This effect was also observed in the load vs. displacement
graph (Fig. 10(b)), where multiple peaks were observed near the
maximum load. It was indicative of lamination failure and se-
paration. The entire sequence of failure is discussed in later
paragraphs of this section (Fig. 13). The second specimen had more
pronounced tensile cracks at the center of the specimen (Fig. 10
(c)), while at the unconfined ends large separations were ob-
served. The third specimen failed at a higher load (Fig. 10(f)) as
compared to other specimens with the failure occurring along
thicker layers of laminations (Fig. 10(e)). The fourth specimen
again showed multiple lamination failure with the maximum load
of 9000 lbs.
When the post-test state of laminated shale specimens and the
observations at a cutter location in a mine entry (Fig. 2) are jux-
taposed, it can be easily pointed out that the lamination separation
is the common denominator in both the cases. The condition of
laminated roof rock as shown in the encircled zone in Fig. 2, de-
teriorates further and caves into the entry if left unsupported. This
creates a cavity in the roof known as a cutter.
Seven more laminated shale specimens were further tested
under similar loading condition to determine the failure mode.
The failure mode indicators (as described in Section 3.3.1) strongly
suggested that specimens underwent splitting parallel to bedding
under biaxial compressive loading (Fig. 11) similar to what is
normally observed in Brazilian tests. This was highlighted by the
fresh cracks developed on the sides sprayed with white enamel
paint. A possible explanation is that the laminated layers are
poorly connected with each other and therefore form the weakest
link. When the load is applied parallel to the lamination, the
specimen will expand, which will create tensile stresses between
the laminated layers, resulting in the splitting of the specimen of
the specimen into thin slabs parallel to the bedding. Also, there
was almost negligible (Fig. 12) distortion of shapes of concentric
squares, indicating the absence of shearing of the laminated layers.
However, it is possible that at the micromechanical level, shear
failure might develop in the specimen, which was not observed in
the post-failed state.
For understanding the movement of the unconfined zone, three
specimens from this group of tests were video recorded using a
digital camera (Section 3.3.2). From frames 1 to 12 in Fig. 13, the
boxed regions indicate the area in which the crack developed and
movement of the outer layers. From frames 9 to 12, crack closure
on the edge of the specimen and bending of the outer layer was
observed. Before failure occurred, the outer rock layer showed
significant deformation. Similar behavior was recorded in addi-
tional tests, verifying that the lamination movement and bending
occurs at the same. From these tests, we also observed that when
the load is transferred from the outer edge to the inner zone of the
specimen, the laminated layers also fail and small microlamina-
tions are developed in the specimen. These inner layers then fall
off when the outer layer is completely removed.
4.3. Effect of biaxial stress on limestone
The following tests on limestone specimens were performed to
verify whether the distinct failure patterns observed in laminated
shale specimens loaded under biaxial stress conditions were not
just an artifact of the experimental design. The tests were per-
formed on specimens prepared from the three-inch cylindrical
cores. The mode of failure was different from the shale specimens.
These specimens failed violently, ejecting the rock chunks (Fig. 14
(a)–(d)). In all the specimens, the core appeared to remain intact
with no visible signs of fractures or splitting. Although the speci-
mens failed initially near the unconfined zone due to lateral ex-
pansion, the core of the specimens remained intact and failed
when their strength was exceeded. This behavior was evident
from the plots of axial force against displacement, which showed
strain softening in the post-failed stage. Consistent strain softening
behavior (Fig. 14(e)–(g)) was observed for all specimens except for
the last, where the loading had to be stopped to prevent the two
platens of the biaxial device from contacting each other and
causing damage to them.
 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123 115

Fig. 9. Post-test pictures and force vs. displacement diagram of specimen P1 (a, e), P2 (b, f), P3 (c, g) and P4 (d, h).15
116 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123

Fig. 10. Post-test pictures and force vs. displacement diagram of specimen P1 (a, b), P2 (c, d), P3 (e, f) and P4 (g, h).15
 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123 117

Fig. 11. Post-test condition of tensile splitting indicator (cracks in the white marked area represents the splitting or tensile failure developed during the test).15
118 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123

Fig. 12. Post-test condition of shear movement indicator.15

 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123
Fig. 13. Separation of the laminated layer and crack development in the laminated
specimen (frames 1–12).15
4.4. Effect of triaxial stress on the laminated rock
As discussed in Section 3.3.3, “pseudo-triaxial loading condi-
tions were confining the exposed surfaces of the specimen with
the help of metal plates pressed against the specimen by c-clamps
119
(Fig. 8).” The plates were placed on the platens instead of the
specimen, which allowed unrestricted movement of the platen in
the axial direction. All the specimens were tested under stroke-
control mode, which allows for controlled failure of the specimen.
In Fig. 15(e), at 0.09-in. displacement there is a drop in the
force–displacement curve, indicating failure of a section in the
specimen. That is followed by rise in the curve, and the maximum
load attained was 53,000 lbs. The second specimen (Fig. 15(f))
showed a non-linear force–displacement curve, which then
dropped sharply at 46,000 lbs., indicating failure of the specimen.
Because of the confinement, the specimen was able to sustain
further load. The third specimen (Fig. 15(g)) showed a non-linear
force–displacement curve that plateaued at 47,000 lbs., followed
by a further rise in the force–displacement curve. When the la-
minated rock is confined, the response of the rock is similar to the
non-laminated intact rock. In the case of the fourth specimen, the
axial force–displacement relationship was non-linear and pla-
teaued at 54,000 lbs., beyond which the test was terminated.
From the tests mentioned above, it was verified that confine-
ment reduces the influence of the lamination and the specimen
shows behavior similar to an intact rock. This restrained the
movement of exposed surfaces and development of localized
failure near them. However, confinement induced a change in the
mode of failure and the presence of lamination did not seem to
have any effect on it. Confinement suppressed splitting, forcing the
rock to fail in shear. The sheared zones are easily identified in two
specimens that failed along conjugate shear planes (Fig. 15(b)–(d)).
This type of failure was earlier reported by Ref. 12, where under
certain confining pressure the rock was found to be at the cusp of
brittle–ductile transitions. It was also observed that the movement
of the lamination—bending and failure of the specimen—was un-
ique to the material and not an artifact of the load system.
5. Conclusions
This research focused on understanding the behavior of coal
mine shale rocks under biaxial and triaxial stresses. Laminated
shale rocks have been found to be more susceptible to delamina-
tion under high biaxial horizontal in-situ stresses, leading to a
variety of ground control issues in underground coal mines, in-
cluding cutter-related roof problems. Case studies on this problem
have shown the failure pattern in which the laminated layer fails
and how the failure progresses into the upper layers of the im-
mediate roof. Our understanding of the mechanism behind this
unique failure mechanism has been limited to numerical models,
which have treated the layer as a “continuum.” Additionally,
strain-softening behavior is typically used without fully under-
standing the mechanistic behavior and the mode of failure of these
rock types. Therefore, it is imperative that the failure mechanism
of these rock types be investigated under different stress condi-
tions to better understand the underlying mechanism behind such
failure. This research project attempted to shed light in this area by
testing shale rocks in biaxial and triaxial stress conditions.
A biaxial device was fabricated that applied equal stresses on
four faces of the specimen. In the triaxial tests, the third dimen-
sions of the specimens were confined using a fixed c-clamp.
Samples were collected from mines located in the Pittsburgh, Il-
linois no. 6, and No. 2 Gas seams. The samples constituted three
rock types: (1) black shale, (2) laminated shale, and (3) limestone.
From these samples, cubic specimens were prepared that were
tested in biaxial platens.
The outcomes of tests on 23 specimens are summarized in
Table 1. The black shale specimen showed extreme brittle type
failure. The failure mode in the four specimens of the black shale
was characterized by the ejection of rock chunks and the
120 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123

Fig. 14. Post-test pictures and force vs. displacement diagram of specimen I5 (a, e), I6 (b, f), I7 (c, g) and I8 (d, h).15
 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123 121

Fig. 15. Post-test pictures and force vs. displacement diagram of specimen P5 (a, e), P6 (b, f), P7 (c, g) and P8 (d, h).15
122 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123

Table 1
Summary of biaxial and triaxial test results.15

Specimen name Rock type Peak load Approximate peak stress Failure type
(lbs.) (psi)a

Biaxial test
1 (P1) Black shale 4009 708 Extremely brittle. Localized failure close to the unconfined surface of
specimen
2 (P2) Black shale 44,778 7915
3 (P3) Black shale 54,759 9680
4 (P4) Black shale 36,118 6384
1 (I1) Laminated shale 5978 1056 Separation of the lamination observed, possibly caused by tensile failure. No
shearing observed at the macrolevel.
2 (I2) Laminated shale 11,403 2015
3 (I3) Laminated shale 45,573 8056
4 (I4) Laminated shale 9385 1659
1 (I5) Limestone 37,902 6700 Core of the specimen intact, severity of the failure close to the exposed
surface.
2 (I6) Limestone 116,882 20,662
3 (I7) Limestone 96,858 17,122
4 (I8) Limestone 27,506 4862
Triaxial test
1 (I5) Laminated shale 53,514 9460 Along multiple shear planes.
2 (I6) Laminated shale 56,865 10,052 Along conjugate shear planes.
3 (I7) Laminated shale 64,936 11,479
4 (I8) Laminated shale 54,145 9571 Along multiple shear planes.

a
Calculated using Eq. (2).

progressive failure of the specimen from the unconfined surface to
the center of the specimen. For the four laminated shale speci-
mens, the failure was marked by the bending and separation of the
outer layers and the subsequent closure of the inner layers. It was
also found that there was no shearing between the laminated
layers at the macrolevel. The four limestone specimens tested in
the biaxial device showed violent failure. It was concluded that the
failure mode of the laminated shale was not an artifact of the
biaxial platen but an inherent property of the rock type and stress
field. Finally, four laminated shale specimens were tested under
triaxial stress and showed conjugate shear failure, validating the
fact that each rock type has a different mode of failure for given
stress conditions.
The main conclusions and contributions of this research are
 Biaxial stress fields on non-laminated shale specimens show
brittle failure mode.
 Laminated shale under biaxial stress condition showed bending
failure accompanied by movement of the laminated layers to-
wards the unconfined ends.
 The laminated layers did not undergo any shearing at the
macro-level. When the rock layers at the unconfined ends failed
in flexure, the load was transferred on to the inner layers, which
prevented them from shearing against each other.
 The mode of complete failure of the specimens are dependent
on rock type. Limestone specimens under biaxial stress showed
a failure mode that was different from the black and laminated
shale specimens.
 The failure mode is also dependent on stress conditions. Lami-
nated shale under triaxial stress showed conjugate shear failure.
The lamination effect on the mode of failure was negligible
under confined conditions and the rock behaved as an intact
rock.
For future work, further instrumentation of the experimental
set-up should be performed to measure the actual load and de-
formation acting on all six sides of the specimen. This will increase
the capability to monitor the response of specimens with the
change in confining pressure. In addition, performing tests on
large rock specimens in an appropriately sized platen will show
the effect of size on failure mode; and more laminations, due to
increased size, will more accurately represent the mine roof. The
biaxial platen applies equal stresses to the four sides of the cubic
specimen, which may not always accurately represent in-situ
stress conditions. Therefore, these tests should be performed in
true triaxial load frames that are capable of applying independent
stresses on 3-sides of the specimen. Different loading paths and
independent stress application might produce more realistic be-
havior of the specimen, which is not captured by the biaxial
device.
Acknowledgments
We thank the Centers for Disease Control and Prevention -
National Institute of Occupational Health and Safety (No. 200-
2011-40676) for funding this project. In addition, special thanks to
all the mine operators who helped us immensely with sample
collection.
References
1. Ahola MP, Donato DA, Kripakov NP. Application of Numerical Modeling Techni-
ques to Analysis of Cutter Roof Failure. USBM IC 9287. Washington D.C; 1991:28.
2. Bajpayee TS, Pappas DM, Ellenberger JL. Roof instability: what reportable non-
injury roof falls in underground coal mines can tell us. Prof Saf. 2014;59(03)57–
62.
3. Control Centers for Disease Control and Prevention. Retrieved February 12,
2015, from 〈http://www.cdc.gov/niosh/mining/UserFiles/statistics/12m11uoc.
svg〉; 2012.
4. Dahl HD, Parsons RC. Ground control studies in the Humphrey no. 7 mine,
Christopher Coal Division, Consolidation Coal Co. Trans Soc Min Eng.
1972;252:211–222.
5. Esterhuizen GS, Bajpayee TS. Horizontal stress related failure in bedded mine
roofs-insight from field observations and numerical models. In: Proceedings of
the 46th US Rock Mechanics/Geomechanics Symposium. Chicago, IL, USA;
2012:68–77.
6. Gadde MM, Peng SS. Numerical simulation of cutter roof failures under weak
roof conditions. In: Proceeding of the SME Annual Meeting, Salt Lake City, Utah,
USA; 2005:459–469.
7. Hill JL III, Bauer ER. An investigation of the causes of cutter roof failure in a
central Pennsylvania coal mine: a case study. In: Proceedings of the 25th Sym-
posium on Rock Mechanics, Evanston, IL, USA; 1984:603–614.
8. Jumkis AR, Jumkis AA. Red Brunswick Shale and Its Engineering Aspects. . New
Brunswick, N.J.: Rutgers University, College of Engineering, Engineering Re-
search Bulletin; 1975. p. 51.
 S. Arora, B. Mishra / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 109–123
9. Kripakov NP. Alternatives: for controlling cutter roof in coal mines. In: Pro-
ceedings of 2nd International Conference on Ground Control in Mining, Morgan-
town, WV, USA; 1982:142–151.
10. Mark C. Horizontal stress and its effects on longwall ground control. Min Eng.
1991;290:1356–1360.
11. Molinda GM, Mark C, Pappas D, Klemetti T. Overview of ground control issues
in the Illinois Basin. Soc Min Eng Trans. 2008;324:41–48.
12. Paterson MS, Wong TF. Experimental Rock Deformation – The Brittle Field. 2nd
ed.,New York: Springer; 2005.
123
13. Peng SS. Ground Control Failures – A Pictorial View of Case Studies. Morgantown,
WV: S.S. Peng Publisher; 2007.
14. Ray AK. Influence of Cutting Sequence and Time Effects on Cutters and Roof Falls in
Underground Coal Mine – Numerical Approach. Morgantown: West Virginia
University; 2007.
15. Arora S. Effect of Biaxial and Triaxial Stresses on Coal Mine Shale Rocks. Mor-
gantown: West Virginia University; 2015 Master's thesis.
16. Su WH, Peng SS. Cutter roof and its causes. Min Sci Technol. 1987;4(02)113–132.