Proceedings of the URECA@NTU 2014-15

High Temperature Impact on Brazilian Tensile Strength of Singapore

Rocks
Chee Khai Yin

Asst Prof Wong Ngai Yuen Louis

Mr Li Zhihuan

School of Civil and Environmental

School of Civil and Environmental

Engineering

Engineering

Abstract Granite being a common concrete aggregate

used in buildings such as nuclear reactors and pavement
is also commonly encountered in tunneling operations
during the construction of underground structures.
Being one of the most abundant rocks types on earth, it
is cheap as a construction material. However, under
prolonged heating (thermal stress) or extreme heat
(which may be encountered in the case of overheating
of nuclear reactors and also event of tunnel fire), the
strength of granite will tend to decrease which may
increase the likelihood of structural failure leading to
collapse and damages to properties as well as casualties.
In this study, the main objective is to establish a
relationship between the Brazilian tensile strength of
granite and heating temperature. The color change of
granite before and after heating for different cooling
rate will also be investigated to assess the impact of
cooling rate on the strength of granite. This will then
provide a medium to establish a correlation between
color change and tensile strength of granite which can
be used to develop a non-destructive testing method to
estimate the strength of granite based on color change
under a short period of time and produce a reliable
result for the purpose of field testing.

Theoretically, the tensile strength of a rock can be

obtained by subjecting the rock to two types of tests
which are the uniaxial tensile test (direct) and the
Brazilian test (indirect). Conventionally, the Brazilian
test is used more widely in this field of study as the
costs are relatively lower than the uniaxial direct tensile
test. Research done on clays have shown that x-rays
could be used to study the thermal expansion of mineral
grains of clay in different lattice directions [2].

Keywords - Granite, Brazilian tensile strength,

heating temperature, color change, cooling rate

Rocks which are exposed to high temperature tend to

produce microcracks due to differential thermal
coefficient of expansion of different minerals
constituting that particular rock and the number of
microcracks that appears is proportional to the
subjected heating temperature [4].

1 Introduction
The existence of micro cracks in rocks particularly
granite rocks after heating (due to differential
coefficient of thermal expansion and contraction) could
significantly influence the tensile strength of the rock
especially in large rocks where thermal gradient is high.
The effect of temperature on mechanical properties of
granite which are abundant especially in tunnels have
raised concerns on the possibility of tunnel collapse due
to the occurrence of tunnel fire which could be caused
by logistic problems (Mont-Blanc tunnel fire) [1].
Rocks are particularly weaker in tension compared to
compression, thus, by evaluating the tensile strength of
granite, a benchmark could be set on the maximum load
a tunnel could support in the event of rock failure under
high temperature.

This method serves as non-destructive testing on the

effect of temperature mechanical properties of clay and
as a mean to determine the expansion coefficient of
various types of clays. Studies done by Mao et al have
shown that sandstone exhibits increasing peak strength
during heating and sudden drop in strength after a
certain temperature is reached measured by methods of
ultrasonic scanning [3]. This anomaly was explained by
the sudden closure of micro cracks as the grains in the
sandstone expand during heating [3]. Therefore, these
factors have to be taken into account when tensile test is
conducted on granite.

2 Theory

This phenomenon can be explained by the means of

thermal gradient theory. Rocks which are subjected to
greater temperature tend to experience a higher thermal
gradient. Therefore the constituent mineral tends to
expand more which causes a greater thermal stress and
eventually more microcracks will be formed [5].
Further, rocks which are subjected to heating will show
a change in color. Rocks which are heated at a greater
temperature will show pronounced change in color.
These color changes can be quantified by carrying out
the L*a*b* color analysis through MATLAB. By
correlating these two features (color change and tensile
strength) through several Brazilian Test and L*a*b*

Proceedings of the URECA@NTU 2014-15

color analysis, a non-destructive test can be developed
which relates color change and tensile strength of rock.

3 Literature Review
The chemical composition of Bukit Mandai granite
used in this experiment consists of 72% silica, 15%
alumina, 5% potassium oxide, 4% sodium oxide and
2% calcium oxide [6]. The mineralogy of the granite
used consists of quartz, alkali feldspar and plagioclase
feldspar.
The different minerals in granite with different thermal
coefficient of thermal expansion causes the
phenomenon of thermal stress; this effect becomes more
severe with increasing heating temperature used. The
different rate at which minerals expand during heating
poses a physical constraint to each other and the
mineral with a lower Young Modulus will tend to
rupture prematurely which causes microcracks to form
around the granite, hence lowering its tensile strength
[7].
Previous researches done on color change of rock upon
heating could be due to mineralogy and wet surfaces on
rocks due to total internal reflection phenomenon [8]. In
this research, the presence of feldspar in granite was
hypothesized to cause color change when heated [8].
L*a*b* analysis was used to record changes in L*, a*
and b* values upon heating and cooling at different rate
by using a MATLAB program [8].
Whereby L* and a*b* represents luminance and
chrominance respectively.
The Brazilian test is an economical method to test
indirect tensile strength of rock as minimal preparation
is required. The Brazilian test can be conducted by
using the MTS hydraulic servo-control testing system
where the load applied can be manually selected
through the MTS system interface. The Brazilian test
was conducted in accordance to the standards set by
American Society for Testing and Materials (ASTM)
whereby the thickness to diameter ratio should be 0.5
[9]. Two opposing normal loads are applied at the
specimen circumference. The loads should be increased
at a constant rate (varies for different specimen) until
specimen failure within the first few minutes. The
tensile stress is then calculated using equation (1) [9]:

t = 2P/LD
Where

The above equation was derived based on the theory of

elasticity for isotropic materials where the tensile stress
is perpendicular to the loaded direction. Hence, the test
is only valid if the initial crack formed during the
Brazilian test originates at the center of the specimen.

4 Methodology
4.1 Cutting granite rock core
Granite cores used were originated from Mandai (Fig.
1) with original length and diameter of around 35cm
and 5cm respectively (Fig. 2).

Fig. 1 Classification on the origin of granite cores

Fig. 2 Granite rock core

In accordance to the standards set by ASTM, specimens
with length to diameter ratio of 0.5 are required
(diameter and length measuring 5cm and 2.5cm
respectively) for the determination of indirect tensile
strength by Brazilian Test [9]. The rock was cut into
smaller segments with length slightly exceeding the
required length (3cm) by using BUEHLER rock cutter
machine (Fig. 3) to account for rough faces produced on
both sides of specimen by the blade (excess of length
was reduced during polishing process).

(1)

= maximum tensile strength (MPa)

= maximum load applied (N)

= thickness of the specimen (mm)

= diameter of the specimen (mm)

Fig. 3 BUEHLER rock cutter machine

Proceedings of the URECA@NTU 2014-15

Fig. 7 Specimen being held into position by metal bars

Fig. 4 Granite rock core being held into position for
cutting

The position of the blade prior to cutting, rotational

speed of the blade and downward displacement of the
blade are adjusted by buttons 3, 1 and 2 respectively,
button 4 was pressed to begin each cycle of cutting
(Fig. 5).
Fig. 8 Control panel of Chiao Bang Polisher

Fig. 5 Operating panel of BUEHLER rock cutter

machine

To correctly align the specimen with the blade prior to

polishing (Fig. 9), the x, y and z coordinate position of
the specimen were adjusted by adjusting lever 2 , lever
3 (Fig. 10) and panel 4 (Fig. 8).

4.2 Polishing both faces of granite specimen

Rough surfaces produced on both sides of the specimen
after the cutting process were removed to obtain the
specified dimensions (5cm diameter, 2.5cm length
according to the ASTM standards) by abrasive
polishing using the Chiao Bang polisher (Fig. 6). 4
metal bars were used to securely position the specimen
(Fig. 7) into place by applying magnetic field (knob 1,
Fig. 8).

Fig. 9 Position of the specimen relative to the blade

Fig. 10 Lever 2 (right) and lever 3 (left)

Fig. 6 Chiao Bang Polisher

The amount of grinding position was adjusted by using

panel 5 (Fig. 8), operation mode was set to abrasive
polishing by selecting button 6 (Fig. 8). The machine
was started by pressing buttons 7, 8 and 9 (water to
reduce friction during polishing) (Fig. 8). Buttons 10,

Proceedings of the URECA@NTU 2014-15

11 (twice) and 12 (Fig. 8) were pressed in sequence to
begin each cycle of polishing.

4.3 Photo taking process

Two light sources were set up on tripod stands with the
light sources tilted at 45 degree with respect to the
ground surface. The light sources were separated at 180
cm along a straight line from each other (Fig. 11) with
the rock specimen placed at the midpoint of the straight
line (90cm away from both light sources) on a clipboard
with contrasting color (Fig. 12).

Fig. 11 Set up of lighting apparatus for photo taking

purposes

constant height. The field of view of the camera is

aligned with points marked on the background with a
constant magnification.

4.4 Color analysis

Photos taken were transferred to a laptop for color
analysis using a MATLAB program. As shown in Fig.
14, the MATLAB program segments the background
from the specimen before obtaining the L*a*b* values
(average L*a*b* values of every pixels contained in the
specimen photo after segmentation). The L*a*b* values
obtained for each specimen were recorded in an excel
spreadsheet.

Fig. 14 Before (left) and after (right) photo of rock

specimen undergo segmentation process through
MATLAB program

Due to differences in photo quality between photos

taken from DSLR camera in an earlier similar study and
smart phone camera (difference in resolution, sensor
size and quality of camera glass material), the L*a*b*
values of specimen photos taken from smart phone
camera will tend to deviate from the DSLR values.
Fig. 12 Red clipboard used as contrasting background
for photo taking
The intensity of the light was selected to be level 1,
consistent for all rock specimens by turning the knob
clockwise once (Fig. 13).

A correction factor (multiplication method) is therefore

needed to account for such deviation and to establish a
correlation between L*a*b* values taken using DSLR
camera and smart phone camera. (Refer to equations
(2), (3) and (4))
L*DSLR = (F L*)(L*smart phone camera )
(2)
a*DSLR
b*DSLR

= (F a*)(a* smart phone camera )

= (F b*)(b* smart phone camera )

(3)
(4)

where F : Correction factor

Fig. 13 Lighting intensity levels for both light sources

Photos were then taken using a smart phone camera

held parallel to the top face of rock specimen at a

To obtain this correction factor, a sub-experiment was

conducted on a separate group of granite rock cores
(used in a previous study by using the DSLR camera)
by taking the photos of 12 such granite specimens.
These photos were fed to the same MATLAB program
to obtain their L*a*b* values. The average of these
L*a*b* values taken from smart phone camera were
obtained and compared with the average L*a*b* values
taken from the DSLR camera.

Proceedings of the URECA@NTU 2014-15

Comparison between the values obtained using different
cameras showed that the L*a*b* values obtained by the
smart phone camera are 14.67%, 8.51% and 25.02%
larger than those obtained by the DSLR camera
respectively. b* obtained from smart phone camera
showed highest deviation from b* value obtained using

DSLR camera. The correction factors of F L*, F a* and F

b* for L*, a* and b* values are 0.8720, 0.9216 and
0.7999 respectively (Table 1).

Table 1 L*a*b* values obtained using smart phone camera and DSLR camera respectively for correction factor
calculation
Specime
n No.
1
2
3
4
5
6
7
8
9
10
11
12
Average

Smart Phone Camera

L*
a*
b*
185.87
139.07
169.26
185.12
139.02
171.48
186.94
138.2
170.1
183.63
139.2
170.89
190.07
137.68
166.55
189.73
138.15
169.78
189.95
137.77
169.88
191.45
136.68
165.78
190.98
136.96
165.4
187.3
139.21
170.73
187.3
137.74
166.09
182.8
138.54
169.38
187.60
138.19
168.78

DSLR Camera
L*
a*
b*
176.25
127.21
135.32
151.99
127.33
134.41
140.36
127.68
134.22
158.33
127.45
136.08
172.91
127.29
135.11
157.92
127.35
134.1
169.75
127.2
134.76
165.16
127.29
135.2
166.17
127.35
135.33
168.53
127.36
135.19
170.16
127.29
135.21
165.54
127.37
135.08
163.59
127.35
135.00

4.5 Heating Process

The temperature variables for this experiment are 200
C, 350 C and 500 C. There are 60 specimens of
which 54 were used for the heating process and the
remaining 6 specimens were used as control specimens
(not heat treated). 18 specimens were used for each
temperature variable. Carbolite furnace (Fig. 15) was
used for heating the specimens. The heating rate for the
furnace was programmed to be 100 C per minute and
heating period of 30 minutes (which includes
temperature dwelling) by adjusting the operating panel,
the same programmed heating rate was used for all 3
temperature variables (Fig. 16). Prior to heating, the
specimens were oven dried at 40 C for 3 days (Fig. 17)
to make sure there are no moisture contained in
specimen air voids (which will cause explosion of
specimen due to the vapor pressure created in the air
voids found in the specimen).

Fig. 15 Carbolite furnace

Fig. 16 Control panel of Carbolite furnace

Fig. 17 Oven for drying specimen

A mini experiment was conducted by heating 4 granite
specimens (discarded specimens) at the maximum
temperature (500 C) to make sure the specimen does
not disintegrate under high temperature. Each
temperature variable was tested by heating each batch
of specimens (18 specimens) once for 30 minutes at the
fixed heating rate. During heating, the time taken for
the temperature in the furnace to reach each of the
temperature variable was recorded. The temperature
incremental rate in the furnace was calculated by
applying equation (5). Data obtained were recorded in
table 2. The average temperature incremental rate was

Proceedings of the URECA@NTU 2014-15

calculated to be 37.173 C per minute compared to the
programmed heating rate of 100 C per minute.
(Temperature variable 27) / Time taken to reach
temperature variable
(5)
Table 2 Time taken to reach the desired temperature
variable and temperature incremental rate in the furnace

Temperature
(C)
200
350
500

Time taken to
reach
temperature
variable
(minute)
4.1
8.5
15.1

Temperature
incremental rate
(C per minute)
42.195
38
31.325

4.6 Cooling process

Upon taken out from the oven, the surface temperature
of specimens were measured for all 3 heating
temperature and was found to be 30 C lower than the
targeted temperature, these temperatures were used in
the analysis of results. 3 types of cooling rates were
chosen for this experiment (slow, average, fast). From a
batch of the same heating temperature, 18 specimens
were equally divided into 6 specimens (for each cooling
rate) to be cooled at 3 different rates. For both the
average and fast cooling rate, silicon mats were used to
insulate heat from the tray (conducted from the
specimen) to prevent any heat related damage on the
table or floor which was used to place the tray
containing the specimens. Specimens for slow cooling
rate were left to cool in the furnace, specimens for
average cooling rate were left to cool under ambient
condition (room temperature) and specimens for fast
cooling rate were fan cooled (Fig. 18).

4.7 Photo taking and color analysis

Steps 3 and 4 were repeated with the heated specimen.
Average L*a*b* changes for heating temperature alone
and heating temperature with cooling rate were
calculated and recorded in an Excel spreadsheet.

4.8 Brazilian tensile test

Prior to loading the granite specimens to failure by
using MTS uniaxial compression machine, the
specimens were kept in an oven heated to 40 C for 7
days to ensure pore water in the specimens are
completely vaporized. The Brazilian test was conducted
on 27 specimens with 9 specimens for each heating
temperature, the remaining 33 specimens were set aside
as back up. The 27 specimens were loaded at 0.2kN/s
according to the ASTM standards until failure. The
MTS program records the axial force at failure for each
specimen; the axial force at failure was interpreted to be
the force just before the first steep drop in axial force
(Fig. 19). High speed camera was used to detect the
initial crack development. The video captured was
transferred to Phantom Camera Control for editing.
Only specimens which shows first crack initiation at the
middle portion were considered as valid test.

Fig. 19 Selection of axial force at failure for one of the

specimen tested.

5 Analysis of results & Discussion

Fig. 18 (From left) Arrangement of specimen and
apparatus for slow, average and fast cooling rates
An infrared temperature gun was used to periodically
monitor the temperature of each specimen. The
specimens were kept and stored once the temperature
reaches around 30 C.

5.1 Color change of specimen versus

heating temperature
Fig. 20 shows the color change of granite after
undergoing different heating temperature. From the
figure, it can be seen that both L* and b* increases as
the heating temperature with a high positive correlation.
Meanwhile a* shows decrement as the heating
temperature increases with a high negative correlation.
From the figure, it can also be noted that L* shows a
more significant change in value when the heating
temperature increases compared to a* and b* values

Proceedings of the URECA@NTU 2014-15

because L* is more prone to changes in lighting
conditions of the set up whereas the chrominance values
are less susceptible to changes in lighting conditions. As
seen in Fig. 20, L* value showed the highest increase,
feldspar being one of the major component in granite
mineralogy tend to become lighter in color (whiter)
upon heat treatment even at relatively low temperature
[10]. a* and b* values (chrominance) showed negligible
change as the other mineral component of granite
requires higher temperature beyond 600 C to show
significant change in color (by which then the granite
will disintegrate) [11-12].

(b)

(c)
Fig. 20 Percentage change in L*a*b* values against
heating temperature

5.2 Color change of specimen versus cooling

rate
Physically, the color change of granite is only visible
for specimen heated above 320 C and no noticeable
color change is observed by naked eyes for different
cooling rates. Fig. 21 shows the color change of
specimen at different cooling rates obtained with a
MATLAB program. From the figure, it can be seen that
L*a*b* values fluctuate for different cooling rates for
all heating temperatures and the percentage changes in
L*a*b* values did not vary by a large margin (less than
~5%) for different cooling rates. The rate of cooling has
negligible effect on color change as it only affects
crystal size and mineral composition [13], both of
which does not alters the individual mineral properties
(i.e. color of individual minerals).

Fig. 21 Percentage change in color versus cooling rate

for heating temperature of (a) 170 C (b) 320 C (c) 470
C

5.3 Brazilian tensile strength

Snapshots obtained using high speed camera and
Phantom Camera Control software shows that all the
specimens were valid tests in which the first crack
propagation originates from the center portion of the
granite (Fig. 22) [9].

Fig. 22 Crack propagation which forms at the center of

one of the granite specimen

(a)

The axial force before the first steep drop was used in
the computation of the indirect tensile strength. The
average of the indirect tensile strength for reach heating

Proceedings of the URECA@NTU 2014-15

temperature is taken and plotted against the heating
temperature. From Fig. 23, the indirect tensile strength
decreases when the heating temperature increases. This
decreasing trend could be explained by the evolution of
thermal cracks in the granite or extension and
connection between existing microcracks in the rock
structure, both of which will weaken the granite (lower
tensile strength). It is also noted that as the heating
temperature increases progressively, the rate of decline
of tensile strength decreases. As the temperature
increases the granite minerals will expand and
temporarily closes the gap formed by thermal cracks
[3], slowing the decrease in tensile strength.

Acknowledgement
I wish to acknowledge the funding support for this
project from Nanyang Technological University under
the Undergraduate Research Experience on CAmpus
(URECA) programme.
I would like to extend my gratitude to my project
supervisor, Asst Prof Wong Ngai Yuen Louis, who has
provided guidance throughout the project and providing
ideas as well as feedbacks as a mean to further improve
this research project.
I would also like to thank Mr Li Zhihuan and Mr Soh
Wei Jie who have provided excellent technical
assistance during this research project as well as
providing advices and support throughout this project.

References

Fig. 23 Indirect tensile strength of granite specimen

against heating temperature

[1]

C. Bailey, Case Studies: Historical Fires:

Mont Blanc Tunnel Fire, Italy/France, n.d. ;
http://www.mace.manchester.ac.uk/project/rese
arch/structures/strucfire/CaseStudy/HistoricFir
es/InfrastructuralFires/mont.htm.

[2]

H. A. McKinstry, Thermal Expansion of Clay

Minerals, The American Mineralogist, Vol.
50, pp. 212- 222, January-February 1965.

[3]

X. Mao, L. Zhang, R. Liu, D. Ma, Mechanical

and Thermal Damage Properties of Sandstone
at High Temperature, Electronic Journal of
Geotechnical Engineering, Vol. 19, pp. 31373150, 2014.

[4]

H. W. Cooper, G. Simmons, The Effect of

Cracks on Thermal Expansion of Rocks,
Earth and Planetary Science Letters, Vol. 36,
pp. 404-412, 1977.

[5]

T. Yin, X. Li, W. Chao, K. Xia, Effects of

Thermal Treatment on Tensile Strength of
Laurentian Granite using Brazilian Test, Rock
Mech Rock Eng, pp. 1-11, 01 February 2015.
doi: 10.1007/s00603-015-0712-3

[6]

H. Blatt, R. J. Tracy, Petrology, 2nd. Ed., W.

H. Freeman, pp. 66, 1997.

[7]

C. Saiang, Influence of Heat on the Physical

and Mechanical Properties of Selected Rock
Types, Fredrag vid Bergmekanikdag i
Stockholm, 12 mars 2012. Stockholm:
Stiftelsen bergteknisk forskning - Befo. 2012,
pp. 15-22, 12 March 2012. (Bergmekanikdag
Stockholm)

[8]

M. Moriizumi, S. Nakashima, S. Okumura, Y.

Yamanoi, Color-Change Processes of a
Plinian Pumice and Experimental Constraints

6 Conclusions
This study shows that the tensile strength of granite
decreases with increase in heating temperature.
Heterogeneities in the mineralogy of granite affects the
expansion of individual mineral to such an extent that it
causes microcrack to extend across the structure of
granite which weakens the granite. A positive
correlation is established between color change and
heating temperature in this project. Thus, a relationship
between color change and indirect tensile strength can
be further derived to serve as a non-destructive test for
the purpose of field test.
Due to limited resources in this study, only granite from
Bukit Mandai source was used. As the geological
formation varies for different regions, the physical and
chemical properties of granite can also vary. Therefore
further studies can be conducted for granite from
different regions. Future studies can also include the
effect of degree of moisture content on strength of
granite to account for the possibility of groundwater.
Other forms of failure such as bearing capacity and
compression strength should also be looked into.

Proceedings of the URECA@NTU 2014-15

of Color-Change Kinetics in Air of an
Obsidian, Bull Volcanol (2009) 71:113, pp.
1-13, 14 March 2008. doi: 10.1007/s00445008-0202-5
[9]

ASTM D3967, 2008. Standard Test Method

for Splitting Tensile Strength of Intact Rock
Core Specimens.

[10]

J. E. Arem, Gem Feldspars and Feldspar

Treatments, April 2011;
http://www.jewelersethicsassociation.com/docs
/GEM%20FELDSPARS.pdf.

[11]

D. B. Clarke, The Mineralogy of

Peraluminous Granite: A Review, Canadian
Mineralogist, Vol. 19, pp. 3-17, 1981.

[12]

H. F. Wang, B. P. Bonner, S. R. Carlson, B. J.

Kowallis, H.C. Heard, Thermal Stress
Cracking in Granite, Journal of Geophysical
Research, Vol. 94, No. B2, pp. 1745-1758, 10
February 1989.

[13]

I. Sunleaf, M. Cotter, M. Booth, The Cooling

Rates of Two Dikes from Granite Falls. MN,
Fall 2012.