Rock Mech Rock Eng (2012) 45:327340

DOI 10.1007/s00603-011-0209-7

ORIGINAL PAPER

Investigating the Effect of Cyclic Loading on the Indirect Tensile

Strength of Rocks
N. Erarslan D. J. Williams

Received: 7 September 2011 / Accepted: 18 November 2011 / Published online: 29 November 2011
Springer-Verlag 2011

Abstract This paper presents the results of laboratory

experiments during the investigation of the stressstrain
characteristics of Brisbane tuff disc specimens under
diametral compressive cyclic loading. Two different cyclic
loading methods were used: namely, sinusoidal cyclic
loading and cyclic loading with increasing mean level. The
first method applied the SN curve approach to the indirect
tensile strength (ITS) of rock specimens for the first time in
the literature, and the second method investigated the effect
of increasing cyclic loading on the ITS of rock specimens.
The ITS of Brisbane tuff disc specimens was measured
using the Brazilian tensile strength test. The reduction in
ITS was found to be 33% with sinusoidal loading tests,
whereas increasing cyclic loading caused a maximum
reduction of 37%. It is believed that the fracturing under
cyclic loading starts at contact points between strong grains
and weak matrices, and that contact points at grain
boundaries are the regions of stress concentration (i.e.,
indenters). Transgranular cracks emanate from these
regions and intergranular cracks sometimes pass through
the contact points. Once cracking begins, there is a steady
progression of damage and a general loosening of the
rock, which is a precursor to the formation of intergranular
cracks.
Keywords Brazilian indirect tensile strength

Cyclic loading
Rock fatigue

N. Erarslan (&)
D. J. Williams

Golder Geomechanics Centre, School of Civil Engineering,
The University of Queensland, St Lucia, QLD 4072, Australia
e-mail: nazife.tiryaki@uqconnect.edu.au

1 Introduction
Rock masses normally consist of blocks of intact rocks
separated by joints, faults or interfaces. Even intact rocks
are categorised as an inhomogeneous, discontinuous
material, on both small and large scales, because they
contain cracks, voids and pores, normally induced by historical thermal, mechanical and/or chemical actions and
reactions. The study of brittle fracture and its relationship
to material damage, fatigue and strength is central in many
rock engineering disciplines, including rock fracture and
damage mechanics, rock cutting, drilling and blasting.
Research on understanding rock cracking, fracture and
fatigue behaviour under various loading conditions, such as
static, cyclic and impact loading, has attracted many
investigators in recent years (Celestino and Bortolucci
1995; Ingraffea 1981; Petros et al. 2003; Zhenyu and
Haihong 1990). Inglis (1913) first confirmed that a maximum tensile stress that is many times larger than the
applied loading stress can be generated at the boundary of
the ellipse modelling crack geometry.
The tensile strength of rocks is much lower than its
compressive strength. Rock is a brittle material and
because all brittle materials are weak in tension, the tensile
strength of rock becomes one of the most important
parameters influencing its deformability and fracture
toughness. The difficulties associated with performing a
direct uniaxial tensile test on a rock specimen have led to a
number of indirect methods for assessing tensile strength.
In 1978, the Brazilian test was officially proposed by the
International Society for Rock Mechanics (ISRM) as a
suggested method for determining the tensile strength of
rock materials (ISRM 1978). The mechanical behaviour of
rock under static loading has been thoroughly investigated.
However, rock reaction to cyclic, repetitive stresses

123

328

resulting from dynamic loads has generally been neglected,

with the exception of a few rather limited studies (Attawel
and Farmer 1973; Bagde and Pedros 2005; Gatelier et al.
2002; Haimson 1978; Zhenyu and Haihong 1990). It is
known that cyclic loading often causes brittle materials,
such as ceramics and rocks, to fail at a stress level lower
than their strength under monotonic conditions (Attawel
and Farmer 1973; Burdine 1963; Evans et al. 1974). This
phenomenon is commonly termed as fatigue. Faults,
joints, bedding planes, tunnel walls, excavation roofs and
ribs, bridge abutments and dam and road foundations are
only a few of the natural and manmade rock structures that
can be weakened by the repetitive loading caused by such
stressors as vehicle-induced vibrations, drilling and blasting and traffic. A principal objective of rock fatigue
research reported in the literature has been to establish a
relationship between the number of cycles (N) and the
reduction in applied stress amplitude (S); the SN curve
approach (Attawel and Farmer 1973; Burdine 1963; Evans
and Fuller 1974; Haimson 1978; Costin and Holcomb
1981; Bagde and Pedros 2009). However, most rock fatigue researches have been focused on uniaxial compressive
strength (UCS) degradation under cyclic loading. Information regarding the dynamic tensile behaviour of rocks is
of considerable importance in assessing the stability of
rock structures under dynamic loads. It is also important in
determining rock breakage and fragmentation under
explosive and percussive excavation (Roxborough 1963,
1968; Hood and Alehossein 2000; Gottlieb and Moore
1981). There is very limited research on the response of the
tensile strength of rocks to cyclic loading (as opposed to
dynamic loading, such as explosive loads and impact
loading). Moreover, relatively little attention has been
given to investigate the damage mechanisms of rock fatigue. A novel outcome of the research described here is the

Fig. 1 Photomicrograph of Feldspar (Fl) and Quartz (Q) mineral

123

N. Erarslan, D. J. Williams

observation of the effect of indirect tensile cyclic loading

on the ITS of rocks and the mechanisms of rock fatigue
damage.

2 Experimental Procedure and Tests

2.1 Sample Preparation
Most of the tests described herein were carried out on
Brisbane tuff, because it is a host rock of Brisbanes first
motorway tunnel, CLEM7, from which core samples were
obtained. Brisbane tuff was chosen since it is a massive
rock type with no bedding (being an ash deposit), making it
easy to handle and prepare for testing, and giving less test
result variability. In addition, it is a targeted rock type in
the Brisbane area for its strength and stability in tunnelling
and excavation applications. Brisbane tuff is a fine-grained,
massive rock of rhyolitic composition with coarser grains
imparting a porphyritic texture. Quartz and feldspar
phenocrysts 13 mm in size are embedded without interlocking in a matrix consisting of polycrystalline silica.
Thin sections of samples were prepared for petrographic
analysis, and the resulting images are given in Figs. 1 and
2. Brisbane tuff is composed largely of quartz and
K-feldspar, with small amount of siderite (Fe-carbonate)
and zeolite minerals. From the thin-section analysis, secondary carbonate cement was observed to replace primary
silica cryptocrystalline cement matrix. Moreover, primary
feldspars were altered to clay minerals (mainly kaolinite
and illite) in some areas, filling the voids in the primary
silica cryptocrystalline cement matrix. The thin sections
also showed that secondary microcrystalline quartz cement
occurs within the primary cryptocrystalline cement matrix
(Fig. 2a), and that some quartz minerals occurred within

Investigating the Effect of Cyclic Loading

329

Fig. 2 Microcrystalline quartz cement (a), and Quartz (Q) mineral in microcrystalline quartz and primary cryptocrystalline cement (b)

the microcrystalline quartz cement (Fig. 2b). For the

mechanical characterisation of Brisbane tuff, UCS and
BTS tests were conducted (see Table 1).
The sample preparation and testing procedure conformed to the requirements of the ISRM (1978). For the
determination of the BTS, samples were cored and trimmed
to form disc specimens that met the ISRM (1978) recommendations; that is, cylindrical specimens 52 mm in
diameter and 26 mm in thickness, to give a thickness to

diameter ratio of 0.5. To improve the contact condition

between the loading arcs and the disc specimens, the disc
specimens were covered with adhesive paper strips as
recommended by the ISRM (1978). Some prepared BTS
disc specimens are shown in Fig. 3a. The Brazilian disc
specimen in Fig. 3b has two strain gauges attached to its
perpendicular faceslateral and axial to the force applied
to the specimen. The vertically placed strain gauge measures the vertical diametral strain, while the horizontally
placed strain gauge measures the horizontal diametral
strain.

Table 1 Mechanical characterisation results for Brisbane tuff

Rock
sample
Brisbane
tuff

Youngs
modulus
(GPa)

Poissons
ratio

UCS (MPa)

BTS
(MPa)

Max = 25

Max = 0.26

Max = 190

Max = 15

Min = 19

Min = 0.22

Min = 97

Min = 8.0

2.2 Static and Cyclic Tests

For the static loading tests, the ISRM suggests that the load
can be applied via two steel loading jaws in contact with a
disk-shaped rock specimen, and that the radius of the jaws
can be 1.5 times the specimen radius (1978). According to

Fig. 3 a Prepared BTS disc specimens, b disc specimen with strain gauges placed between loading jaws

123

330

the ISRM suggestions, the two steel loading jaws are

designed to make contact with the disc-shaped rock specimen at diametrically opposed surfaces over an arc of
contact of approximately 10 at failure (see Fig. 1b). The
static load was applied by a stiff hydraulic Instron loading
frame, with a loading rate of 200 N/s on to the standard
loading jaws, as suggested by the ISRM (1978). The tensile
strength of the five rock specimens tested using the standard Brazilian jaws was calculated using the formula given
by the ISRM (1978).
Two types of cyclic loading were used in this research;
cyclic loading with constant mean level and constant
amplitude, termed sinusoidal cyclic loading (see Fig. 4a);
and cyclic loading with increasing mean level and constant
amplitude, termed increasing cyclic loading (see Fig. 4b).
Loading amplitudes are constant in both types of cyclic
loading. However, the mean level of each cycle increases at
a constant rate in increasing cyclic loading tests, whereas
the mean level of each cycle is constant in sinusoidal cyclic
loading tests. The cyclic loading used in these tests was
ramp waveform type cyclic compressive loading. The
loading frequency was 1 Hz for all tests. The amplitude is
expressed as an absolute (value) equal to the total range.
Four different amplitudes were chosen to investigate the
effect of fatigue on the ITS of Brisbane tuff: 2.3 kN, at
10% of the static ultimate load (SUL); 4.6 kN (20% SUL);
6.9 kN (30% SUL) and 9.2 kN (40% SUL).
Sinusoidal cyclic loading tests were carried out to obtain
SN curves illustrative of the continuous weakening of
rock with the increase in number of cycles (N) required for
failing a specimen loaded to a certain upper peak stress (S).
In the literature, the SN curve concept has been used for
fatigue research under uniaxial compressive loading with
cylindrical rock specimens (Haimson and Kim 1971;
Atkinson 1984; Singh 1989). However, this research represents the SN curve for ITS degradation due to fatigue
under cyclic loads have been reported in the literature for
the first time.

Fig. 4 a Sinusoidal cyclic loading, b increasing cyclic loading

123

N. Erarslan, D. J. Williams

3 Results of Tests
3.1 Results of Static Tests
The tested Brisbane tuff specimens under loading arcs are
shown in Fig. 5. In the present tests on Brisbane tuff disc
specimens, loading with the Brazilian jaws caused central
tensile cracking at the centre of the disc specimens with the
occurrence of several fractures branching from the diametral plane.
The details of the test results for the two different
Brisbane tuff specimens, NST35 and NST207, are given in
Table 2. Two typical stressstrain curves from standard
Brazilian tests are shown in Fig. 6.
3.2 Results of Cyclic Tests
During the monotonic tests on Brazilian disc specimens, it
was found that the standard BTS test caused catastrophic
failure of the disc specimens, rather than the expected
tensile splitting failure initiated by a central crack. For this

Fig. 5 Tested Brisbane tuff disc specimens

Investigating the Effect of Cyclic Loading

331

Table 2 Results of indirect tensile tests on Brisbane tuff disc

specimens
Failure load (kN)

Indirect tensile strength (MPa)

NST35

NST207

NST35

NST207

Replicate 1

26.29

24.58

12.36

11.56

Replicate 2

24.39

19.45

11.47

9.15

Replicate 3
Replicate 4

25.42
23.49

20.98
21.86

11.96
11.05

9.87
10.28

Replicate 5

25.97

25.06

12.22

11.79

Average

25.11

22.38

11.81

10.53

reason, diametral loading arcs with various angles were

used in a separate static test series. However, because there
is no international standard or formulation for determining
the ITS of a rock specimen under different diametral loads,
the standard Brazilian jaws proposed by the ISRM (1978)
were also used in the cyclic loading tests.
The first important conclusion of the sinusoidal cyclic
loading tests was that the failure load used in the calculation of ITS is definitely weakened 33% by repetitive
loading. Results are given in Table 3.
Figure 7 shows the SN curve for the ITS testing of
Brisbane tuff, from which the ITS of 11.17 MPa under
monotonic loading was determined. After the sinusoidal
cyclic tests, it was observed that the ITS of Brisbane tuff
was reduced from the monotonic value of 11.17 to
7.5 MPa. The percentage of the static ITS at which the
specimen did not fail, even after 800,000 cycles, was
considered to be the fatigue stress level of the ITS.
In addition, a series of diametral compressive increasing
cyclic loading tests was performed on 12 BTS Brisbane
tuff disc. The ITS of Brisbane tuff was reduced from the
monotonic value of 11.17 to 7 MPa as a result of the
increasing cyclic loading. The BTS reduction due to cyclic
loading is given in Table 4 in comparison with the static

Table 3 Results of sinusoidal cyclic loading

Number of
cycles to failure
(N)

Percentage
decrease in
failure load

Smax (as % of
static failure
load)

Fmax
(failure
load, kN)

100

23

95

21.8

90
90

20.7
20.7

98
268

10
10

90

20.7

104

10

90

20.7

190

10

80

18.4

3,456

20

80

18.4

775

20

80

18.4

18,723

20

80

18.4

121,000

20

80

16.1

12,047

30

70

16.1

74,515

30

70

16.1

500,000

30

70

16.1

560,000

30

70

16.1

800,000

30

70

16.0

800,000a

30

No failure

BTS values. The main purpose of this comparison is to

show the clear reduction in ultimate failure load, resulting
in a reduction in the ITS due to rock fatigue. An attempt
was made to show that tensile crack propagation causing
failure is possible at lower stress values than the ITS values
calculated using the ISRM suggested methods.
The experimental observations of cyclic loading tests in
this research have indicated that increasing cyclic loading
causes a greater ITS degradation of rock than sinusoidal
compressive cyclic loading. The reduction in ITS was
determined to be 33% with sinusoidal loading tests,
whereas increasing cyclic loading caused a maximum
reduction in ITS of 37%.

Fig. 6 Stressstrain curves of standard Brazilian tests

123

332

N. Erarslan, D. J. Williams

Fig. 7 SN curve for indirect tensile strength of Brisbane tuff

Table 4 Results of increasing
cyclic loading tests performed
on Brazilian disc specimens

For a clear understanding of the effect of rock fatigue on

damage mechanisms, a comparison between static and
dynamic cyclic loading tests is shown in Fig. 8 by plotting
both results on the same axes. Comparing these plots,
diametral axial strain under both static and cyclic loading is
greater when compared with diametral lateral strain.
However, the contribution of plastic strain to the diametral
lateral strain plot under cyclic loading is much more than to
that of the diametral lateral strain under monotonic loading.
Moreover, diametral lateral plastic strain accumulation at
the highest level is a result of the loosening of the hysteresis loop during the final cycles up to failure. During
cyclic loading, the diametral lateral deformation develops
faster than the diametral axial deformation. This behaviour

Sample

Amplitude
% SUL

Ultimate
load (kN)

Number of
cycles
up to failure

Indirect
tensile
strength (ITS)

Reduction in
ITS (%)

NST207-Rp1

10

17.4

2,583

8.2

24.1

NST207-Rp2

10

15.6

2,384

7.3

32.5

NST207-Rp3

10

18.1

2,723

8.5

21.3

17.03

2,563

8.0

25.9

Average
NST207-Rp1

20

17.5

2,195

8.2

24.1

NST207-Rp2

20

15.4

1,887

7.2

33.4

NST207-Rp3
Average

20

14.6
15.9

1,679
1,920

6.8
7.4

37.1
31.5

NST207-Rp1

30

17.3

1,906

8.1

25.0

NST207-Rp2

30

15.2

1,334

7.1

34.3

NST207-Rp3

30

16.1

1,556

7.5

31.0

16.2

1,598

7.6

30.1
42.0

Average
NST207-Rp1

40

14.0

738

6.5

NST207-Rp2

40

14.4

992

6.7

40.0

NST207-Rp3

40

16.8

1227

7.9

29.3

15.0

985

7.0

37.1

Average

Fig. 8 Comparison of loadstrain curves of Brazilian disc specimens tested under static and increasing cyclic loading with various amplitudes

123

Investigating the Effect of Cyclic Loading

indicates the development of crack in the specimen. This is

the reason for the dilatancy of the rock. The higher the
stress level or the larger the cycle amplitude, the more
obvious this behaviour is.
Figure 9 shows the tested Brazilian disc specimens
under static and cyclic loading. It is obvious that much
more crushed rock material and debris were produced
under cyclic loading than under static loading. The mechanisms of rock fatigue under cyclic loading induce many
more microcracks compared with the failure mechanism
under static loading, as explained in the literature (Costin
and Holcomb 1981; Hadly 1976). This experimental outcome is very interesting for investigating the effect of
fatigue on the damage mechanisms of rocks. During both
monotonic and cyclic tests, failure of specimens occurred
very quickly without gradual fracturing or any sign of predamage or cracking.
3.3 Mechanisms of Rock Fatigue Damage
Load strain and strain-number of cycles (SN) plots are
given in Fig. 8a, b, respectively. Measurements of

333

permanent strain show that both the axial and lateral strain
curves increase with the number of cycles, but at different
rates. A slight deviation from linearity is evident only in
the final stage of vertical diametral strain prior to failure,
whereas remarkable deviation from linearity is evident in
horizontal diametral strain (see Fig. 10a). The axial stress
level at which the axial strain departs from linearity marks
the beginning of the onset of unstable crack growth, as
defined by Bieniawskie (1967). This generally occurs at an
axial stress level between 70 and 85% of the short-term
peak strength of cylindrical specimens. The dominant
mechanism causing this increase in axial strain is sliding
along inclined surfaces. These occur in the region characterised by the most significant structural changes to the
specimen, with the density of microcracks increasing sevenfold. As seen from Fig. 10b, the fatigue damage of rock
can be attributed to the continuous accumulation of stressinduced fracture damage. Examinations of the horizontal
diametral strain changes reveal relatively high stiffness
values up to 9 kN when the load is first applied to the
specimen (the lateral stiffness term represents the change in
the lateral strain rate under diametral compressive loading).

Fig. 9 Tested Brazilian disc specimens under: a static loading, and b, c cyclic loading

123

334

N. Erarslan, D. J. Williams

Fig. 10 a Increasing diametral cyclic loading; amplitude = 2.3 kN (10% SUL)

The vertical diametral strain is almost double that of the

horizontal diametral strain, up to 70% of the peak load
(12 kN). After this point, the horizontal diametral strain
rate increases very rapidly up to failure.
Examples of each cyclic test repeated under various
amplitudes are given in Figs. 11, 12, and 13. No significant differences were found in the total amount of strain
produced in a specimen across the various amplitude
cyclic loading tests. However, measurements of permanent strain damage show that both the axial and the lateral strain damage curves increased with increasing
damage increments, but at different rates because of the
reduced number of cycles. The converse trends near the
failure points result from lateral strain increasing more
rapidly than axial strain. It can be concluded that high
amplitude cyclic loading produces the same amount of
irreversible damage as low amplitude cyclic loading in a
specimen, meaning that equivalent damage in rock

material is possible with fewer cycles and less energy

with high amplitude cyclic loading compared with low
amplitude cyclic loading.
The damage mechanism caused by the loading ratedependent dynamic tensile tests on rock has been discussed
by various researchers (Zhang et al. 1999; Zhao and Li
2000). It has generally been concluded that Youngs
modulus and the tensile strength of rocks obtained by the
Brazil tests tend to increase with increasing loading rate.
However, the studies mentioned above were concerned
with the mechanical properties of rock material under high
loading rates (strain rates). In contrast, for moderate to low
loading rates, the proposed theoretical results were not
deemed applicable (Nemat-Nasser and Deng 1994; Zhao
and Li 2000). The result being that cyclic loading is not a
rate-dependent type of loading. Therefore, the mechanism
of tensile failure of rock material under cyclic loading
remains in need of study.

Fig. 11 Increasing diametral cyclic loading; amplitude = 4.6 kN (20% SUL)

123

Investigating the Effect of Cyclic Loading

335

Fig. 12 Increasing diametral cyclic loading; amplitude = 6.9 kN (30% SUL)

Fig. 13 Increasing diametral cyclic loading; amplitude = 9.2 kN (40% SUL)

To examine the major characteristics of the damage

process at the microscopic level, scanning electron
microscope (SEM) imaging was used. The failure surfaces
of the discs tested under both static and cyclic tests were
examined by SEM. A JEOL JSM-6460 LA brand SEM
device was used for tungsten low vacuum (LV) analytical
scanning electron microscopy. In this study, all rock fracture images were obtained under LV chamber pressures
(i.e., 150 Pa with adjustable pressure between 10 and
270 Pa). This allowed certain specimens to be observed
uncoated, which reduced damage to the specimens from the
effects of high vacuum. In addition, failure surfaces could
be examined by SEM directly, without the preparation of
thin sections, avoiding the creation of extra microcracks
during the preparation of thin-section specimens.
Scanning electron micrographs of fracture surfaces
of failed Brisbane tuff disc specimens tested under
cyclic loading are shown in Fig. 14. According to the

petrographic analysis results, quartz and feldspar minerals

are infrequently packed in the silica cryptocrystalline
matrix. Observation of the failed specimens by SEM
showed that massive amounts of dust and small grains were
produced in specimens tested under cyclic loading, but not
with monotonic loading. The damaged matrix of Brisbane
tuff and the very small grains are shown in Fig. 14.
Loosening and de-cohesion mechanisms seem to be very
effective damage mechanisms in the matrix. The damage
and fracturing mechanisms in the matrix of Brisbane tuff
disc specimens tested under monotonic loading are shown
in Fig. 15. Micro- and macro-scale cleavage cracks are
abundant, and represent failure by splitting (pure opening
mode) on the fracture surface of specimens tested under
monotonic loading.
Most grains were crack free, and others relatively so, in
the fracture surfaces of specimens tested under cyclic
loading, whereas almost all grains were highly cracked in

123

336

N. Erarslan, D. J. Williams

Fig. 14 Damage and small grains in the matrix of failed Brisbane tuff disc specimen tested under cyclic loading

Fig. 15 Damage and fracturing in the matrix of failed Brisbane tuff disc specimens tested under monotonic loading

the failure surfaces of specimens tested under monotonic

loading. The cracking mechanism is brittle, and there is a
cleavage cracking through minerals (see Fig. 16a, b). As

123

the quartz and feldspar minerals are randomly located in

the matrix of Brisbane tuff, a mismatch in elastic stiffness
at the interface between the feldspar and quartz grains,

Investigating the Effect of Cyclic Loading

337

Fig. 16 a, b Typical transcrystalline cracks, and c, d intercrystalline and intracrystalline cracks on the failure surfaces of Brisbane tuff disc
specimens tested under monotonic loading

inducing a greater tensile stress in the stiffer grain, is not

possible. This may mean that the cleavage mechanism, due
to the high tensile stress produced in the centre of a disc
specimen, causes intergranular (intercrystalline) cracks in a
disc specimen tested under monotonic loading. The intergranular cracks continue to grow in the grains, propagated
through the matrix in the specimen (see Fig. 16b).
Almost all grain boundaries are cracked along their
entire length and some are cracked at small angles to the
maximum stress direction on the failure surfaces of the
Brisbane tuff disc specimens tested under cyclic loading
(see Fig. 17a, b). The extracted loosened minerals shown in
Fig. 17c, d show general fatigue effects observed on the
failure surfaces of the specimens used in this research.
Intercrystalline cracks were revealed as the principal
cracking form in the fracture surfaces of cyclically fractured specimens. It is already known that the fracturing
mechanism in a disc specimen under diametral monotonic
compressive loading is splitting (opening). However, the
dust and small grains around the loosened grains may

indicate the fracturing mechanism of rock fatigue is produced by the combination of splitting (opening) due to
stress-induced fatigue cracks and frictional sliding (shear),
causing crushing at the grain boundaries during each
loading and unloading cycle around loosened grains.
Some dust grains were observed around the loosened
grains using a higher magnification (see Fig. 18). In general, contact points at grain boundaries are regions of stress
concentration. Transgranular cracks were often found
emanating from these regions. However, these small particles around the grains may also be explained by the
indentation of the grains. It is possible for stiffer grains
(indenters) to exert an indentation pressure on the surface
of the weaker matrix during each loadingunloading cycle.
It is well known that some special cracks, such as radial
cracks, propagate through the surface of a material during
unloading of indenters (Tiryaki, 2008; Whittaker et al.
1992).
From loadstrain plots it was determined that fatigue
fracturing is a continuous process of ongoing and

123

338

N. Erarslan, D. J. Williams

Fig. 17 Scanning electron microscope images of the fractured surfaces of the Brisbane tuff disc specimens: a, b loosened grains and typical
intercrystalline cracks around grains, and c, d extracted grains from the tuff matrix

Fig. 18 Scanning electron microscopic images of the fractured surfaces of Brisbane tuff disc specimens

123

Investigating the Effect of Cyclic Loading

irreversible mechanical and physical deformation. Findings

based on SEM images indicate that the bonds between the
grains and the matrix are responsible for the fatigue damage of rock under cyclic loading.

4 Discussion
Monotonic loading, sinusoidal cyclic loading and increasing cyclic loading tests were conducted on Brazilian disc
specimens. The reduction in ITS was determined to be 30%
with sinusoidal loading, whereas increasing cyclic loading
caused a maximum reduction in ITS of 35%; these percentages are in comparison to the ITS values obtained
under monotonic loading. Besides the decrease in ITS
value, the fracturing mechanism of rock material was found
to be very different under monotonic and cyclic loading,
with excessive amounts of dust and small grains found in
cyclically failed specimens. This observation suggests that
opening tensile fracturing is not the only effective fracturing mechanism in rock fatigue damage. As the preexisting cracks and deformities in the rock disc specimen
do not necessarily align with the principal stress direction,
shear type or mixed type fracturing mechanisms are also
possible around inclined pre-existing cracks in a disc
specimen. Consequently, sliding and shear effects of crack
surfaces are believed to be the principal source of excessive
microfracturing in cyclic tests. The dilation of the cracks
along the diametral axial direction results in greater lateral
strain compared with diametral axial strain. This shows that
unstable crack propagation has been started in the specimen. Unstable crack propagation continues to the point at
which the numerous fatigue-induced microcracks coalesce
into larger cracks, and the rock can no longer support
loading cycles. It is believed that the subcritical failure of
rock, which is characterised by the development of a large
number of microcracks, is possible even prior to the
appearance of the main fatigue cracks in cyclically loaded
specimens.
The SEM images showed that transgranular and intergranular cracks are the main mechanisms of rock fatigue.
This may make an explanation to fatigue researches, which
has reported that a massive amount of microfractures and
irreversible damage accumulation are the main causes of
rock fatigue (Carfagni and Salvatore 2000; Holcomb 1981;
Pino et al. 1999). The SEM images also revealed many
small particles on failure surfaces, small grains and dust
around grain boundaries. However, it is still unknown how
the loading cycles break the bonds between the grains and
the rock matrix to produce intergranular cracks around
grain boundaries. Two possible mechanisms can be inferred from the SEM image analysis in this research: (1)
stiffer grains (i.e., quartz and feldspar minerals) act on the

339

surface of the weaker matrix as indenters under principal

compressive stress and (2) there is an effective sliding and
shear mechanism between grains and rock matrix during
each loading and unloading cycle in which internal stress is
produced between grains and matrix as a reaction to the
external diametral compressive stress.
The effect of increasing cyclic loading on the tensile
strength of rocks was investigated in this study by varying
the loading amplitude at a constant frequency of 1 Hz,
dictated by the limitations of the loading machine. The
effect of loading frequency on rock fatigue is well known
in terms of the fatigue limit of rocks (Burdine 1963;
Attawel and Farmer 1973; Bagde and Pedros 2005).
However, the increasing cyclic loading used in this study is
different from the constant mean level cyclic loading
applied in the determination of the fatigue limit of rocks
reported in the literature. Bagde and Pedros (2009)
observed that the fatigue strength of rock reduced, while
the rock modulus increased with increasing frequency and
decreased with increasing amplitude, under increasing
uniaxial compressive cyclic loading. Hence, frequency is
considered to be the prime parameter affecting the fatigue
strength of BTS specimens under increasing cyclic loading.

5 Conclusion
The ITS response to cyclic loading was found to be different from that under static loading in terms of the lower
ultimate load and different fracturing mechanisms. A
macro-scale splitting crack with a sharp and rough surface
was found along the diametral compressive loading direction under monotonic loading, whereas an excessive
amount of small particles and dust was produced under
cyclic loading. In this research, various experimental
observations of cyclic loading tests with varying amplitudes have indicated that increasing cyclic loading is more
effective at ITS degradation of rocks than sinusoidal
compressive cyclic loading. The reduction in ITS was
found to be 33% with sinusoidal loading, whereas
increasing cyclic loading caused a maximum reduction in
ITS of 37%. In addition, no post-peak behaviour was
observed in the loadstrain plots of either the monotonic or
the cyclic loading tests. The Brazilian disc specimens
failed violently during both the monotonic and cyclic tests.
The SEM imaging results enable some of the qualitative
features of the fatigue damage process in Brisbane tuff to
be inferred. SEM investigations revealed that grain
boundaries play an important role in rock fatigue. The
largest cracks follow the grain boundaries and microcracks
generally start to form at the grain boundaries or voids.
Grain de-cohesion seems to be the source of large cracks.
Further, transgranular cracks were observed during rock

123

340

fatigue. Secondary cracks were seen on the fracture surfaces of one cyclically fractured specimen, but no such
secondary cracks were found on any of the static fracture
surfaces. The main characteristic was grain breakage under
cyclic loading, which probably starts at points of contact
between grains and is accompanied by the production of
very small fragments, probably due to frictional sliding
within the weak matrix. It is believed that contact points at
grain boundaries are the regions of stress concentration
(i.e., indenters). Transgranular cracks may emanate from
these regions, and intergranular cracks sometimes pass
through these contact points. Once cracking starts, there is
a steady progression of damage and a general loosening
of the rock, which is a precursor to the formation of
intergranular cracks.
Acknowledgments Acknowledgement is made to Leighton Contractors who provided core samples of Brisbane tuff from the CLEM7
Project and to Professor Ted Brown AC, Les McQueen, Mark Funkhauser and Rob Morphet of Golder Associates Pty Ltd for their
assistance and advice. The work described forms part of the first
authors PhD research carried out within the Golder Geomechanics
Centre at The University of Queensland. The first author was supported by an Australian Postgraduate Award/UQRS and the Golder
Geomechanics Centre.

References
Atkinson BK (1984) Subcritical crack growth in geological materials.
J Geophys Res 89:40774114
Attawel PB, Farmer IW (1973) Fatigue behaviour of rock. Int J Rock
Mech Min 10:19
Bagde MN, Pedros V (2005) Fatigue properties of intact sandstone
specimens subjected to dynamic uniaxial cyclical loading. Int J
Rock Mech Min 42(2):237250
Bagde MN, Pedros V (2009) Fatigue and dynamic energy behaviour
of rock subjected to cyclical loading. Int J Rock Mech Min
46:200209
Bieniawskie ZT (1967) Mechanism of brittle rock fracture: part II
experimental studies. Int J Rock Mech Min 4(4):407423
Burdine NT (1963) Rock failure under dynamic failure conditions.
SPE 3:18
Carfagni S, Salvatore W (2000) The characterisation of marble by
cyclic compression loading: experimental results. Mech Cohes
Frict Mat 5:535547
Celestino TB, Bortolucci AA (1995) Determination of rock fracture
toughness under creep and fatigue. In: Proceedings of the 35th
U.S. symposium on rock mechanics. Reno, New York,
pp 147152
Costin LS, Holcomb DJ (1981) Time-dependent failure of rock under
cyclic loading. Tectonophys 79:279296

123

N. Erarslan, D. J. Williams
Evans AG, Fuller ER (1974) Crack propagation in ceramic materials
under cyclic loading conditions. Metall Trans 5:2729
Evans AG, Linzer M, Russell LR (1974) Acoustic emission and crack
propagation in polycrystalline alumina. Mater Sci Eng
15:253261
Gatelier N, Pellet F, Loret B (2002) Mechanical damage of an
anisotropic porous rock in cyclic triaxial tests. Int J Rock Mech
Min 39:335354
Gottlieb L, Moore PJ (1981) Vibratory cutting of brown coal. Int J
Rock Mech Min 18:335339
Hadly K (1976) The effect of cyclic stress on dilatancy: another look.
J Geophys Res 81:24712474
Haimson BC (1978) Effect of cyclic loading on rock. In: Dynamic
geotechnical testing. ASTM STP 654:228245
Haimson BC, Kim CM (1971) Mechanical behaviour of rock under
cyclic fatigue. Rock Mech 3:845863
Holcomb DJ (1981) Memory, relaxation, and microfracturing in
dilatant rock. J Geophys Res 86(B7):62356248
Hood M, Alehossein HA (2000) Development in rock cutting
technology. Int J Rock Mech Min 37(12):297305
Inglis CE (1913) Stresses in a plate due to the presence of cracks and
sharp corners. T Inst Nav Archit 55:219246
Ingraffea AR (1981) Mixed mode fracture initiation in Indiana
limestone and westerly granite. In: Proceedings of the 22nd U.S.
symposium on rock mechanics. Cambridge, MA, pp 186191
ISRM (1978) Suggested methods for determining tensile strength of
rock materials. Int J Rock Mech Min 15:99103
Nemat-Nasser S, Deng H (1994) Strain rate effect on brittle failure in
compression. Acta Metall Mater 42:10131024
Petros V, Bagde MN, Holub K, Michalcik P (2003) Comparison of
changes in the strength and the deformation behavior of rocks
under static and dynamic loading. In: The 10th Congress of the
ISRM, Technology Roadmap for Rock Mechanics, Johannesburg, South Africa, pp 899902
Pino DR, Narducci P, Royer-Carfagni G (1999) A SEM investigation
on fatigue damage of marble. J Mater Sci Lett 18:16191622
Roxborough FF (1963) Cutting rocks with picks. Min Eng (Littleton)
132:445452
Roxborough FF (1968) Coal ploughing. Col Eng 12(16):6671
Singh SK (1989) Fatigue and strain hardening behaviour of
greywacke from the flagstaff formation, NSW. Eng Geol
26:171179
Tiryaki N (2008) Numerical analysis of rock fracturing by various
indenters. Dissertation, Hacettepe University, Ankara, Turkey
Whittaker BN, Singh RN, Sun G (1992) Rock fracture mechanics,
principles, design and applications. Elsevier, Amsterdam
Zhang ZX, Kou SQ, Yu J, Yu Y, Jiang LG, Lindqvist PA (1999)
Effects of loading rate on rock fracture. Int J Rock Mech Min
36(5):597611
Zhao J, Li HB (2000) Experimental determination of dynamic tensile
properties of a granite. Int J Rock Mech Min 37:851861
Zhenyu T, Haihong M (1990) An experimental study and analysis of
the behaviour of rock under cyclic loading. Int J Rock Mech Min
27(1):5156