International Journal of Rock Mechanics & Mining Sciences 130 (2020) 104309

Contents lists available at ScienceDirect

International Journal of Rock Mechanics and Mining Sciences

journal homepage: http://www.elsevier.com/locate/ijrmms

Failure properties of cubic granite, marble, and sandstone specimens under

true triaxial stress
Kun Du a, b, Chengzhi Yang a, Rui Su a, Ming Tao a, Shaofeng Wang a, *
a
School of Resources and Safety Engineering, Central South University, Changsha, Hunan, 410083, PR China
b
Advanced Research Center, Central South University, Changsha, Hunan, 410083, PR China

A R T I C L E I N F O A B S T R A C T

Keywords: Cubic specimens of granite, marble, and sandstone, with side lengths of 50 mm, were prepared in the laboratory.
Rock lithology True triaxial compression tests were conducted on each rock type with minor principal stress (σ 3) magnitudes of
True triaxial failure strength 10, 20, 30, 50, and 100 MPa and the intermediate principal stress (σ 2) varying from σ 2 ¼ σ3 to σ 2 ¼ σ1 (where σ1
Fracture dip angle
represents the major principal stress). The experimental results demonstrated that the failure strength, fracture
Non-linear behavior
Failure mode
dip angle, non-linear behavior and failure mode of these rocks under true triaxial conditions were affected by
both σ 2 and σ 3, as well as by the rock lithology. When σ3 was kept constant and σ2 increased, failure strength
showed an initial increase followed by a decrease, whereas the fracture dip angle increased monotonically. The
non-linear behavior before the peak strength point of granite became more apparent by the increase of σ 3,
whereas σ 2 had an opposite effect on the non-linear behavior of granite under high σ 3 (e.g. 100 MPa). The non-
linear behavior before the peak strength point was always found in marble and sandstone for all of the stress state
conditions. Analyses of fracture dip angle and failure mode showed that when σ 3 ¼ 10, 20, and 30 MPa and σ2/
σ3 � 5, 7.5, and 10, respectively, the failure mode of granite changed from shear to slabbing, whereas marble and
sandstone always fractured in the shear direction. This study confirms that rocks may experience slabbing failure
under true triaxial stress, and slabbing failure mode is likely to be induced in high-strength and brittle rocks
under low σ3 with the stress ratio of σ 2/σ 3 exceeding a particular threshold value.

1. Introduction state acting on the rock and the associated stress variation.27,28
Many studies have focused on true triaxial strength, fracture dip
Stress condition significantly influences the strength property and angle, and the brittle–ductile transition of rocks under true triaxial
failure behavior of rock in underground engineering. Rocks below the stress. Mogi stated that the fracture dip angle, which expresses the angle
subsurface are originally subjected to three-dimensional compressive between the failure plane and the major principal stress, and the peak
stress (σ1 � σ 2 � σ 3), where σ 1, σ2, and σ3 are the major, intermediate, strength are dependent on the magnitude of σ 2.15 This work was fol­
and minor principal stresses, respectively. Stress conditions will change lowed by several studies that validated his findings and offered more
with excavation-induced stress redistributions including stress transfer, in-depth analysis.16,19,24,29,30 Under different stress states, rocks may
stress concentration, stress release and stress relief in rock around un­ exhibit distinct failure modes and strength characteristics. Specifically,
derground openings.1–8 The stress state acting on rock in the ground (σ1 many studies have revealed that σ 2 has an indisputable influence on rock
� σ 2 � σ3) is often referred to as true triaxial stress, meaning that the strength and rock failure.31 In order to quantitatively determine the
principal stresses are unequal, the level and formation of which is effect of σ2 on true triaxial strength of rock, some three-dimensional
influenced by underground excavation activities such as mining, failure criteria were proposed by extensions from classical
tunnelling and drilling.9–14 True triaxial stress might play a key role in two-dimensional failure criteria, such as Mohr–Coulomb criterion and
influencing fracture properties within rocks and make rock present Hoek–Brown criterion without considering the σ 2 effect, and by empir­
different strength properties.3,15–26 In engineering practices, many rock ical regressions from true triaxial testing data.16,31–39 However, some­
failure phenomena, such as slabbing, spalling, rockburst, large defor­ times the existing criteria still overestimate or underestimate the true
mation and zonal disintegration, are controlled by the true triaxial stress triaxial rock strength, although the excellent estimation results have

* Corresponding author.
E-mail addresses: [email protected], [email protected] (S. Wang).

https://doi.org/10.1016/j.ijrmms.2020.104309
Received 12 November 2019; Received in revised form 19 March 2020; Accepted 24 March 2020
Available online 13 April 2020
1365-1609/© 2020 Elsevier Ltd. All rights reserved.
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Table 1
Basic physical and mechanical parameters of the selected rocks.
Rock lithology Density (g�cm 3) σucs (MPa) Fracture angle (� ) Young’s modulus E (GPa) Possion’s ratio μ

Cylinder Cubic Cylinder Cubic Cylinder Cubic Cylinder Cubic

Granite 2.38 113.7 123.2 75 79 45.3 48.9 0.21 0.22

Marble 2.31 44.3 52.4 64 64 17.4 19.05 0.28 0.27
Sandstone 2.25 24.9 32.1 71 65 9.8 10.83 0.33 0.34

Table 2
Mineralogical compositions, microstructures and geological conditions of the selected rocks.
Property Granite Marble Sandstone

Mineralogical composition 38% plagioclase (Pl), Size 0.1–3.6 mm; 98% calcite (Cal), Size 0.05–1.2 mm; 12% sand debris (Sd), 0.05–0.6 mm;
25% K-feldspar (Kfs), Size 0.1–3.2 mm; 2% others 30% plagioclase (Pl), 0.05–0.4 mm;
5% albite (Ab), Size 0.1–1.2 mm; 12% K-feldspar (Kfs), 0.05–0.4 mm;
18% quartz (Qtz), Size 0.05–1.2 mm; 35% quartz (Qtz), 0.05–0.7 mm;
5% hornblende (Hbl), Size 0.02–3.0 mm; 1% calcite (Cal), Size 0.01–0.05 mm;
2% calcite (Cal), Size 0.01–0.3 mm; 10% others
7% others
Microstructure

Massive structure Massive structure Pore cementation and massive structure

Geological condition

Homogeneous and no geological defects Homogeneous and no geological defects Homogeneous and no geological defects

Therefore, the existing failure criteria can not represent the complete
failure strengths in the full range of intermediate principal stress (σ2)
varying from σ2 ¼ σ 3 to σ2 ¼ σ1. In addition, Ma et al.30,41 conducted a
suite of true triaxial compression experiments on two porous sandstones,
showing that the effect of σ2 on failure strength, fracture dip angle, and
the brittle–ductile transition was different for each scenario. Other re­
searchers have dedicated their efforts by numerical simulations to
investigate the correlation between true triaxial stress and rock fail­
ure.42–44 Using EPCA3D software, Pan et al.42 concluded that the het­
erogeneous stress distribution induced by rock heterogeneity and the
effect of loading platens were the main factors that induced the σ 2 effect,
in which the peak strength presents a trend of initial increase and sub­
sequent decrease. Duan et al.43 used the discrete element method to
examine the effects of σ 2 on factors such as peak strength, Young’s
modulus, the dip angle of the main fracture plane, the brittle–ductile
transition, and the evolution of failure mode. The authors found that σ2
plays an important role in the macroscopic failure response of rock.
However, Vachaparampil and Ghassemi45 investigated the failure
characteristics of three types of shale under true triaxial stress, and
found no significant change in fracture dip angle while changing σ 2. This
finding was in contrast to previous observations of a modest increase in
fracture dip angle with increasing σ 2. The inconsistent results mentioned
above could have been caused by the influence of rock lithology on
Fig. 1. Photographs of fracture fragments of (a) granite, (b) marble and (c) fracture dip angle, and the limited range over which σ2 was varied may
sandstone in uniaxial compression tests (the ratios in the figure indicate the have concealed the relationship between σ2 and fracture dip angle.
aspect ratios of the rock specimens). Therefore, the influence of σ 2 on fracture behavior and strength char­
acteristics of rock under true triaxial stress has not yet been clearly
been produced over the limited range of σ 2 at constant σ3.31,40 known.
As the depth to an underground structure increases, the occurrence

2
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Fig. 2. True triaxial testing system: (a) overview of the testing system; (b) loading apparatus; (c) loading box and measures taken to reduce friction; (d) loading path.

Table 3
Rock specimen labelling system in terms of the experimental stress conditions.
Rock type σ3 (MPa) Rock specimen number

Granite 10 g-10-10 g-10-30a g-10-50 g-10-100 g-10-150 g-10-175 g-10-200 g-10-240

20 g-20-20 g-20-50 g-20-100 g-20-150 g-20-200 g-20-250 g-20-275 g-20-340
30 g-30-30 g-30-50 g-30-100 g-30-150 g-30-200 g-30-300 g-30-344
50 g-50-50 g-50-100 g-50-150 g-50-200 g-50-300 g-50-400 g-50-440
100 g-100-100 g-100-180 g-100-260 g-100-340 g-100-420 g-100-490 g-100-600
Marble 10 m-10-10 m-10-30 m-10-50 m-10-75 m-10-100 m-10-150 m-10-175
20 m-20-20 m-20-50 m-20-75 m-20-100 m-20-125 m-20-150 m-20-175 m-20-200
30 m-30-30 m-30-50 m-30-100 m-30-130 m-30-150 m-30-175 m-30-200 m-30-212
50 m-50-50 m-50-100 m-50-130 m-50-175 m-50-200 m-50-250 m-50-283
100 m-100-100 m-100-150 m-100-200 m-100-250 m-100-300 m-100-350 m-100-460 m-100-494
Sandstone 10 s-10-10 s-10-30 s-10-50 s-10-75 s-10-100 s-10-120 s-10-132
20 s-20-20 s-20-63 s-20-105 s-20-150 s-20-170
30 s-30-30 s-30-60 s-30-100 s-30-140 s-30-175
50 s-50-50 s-50-100 s-50-150 s-50-210 s-50-234 s-50-311
100 s-100-100 s-100-170 s-100-230 s-100-300 s-100-365 s-100-413
a
Note: Taking specimen “g-10-30” as an example, “g” represents the granite group, “10” is the predefined value of σ 3, and “30” is the predefined value of σ2. In
addition, this labelling also represents the associated load path in true triaxial compression test, which indicates that σ 3, σ2 and σ1 are first applied to predefined values
of 10 MPa at a loading rate of 0.2 MPa/s, σ 2 and σ 1 are then increased to predefined values of 30 MPa at a loading rate of 0.2 MPa/s, and finally σ1 are increased until
rock failure at a loading rate of 0.3–0.4 MPa/s.

frequency and severity degree of fractures in typical hard rock also in­ conducted on cubic rock specimens (granite, marble and sandstone)
crease. Slabbing, which is a type of extension fracture, has been using a true triaxial machine with solid piston loading in all three
frequently found in rock failures in laboratory and field observations, principal stress directions. The influences of σ 2 and σ3 on the failure
and has been studied in comparison with shear fractures.2,26,46–53 strength, brittle–ductile transition and fracture dip angle were
Slabbing is generally defined as the formation of stress-induced rock adequately investigated and comprehensively analyzed under the minor
slabs. Such slabs are easily observed at the boundaries of underground principal stress (σ 3) magnitudes of 10, 20, 30, 50, and 100 MPa and in
excavations where they are parallel to the excavation walls.54 The pre­ the full change range of intermediate principal stress (σ2) varying from
vious efforts found that the occurrence of slabbing was correlated to the σ 2 ¼ σ 3 to σ 2 ¼ σ 1. More specifically, the transitions in failure modes for
size scale such as height-to-width ratio of rock and the level of inter­ different rocks under true triaxial stress conditions were compared and
mediate principal stress.25,47 However, the detailed comparisons and analyzed to shed light on the fracture mechanism of rocks and the
analyses of rock failure properties of different rock lithologies under true associated correlation with stress conditions.
triaxial stress conditions are rare, and the mechanism that underlies the
transition from shear failure to slabbing failure remains unclear.
In this study, a series of true triaxial compression tests were

3
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Fig. 3. True triaxial failure strengths: (a–c) three-dimensional views; and (d–f) section views on the σ1–σ3 planes.

2. Experimental program were 163.2, 69.4, and 32.1 MPa, respectively. In all uniaxial compres­
sion tests, the failure mode of the cylindrical specimens was shear. For
2.1. Rock specimens the cubic specimens, the failure mode of granite and marble was tension
and that of sandstone was shear. The fracture photographs of rock
The rock types used in this study were granite, marble, and sand­ specimens under uniaxial compression tests are shown in Fig. 1.
stone. Each rock type had different basic physical and mechanical In addition to uniaxial tests, cubic specimens with a side length of 50
properties, including density, uniaxial compressive strength (σ ucs), mm (size dimensions of 50 � 50 � 50 mm) were subjected to true
Young’s modulus (E), and Poisson’s ratio (μ) listed in Table 1. The triaxial compression tests. Samples that are subjected to laboratory tests
mineralogical compositions, microstructures and geological conditions may experience size effects and end effects. In cubic specimens, the size
of the selected rock materials were depicted in Table 2. The cylindrical and end effects are the same along the directions of σ1, σ 2, and σ 3;
specimens with diameter of 50 mm and length of 100 mm (aspect ratio consequently, we can ignore the influence of the difference in the size
¼ 2) and cubic specimens with a side length of 50 mm (shape ratio ¼ 1) and end effects along the directions of σ 1, σ2, and σ 3 on the failure
were subjected to uniaxial compression tests. The σ ucs values of the cy­ properties of rocks in true triaxial compression tests.
lindrical specimens of granite, marble, and sandstone were 113.7, 44.3, To avoid stress concentration due to inadequate contacts between
and 24.9 MPa, respectively, and the σucs values of the cubic specimens rock and loading platens, all rock specimens used in this study were

4
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Table 4 Table 5
Failure strength, fracture dip angle, and failure mode of granite under a true Failure strength, fracture dip angle, and failure mode of marble under a true-
triaxial stress state. triaxial stress state.
Rock Predefined σtts Fracture dip Failure mode Rock Predefined σtts Fracture dip Failure mode
number principal stress (MPa) anglea(� ) number principal stress (MPa) anglea(� )

σ3 σ2 σ3 σ2
(MPa) (MPa) (MPa) (MPa)

g-10-10 10 10 217.9 65 Shear m-10-10 10 10 107.0 65 Shear

g-10-30 10 30 281.0 72 Tension with m-10-30 10 30 129.0 70 Shear
shear m-10-50 10 50 148.1 74 Shear
g-10-50 10 50 323.0 80 Slabbing m-10-75 10 75 166.4 75 Shear
g-10-100 10 100 351.0 82 Slabbing m-10-100 10 100 207.3 79 Shear
g-10-150 10 150 360.0 85 Slabbing m-10-150 10 150 198.0 82 Tension with
g-10-175 10 175 370.2 85 Slabbing shear
g-10-200 10 200 352.0 86 Slabbing m-10-175 10 175 176.3 81 Tension with
g-10-240 10 240 287.0 83 Slabbing shear
g-20-20 20 20 293.2 62 Shear m-20-20 20 20 147.2 60 Shear
g-20-50 20 50 348.4 66 Shear m-20-50 20 50 195.3 66 Shear
g-20-100 20 100 371.7 74 Tension with m-20-75 20 75 212.2 69 Shear
shear m-20-100 20 100 225.4 72 Shear
g-20-150 20 150 390 82 Slabbing m-20-125 20 125 239.9 73 Shear
g-20-200 20 200 423.9 83 Slabbing m-20-150 20 150 242.5 73 Shear
g-20-250 20 250 436.9 84 Slabbing m-20-175 20 175 226.1 74 Shear
g-20-275 20 275 399.2 83 Slabbing m-20-200 20 200 226.6 76 Shear
g-20-340 20 340 400.0 84 Slabbing m-30-30 30 30 186.9 58 Shear
g-30-30 30 30 340.3 61 Shear m-30-50 30 50 209.4 63 Shear
g-30-50 30 50 365.0 67 Shear m-30-100 30 100 243.9 65 Shear
g-30-100 30 100 400.0 69 Shear m-30-130 30 125 291.8 68 Shear
g-30-150 30 150 455.0 72 Tension with m-30-150 30 150 269.6 70 Shear
shear m-30-175 30 175 272.5 72 Shear
g-30-200 30 200 500.0 75 Tension with m-30-200 30 200 260.4 73 Shear
shear m-30-212 30 212 212.8 73 Shear
g-30-300 30 300 474.7 77 Slabbing m-50-50 50 50 252.1 60 Shear
g-30-344 30 344 430.0 80 Slabbing m-50-100 50 100 304.0 62 Shear
g-50-50 50 50 421.9 57 Shear m-50-130 50 130 350.0 64 Shear
g-50-100 50 100 447.4 66 Shear m-50-175 50 175 364.3 66 Shear
g-50-150 50 150 527.1 68 Shear m-50-200 50 300 346.3 70 Shear
g-50-200 50 200 535.2 71 Tension with m-50-250 50 250 334.0 72 Shear
shear m-50-283 50 284 283.8 74 Shear
g-50-300 50 300 570.1 73 Tension with m-100- 100 100 392.7 56 Shear
shear 100
g-50-400 50 400 533.0 78 Tension with m-100- 100 150 435.1 60 Shear
shear 150
g-50-440 50 440 453.2 78 Tension with m-100- 100 200 496.5 63 Shear
shear 200
g-100-100 100 100 592.2 55 Shear m-100- 100 250 510.4 67 Shear
g-100-180 100 180 650 61 Shear 250
g-100-260 100 260 750 64 Shear m-100- 100 300 531.4 67 Shear
g-100-340 100 340 783.2 70 Shear 300
g-100-420 100 420 815 71 Tension with m-100- 100 350 552.0 68 Shear
shear 350
g-100-490 100 490 800 72 Tension with m-100- 100 460 511.2 69 Shear
shear 460
g-100-600 100 600 754 72 Tension with m-100- 100 494 497.2 69 Shear
shear 494
a a
Note: The fracture dip angle listed in this table is the largest angle measured Note: The fracture dip angle listed in this table is the largest angle measured
in each specimen. in each specimen.

prepared under strict tolerances to meet the experimental requirements, hydraulically driven pistons with maximum loads of 3000, 2000, and
and the squareness of the specimen surfaces was controlled within 0.02 2000 kN along the vertical (Z) and horizontal (X and Y) axes, respec­
mm. tively, as shown in Fig. 2b.
Prior to testing, the rock specimens were placed in a custom-made
loading apparatus, shown in Fig. 2b and c, which comprised a box
2.2. True triaxial testing system made of hard resin and six platens made of manganese steel. Butter and
polythene sheets were applied evenly to the six surfaces of each rock to
True triaxial compression tests were conducted using a novel true reduce the frictional resistance from the platens as far as possible, as
triaxial testing system (Model TRW 3000) at Central South University, shown in Fig. 2c. The selected platens had extremely high stiffness and
Changsha, China. The loading system is shown in Fig. 2a and b. The rigidity in order to adequately transfer the load in each direction. To
testing apparatus system consisted of the main machine frame, servo- avoid mutual extrusion among platens during the loading phase (espe­
control hardware, and software to control the machine operation, as cially when using softer rocks), the cross-sectional area of platens should
shown in Fig. 2a. The strain of the specimens in the σ 1 direction was be slightly smaller than that of the rock specimens. This was achieved by
measured using a linear variable differential transformer (LVDT) with a using a chamfer edge.19
resolution of 0.5 μm. This testing system can apply quasi-static loading
along the three principal stress directions independently through

5
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Table 6
Failure strength, fracture dip angle, and failure mode of sandstone under a true
triaxial stress state.
Rock Predefined σtts Fracture dip Failure
number principal stress (MPa) anglea(� ) mode

σ3 σ2
(MPa) (MPa)

s-10-10 10 10 92.1 55 Shear

s-10-30 10 30 122.2 56 Shear
s-10-50 10 50 134.5 63 Shear
s-10-75 10 75 153.1 69 Shear
s-10-100 10 100 155.9 62 Shear
s-10-120 10 120 146.0 70 Shear
s-10-132 10 132 150.3 72 Shear
s-20-20 20 20 133.5 54 Shear
s-20-63 20 63 178.8 58 Shear
s-20-105 20 105 191.5 58 Shear
s-20-150 20 150 213.1 64 Shear
s-20-170 20 170 191.0 67 Shear
s-30-30 30 30 167.9 53 Shear
s-30-60 30 60 198.9 56 Shear
s-30-100 30 100 232.4 58 Shear
s-30-140 30 140 255.4 60 Shear
s-30-175 30 175 211.3 65 Shear
s-50-50 50 50 236.3 53 Shear
s-50-100 50 100 298.4 55 Shear
s-50-150 50 150 305.0 53 Shear
s-50-210 50 210 336.6 57 Shear
s-50-234 50 234 345 63 Shear
s-50-311 50 311 334.9 63 Shear
s-100-100 100 100 366.0 52 Shear
s-100-170 100 170 401.4 56 Shear
s-100-230 100 230 439.5 59 Shear
s-100-300 100 300 449.0 58 Shear
s-100-365 100 365 492.1 58 Shear
s-100-413 100 413 431.7 59 Shear
a
Note: The fracture dip angle listed in this table is the largest angle measured
in each specimen.

2.3. Experimental schemes

Throughout the true triaxial compression tests, σ1 was applied in the

Z direction, σ 2 in the Y direction, and σ3 in the X direction. The loading
path is shown in Fig. 2d and can be summarized as follows. First, σ 1, σ2,
and σ 3 were applied at their predefined values by loading at a rate of 0.2
MPa/s. Next, σ2 and σ 3 were kept constant at their predefined values,
and σ1 was monotonically increased at a rate of 0.3–0.4 MPa/s until rock
failure.
Rock specimen labelling system in terms of the experimental stress
conditions was defined as “lithology-predefined load value of σ3 - pre­
defined load value of σ 2”. Taking specimen “g-10-30” as an example, “g”
represents the granite group, “10” is the predefined value of σ3, and “30”
is the predefined value of σ 2. In addition, this labelling also represents
the associated load path in true triaxial compression test, which in­
dicates that σ3, σ2 and σ 1 are first applied to predefined value of 10 MPa
at a loading rate of 0.2 MPa/s, σ2 and σ1 are then increased to predefined
value of 30 MPa at a loading rate of 0.2 MPa/s, and finally σ1 are
increased until rock failure at a loading rate of 0.3–0.4 MPa/s. All labels
of rock specimens tested in true triaxial compression tests are listed in
Table 3.

3. Experimental results

3.1. True triaxial failure strength

Fig. 4. Regressed results of relationships (a) between the increase index υ and
σ3, (b) between the position index λ and σ3, and (c) among the increase index υ,
The true triaxial failure strength σtts (i.e. σ1 at rock failure) observed
position index λ and σ 3.
for the various magnitudes of σ2 and σ 3 are shown in Fig. 3. The σtts
values of granite, marble, and sandstone are listed in Tables 4–6,
respectively. The σtts follows a similar trend for each of the rocks in this
study; i.e., an initial increase followed by a decrease in σ tts as σ2

6
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Fig. 5. Typical pre-peak σ 1–ε1 curves of granite specimens for different σ 2 under σ 3 magnitudes of (a) 20 MPa, (b) 50 MPa and (c) 100 MPa, curves of marble
specimens for different σ2 under σ 3 magnitudes of (d) 20 MPa, (e) 50 MPa and (f) 100 MPa, and curves of sandstone specimens for different σ 2 under σ 3 magnitudes of
(g) 20 MPa, (h) 50 MPa and (i) 100 MPa.

increases for each predefined σ3 (Fig. 3a–c). When σ2 is initially loaded

beyond σ 3, the value of σ tts increases with increasing σ 2, which is due to
the increasing restricted expansion of the rock specimen in the direction
of σ 2 with increase in σ2. However, as σ 2 continues to increase, the rock
specimens occur pre-damage before failure, and so a trend of decreasing
σ tts is evident once σ2 exceeds a certain threshold. Similar trends have
also been reported in previous studies.18,30,41,42,44,55 In the true triaxial
compression tests, tests with σ2 ¼ σ 3 were termed conventional triaxial
compression tests, whereas tests with σ1 ¼ σ2 were termed conventional
triaxial extension tests.15 The failure strengths of specimens under σ1 ¼
σ 2 was higher than that of specimens under σ2 ¼ σ3 (Fig. 3d–f),
demonstrating the σ2 effect.21,56 In this study, the increase indicator (υ)
and the position indicator (λ) are proposed to evaluate the effects of σ2
on rock failure strength at each magnitude of σ3. The increase indicator
is calculated using
σtts ðmaxÞ σ tts ðminÞ
v¼ (1)
σtts ðminÞ

Fig. 6. Calculation scheme for the plastic and elastic strain of rocks. OA: nat­ where σ tts(max) and σtts(min) are the maximum and minimum values of
ural crack closure and compaction stage; AB: elastic stage; BC: yield stage. CD: the true triaxial strength for a constant σ3, respectively. The value of
oblique line with a slope equal to that of the elastic stage of the σ 1–ε1 curve; OD:
σ tts(min) is the σtts in the conventional triaxial compression test (σ2 ¼
plastic strain; DE: elastic strain; OE: total strain; C: peak strength point.
σ 3). Likewise, the position indicator is calculated as follows:

7
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Table 7
Plastic strain and total strain before peak strength for each rock type at various σ2 magnitudes, when σ 3 ¼ 20, 50, and 100 MPa.
Granite Marble Sandstone

Specimen Permanent strain Strain at peak Specimen Permanent strain Strain at peak Specimen Permanent strain Strain at peak
number (%) point (%) number (%) point (%) number (%) point (%)

g-20-20 0.055 0.603 m-20-20 0.281 0.588 s-20-20 0.323 0.718

g-20-50 0.044 0.647 m-20-50 0.260 0.664 s-20-63 0.301 0.611
g-20-100 0.038 0.671 m-20-75 0.245 0.697 s-20-105 0.403 0.895
g-20-150 0.045 0.725 m-20-100 0.202 0.615 s-20-150 0.410 0.951
g-20-200 0.041 0.685 m-20-125 0.205 0.648 s-20-170 0.34 0.887
g-20-250 0.037 0.689 m-20-150 0.186 0.552 s-50-50 0.810 1.370
g-20-275 0.040 0.679 m-20-175 0.156 0.621 s-50-100 0.814 1.509
g-20-340 0.035 0.620 m-20-200 0.162 0.512 s-50-150 0.745 1.609
g-50-50 0.103 0.770 m-50-50 0.331 0.934 s-50-210 0.786 1.470
g-50-100 0.095 0.744 m-50-100 0.210 0.800 s-50-234 0.697 1.385
g-50-150 0.110 0.787 m-50-130 0.350 0.956 s-50-311 0.732 1.440
g-50-200 0.076 0.792 m-50-175 0.367 0.843 s-100-100 0.912 1.653
g-50-300 0.065 0.701 m-50-200 0.255 0.906 s-100-170 0.845 1.601
g-50-400 0.048 0.850 m-50-250 0.286 0.870 s-100-230 0.713 1.648
g-50-440 0.033 0.645 m-50-284 0.205 0.783 s-100-300 0.726 1.545
g-100-100 0.302 1.312 m-100-100 0.472 1.178 s-100-365 0.622 1.335
g-100-180 0. 310 1.475 m-100-150 0.751 1.454 s-100-413 0.645 1.563
g-100-260 0.295 1.423 m-100-200 0.543 1.377
g-100-340 0.176 0.879 m-100-250 0.772 1.371
g-100-420 0.105 1.021 m-100-300 0.765 1.474
g-100-490 0.064 0.880 m-100-350 0.774 1.473
g-100-600 0.084 1.085 m-100-460 0.696 1.523
m-100-494 0.593 1.379

σ2 ðmaxÞ σ 2 ðminÞ mlnðσ3 Þ þ nλ þ t, where m, n and t were regressive coefficients corre­

λ¼ (2)
σ3 sponding to rock material properties, was regressed using calculated
data by experimental strengths under true triaxial stresses to further
where σ2(max) is the value of σ 2 which corresponds to σtts(max) for a understand the relationships between true triaxial strength and
constant σ 3, and σ2(min) is the value of σ 2 which corresponds to σ tts(min) confining stress conditions, the results of which were shown in Fig. 4c.
for a constant σ 3 such that σ 2(min) is equal to σ3. The increment of peak strength has a linearly positive correlation with
In the current study, the index υ was used to illustrate the influence of the corresponding increment of σ2, and has a linearly negative correla­
the σ 2 effect on σtts under a constant σ3. For example, a larger υ indicates tion with the natural logarithm of σ 3. This result indicates that the in­
that σ2 presents a more significant influence on σtts under a constant σ3. crease in σ 2 can obviously enlarge the peak strength of rock under low
For granite, υ changed from 69.9% when σ 3 ¼ 10 MPa to 37.6% when σ 3 σ 3, while that increment effect will weaken under high σ 3.
¼ 100 MPa. For marble, υ changed from 93.7% when σ 3 ¼ 10 MPa to
40.6% when σ 3 ¼ 100 MPa. Finally, for sandstone, υ decreased from
69.3% when σ 3 ¼ 10 MPa to 34.5% when σ 3 ¼ 100 MPa. The υ values 3.2. Stress–strain curves
show a monotonically decreasing trend as σ 3 increases for each rock
type, which indicates that the increment of σtts is gradually reduced The main compressive strain for rock failure occurs at the maximum
when the magnitude of σ 3 is increased, and that the influence of σ2 on σtts principal stress direction under true triaxial compression test, which can
gradually decreases with increasing σ3. The relationship between υ and significantly reflect the deformation behavior of rock. Therefore, the
σ 3 could be expressed as a natural logarithmic function; i.e., υ ¼ typical pre-peak σ 1–ε1 (i.e. the strain along the σ1 loading direction)
alnðσ 3 Þ þ b, where a and b are material constants. The calculated values curves of granite, marble and sandstone specimens for various σ 2 mag­
using experimental data and regressive curves of υ corresponding to σ3 nitudes under a constant σ 3 of 20, 50 or 100 MPa were investigated and
are plotted in Fig. 4a. The coefficients of determination R2 values of the plotted in Fig. 5. The pre-peak σ1–ε1 curves can be divided into three
fitting curves in Fig. 4a are all higher than 0.80 (i.e., 0.8146, 0.8964, and stages as shown in Fig. 6, in which specimen g-100-260 is shown as an
0.9975 for granite, marble, and sandstone, respectively). example. The stages comprise a natural crack closure and compaction
The index λ was also used in the present study to quantify the diffi­ stage (OA), an elastic stage (AB), and a yield stage (BC). The non-linear
culty degree of employing σ2 in increasing σtts from σ tts(min) to σtts behaviors of the rocks were analyzed based on the proportions of plastic
(max) for a constant σ3, whereby a higher λ indicates that a larger ab­ and elastic strain observed for each specimen before peak stress. From
solute multiplier between σ 2 and σ 3 is needed to obtain σtts(max) for a the strength point C, two linear reference lines, CE and CD, were drawn.
constant σ 3. For granite, λ changed from 16.5 when σ 3 ¼ 10 MPa to 3.2 CE is a vertical line, and CD is an oblique line with a slope equal to that
when σ 3 ¼ 100 MPa. For marble, λ changed from 9.0 when σ3 ¼ 10 MPa of the elastic stage of the σ 1–ε1 curves. The intersections between the x-
to 2.5 when σ 3 ¼ 100 MPa. For sandstone, λ decreased from 9.0 when σ 3 axis and the two reference lines are denoted by points D and E,
¼ 10 MPa to 2.65 when σ3 ¼ 100 MPa. Therefore, λ also followed a respectively. The line OD represents plastic strain, line DE represents
monotonically decreasing trend as σ 3 increases for each rock type. elastic strain, and line OE represents the total strain. The plastic strain
Therefore, for each σ 3 group, the increment of σ 2(max) was gradually and the total strain before peak-point of the specimens for σ3 magnitudes
reduced when the magnitude of σ 3 was increased. The relationship be­ of 20, 50, and 100 MPa are shown in Table 7.
tween λ and σ 3 can also be expressed as a natural logarithmic function; i. It can be drawn from analyses of strain data that the non-linear be­
e., λ ¼ clnðσ3 Þ þ d, where c and d are material constants. The calculated haviors of rocks in true triaxial compression tests are influenced by three
values using experimental data and the regressive curves of λ corre­ factors: rock lithology, σ2 level and σ 3 level. High-strength rocks (e.g.,
sponding to σ 3 are shown in Fig. 4b. The coefficients of determination R2 granite) experienced a transition from minor plastic strain to large
of the fitting curves in Fig. 4b are all larger than 0.86 for each rock type. plastic strain with increasing σ3, while low-strength rocks (e.g. marble
In addition, the three-dimensional linear surface expressed as υ ¼ and sandstone) experienced strong plastic deformation failure at each σ3

8
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Table 8
Increments of fracture dip angle with increase in σ3 from σ2 ¼ σ3 to σ 2 ¼ σ 1 under
each magnitude of σ3.
σ3 (MPa) Increment of fracture dip angle under each σ3 (� )

Granite Marble Sandstone

10 21 17 17
20 22 16 13
30 19 15 12
50 21 14 10
100 17 13 7

Fig. 8. Relationship between the increment index ω of the fracture dip angle
and σ3 for granite, marble, and sandstone.

Fig. 9. Plastic strain and fracture dip angle observed in granite, marble, and
sandstone under σ2 ¼ σ3 ¼ 20, 50, and 100 MPa.

which indicates that increased σ2 favors the brittle failure of rocks under
high-stress conditions.

3.3. Fracture dip angle

The fracture dip angles for each rock type under true triaxial
Fig. 7. Relationships between facture dip angle and σ 2 for different values of σ3 compression tests are listed in Tables 4–6. The relationship between
in (a) granite, (b) marble and (c) sandstone. fracture dip angle (i.e., the angle between the fracture plane and the
horizontal direction) and σ2 for each rock type under a given value of σ3
level. The plastic strain increases with increasing σ3 for all rock types, was analyzed and shown in Fig. 7. Note that there may be several
and the increment of plastic strain in granite is much larger than that in fractures within a given specimen and that the fracture dip angles are
marble and sandstone. The plastic strain decreases with increasing σ2 for expected to vary. To illustrate the influences of σ 2 and σ3 on the fracture
all rock types under a given σ3, and this effect is enhanced at higher σ3, dip angles, the largest fracture dip angle within each specimen was
selected and seen as the fracture dip angle of rock for analysis. As σ2

9
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Fig. 10. Typical fracture fragments of granite for σ3 magnitudes of (a) 10 MPa, (b) 20 MPa, (c) 30 MPa, (d) 50 MPa, and (e) 100 MPa.

increases from σ2 ¼ σ3 to σ 2 ¼ σ1, the fracture dip angles for a given The fracture dip angle increases with respect to increasing σ2 in all σ3
value of σ3 gradually increase. The increment of failure dip angle (i.e., groups, and the associated increments of fracture dip angles are lower
the difference between the maximum and minimum values of the frac­ for lower-strength rock and under higher magnitude of σ 3. For a σ3
ture dip angle for each σ 3) shows a general decrease with increasing σ3, magnitude of 10 MPa, the fracture dip angles within the granite, marble,
as shown in Table 8. In addition, an increment index ω of the fracture dip and sandstone specimens increased with increasing σ2 from σ2 ¼ σ3 to σ2
angle was defined using Eq. (3) to reflect the increment ratio of fracture ¼ σ 1 by 21� , 17� , and 17� , respectively. For a σ3 magnitude of 100 MPa,
dip angle under σ2 from σ 2 ¼ σ3 to σ2 ¼ σ 1 for ach magnitude of σ 3. The the fracture dip angles increased by 17� , 13� , and 7� , respectively.
relationships between index ω and σ 3 for granite, marble, and sandstone The increment ratio of fracture dip angle generally decreases with
are shown in Fig. 8. increases in σ 3. The associated decrements of increment ratios are
unobvious for granite, slight for marble, and obvious for sandstone.
θðmaxÞ θðminÞ
ω¼ (3)
θðminÞ
3.4. Failure modes
where, for each magnitude of σ 3, θ(max) and θ(min) are the maximum
and minimum values of the fracture dip angles under σ 2 from σ 2 ¼ σ3 to
The fragments of the specimens that failed under different stress
σ 2 ¼ σ1, respectively.
conditions were collected and documented, the photographs of which
Based on the experimental data and the associated analysis results,
for granite, marble and sandstone were shown in Fig. 10, Fig. 11 and
the fracture dip angle is influenced by rock lithology, σ2 and σ3, and the
Fig. 12, respectively. The transition in fracture mode from macroscopic
following specific conclusions can be drawn.
shearing to slabbing is assessed based on the images of the fragments
All three tested rock types, the fracture dip angles gradually
and the fracture dip angles described above. For granite, the fracture dip
decreased with increasing σ 3. Taking granite as an example, the fracture
angles of the specimens are greater than 80� where σ2 � 50 MPa at σ3 ¼
dip angles (for the case of σ2 ¼ σ3) corresponding to σ3 magnitudes of 10,
10 MPa, σ 2 � 150 MPa at σ 3 ¼ 20 MPa, and σ 2 � 300 MPa at σ3 ¼ 30
20, 30, 50, and 100 MPa were 65� , 62� , 61� , 57� , and 55� , respectively.
MPa. Under these stress conditions, the fracture orientations in the
This also reflects the enhanced plastic behavior with higher σ 3, as shown
granite specimens are distributed approximately along the σ 1–σ 2 plane,
in Fig. 9.
which is similar to slabbing failure. Slabbing failure is defined as densely

10
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Fig. 11. Typical fracture fragments of marble for σ3 magnitudes of (a) 10 MPa, (b) 20 MPa, (c) 30 MPa, (d) 50 MPa, and (e) 100 MPa.

spaced “onion-skin” fractures or slabs [2, 47, 51 57]. The transition to rocks show a similar trend in υ which decreases monotonically as σ3
slabbing fractures occurs at σ3 ¼ 10 MPa and a stress ratio σ2/σ3 of 5, at increases. The degree to which σ2 affects the failure strength at different
σ 3 ¼ 20 MPa and a stress ratio of 7.5, and at 30 MPa and a stress ratio of values of σ3 varies, and when σ3 is lower, the effect of σ 2 is larger. For a
10, as shown in Fig. 13. In this study, slabbing failure is no longer constant value of σ 3, there is an explicit value of σ2 that results in the
observed when σ 3 reaches 50 MPa for granite. Specimens under the maximum value of the failure strength, and σ2 can be noted as σ2(max).
conditions of σ3 ¼ 50 MPa and 100 MPa experienced shearing fractures, When σ2 reaches beyond this noted σ2(max), the true triaxial compres­
where macro shear bands were the main failure mode. For marble and sive strength instead drops, resulting from the high deviatoric stress
sandstone, shear failure was always the dominant failure mode in a true between σ2 and σ3 can induce more prominent pre-fracturing that
triaxial stress state. Even though some of the fractures within the marble reduce the rock strength. It is predictable that σ 2(max) increases as σ3
specimens in the σ1–σ2 plane (i.e., specimens m-10-150 and m-10-175 in increases for obtaining the required deviatoric stress for pre-damaging
Fig. 11 and Table 5) had fracture dip angles exceeding 80� , the rest of the rock, which can be confirmed by the statistical data shown in
fractures within these specimens had smaller dip angles near 70� . Fig. 14b. However, the position indicator λ ¼ ðσ 2 ðmaxÞ σ 3 Þ=σ3 de­
Therefore, the failure modes of these specimens including many frac­ creases as σ 3 increases, as shown in Fig. 14c. Therefore, when we pursue
tures with fracture dip angels beyond 80� and some fractures with a mathematical relationship between the failure strength, σ2, and σ 3, the
fracture dip angles below 80� were characterized as tension with shear. indicators υ and λ can be considered to reflect the influence of σ 2 and σ3
on failure strength of rock.
4. Discussion
4.2. True triaxial compressive deformation properties
4.1. True triaxial compressive strength
The effect of confining stress on the deformability of rocks has been
The increase indicators υ ¼ ðσtts ðmaxÞ σtts ðminÞÞ=σ tts ðminÞ of failure widely reported. In conventional triaxial compression tests, if the
strengths for different rocks are shown in Fig. 14a, according to the confining stress applied to cylindrical specimens is increased, then the
statistical analyses from previous and present studies. Most types of axial strain required to cause rock failure will increase, and the rock will

11
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Fig. 12. Typical fracture fragments of sandstone for σ 3 magnitudes of (a) 10 MPa, (b) 20 MPa, (c) 30 MPa, (d) 50 MPa, and (e) 100 MPa.

show a tendency toward greater ductility.58 In general, for sufficiently

low confining stresses, brittle fractures occur in rocks. However, at
higher confining stresses, the rock can experience a large strain with no
significant loss in its ability to support a load, and it can exhibit ductile
behavior. These conclusions are inferred as the effect of σ3 on the
deformation properties of rock during failure. In fact, σ2 has the opposite
effect on the deformability of rocks compared to σ3. When σ3 is constant,
the brittleness of a rock failure will increase as σ2 increases. Thus, based
on the change trends in the fracture dip angle, when σ3 was less than
100 MPa, the Coconino sandstone specimens experienced ductile failure
followed by brittle failure as σ2 increases, and the cut-off point between
the ductile and brittle failure of the sandstone was approximately σ 2 ¼
200 MPa.30,41 Maybe the reason is that the pre-fracturing, induced by
the increasing deviatoric stress with increase in σ2, shortens or replaces
the initiation and propagation of cracks in plastic deformation, and then
promotes the transition from ductile failure to brittle failure. In this
study, we analyzed the deformation properties of three rocks below the
Fig. 13. Conditions of the transition in fracture mode for granite during true peak strength. Rock lithology is the main factor influencing the defor­
triaxial compression tests.
mation properties of rock, and ductile failure is always dominant for
medium strong rocks and weak rocks. However, σ2 can shorten the
ductile stage of typical strong and hard rocks (such as granite) when σ3 is
high, e.g. 100 MPa, and in this study the cut-off point between ductile

12
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

Fig. 14. Relationships between (a) the increase indicator (υ) and σ 3, (b) σ 2 (max) and σ3, and (c) the position indicator (λ) and σ3.

and brittle failures of granite as approximately σ2 ¼ 340 MPa, as shown during true triaxial compression tests is dependent on two key factors:
in Fig. 5c. (a) the presence of brittle and high-strength rocks; and (b) low σ 3 values
and high σ2/σ 3 ratios. We also found that the anticipated threshold
leading to slabbing failure increases with increase in σ3. Moreover, it
4.3. Failure pattern seems that the transition from shear to slabbing failure only occurs in
high-strength rocks. This is a strong indication that failure mode is not
The previous studies in the literature have demonstrated that slab­ only affected by stress conditions, but also by the rock lithology and
bing failure is observed mainly in rock specimens during uniaxial texture.
loading, biaxial loading or true triaxial unloading.2,57–66 In the present Slabbing failure in hard rocks is seldom observed in true triaxial
study, slabbing failure was induced under true triaxial compression (σ 3 compression tests, even with low σ 3 values and high σ2/σ3
6¼ 0, σ2 6¼ 0). The realization of such a failure mode in rock specimens

13
K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

values16,19,22,45,55,67 The discrepancy between previous results and Appendix A. Supplementary data
those of the present study is attributed to the fact that the previously
used specimens had a height-to-width ratio of 2, whereas in this study Supplementary data to this article can be found online at https://doi.
the height-to-width ratio was 1 (for the cubic specimens). According org/10.1016/j.ijrmms.2020.104309.
to47,51, in shorter specimens, tensile cracks can easily propagate to­
wards the ends of specimen. Because of the limited specimen height and References
the special stress state comprising a low σ3 and high σ 2/σ 3, this even­
tually leads to slabbing failure. Our results also indicate that σ3 ¼ 0 is not 1 Du K, Li XB, Li DY, Weng L. Failure properties of rocks in true triaxial unloading
compressive test. Trans. Nonferr. Metal. Soc. 2005;25:571–581.
a prerequisite for slabbing failure. These results further our under­ 2 Du K, Tao M, Li XB, Zhou J. Experimental study of slabbing and rockburst induced by
standing of rock failure phenomena in the walls of underground open­ true-triaxial unloading and local dynamic disturbance. Rock Mech Rock Eng. 2016;49:
ings at greater depth. 3437–3453.
3 Feng XT, Kong R, Zhang XW, Yang CX. Experimental study of failure differences in
hard rock under true triaxial compression. Rock Mech Rock Eng. 2019;52:2109–2122.
5. Conclusions 4 Wang SF, Li XB, Yao JR, et al. Experimental investigation of rock breakage by a
conical pick and its application to non-explosive mechanized mining in deep hard
rock. Int J Rock Mech Min Sci. 2019;122:104063.
The main effect of σ2 on rock failure characteristics can be described 5 Tao M, Ma A, Cao WZ, Li XB, Gong FQ. Dynamic response of pre-stressed rock with a
as follows: (a) as σ 2 increases, the failure strength initially increases and circular cavity subject to transient loading. Int J Rock Mech Min Sci. 2017;99:1–8.
then decreases for a given value of σ3; (b) the degree to which σ2 affects 6 Tao M, Li ZW, Cao WZ, Li XB, Wu CQ. Stress redistribution of dynamic loading
incident with arbitrary waveform through a circular cavity. Int J Numer Anal Methods
the failure strength shows a decrease with increasing σ 3; (c) for a con­
GeoMech. 2019;43(6):1279–1299.
stant value of σ 3, there is an explicit value of σ2 that results in the 7 Peng K, Zhou J, Zou Q, Song X. Effect of loading frequency on the deformation
maximum value of failure strength, and, in addition, the net increase behaviours of sandstones subjected to cyclic loads and its underlying mechanism. Int
multiplier of σ2 (resulting in the maximum strength for each constant σ 3) J Fatig. 2020;131:105349.
8 Zhu D, Wu Y, Liu Z, Dong X, Yu J. Failure mechanism and safety control strategy for
varies with σ 3. laminated roof of wide-span roadway. Eng Fail Anal. 2020;111:104489.
Compared with σ 3, σ2 has the opposite effect on the deformability of 9 Zhang QB, Zhao J. A review of dynamic experimental techniques and mechanical
rocks. Rock lithology is also a key influence factor on the deformation behaviour of rock materials. Rock Mech Rock Eng. 2014;47:1411–1478.
10 Cai X, Zhou ZL, Liu K, Du XM, Zang HZ. Water-weakening effects on the mechanical
properties of rock under true triaxial conditions. The yield stage (rep­ behavior of different rock types: phenomena and mechanisms. Appl Sci. 2019;9(20):
resenting the rock’s plastic properties) becomes pronounced during the 4450.
pre-peak stage in all cases for marble and sandstone. However, 11 Zhou ZL, Cai X, Li XB, Cao WZ, Du XM. Dynamic response and energy evolution of
sandstone under coupled static–dynamic compression: insights from experimental
increasing σ 2 can reduce the proportion of plastic strain before peak study into deep rock engineering applications. Rock Mech Rock Eng. 2019. https://
strength is reached in granite under high σ 3 (e.g. 100 MPa). doi.org/10.1007/s00603-019-01980-9.
The fracture dip angle is influenced by rock lithology, σ 2 and σ 3. The 12 Wang YT, Zhou XP, Kou MM. Three-dimensional numerical study on the failure
characteristics of intermittent fissures under compressive-shear loads. Acta Geotech.
fracture dip angle gradually decreases with increasing σ3 in all three 2019;14(4):1161–1193.
tested rock types, which reflects the enhanced non-linear properties of 13 Dang W, Konietzky H, Frühwirt T, Herbst M. Cyclic frictional responses of planar
rock under a high σ 3. The fracture dip angle increases with increasing σ 2 joints under cyclic normal load conditions: laboratory tests and numerical
simulations. Rock Mech Rock Eng. 2020;53:337–364.
in all σ 3 groups, and the associated increments are lower for lower-
14 Peng K, Liu Z, Zou Q, Wu Q, Zhou J. Mechanical property of granite from different
strength rock and under higher magnitude of σ3. buried depths under uniaxial compression and dynamic impact: an energy-based
The occurrence of slabbing failure in true triaxial compression tests investigation. Powder Technol. 2020;362:729–744.
requires several key factors, including the presence of brittle and high- 15 Mogi K. Effect of the intermediate principal stress on rock failure. J Geophys Res.
1967;72:5117–5131.
strength rocks, a stress ratio σ2/σ 3 equal to or greater than a critical 16 Mogi K. Fracture and flow of rocks under high triaxial compression. J Geophys Res.
value, and low σ3. 1971;76:1255–1269.
Additional testing and concurrent numerical analysis are underway 17 Reches Z, Dieterich JH. Faulting of rocks in three-dimensional strain fields I. Failure
of rocks in polyaxial, servo-control experiments. Tectonophysics. 1983;95:111–132.
to verify these results and to expand on the range of stress conditions and 18 Michelis P. Polyaxial yielding of granular rock. J Eng Mech. 1985;111:1049–1066.
rock types investigated. The conclusions in this study are therefore 19 Haimson BC, Chang C. A new true triaxial cell for testing mechanical properties of
limited to the rock types and stress ranges covered in the experimental rock, and its use to determine rock strength and deformability of Westerly granite. Int
J Rock Mech Min Sci. 2000;37:285–296.
program. Following the recent advancements in rock engineering 20 Al-Ajmi AM, Zimmerman RW. Relation between the mogi and the coulomb failure
practices, one would also consider the implementation of modern ap­ criteria. Int J Rock Mech Min Sci. 2005;42:431–439.
proaches such as nuclear magnetic resonance imaging for better char­ 21 Haimson BC. True triaxial stresses and the brittle fracture of rock. Pure Appl Geophys.
2006;163:1101–1130.
acterization of fractures under uniaxial and triaxial stress 22 Lee H, Haimson BC. True triaxial strength, deformability, and brittle failure of
conditions.68,69 granodiorite from the San Andreas Fault Observatory at Depth. Int J Rock Mech Min
Sci. 2011;7:1199–1207.
23 Nasseri MHB, Goodfellow SD, Lombos L, Young RP. 3-D transport and acoustic
Declaration of competing interest
properties of Fontainebleau sandstone during true-triaxial deformation experiments.
Int J Rock Mech Min Sci. 2014;69:1–18.
The authors declare that they have no known competing financial 24 Feng XT, Zhang XW, Kong R, Wang G. A novel mogi type true triaxial testing
interests or personal relationships that could have appeared to influence apparatus and its use to obtain complete stress-strain curves of hard rocks. Rock Mech
Rock Eng. 2017;49:1649–1662.
the work reported in this paper. 25 Wang S, Li X, Du K, Wang S, Tao M. Experimental study of the triaxial strength
properties of hollow cylindrical granite specimens under coupled external and
Acknowledgements internal confining stresses. Rock Mech Rock Eng. 2018;51:2015–2031.
26 Feng F, Li XB, Rostami J, Peng DX, Li DY, Du K. Numerical investigation of hard rock
strength and fracturing under polyaxial compression based on Mogi-Coulomb failure
This study was funded by the National Natural Science Foundation of criterion. Inter. J. Geomech.. 2019 https://doi.org/10.1061/(ASCE)GM.1943-
China (Nos: 51774326 and 51904333), the open fund of Hebei Mining 5622.0001352.
27 Qian QH, Zhou XP, Yang HQ, Zhang YX, Li XH. Zonal disintegration of surrounding
Development and Safety Technology Laboratory (No: 2016kykf03), the rock mass around the diversion tunnels in Jinping II Hydropower Station,
open fund of Mining Disaster Prevention and Control Ministry Key Southwestern China. Theor Appl Fract Mech. 2009;51:129–138.
Laboratory at Shandong University of Science and Technology (No: 28 Wang S, Huang L, Li X. Analysis of rockburst triggered by hard rock fragmentation
using a conical pick under high uniaxial stress. Tunn Undergr Space Technol. 2020;96:
MDPC201917), and the Fundamental Research Funds for the Central 103195.
Universities of Central South University (No: 2018zzts749). 29 Chang C, Haimson BC. True triaxial strength and deformability of the German
Continental Deep Drilling Program (KTB) deep hole amphibolite. J. Geophys. Res-Sol.
Ea. 2000;105:18999–19013.

14
 K. Du et al. International Journal of Rock Mechanics and Mining Sciences 130 (2020) 104309

30 Ma XD, Haimson BC. Failure characteristics of two porous sandstones subjected to 51 Li XB, Feng F, Li DY, Du K, Ranjith PG, Rostami J. Failure characteristics of granite
true triaxial stresses. J. Geophys. Res-Sol. Ea. 2016;121:6477–6498. influenced by sample height-to-width ratios and intermediate principal stress under
31 Feng XT, Kong R, Yang C, et al. A three-dimensional failure criterion for hard rocks true-triaxial unloading conditions. Rock Mech Rock Eng. 2018;51:1321–1345.
under true triaxial compression. Rock Mech Rock Eng. 2020;53:103–111. 52 Du K, Su R, Tao M, Yang CZ, Momeni A, Wang SF. Specimen shape and cross-section
32 Benz T, Schwab R, Kauther RA, Vermeer PA. A Hoek–Brown criterion with intrinsic effects on the mechanical properties of rocks under uniaxial compressive stress. Bull
material strength factorization. Int J Rock Mech Min Sci. 2008;45:210–222. Eng Geol Environ. 2019;78:6061–6074.
33 You MQ. True-triaxial strength criteria for rock. Int J Rock Mech Min Sci. 2009;46: 53 Feng F, Li XB, Rostami J, Li DY. Modeling hard rock failure induced by structural
115–127. planes around deep circular tunnels. Eng Fract Mech. 2019;205:152–174.
34 Singh M, Raj A, Singh B. Modifed Mohr–Coulomb criterion for non-linear triaxial and 54 Liu G, Feng XT, Jiang Q, Yao Z, Li S. In situ observation of spalling process of intact
polyaxial strength of intact rocks. Int J Rock Mech Min Sci. 2011;48:546–555. rock mass at large cavern excavation. Eng Geol. 2017;226:52–69.
35 Labuz JF, Zang A. Mohr–Coulomb failure criterion. Rock Mech Rock Eng. 2012;45: 55 Haimson BC, Chang C. True triaxial strength of the KTB amphibolite under borehole
975–979. wall conditions and its use to estimate the maximum horizontal in situ stress.
36 Chang C, Haimson B. A failure criterion for rocks based on true triaxial testing. Rock J. Geophys. Res-Sol. Ea. 2020;107. ETG–15.
Mech Rock Eng. 2012;45:1007–1010. 56 Murrell S. A criterion for brittle fracture of rocks and concrete under triaxial stress
37 Meyer JP, Labuz JF. Linear failure criteria with three principal stresses. Int J Rock and the effect of pore pressure on the criterion. Rock Mech. 1963:563–577.
Mech Min Sci. 2013;60:180–187. 57 Cai M. Influence of intermediate principal stress on rock fracturing and strength near
38 Jiang H, Zhao J. A simple three-dimensional failure criterion for rocks based on the excavation boundaries—insight from numerical modeling. Int J Rock Mech Min Sci.
Hoek-Brown criterion. Rock Mech Rock Eng. 2015;48:1807–1819. 2008;45:763–772.
39 Labuz JF, Zeng F, Makhnenko R, Li Y. Brittle failure of rock: a review and general 58 Jaeger JC, Cook NG, Zimmerman R. Fundamentals of Rock Mechanics. fourth ed.
linear criterion. J Struct Geol. 2018;112:7–28. Oxford: Blackwell Publishing; 2007.
40 Jimenez R, Ma X. A note on the strength symmetry imposed by Mogi’s true-triaxial 59 Peng S, Johnson A. Crack growth and faulting in cylindrical specimens of Chelmsford
criterion. Int J Rock Mech Min Sci. 2013;64:17–21. granite. Int J Rock Mech Min Sci Geomech Abstr. 1972;9(1):37–42.
41 Ma XD, Rudnicki JW, Haimson BC. Failure characteristics of two porous sandstones 60 Horii H, Nemat-Nasser S. Compression-induced microcrack growth in brittle solids:
subjected to true triaxial stresses: applied through a novel loading path. J. Geophys. axial splitting and shear failure. J. Geophys. Res-Sol. Ea. 1985;90:3105–3125.
Res-Sol. Ea. 2017;122:2525–2540. 61 Bazant ZP, Xiang Y. Size effect in compression fracture: splitting crack band
42 Pan PZ, Feng XT, Hudson JA. The influence of intermediate principal stress on rock propagation. J Eng Mech. 1997;123:162–172.
failure behavior: a numerical study. Eng Geol. 2012;124(4):109–118. 62 Fakhimi A, Hemami B. Axial splitting of rocks under uniaxial compression. Int J Rock
43 Duan K, Kwok C, Ma XD. DEM simulations of sandstone under true triaxial Mech Min Sci. 2015;79:124–134.
compressive tests. Acta Geotech. 2017;12:495–510. 63 Gong QM, Yin LJ, Wu SY, Zhao J, Ting Y. Rockburst and slabbing failure and its
44 He PF, Kulatilake PHSW, Yang XX, Liu DQ, He MC. Detailed comparison of nine influence on TBM excavation at headrace tunnels in Jinping II hydropower station.
intact rock failure criteria using polyaxial intact coal strength data obtained through Eng Geol. 2012;124(1):98–108.
PFC 3D simulations. Acta Geotech. 2018;13:419–445. 64 Gong WL, Peng YY, Wang H, He MC, Sousa LRE, Wang J. Fracture angle analysis of
45 Vachaparampil A, Ghassemi A. Failure characteristics of three shales under true- rock burst faulting planes based on true-triaxial experiment. Rock Mech Rock Eng.
triaxial compression. Int J Rock Mech Min Sci. 2017;100:151–159. 2015;48:1017–1039.
46 He MC, Miao JL, Feng J. Rock burst process of limestone and its acoustic emission 65 Zhu WS, Yang WM, Li XJ, Xiang L, Yu DJ. Study on splitting failure in rock masses by
characteristics under true-triaxial unloading conditions. Int J Rock Mech Min Sci. simulation test, site monitoring and energy model. Tunn Undergr Space Technol. 2014;
2010;47:286–298, 2010. 41:152–164, 2014.
47 Li DY, Li CC, Li XB. Influence of sample height-to-width ratios on failure mode for 66 Zhao F, He M. Size effects on granite behavior under unloading rockburst test. Bull
rectangular prism samples of hard rock loaded in uniaxial compression. Rock Mech Eng Geol Environ. 2017;76(3):1183–1197.
Rock Eng. 2011;44:253–267. 67 Takahashi M, Koide H. Effect of the intermediate principalstress on strength and
48 Zhao XG, Cai M. Influence of specimen height-to-width ratio on the strainburst deformation behavior of sedimentary rocks at the depth shallower than 2000 m. In:
characteristics of Tianhu granite under true-triaxial unloading conditions. Can International Symposium on Rock at Great Depth. August 1989:19–26. Pau, France; 28-
Geotech J. 2014;52:890–902. 31.
49 Zhao XG, Wang J, Cai M, et al. Influence of unloading rate on the strainburst 68 Golsanami N, Sun GM, Liu Y, et al. Distinguishing fractures from matrix pores based
characteristics of Beishan granite under true-triaxial unloading conditions. Rock on the practical application of rock physics inversion and NMR data: a case study
Mech Rock Eng. 2014;47:467–483. from an unconventional coal reservoir in China. J. Nat. Gas Sci. 2019;65:145–167.
50 Li XB, Du K, Li DY. True triaxial strength and failure modes of cubic rock specimens 69 Golsanami N, Sun GM, Zhang ZY. A review on the applications of the nuclear
with unloading the minor principal stress. Rock Mech Rock Eng. 2015;48:2185–2196. magnetic resonance (NMR) technology for investigating fractures. J Appl Geophys.
2016;133:30–38.

Du et al. (2020) conducted a series of true triaxial compression tests on cubic rock specimens i.e granite, sandstone
and marble. the influence of ()2 and ()3 on the failure strength were investigated. the mechanical response of the
specimens under various true triaxial stress conditions were presented. From the study, it is understood that as ()2
increases, the failure strength initially increases and then decreases for a given value of ()3. For a constant value of ()
3, there is an explicit value of ()2 that results in the maximum value of failure strength.

15