1  Cyclic flattened Brazilian disc tests for measuring the tensile

2  fatigue properties of brittle rocks

3 

4  Yi Liu, Feng Dai*, Nuwen Xu, Tao Zhao

5   
6  State Key Laboratory of Hydraulics and Mountain River Engineering, College of
7  Water Resource and Hydropower, Sichuan University, Chengdu, Sichuan 610065,
8  China
9  * Corresponding author Email: fengdai@scu.edu.cn (F. Dai)
10 

11  Abstract

12  We propose a cyclic flattened Brazilian disc testing method to measure the tensile
13  fatigue properties of brittle rocks. Our method has obvious merits in its specimen
14  preparation and experimental operation. Two parallel flattens are introduced in disc
15  specimen, which facilitate easily and uniformly loading the specimen without special
16  loading devices required. Moreover, the contact regions between two flattens and
17  loading planes barely change during the entire loading and unloading process,
18  ensuring a consistent contact condition. With certain appropriate loading angles, this
19  method guarantees the very first breakage of the specimen occurs at the center of the
20  disc, which is the prerequisite of the Brazilian-type indirect tensile tests. To
21  demonstrate our new method, nine cyclic FBD tensile tests are conducted. The fatigue
22  load-deformation characteristics of FBD specimens are revealed. The tensile fatigue
23  lives of tested specimens are observed to increase with the increase of cyclic loading
24  frequency. Our proposed method provides a convenient and reliable approach to
25  indirectly measure the fatigue tensile properties of brittle rocks and other brittle solids
26  subjected to cyclic tensile loadings.

27  Keywords: Cyclic loadings; Flattened Brazilian disc; Loading frequency; Tensile
28  fatigue properties

‐ 1 ‐ 
 
29  I. Introduction

30  In numerous mining and civil engineering projects, such as underground tunnels,
31  dam foundations and bridge abutments, rocks are likely to be subjected to dynamic
32  cyclic loadings  resulting from earthquakes, rock cutting and rock bursts, etc1. Since
33  rocks are considerably weaker in tension than in compression2, accurate
34  characterizations of tensile fatigue properties and response of rocks subjected to
35  cyclic loading conditions are thus crucial for the stability analysis and rational design
36  of rock engineering structures.
37  A straightforward approach to measure tensile properties of rocks is to conduct
38  uniaxial tensile tests. Due to difficulties of performing direct tensile tests including
39  experimental instrumentations and operation, a variety of indirect methods are derived
40  for measuring the tensile strength of brittle solids3-6, such as the Brazilian test, ring
41  test and three-point or four-point bend test. The Brazilian disc (BD) test, suggested by
42  the International Society for Rock Mechanics (ISRM)7 and standardized by the
43  American Society for Testing and Materials (ASTM)8, has been receiving wide
44  acceptance due to its convenience in both specimen preparation and experimental
45  procedure9-15. Figure 1a shows the typical Brazilian test. A thin disc specimen is
46  diametrically compressed, and the compression induces the tensile stresses 
47  perpendicular to the loading direction which eventually results in the failure of testing
48  specimens. In general, the indirect tensile strength can be determined if the tensile
49  failure occurs at the center of the disc specimen where the local tensile stress is the
50  maximum. However, in practice, the failure has often been observed to initiate at
51  some points away from the center of tested disc, such as the contact point between
52  loading device and the circular boundary of BD specimen16-18, which violates the
53  calculation pre-condition of indirect tensile tests and results in uncertain
54  measurements.

‐ 2 ‐ 
 
55 

56  Fig. 1 Representative loading configurations of Brazilian-type tensile tests: (a) flat loading
57  platen, (b) flat loading platen with two small-diameter steel rod, (c) flat loading platen with two
58  cushions, (d) curved loading jaw and (e) the flattened Brazilian disc test

59  In order to guarantee a reasonable failure mode of rock discs, some improvements
60  in the loading configurations have been made, including the flat loading platens with
61  two small-diameter steel rods or two cushions, and curved loading jaw, as shown in
62  Figs 1b-d19. Although the modified loading configurations are beneficial for forming
63  the central crack, the design and manufacture of these special loading devices are
64  tedious and difficult, and different devices are required for matching different
65  specimen sizes. Wang et.al20, 21 proposed the  flattened Brazilian disc (FBD) testing
66  method shown in Fig 1e. Instead of improving the loading devices, the disc specimen
67  are fabricated with two parallel flattens to fit the loading platens. They concluded that
68  when the loading angle of FBD specimen corresponding to the flat end width is
69  greater than a certain critical value (2α > 20°), the stress concentration at the loading
70  ends can be avoided, and that the central crack initiation can be guaranteed.
71  Most studies in the literature have been concentrated on the fatigue mechanical
72  properties of rocks under cyclic uniaxial compression22-25. However,  the study on
73  fatigue properties of rocks subjected to cyclic tensile loadings is rather limited. Only
74  Erarslan and Williams26 conducted cyclic tensile tests on Brisbane tuff using the BD
75  specimens, and they reported the effect of cyclic loading on the indirect fatigue tensile
‐ 3 ‐ 
 
 76  strength. Nevertheless, the fatigue reactions of brittle rocks to cyclic tensile loadings
77  remain far from being understood, and the influences of the cyclic loading frequency,
78  the prime parameters affecting the fatigue tensile strength of rocks26, on these fatigue
79  properties have never been reported in the literature. Most importantly, there are
80  several critical issues remaining to be addressed in the cyclic tensile tests using BD
81  specimens, involving the inconsistent contact condition at the boundary and the
82  tedious manufacture of loading devices.
83  Given the difficulties in direct tensile tests and conventional Brazilian tests for
84  characterizing cyclic tensile properties, it is thus our intension in this study to develop
85  an applicable method to characterize the tensile fatigue properties of brittle rocks
86  under cyclic loadings. For the first time, cyclic flattened Brazilian disc tests have been
87  proposed in this study for measuring the tensile fatigue properties of brittle rocks.
88  Tests with different cyclic frequencies are conducted to validate our method and
89  illustrate the influences of loading frequency on fatigue tensile properties. The fatigue
90  load-deformation characteristics of tested specimens under cyclic tensile loadings are
91  obtained and the tensile fatigue life of rocks is observed to increase with increasing
92  cyclic frequency. Our testing method provides a convenient and available approach to
93  indirectly measure the tensile fatigue properties of brittle solids including rocks.
94  The remainder of this paper is organized as follows. Section II presents the
95  experimental methodology, including the cyclic flattened Brazilian disc testing
96  method and the finite element analysis, followed by an introduction of the testing
97  equipment, the specimen preparation, and the test schemes in Section III. Section IV
98  synthetically analyzes and discusses the representative experimental results. The
99  entire study is summarized in Section V.

100  II. Experimental methodology

101  A. Cyclic flattened Brazilian disc testing method

102  The FBD method has been developed for measuring the indirect tensile strength
103  under quasi-static loading conditions; for the first time, this concept has been adopted

‐ 4 ‐ 
 
104  in our cyclic tensile tests. The schematic of the cyclic flattened Brazilian disc tests
105  and the geometry of the FBD specimen are shown in Fig. 2, where Pc and P are the
106  cyclic loading and line loading applied on the flat ends, respectively, D and B are the
107  diameter and thickness of the disc, respectively. Our method has certain obvious
108  merits than previous cyclic tensile test using BD specimen. First, no special loading
109  devices are required since two parallel flattens of equal width are introduced in the
110  specimen, which avoids manufacturing the complicate jaws for BD specimens and
111  simplifies the experimental operation. Second, the contact condition between the
112  circular boundary of BD specimen and the arc-loading jaws is not clear27; while the
113  contact boundary between the loading platen and FBD specimen facilitates easily and
114  uniformly loading the specimen, and the contact regions will barely change during the
115  entire loading and unloading process, ensuring the consistent contact condition at the
116  boundary. Third, for FBD specimens, the crucial prerequisite for a valid indirect
117  tensile strength test, namely, first breakage at the center of the disc, can be guaranteed
118  with appropriate loading angles. According to previous researches20, the flat end angle
119  2α (i.e., loading angle) should satisfy the condition of 20°≤2α≤30° to guarantee the
120  testing validity. A loading angle of 2α=25° is thus chosen for all our cyclic tests.

121 
122  Fig. 2 Schematic of the cyclic flattened Brazilian disc tests and the geometry of the FBD
123  specimen

124  B. Determining the tensile stress in FBD specimens

125  Since two flattens are machined on the rock disc, the original equation to calculate
126  the tensile stresses for the full Brazilian disc (with no flattens) is no longer applicable
127  for FBD specimen. Based on Griffith’s strength criterion28, the following new formula
‐ 5 ‐ 
 
128  should be adopted for the FBD specimen with different loading angles:
2 Pf
129  t  k                                    (1)
 DB

130  where σt is the indirect tensile stress, Pf is the failure load, and k is the coefficient

131  correlated to the loading angle.

132  In our study, the key coefficient k is calibrated by finite element analysis with a
133  commercial software ANSYS. The numerical model of the FBD specimen consists of
134  4028 quadrilateral eight-node elements and 12297 nodes (Fig. 3a). For the FBD
135  specimen with a loading angle of 2α=25°, the variations of dimensionless tangential
136  normal stress σθ and radial normal stress σr are depicted in Fig. 3b, where r is the
137  distance from the calculated point on the loading diameter to the center of the disc
138  specimen, and R is the radius of the disc. Under this stress condition composing two
139  non-zero orthogonal stress components, the Griffith failure criterion can be
140  employed25. The failure occurs once the equivalent stress σG equals to the tensile
141  strength σt, in which σG can be calculated according to the maximum principal stress
142  σ1 and the minimum principal stress σ3. For the FBD specimen studied herein, we
143  have σ1 = σθ, σ3 = σr and 3σθ + σr < 0. If the tensile stress is considered positive herein,
144  σG can thus be calculated with the following formula:

(    r ) 2
145  G                                                             (2)
8(    r )

146  The variation of the dimensionless equivalent stress with the relative position of
147  calculated point is also shown in Fig. 3b. For the loading angle 2α=25°, the center of
148  the FBD specimen has the maximum value of σG than any point elsewhere, which
149  indicates that the crack is most likely to initiate at this center point.
150  Since the failure occurs when σG = σt, the coefficient k can thus be calculated as:

 G  ( DB)
151  k                                        (3)
2Pf

152  Substitute σG from Eq. (2) into Eq. (3), the value of k can be determined as 0.9445
153  when 2α=25°; this calibrated value is in good agreement with the value of k=0.9425
154  calculated via analysis method proposed by Wang et al21. Hence, for the FBD
‐ 6 ‐ 
 
155  specimen with a loading angle of 25°, the indirect tensile strength σt can be
156  determined using Eq. (4):
2 Pf
157   t  0.9445                                                             (4)
 DB

158 

159  Fig. 3 (a) Meshing schematic of the FBD specimen for finite element analysis, and (b) σθ, σr and
160  σG at the vertical diameter of FBD specimen calculated with ANSYS

161  III. Experimental setup and scheme

162  A. Test equipment and specimen preparation

163  The MTS-322 Rock and Concrete Test System shown in Fig. 4a are employed for
164  our tests. It is composed of a compression loading frame, an axial dynamic loading
165  system and a data acquisition system. The loading system has a 500 KN compression
166  load capacity. Specifically, for cyclic tests, the loading frequency can be varied in the
167  range of 0.01 to 3 Hz and the loading waveform can be selected from sinusoidal, ramp
168  and square waveforms, etc. The loading mode can be selected by automated dynamic
169  control mode switching between any connected transducers. Via the data acquisition
170  system consisting of signal conditioning and acquisition units, the axial load and
171  displacement of rock specimens can be automatically recorded during the entire tests.

172  Rock cores with a normal diameter of D=50 mm are first drilled from a rock block
173  of Neijiang sandstone (from Neijiang region of Sichuan province, China) and then
174  sliced into disc specimen with an average thickness of B=20 mm. All these specimens
175  are polished to a surface roughness better than 0.4% of the average thickness. Two
‐ 7 ‐ 
 
176  parallel flat ends of equal width are machined on the disc circumference; then the
177  complete disc specimens are fabricated to FBD specimens; the prepared virgin FBD
178  specimens are shown in Fig. 4b. In our tests, the contact surfaces between the loading
179  device and tested specimen are lubricated with high vacuum grease to reduce friction.

180 
181  Fig. 4 (a) The MTS-322 Rock and Concrete Test System, and (b) the prepared virgin FBD
182  specimens

183  B. Test schemes and procedure

184  Quasi-static indirect tensile tests on FBD specimens are first performed to obtain
185  quasi-static tensile strength as a reference for further cyclic loading parameters. Four
186  FBD specimens are used to perform the quasi-static tensile tests in an axial
187  displacement control mode with a strain rate of 5×10-5 s-1, and the indirect tensile
188  strengths are calculated using Eq. (4). For cyclic tensile tests, a periodic sinusoidal
189  waveform is specified with a loading path shown in Fig. 5, where Pmax, Pmin and Pave
190  are the maximum cyclic load, the minimum cyclic load and the average cyclic load,

191  respectively, P  ( Pmax  Pmin ) / 2 is the cyclic load amplitude, T is the loading

192  period, and the cyclic frequency F is defined as F=1/T. Initially, the axial load
193  gradually increases from 0 to Pave in an axial load control mode with the rate of 0.1
194  KN/s, and then cyclic loading and unloading processes are performed following
195  specified loading parameters. In our study, the cyclic loading was applied with a fixed
196  maximum load level of 0.9 (i.e., the ratio of the maximum cyclic load to the failure
197  load in quasi-static indirect tensile test) and a fixed amplitude level of 0.5 (i.e., the
198  cyclic amplitude to the failure load in quasi-static indirect tensile test). Three loading
‐ 8 ‐ 
 
199  frequencies are selected, i.e., 0.1, 0.5 and 1 Hz, to investigate the effect of cyclic
200  loading frequency on the tensile fatigue properties of brittle rocks.

201 
202  Fig. 5 A sketch of the load path and the characteristic parameters of cyclic loading

203  IV. Results and discussion

204  A. Results of quasi-static indirect tensile tests using FBD specimens

205  The  quasi-static indirect tensile tests are performed on four FBD specimens to
206  determine the quasi-static indirect tensile strength, providing a reference loading
207  parameter for subsequent cyclic tensile tests. Figure 6 shows the load-displacement
208  curves of tested specimens for the quasi-static tensile tests. The failure process of
209  FBD specimen consists of two stages. Stage I is the elastic deformation stage, and this
210  stage ends with the peak stress, i.e., point a, where the primary crack initiates.
211  Subsequently, the tensile stress drops to the local minimum stress; this stage
212  represents the unstable crack development. The details of experimental results are
213  summarized in Table 1, in which the quasi-static tensile strength is calculated
214  according to Eq. (4). The measured average tensile strength of tested FBD rock
215  specimens is 7.40 MPa. Figure 7 depicts that recovered FBD specimens from the
216  quasi-static tensile tests. A clear splitting crack is evident in all specimens, featuring a
217  tensile failure mode for a valid Brazilian-type rock test.

‐ 9 ‐ 
 
218 
219  Fig. 6 Load-displacement curves of FBD specimens obtained from quasi-static indirect tensile
220  tests

221 
222  Fig. 7 Failure modes of four FBD specimens recovered from quasi-static indirect tensile tests

223  B. Results of cyclic flattened Brazilian disc tests

224  In this section, cyclic tensile tests with three loading frequencies are conducted on
225  FBD rock specimens and the influence of cyclic frequency on the tensile fatigue
226  properties has been reported. For each loading frequency, three FBD specimens are
227  tested, and the representative results are presented and discussed herein.

228  Figures 8a-c shows the representative fatigue load-displacement curves of testing
229  FBD specimens under cyclic tensile loadings with a fixed maximum load level of 0.9,
230  a fixed amplitude level of 0.5 and three frequencies (i.e., 0.1, 0.5 and 1 Hz). These
231  curves depict similar evolution law. For demonstration, the test on specimen F5-2 is
232  chosen for detailed analysis. The development of the axial displacement in this FBD
‐ 10 ‐ 
 
233  specimen under cyclic tensile loadings can be divided into four stages, namely the
234  initial, unstable, stable and accelerated stages. As shown in Fig. 8d, in the initial stage,
235  the hysteresis loop in dynamic curves distributes sparsely and the axial deformation
236  increases quickly, which is likely to be induced by the closure of internal pores and
237  the hardening of specimen; this stage ends with the crack initiation on the surface of
238  tested specimen. Subsequently, the axial deformation suddenly increases, and there is
239  no hysteresis loop generated in this stage, probably representing an unstable crack
240  development. After the second stage, the distribution of hysteresis loop in fatigue
241  curves becomes dense, and the axial deformation increases stably; there is almost no
242  change of the formed surface crack in this stable stage. Eventually, the distribution of
243  hysteresis loop is looser and looser, and the increase of axial deformation is
244  accelerated until the fatigue failure occurs. In this accelerated stage, we interpret that
245  the formed crack thoroughly coalesces and leads to the structural fatigue failure of
246  tested specimen. The cyclic load-deformation characteristics obtained in our
247  experiments reflects the fatigue failure processes of brittle rocks.

248 

249  Fig. 8 Representative load-displacement curves of FBD specimens subjected to indirect cyclic
250  tensile loadings with different frequencies

251  Table 2 summarizes the details of our experimental results. Comparing the axial
252  displacement at the crack initiation point, i.e., point a, in quasi-static tensile tests
‐ 11 ‐ 
 
253  (Table 1) and cyclic tests, the deviations between them are less than 8%. It can thus be
254  concluded that the fatigue deformation of FBD specimen under cyclic tensile loadings
255  is dominated by its quasi-static deformation behavior. A similar fatigue deformation
256  scenario was also revealed for rocks under cyclic uniaxial compression in previous
257  literatures 29-30.

258  Figure 9 shows the representative recovered FBD specimens subjected to cyclic
259  tensile loadings with different loading frequencies. Dramatically differing from the
260  failure mode under quasi-static tensile tests shown in Fig.7, there are more crushed
261  rock fragments in these cyclically failed FBD specimens. With decreasing loading
262  frequency, the recovered FBD specimen is more fragmented. Figure 10 illustrates the
263  tensile fatigue life (i.e., the number of cycles to fatigue failure) of tested specimens
264  and its dependence on the loading frequencies. As the loading frequency increases
265  from 0.1, 0.5 to 1 Hz, the fatigue life of specimens increases exponentially from 19,
266  38 to 85 cycles. We interpret that such influences of loading frequency on fatigue
267  tensile responses of FBD specimens may be explained by considering the duration of
268  cyclic loading. Under a lower loading frequency, the variation of cyclic loads is
269  slower than that at a higher frequency, and thus the duration of loads acting on the
270  specimens is longer. Consequently, there is more sufficient time for micro-cracks to
271  develop and coalesce. As a result, fewer cycles are required to break the specimen and
272  more crushed rock fragments are produced in cyclically failed specimens.

273 
274  Fig. 9 Representative fatigue failure modes of FBD specimens recovered from cyclic tensile
275  tests with different loading frequencies: (a) 0.1Hz, (b) 0.5Hz and (c) 1Hz

‐ 12 ‐ 
 
276 

277  Fig. 10 Influences of cyclic loading frequency on the tensile fatigue life of FBD specimens

278  V. Conclusions

279  In this study, for the first time, the cyclic tensile testing method using FBD
280  specimen is proposed to measure the tensile fatigue properties of brittle rocks, and the
281  influences of cyclic loading frequency on fatigue properties are investigated. Our new
282  method has certain distinct advantages than previous cyclic tensile test methods. First,
283  the specimen preparation and experimental operation are simpler, and no special
284  complicate loading devices are required. Second, two introduced flattens facilitate
285  easily loading the specimen, and the contact region between the tested specimen and
286  machine barely change during the entire loading and unloading process, guaranteeing
287  a consistent loading boundary. Third, the key prerequisite of indirect tension tests, i.e.,
288  the central crack initiation can be satisfied by applying an appropriate loading angle.
289  Our experiments demonstrate that this proposed method can obtain tensile fatigue
290  characteristics of brittle rocks under cyclic loadings. Different from the failure mode
291  under quasi-static tensile loading, the fatigue failure of FBD specimens under cyclic
292  tensile loadings is more sudden and catastrophic with more fragments. In addition, the
293  loading frequency significantly affects the fatigue tensile properties of rocks. With
294  increasing frequency, the tensile fatigue life increases, and the rock specimen is less

‐ 13 ‐ 
 
295  fragmented. Our proposed cyclic FBD testing method provides a convenient and
296  reliable way to measure the tensile fatigue properties of rocks and other brittle solids
297  subjected to cyclic tensile loadings.

298  Acknowledgments

299  The authors are grateful for the financial support from the National Program on
300  Key Basic Research Project (No. 2015CB057903) and National Natural Science
301  Foundation of China (No. 51374149).

302  References

1
303  M.N. Bagde and V. Petroš, Int. J. Rock Mech. Min. Sci. 46, 200 (2009)
2
304  F. Dai, K.W. Xia and S.N. Luo, Rev. Sci. Instrum. 79, 123903 (2008)
3
305  J.A. Hudson, Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 6, 91 (1969)
4
306  M. Mellor and I. Hawkes, Eng. Geol. (Amsterdam) 5, 173 (1971)
5
307  Z. Bieniawski and I. Hawkes, Int. J. Rock Mech. Min. Sci. 15, 99 (1978)
6
308  A. Coviello, R. Lagioia and R. Nova, Rock Mech. Rock Eng. 38, 251 (2005)
7
309  ISRM, Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 15, 99 (1978)
8
310  ASTM, In Annual Book of ASTM, Standards, Philadelphia: ASTM 42, 336 (1984)
9
311  G.E. Andreev, Min. Sci. Technol. 13, 445 (1991)
10
312  G.E. Andreev, Min. Sci. Technol. 13, 457 (1991)
11
313  C.S. Chen, E. Pan and B. Amadei, Int. J. Rock Mech. Min. Sci. 35, 43 (1998)
12
314  J. Claesson, and B. Bohloli, Int. J. Rock Mech. Min. Sci. 39, 991 (2002)
13
315  N. Erarslan, Z.Z. Liang and D.J. Williams, Rock Mech. Rock Eng. 45, 739 (2012)
14
316  H. Guo, N.I. Aziz and L.C. Schmidt, Eng. Geol. 33, 177 (1993)
15
317  Z. Zhao, J. Yang, D. Zhang and H. Peng, Rock Mech. Rock Eng. 50, 485 (2017)
16
318  J.A. Hudson, E. T. Brown and F. Rummel, Int. J. Rock Mech. Min. Sci. 9, 241
319  (1972)
17
320  Y. Yu, J.X. Zhang and J.C. Zhang, Int. J. Rock Mech. Min. Sci. 46, 421 (2009)
18
321  J.J. Swab, J. Yu, R. Gamble and S. Kilczewski, Int. J. Fract. 172, 187 (2011)
19
322  D.Y. Li, and L.N.Y. Wong, Rock Mech. Rock Eng. 46, 269 (2013)
‐ 14 ‐ 
 
 20
323  Q.Z. Wang, and X. Lei, Eng. Fract. Mech. 64, 193 (1999)
21
324  Q.Z. Wang, X.M. Jia, S.Q. Kou, Z.X. Zhang and P.A. Lindqvist, Int. J. Rock Mech.
325  Min. Sci. 41, 245 (2004)
22
326  N.T. Burdine, Soc. Petr. Eng. J. 3, 1 (1963)
23
327  P.B. Attewell and I.W. Farmer, Int. J. Rock Mech. Min. Sci. 10, 1 (1973)
24
328  M.N. Bagde and V. Petroš, Int. J. Rock Mech. Min. Sci. 42, 237 (2005)
25
329  A. Momeni, M. Karakus, G.R. Khanlari and M. Heidari, Int. J. Rock Mech. Min.
330  Sci. 77, 89 (2015)
26
331  N. Erarslan and D.J. Williams, Rock Mech. Rock Eng. 45, 327 (2012)
27
332  K.L. Johnson, Cambridge University Press, 1985
28
333  A.A. Griffith, Proceedings of the First International Congress on Applied
334  Mechanics, Delft: 55 (1924)
29
335  J.Q. Xiao, D.X.  Ding, F.L. Jiang and G. Xu, Int. J. Rock Mech. Min. Sci. 47, 461
336  (2010)
30
337  Y. Liu, F. Dai, P.X. Fan, N.W. Xu and L. Dong, Rock Mech. Rock Eng. 50, 1453
338  (2017)
339 

‐ 15 ‐ 
 
340  Table 1 Results of quasi-static indirect tensile tests on FBD specimens

Specimen Failure load Quasi-static tensile Axial displacement at

no. (KN) strength (MPa) point a (mm)
T1 12.72 7.61 0.29
T2 12.38 7.41 0.28
T3 11.90 7.12 0.28
T4 12.43 7.44 0.29
Average 12.36 7.40 0.29
341   

342  Table 2 Results of cyclic flattened Brazilian disc tests

Specimen Loading Cyclic tensile Axial displacement Deviation Tensile

no. frequency (Hz) strength (MPa) at point a (mm) (%) fatigue life
F0-1 0.1 6.66 0.30 3.45 19
F5-2 0.5 6.66 0.27 6.90 38
F1-2 1 6.66 0.29 0.00 85
343   
344   
345 

‐ 16 ‐ 
 
346  Captions

347  Fig. 1 Representative loading configurations of Brazilian-type tensile tests: (a) flat
348  loading platen, (b) flat loading platen with two small-diameter steel rod, (c) flat
349  loading platen with two cushions, (d) curved loading jaw and (e) the flattened
350  Brazilian disc test

351  Fig. 2 Schematic of the cyclic flattened Brazilian disc tests and the geometry of the
352  FBD specimen

353  Fig. 3 (a) Meshing schematic of the FBD specimen for finite element analysis, and (b)
354  σθ, σr and σG at the vertical diameter of FBD specimen calculated with ANSYS

355  Fig. 4  (a) The MTS-322 Rock and Concrete Test System, and (b) the prepared virgin
356  FBD specimens

357  Fig. 5 A sketch of the load path and the characteristic parameters of cyclic loading

358  Fig. 6 Load-displacement curves of FBD specimens obtained from quasi-static

359  indirect tensile tests

360  Fig. 7 Failure modes of four FBD specimens recovered from quasi-static indirect
361  tensile tests

362  Fig. 8 Representative load-displacement curves of FBD specimens subjected to

363  indirect cyclic tensile loadings with different frequencies

364  Fig. 9 Representative fatigue failure modes of FBD specimens recovered from cyclic
365  tensile tests with different loading frequencies: (a) 0.1Hz, (b) 0.5Hz and (c) 1Hz

366  Fig. 10 Influences of cyclic loading frequency on the tensile fatigue life of FBD
367  specimens

‐ 17 ‐ 
 
Flat loading platen
Steel rod Cushion

(a) (b) (c)

Flatten

2α
Curved loading jaw

2α

(d) (e)
Pc P

D
2α 2α

B
2α 2α

Pc P
 2 1.0

σθ
Dimensionless stress component

Dimensionless equivalent stress

0 0.8
2P/πDB
R
r -2 σG 0.6
2P/πDB
-4
σr 0.4
2P/πDB

-6 Dimensionless stress component 0.2

Dimensionless equivalent stress
-8 0.0
0.0 0.2 0.4 0.6 0.8 1.0
Relative distance r/R
(a) (b)
(a) (b)
Load
Pmax

Pave △P

Pmin
T
Time
 14.0
T1 a: Crack initiation
T2
10.5 T3
T4
Load (KN)

7.0

3.5

0.0
0.0 0.1 0.2 0.3 0.4
Displacement (mm)
 12 12
Frequnency : 0.1 Hz Frequnency : 0.5 Hz

9 9

Load (KN)
Load (KN)

6 6

3 3

0 0
0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00
Deformation (mm)
Displacement (mm) Deformation (mm)
Displacement (mm)

(a) (b)
12 16
Frequnency : 1 Hz Unstable stage
a: Crack initiation b: Failure point
9 12
Initial stage
Load (KN)

Load (KN)

6 8

3 4

Stable stage Accelerated stage

0 0
0.00 0.25 0.50 0.75 1.00 0.0 0.2 0.4 0.6
Deformation
Displacement(mm)
(mm) Deformation (mm)
Displacement (mm)

(c) (d)
(a) (b) (c)
 100

75
life life

1.52F
fatigue

N = -3.63+ 19.45e
Fatigue

50
Tensile

25

0
0.0 0.4 0.8 1.2
Frequency(Hz)
Frequency (Hz)