You are on page 1of 56

Accepted Manuscript

The effect of mineralogical parameters on the mechanical


properties of granitic rocks

S. Cowie, G. Walton

PII: S0013-7952(17)30747-0
DOI: doi:10.1016/j.enggeo.2018.04.021
Reference: ENGEO 4832
To appear in: Engineering Geology
Received date: 3 June 2017
Revised date: 21 March 2018
Accepted date: 27 April 2018

Please cite this article as: S. Cowie, G. Walton , The effect of mineralogical parameters
on the mechanical properties of granitic rocks. The address for the corresponding author
was captured as affiliation for all authors. Please check if appropriate. Engeo(2017),
doi:10.1016/j.enggeo.2018.04.021

This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
ACCEPTED MANUSCRIPT

TITLE: The effect of mineralogical parameters on the mechanical properties of granitic rocks

AUTHORS: *Cowie, S. and Walton, G.


Department of Geology and Geological Engineering, Colorado School of Mines, Golden, Colorado

*Corresponding Author E-mail: seanrcowie@gmail.com

T
IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

ABSTRACT

The mechanical properties of a given rock volume are dependent on its geological composition.
Being able to predict these mechanical properties based on mineralogical parameters could prove
useful in engineering and construction settings. The aim of this study is to propose relationships
between mineralogical parameters and mechanical properties of granitic rocks. Furthermore, relevant
correlations proposed previously in literature were analyzed within the context of a larger dataset.
An extensive literature search was performed to compile a database of laboratory compression testing

T
data for granitic rock. Mineralogical parameters and mechanical properties of 58 distinct granitic rock

IP
types from 12 countries were analyzed in order to obtain a more complete view of their

CR
interrelationships. The following geological properties were compiled: grain sizes (overall and
mineral-specific), modal percentages of minerals (quartz, alkali feldspar, plagioclase, and micas),
density, and average Mohs Hardness. The following mechanical properties were assessed:

US
Unconfined Compressive Strength (UCS), Brazilian Tensile Strength (BTS), crack initiation stress
(CI), crack damage stress (CD), Young’s Modulus (E), Poisson’s ratio (ν), and the Hoek-Brown
AN
constant, mi . The modal percentage of mica minerals was found to have a strong negative correlation
with crack initiation stress, crack damage stress, and tensile strength. Additionally, minimum overall
M

grain size is strongly correlated with the CD/CI ratio. A multivariate regression analysis utilized
various combinations of five independent variables (minimum overall grain size, average overall
ED

grain size, standard deviation of overall grain size, quartz modal percentage, and mica modal
percentage) to predict CI, CD, and tensile strength. Multivariate analysis showed that the use of two
PT

geological predictor variables typically improved models for CI, CD, and BTS; however, the addition
of further predictor variables produced minimal additional improvement. The most important finding
of this analysis was that joint consideration of mica modal percentage and grain size parameters
CE

allows for improved predictions of CI, CD, and tensile strength to be made for granitic rock.
Additionally, it was found that correlations proposed previously in literature appear to be spurious
AC

when considered in the context of the larger data set compiled in this research. This occurrence is
most likely a consequence of unknown additional variables (confounding variables) that is correlated
to both the independent and dependent variables used in forming the original correlations.

KEYWORDS: Granite, crack initiation, crack damage, rock strength


ACCEPTED MANUSCRIPT

1. INTRODUCTION
Knowledge of the mechanical properties of rock materials is of paramount importance in engineering
applications, such as foundation design, slope stability analysis, underground construction, and tunneling.
In engineering and construction practice, a laboratory-based determination of relevant mechanical
properties is critical in most projects. Fundamentally, these mechanical properties of intact rock are
controlled by their grain-scale geological composition.

Granitic rocks are the focus of this study, specifically, because of the availability of ample data and the

T
relative simplicity and independence of the geological features which define these rocks (i.e. mineral

IP
composition and grain size characteristics). Recent studies (Tuǧrul and Zarif, 1999; Güneş Yilmaz et al.,

CR
2011; Keikha and Keykha, 2013; Yesiloglu-Gultekin et al., 2013) have attempted to identify relationships
between geological and mechanical properties of granitic rock based on analyses of thin sections and
associated mechanical tests. The following relationships between granitic geological and mechanical

US
properties were postulated from these recent studies: 1) Average quartz grain size versus UCS, 2)
Plagioclase average grain size versus UCS, 3) Alkali feldspar average grain size versus UCS, 4) Mica
AN
average grain size versus BTS, 5) Quartz: feldspar modal percentage versus UCS, 6) Quartz: feldspar
modal percentage versus BTS and 7) Alkali feldspar modal percentage versus UCS. This work builds on
M

these studies by developing a broader database of information with respect to the geological controls on the
strength of granitic rocks. A literature review was performed by searching for studies containing both
ED

geological and mechanical data for granitic rocks. Such data were obtained for 58 granitic rock types
from 18 literature sources (Arzúa and Alejano, 2013; Duevel and Haimson, 1997; Eberhardt et al., 1998;
Eloranta and Hakala, 1999; Ghazvinian, 2015; Ghazvinian et al., 2012; Güneş Yilmaz et al., 2011;
PT

Heikkila and Hakala, 1998; Keikha and Keykha, 2013; Liang et al., 2015; Nasseri and Mohanty, 2008;
Nicksiar and Martin, 2014; Takarli et al., 2008; Tuǧrul and Zarif, 1999; Yesiloglu-Gultekin et al., 2013;
CE

Zhao et al., 1995, 2015, 2013) and from the CanmetMINING Rock Mechanics Laboratory in Ottawa,
Canada in order to create the database. This database has been used to scrutinize the relationships
AC

proposed in previous studies, as well as to identify novel relationships. In addition to bi-variate regression
analyses, a principal component analysis and subsequent multivariate regression analyses were conducted
in an attempt to quantify and understand the complex geological factors that control mechanical properties
in granitic rocks.

2. DATABASE
For this study, the majority (39) of the granitic rock types studied are considered syeno-granite, monzo-
granite, or granodiorite when plotted on a Quartz-Alkali Feldspar-Plagioclase Feldspar (QAP) diagram.
The other 19 granites are considered alkali feldspar granite, tonalite, quartz alkali feldspar syenite, quartz
ACCEPTED MANUSCRIPT

syenite, quartz monzonite, quartz monzodiorite/monzogabbro, or quartz diorite/gabbro/anorthosite. The


mineralogical distribution of the types of granite used in this database is shown in Figure 1 as red circles,
and a breakdown of the types of granites used in the database is shown in Table 1.
Table 1: Distribution of granitic rocks in the database.
C: D: E: F: G: H: I: J: K: L
Quartz :
Alkali Alkali Quartz Quartz Quartz
Feldsp a Syeno Monzo Grano Feldsp a Quartz Monz o Monzodiorit Diorite
r - - - Tonalit r Syenit - nite e/ /

T
Granite granit granite diorit e Syenite e Monzogabbr Gabbro /
e e o Anorthosit

IP
e
1 9 22 8 1 1 1 5 6 3

CR
The rocks used in this study are from a variety of locations worldwide, shown in Figure 2. Many of the
locations for granites are not exact, because in some cases the author of the publication containing data

US
used provided only the country of origin. Also, Güneş Yilmaz et al. (2011) listed the countries of origin
for the twelve reported granites but did not specify how many granites came from each country. Much of
AN
the data in this study is associated with rocks originating from Turkey, Canada, and Scandinavia.

Figure 3 shows seven hand samples which are representative of the variability in the rocks considered in
M

this study. These photos also illustrate the wide range of grain sizes and mineralogies of the granitic rocks
included in the database.
ED

Mineralogical parameters and mechanical properties were both compiled for the database.
PT

2.1 Mineralogical Descriptions


The mineralogical parameters that were collected through the literature review were the modal
CE

percentages of quartz, alkali feldspar, plagioclase, and micas, the average grain size and standard
deviation of grain sizes of the previously mentioned minerals, and the minimum, maximum, average and
standard deviation of the overall sample grain size. The granitic rocks were classified according to their
AC

mineralogical composition using Streckeisen’s classification system (Streckeisen, 1976), which follows
the IUGS classification system for plutonic rocks. Table A1 in Appendix A shows the modal percentages
of quartz, alkali feldspar, plagioclase, and micas for each granite. In the case of the Smaland Granite,
which is lightly metamorphosed and technically considered a metagranite, mineralogical information
could not be obtained (Ghazvinian et al., 2012). Additionally, Yesiloglu-Gultekin et al. (2013) only
reported the modal percentage of quartz, alkali feldspar, and plagioclase. The remainder percentage is not
only attributed to mica, but to all accessory minerals. Therefore, this remainder was not considered solely
mica modal percentage in this analysis. Blank entries in Table A1 represent values that were not provided
ACCEPTED MANUSCRIPT

by the author(s) of the study in question and could not be estimated or back-calculated. The average Mohs
Hardness was calculated as a weighted average based on modal percentages, using Tabor’s classification
system (Tabor, 1954).

In cases when information concerning grain sizes (either by mineral or overall) was not explicitly or
pictorially reported (Arzúa and Alejano, 2013; Duevel and Haimson, 1997; Eberhardt et al., 1998;
Eloranta and Hakala, 1999; Ghazvinian, 2015; Ghazvinian et al., 2012; Heikkila and Hakala, 1998;
Nicksiar and Martin, 2014; Takarli et al., 2008; Yesiloglu-Gultekin et al., 2013; Zhao et al., 1995, 2015),

T
grain size parameters for these granites were not considered in analysis. In cases where grain size

IP
information was reported as a range (Keikha and Keykha, 2013; Tuǧrul and Zarif, 1999), the average
grain size was taken to be the average of the minimum and maximum values, and the standard deviation

CR
was calculated based on a four-sigma minimum-maximum range (assuming a normal distribution). Where
the average overall grain size was not provided, it was calculated as an average of the individual mineral

US
average grain sizes weighted by their respective modal percentages (Güneş Yilmaz et al., 2011; Keikha
and Keykha, 2013; Nasseri and Mohanty, 2008; Tuǧrul and Zarif, 1999). Some authors (Güneş Yilmaz et
AN
al., 2011; Liang et al., 2015; Nasseri and Mohanty, 2008; Zhao et al., 2013) provided thin - section
images for rock samples; therefore, manual analysis of grain size distributions was performed where grain
size information was not directly reported. Figure 4 shows nine thin sections collected from various
M

sources, illustrating the variety of grain sizes and mineralogies for the granitic rocks included in the
ED

database. Note that hand samples (Figure 3) and thin sections (Figure 4) labelled with the same letter do
not necessarily come from the same sample. Information concerning grain size distributions for individual
mineral components and overall samples is shown in Table A2 in Appendix A. Blank entries in Table A2
PT

represent values that were not provided by the author(s) of the study in question and could not be
estimated based on thin section images. Tables 2, 3, 4, 5, and 6 illustrate the variety and range of
CE

mineralogical parameters within granites included in the database. The amount, maximum, minimum,
average, and standard deviation of overall average grain size (Table 2), quartz average grain size (Table
AC

3), alkali feldspar average grain size (Table 4), plagioclase grain size (Table 5), and mica grain size
(Table 6) are divided by granitic rock classification.
ACCEPTED MANUSCRIPT

Table 2: Compiled Overall Average Grain Size statistics divided by granitic rock classification. The
bolded bottom row corresponds to the overall statistic values considering all granitic rock classifications.
Overall Average Grain Size (mm)
Rock Type Count Maximum Minimum Average Std. Dev.
C: Alkali Feldspar Granite 1 3.5 3.5 3.5 N/A
D: Syeno-granite 9 6.1 15.1 0.8 5.6
E: Monzo-granite 19 2.2 6.5 0.4 1.8
F: Granodiorite 4 1.5 2.1 0.9 0.6

T
G: Tonalite
H: Quartz Alkali Feldspar Syenite 1 2.9 2.9 2.9 N/A

IP
I: Quartz Syenite 1 6.6 6.6 6.6 N/A
5 1.6 2.8 1.2 0.7

CR
J: Quartz Monzonite
K: Quartz Monzodiorite / Monzogabbro 6 1.9 3.8 0.9 1.0
L: Quartz Diorite / Gabbro / Anorthosite 2 1.5 2.1 0.8 0.9
Not Specified (Smaland Granite) 1 0.8 0.8 0.8 N/A

US
49 AN 2.8 15.1 0.4 3.1

Table 3: Compiled Quartz Average Grain Size statistics divided by granitic rock classification type.
The bolded bottom row corresponds to the overall statistic values considering all granitic rock
classifications.
M

Quartz Average Grain Size (mm)


Count Maximum Minimum Average Std. Dev.
ED

Rock Type
C: Alkali Feldspar Granite
D: Syeno-granite 7 7.5 0.7 4.0 2.6
PT

E: Monzo-granite 12 5.0 0.5 1.6 1.5


F: Granodiorite 3 1.3 0.4 1.0 0.5
G: Tonalite
CE

H: Quartz Alkali Feldspar Syenite 1 0.7 0.7 0.7 N/A


I: Quartz Syenite 1 3.5 3.5 3.5 N/A
J: Quartz Monzonite 5 4.5 0.5 1.7 1.7
AC

K: Quartz Monzodiorite / Monzogabbro 6 2.7 0.5 1.3 1.0


L: Quartz Diorite / Gabbro / Anorthosite 2 1.6 0.7 1.1 0.6
Not Specified (Smaland Granite)
37 7.5 0.4 2.0 1.9
ACCEPTED MANUSCRIPT

Table 4: Compiled Alkali Feldspar Average Grain Size statistics divided by granitic rock classification
type. The bolded bottom row corresponds to the overall statistic values considering all granitic rock
classifications.
Alkali Feldspar Average Grain Size (mm)
Granite Type Count Maximum Minimum Average Std. Dev.
C: Alkali Feldspar Granite
D: Syeno-granite 7 25.0 1.0 11.1 9.3

T
E: Monzo-granite 12 10.0 0.3 3.0 3.4

IP
F: Granodiorite 3 2.3 1.6 1.9 0.3
G: Tonalite

CR
H: Quartz Alkali Feldspar Syenite 1 3.8 3.8 3.8 N/A
I: Quartz Syenite 1 12.0 12.0 12.0 N/A
J: Quartz Monzonite 5 4.7 1.3 2.6 1.2

US
K: Quartz Monzodiorite / Monzogabbro 6 9.0 0.6 3.5 3.0
L: Quartz Diorite / Gabbro / Anorthosite 2 2.2 1.5 1.8 0.5
Not Specified
AN
37 25.0 0.3 4.7 5.6

Table 5: Compiled Plagioclase Average Grain Size statistics divided by granitic rock classification type.
M

The bolded bottom row corresponds to the overall statistic values considering all granitic rock
classifications.
ED

Plagioclase Average Grain Size (mm)


PT

Granite Type Count Maximum Minimum Average Std. Dev.


C: Alkali Feldspar Granite
D: Syeno-granite 7 8.0 0.4 2.8 2.9
CE

E: Monzo-granite 11 5.0 0.4 2.1 1.8


F: Granodiorite 3 2.9 1.2 2.3 0.9
G: Tonalite
AC

H: Quartz Alkali Feldspar Syenite


I: Quartz Syenite 1 1.9 1.9 1.9 N/A
J: Quartz Monzonite 5 1.6 0.8 1.0 0.4
K: Quartz Monzodiorite / Monzogabbro 6 3.5 1.1 1.8 1.0
L: Quartz Diorite / Gabbro / Anorthosite 2 2.5 0.9 1.7 1.1
Not Specified
35 8.0 0.4 2.0 1.7
ACCEPTED MANUSCRIPT

Table 6: Compiled Mica Average Grain Size statistics divided by granitic rock classification type. The
bolded bottom row corresponds to the overall statistic values considering all granitic rock classifications.
Mica Average Grain Size (mm)
Granite Type Count Maximum Minimum Average Std. Dev.
C: Alkali Feldspar Granite
D: Syeno-granite 7 2.2 0.2 0.8 0.7
E: Monzo-granite 12 0.8 0.2 0.4 0.2

T
F: Granodiorite 3 0.6 0.3 0.4 0.2

IP
G: Tonalite
H: Quartz Alkali Feldspar Syenite

CR
I: Quartz Syenite 1 1.5 1.5 1.5 N/A
J: Quartz Monzonite 4 0.8 0.4 0.6 0.1
K: Quartz Monzodiorite / Monzogabbro 6 1.6 0.4 1.0 0.4

US
L: Quartz Diorite / Gabbro / Anorthosite 2 0.6 0.5 0.5 0.1
Not Specified
35 2.2 0.2 0.7 0.5
AN
2.2 Mechanical Properties
M

The following mechanical properties were compiled for analysis: Unconfined Compressive Strength
(UCS), Poisson’s ratio (ν), Elastic Modulus (E), Crack Initiation Stress (CI), Crack Damage Stress (CD),
ED

Brazilian Tensile Strength (BTS), and the Hoek-Brown constant, mi . The reported number of samples,
average value, and standard deviation for each of these properties from each of the studies considered
were compiled into Table A3 in Appendix A. Statistics for each mechanical property separated by granitic
PT

rock classification are provided in Tables 7 – 13.


CE

Of the 19 different sources data sources used for this study, 18 followed ISRM suggested methods
(Brown, 1981; Fairhurst and Hudson, 1999; Ulusay and Hudson, 2007) to test for UCS and BTS. For the
determination of UCS and BTS, Güneş Yilmaz et al. (2011) followed Turkish Standard TS EN 1926,
AC

Natural stone test methods - determination of compressive strength,” 2000 and “Turkish Standard TS 699,
Natural stone test methods - determination of splitting tensile strength,” 1987.

Bieniawski and Bernede (1979) suggest that (for UCS testing) “the specimen should be 10 times larger
than the size of the largest grain (10:1).” ISRM testing standards now adopt this suggestion into their
suggested methods. Blanks and McNamara (1935) found that “for a ratio larger than 10:1 the UCS is only
slightly affected by the grain size while for smaller ratios a large variability in UCS is to be anticipated.”
Therefore, due to 18 of 19 sources following ISRM procedures and the other following Turkish Standard
testing methods, UCS testing results reported from all sources are considered valid, and any grain size
ACCEPTED MANUSCRIPT

effects observed are expected to represent fundamental geological effects on mechanical behavior, as
opposed to scale effects.

CI is defined as the stress level where microfracturing begins and is the first stage of stress-induced
damage in low-porosity rocks (Eberhardt et al., 1998; Nicksiar and Martin, 2014). It also marks the point
where the lateral and volumetric strain curves depart from linearity (Eberhardt et al., 1998). CD is defined
as the stress level where unstable crack propagation begins within a rock. It also marks the point of
reversal in the volumetric strain curve under unconfined testing conditions. As stresses increase beyond

T
CD, unstable propagation continues to the point where microfractures have coalesced and the rock can no

IP
longer support an increase in the load (peak strength) (Eberhardt et al., 1998). Although ISRM does not
provide guidelines for the determination of the CI and CD thresholds, there have been several studies (

CR
Brace et al., 1966; Bieniawski, 1967; Lajtai, 1974; Stacey, 1981; Eberhardt et al., 1998; Martin and
Chandler, 1994; Diederichs et al., 2004; Nicksiar and Martin, 2014) that suggest methods to determine

US
these stresses. Nicksiar and Martin (2014) studied these methods and concluded that all approaches
showed reasonable statistical agreement with one another. The values used in this study were originally
AN
obtained by each set of authors using their own preferred approach.

The brittle deformation process of low-porosity rocks is portrayed in Figure 5. The failure process can be
M

broken into four stages. First, crack closure occurs during the initial stages of uniaxial loading; it
represents the stress level at which pre-existing microfractures generally aligned with the loading
ED

direction are closed (Eberhardt et al., 1998). Linear elastic deformation then takes place until
microfracturing begins (CI), followed by stable crack growth until the CD stress level, where unstable
PT

crack growth initiates.


CE

Table 7: Compiled Unconfined Compressive Strength (UCS) statistics divided by granitic classification.
The bolded bottom row corresponds to the overall statistical value considering all granitic rock
classifications. Note that all statistics shown were derived considering the average values obtained from
AC

several specimens of a given rock type.

Unconfined Compressive Strength (MPa)


Granite Type Count Maximum Minimum Average Std. Dev.
C: Alkali Feldspar Granite 1 186.0 186.0 186.0 N/A
D: Syeno-granite 9 219.0 86.7 156.4 38.6
E: Monzo-granite 22 259.2 76.9 169.6 47.0
F: Granodiorite 8 211.8 73.5 132.8 53.2
G: Tonalite 1 120.7 120.7 120.7 N/A
H: Quartz Alkali Feldspar Syenite 1 142.4 142.4 142.4 N/A
ACCEPTED MANUSCRIPT

I: Quartz Syenite 1 121.4 121.4 121.4 N/A


J: Quartz Monzonite 5 172.3 129.0 149.7 19.4
K: Quartz Monzodiorite / Monzogabbro 6 162.5 64.0 125.9 33.8
L: Quartz Diorite / Gabbro / Anorthosite 3 169.0 109.2 145.3 31.8
Not Specified 1 240.1 240.1 240.1 N/A
58 259.2 64.0 154.3 44.4

Table 8: Compiled Poisson’s ratio statistics divided by granitic classification. The bolded bottom row

T
corresponds to the overall statistical value considering all granitic rock classifications. Note that all
statistics shown were derived considering the average values obtained from several specimens of a given

IP
rock type.
Poisson's Ratio

CR
Granite Type Count Maximum Min Average Std. Dev.
C: Alkali Feldspar Granite 1 0.25 0.25 0.25 N/A
D: Syeno-granite 2 0.25 0.25 0.25 0.00

US
E: Monzo-granite 10 0.41 0.14 0.24 0.10
F: Granodiorite 1 0.18 0.18 0.18 N/A
G: Tonalite 0 -- -- -- --
AN
H: Quartz Alkali Feldspar Syenite 0 -- -- -- --
I: Quartz Syenite 0 -- -- -- --
J: Quartz Monzonite 0 -- -- -- --
M

K: Quartz Monzodiorite / Monzogabbro 1 0.14 0.14 0.14 N/A


L: Quartz Diorite / Gabbro / Anorthosite 1 0.22 0.22 0.22 N/A
ED

Not Specified 1 0.27 0.27 0.27 N/A


17 0.41 0.14 0.23 0.08
PT

Table 9: Compiled Young’s Modulus (E) statistics divided by granitic classification. The bolded bottom
row corresponds to the overall statistical value considering all granitic rock classifications. Note that all
CE

statistics shown were derived considering the average values obtained from several specimens of a given
rock type.
AC

Young’s Modulus, E (GPa)


Granite Type Count Maximum Minimum Average Std. Dev.
C: Alkali Feldspar Granite 1 84.0 84.0 84.0 N/A
D: Syeno-granite 4 71.0 45.3 56.2 10.8
E: Monzo-granite 15 71.5 19.0 50.9 16.3
F: Granodiorite 5 64.0 38.8 50.7 9.1
G: Tonalite 0 -- -- -- --
H: Quartz Alkali Feldspar Syenite 1 38.7 38.7 38.7 N/A
I: Quartz Syenite 0 -- -- -- --
J: Quartz Monzonite 4 44.8 29.8 38.3 7.0
ACCEPTED MANUSCRIPT

K: Quartz Monzodiorite / Monzogabbro 4 63.5 30.1 39.7 15.9


L: Quartz Diorite / Gabbro / Anorthosite 3 61.0 30.3 46.6 15.4
Not Specified 1 70.6 70.6 70.6 N/A
38 84.0 19.0 49.7 15.3

Table 10: Compiled Crack Initiation Stress (CI) statistics divided by granitic classification. The bolded
bottom row corresponds to the overall statistical value considering all granitic rock classifications. Note
that all statistics shown were derived considering the average values obtained from several specimens of a

T
given rock type.

IP
CR
Crack Initiation Stress, CI (MPa)
Granite Type Count Maximum Minimum Average Std. Dev.
C: Alkali Feldspar Granite 0 -- -- -- --

US
D: Syeno-granite 1 72.0 72.0 72.0 N/A
E: Monzo-granite 9 92.7 26.7 63.6 24.2
F: Granodiorite 1 80.9 80.9 80.9 N/A
AN
G: Tonalite 0 -- -- -- --
H: Quartz Alkali Feldspar Syenite 0 -- -- -- --
I: Quartz Syenite 0 -- -- -- --
M

J: Quartz Monzonite 0 -- -- -- --
K: Quartz Monzodiorite / Monzogabbro 1 74.2 74.2 74.2 N/A
L: Quartz Diorite / Gabbro / Anorthosite 1 64.0 64.0 64.0 N/A
ED

Not Specified 1 121.5 121.5 121.5 N/A


14 121.5 26.7 70.4 24.6
PT

Table 11: Compiled Crack Damage Stress (CD) statistics divided by granitic classification. The bolded
CE

bottom row corresponds to the overall statistical value considering all granitic rock classifications. Note
that all statistics shown were derived considering the average values obtained from several specimens of a
given rock type.
AC

Crack Damage Stress, CD (MPa)


Granite Type Count Maximum Minimum Average Std. Dev.
C: Alkali Feldspar Granite 0 -- -- -- --
D: Syeno-granite 1 146.0 146.0 146.0 N/A
E: Monzo-granite 9 157.3 51.1 114.6 37.0
F: Granodiorite 1 130.7 130.7 130.7 N/A
G: Tonalite 0 -- -- -- --
H: Quartz Alkali Feldspar Syenite 0 -- -- -- --
I: Quartz Syenite 0 -- -- -- --
ACCEPTED MANUSCRIPT

J: Quartz Monzonite 0 -- -- -- --
K: Quartz Monzodiorite / Monzogabbro 1 113.7 113.7 113.7 N/A
L: Quartz Diorite / Gabbro / Anorthosite 1 146.0 146.0 146.0 N/A
Not Specified 1 202.1 202.1 202.1 N/A
14 202.1 51.1 126.4 38.1

Table 12: Compiled Brazilian Tensile Strength (BTS) statistics divided by granitic classification. The
bolded bottom row corresponds to the overall statistical value considering all granitic rock classifications.
Note that all statistics shown were derived considering the average values obtained from several

T
specimens of a given rock type.

IP
CR
BTS (MPa)
Granite Type Count Maximum Minimum Average Std. Dev.
C: Alkali Feldspar Granite 1 11.4 11.4 11.4 N/A

US
D: Syeno-granite 9 25.9 3.1 11.7 7.4
E: Monzo-granite 15 28.0 6.1 13.3 8.1
F: Granodiorite 4 26.1 5.1 19.2 9.6
AN
G: Tonalite 0 -- -- -- --
H: Quartz Alkali Feldspar Syenite 1 19.6 19.6 19.6 N/A
I: Quartz Syenite 1 6.9 6.9 6.9 N/A
M

J: Quartz Monzonite 5 21.8 11.0 18.3 4.3


K: Quartz Monzodiorite / Monzogabbro 5 21.6 3.4 13.0 7.2
L: Quartz Diorite / Gabbro / Anorthosite 2 22.2 14.9 18.5 5.1
ED

Not Specified
43 28.0 3.1 14.3 7.4
PT

Table 13: Compiled mi statistics divided by granitic classification. The bolded bottom row corresponds
CE

to the overall statistical value considering all granitic rock classifications. Note that all statistics shown
were derived considering the average values obtained from several specimens of a given rock type.
mi
AC

Granite Type Count Maximum Minimum Average Std. Dev.


C: Alkali Feldspar Granite -- -- -- -- --
D: Syeno-granite 1 37.2 37.2 37.2 N/A
E: Monzo-granite 6 41.0 28.2 34.8 5.7
F: Granodiorite -- -- -- -- --
G: Tonalite -- -- -- -- --
H: Quartz Alkali Feldspar Syenite -- -- -- -- --
I: Quartz Syenite -- -- -- -- --
J: Quartz Monzonite -- -- -- -- --
K: Quartz Monzodiorite / Monzogabbro -- -- -- -- --
L: Quartz Diorite / Gabbro / Anorthosite -- -- -- -- --
ACCEPTED MANUSCRIPT

Not Specified -- -- -- -- --
7 41.0 28.2 35.1 5.3

3 CORRELATIONS BETWEEN MINERALOGICAL PARAMETERS AND MECHANICAL


PROPERTIES

3.1 Statistical analysis of collected data


Once all of the relevant data were collected for the database, linear regression analysis was performed in

T
order to assess previously proposed relationships against a broader database as well as to find new

IP
relationships between geological and mechanical properties. The correlations proposed in the literature
were tested in order to verify the correlation using a more extensive database. Bi-variate regression was

CR
performed for all pairs of mineralogical parameters and mechanical properties using MATLAB (The
MathWorks Inc., 2016). In addition, some ratios of characteristics were tested against each other (e.g.

US
Quartz % / Mica % versus UCS / CD) and were analyzed to see if any statistically significant relationship
was present. In addition to the raw data, transformed versions of the variables (i.e. log10 , ln, and power
AN
transformations) were also considered in the regression analysis to identify potential non-linear
correlations. Furthermore, due to the relatively high number of monzo-granites (22 of 58 granitic rock
types considered), an additional analysis was performed in which monzo-granites were isolated and a
M

separate regression analysis was performed considering the mechanical properties for which correlations
had been identified using the broader database (CI, CD, and BTS).
ED

3.2 Comparative correlations with previous studies


Previous studies (Tuǧrul and Zarif, 1999; Güneş Yilmaz et al., 2011; Keikha and Keykha, 2013;
PT

Yesiloglu-Gultekin et al., 2013) postulated the existence of 7 relationships between granitic geological
and mechanical properties: 1) Average quartz grain size versus UCS, 2) Plagioclase average grain size
CE

versus UCS, 3) Alkali feldspar average grain size versus UCS, 4) Mica average grain size versus BTS, 5)
Quartz: feldspar modal percentage versus UCS, 6) Quartz: feldspar modal percentage versus BTS and 7)
AC

Alkali feldspar modal percentage versus UCS. Of these relationships, six of them were studied by Tuǧrul
and Zarif (1999) and two were studied by Güneş Yilmaz et al. (2011).

Figures 6 and 7 shows seven graphs containing scatter plots with lines of best fit for each relationship. For
each relationship, Table 14 shows the regression equation, r2 value, and number of observations for each
dataset. In comparative relationships where ‘all other data’ is referenced, these regressions include all
possible data other than that included in Tuǧrul and Zarif (1999) and Güneş Yilmaz et al. (2011), except
when noted otherwise.

Figure 6a shows average alkali feldspar grain size (AKGS) versus UCS and Figure 6b shows average
ACCEPTED MANUSCRIPT

plagioclase feldspar grain size (PlagGS) versus UCS data plotted from Tuǧrul and Zarif (1999). Both have
strong negative correlations (r2 = 0.822 and r2 = 0.607, respectively). However, the data from all other
sources associated with AKGS and PlagGS both produced extremely small r2 values. This suggests that
although the dataset compiled by Tuǧrul and Zarif (1999) contains statistically significant correlations,
these correlations are not representative of more broadly applicable causal trends. A possible cause of
this notable difference in results is that all of the granites tested by Tuǧrul and Zarif (1999) were collected
in Turkey, limiting the range of conditions under which they formed. Additionally, AKGS only ranged

T
from 0.95 mm – 4.65 mm, whereas the larger dataset considering data collected worldwide (various

IP
geological environments), AKGS ranged from 0.3 mm – 25 mm. Similar issues were found for the
relationships of Quartz : Feldspar modal percentage ratio (QFR) versus UCS, shown in Figure 6c, and

CR
QFR versus BTS, shown in Figure 6d (Tuǧrul and Zarif, 1999), as well as between alkali feldspar modal
percentage (AK%) versus UCS, shown in Figure 7b (Güneş Yilmaz et al., 2011).

US
The plot of average quartz grain size (QtzGS) versus UCS (Figure 7a) contains data from Güneş Yilmaz et
al. (2011), Tuǧrul and Zarif (1999), and from various other sources. QtzGS versus UCS data from Tuǧrul
AN
2
and Zarif (1999) has a fairly high r value of 0.652; however, their line of best fit has significantly steeper
slope than is suggested by the data from Güneş Yilmaz et al. (2011) and other sources. In the context of a
M

larger dataset, this dramatic dissimilarity is possibly a result of a confounding variable present in the
relationship, meaning that there is an unconsidered external variable correlated with average quartz grain
ED

size that is controlling the behaviour of the granites studied by Tuǧrul and Zarif (1999). Overall, it
appears unlikely that average quartz grain size has a meaningful causal impact on UCS. Although average
PT

mica grain size (Mica GS) versus BTS (Figure 7c) data from both Tuǧrul and Zarif (1999), and all other
sources (including Güneş Yilmaz et al. (2011)) show a weak negative correlation, each of the individual
CE

datasets are highly scattered relative to one another.

In all cases examined, the correlations identified based on individual datasets were found to be inaccurate
AC

in the context of a larger database. This highlights the need to a consider a broad data set which
incorporates substantially distinct rock types with a wide range of mineralogical parameters when
attempting to capture meaningful correlations between mineralogical parameters and mechanical
properties.
Table 14: Previously proposed relationships compared to all other existing data. Previous relationships
shown here were not strengthened by the addition of data from other sources. For each model, data from
“various” sources does not include data from sources explicitly listed.

Model Predictor Variable Response Regression Equation r2 No. Reference(s)


No.
Variable Obs.
ACCEPTED MANUSCRIPT

UCS = 247.7 – 128.7 0.652 19 (T uǧrul and


Quartz Average Unconfined Zarif, 1999)
QtzGS
Grain Size - Compressive (Güneş Yilmaz
1 UCS = 190.2 – 8.0 0.347 12
QtzGS (mm) Strength - et al., 2011)
QtzGS
UCS (MPa)
UCS = 182.3 – 26.0 0.491 6 Various

QtzGS
Plagioclase Unconfined UCS = 198.8 – 48.8 0.607 19 (T uǧrul and
2 Zarif, 1999)
Average Grain - Compressive PlagGS
Size PlagGS Strength - UCS = 165.4 – 4.6 0.051 17 Various
(mm) UCS (MPa) PlagGS

T
Alkali Unconfined UCS = 204.9 – 22.1 0.822 19 (T uǧrul and

IP
3 AKGS
Zarif, 1999)
Feldspar Compressive
Average Strength - UCS = 157.0 – 0.6 0.013 18 Various

CR
Grain Size - UCS (MPa) AKGS
AKGS (mm)
BTS = 26.2 – 7.8 0.316 18 (T uǧrul and
Mica Average Brazilian MicaGS Zarif, 1999)

US
Grain Size - Tensile Strength
4 Various
MicaGS (mm) - BTS (MPa) BTS = 10.7 – 3.4 0.553 14
(including
MicaGS
Güneş Yilmaz
AN
et al., 2011)
Quartz : Unconfined UCS = 115.5 + 120.3 0.619 19 (T uǧrul and
5 QFR
Zarif, 1999)
Feldspar Compressive
Modal Strength - Various
M

UCS = 147.4 + 12.8 0.002 37


Percentage - UCS (MPa) QFR
QFR
ED

Quartz : Brazilian BTS = 15.4 + 19.5 0.640 19 (T uǧrul and


6 QFR
Zarif, 1999)
Feldspar Tensile Strength
Modal - BTS (MPa) BTS = 10.0 – 4.2 0.062 23 Various
Percentage - QFR
PT

QFR
Alkali Unconfined UCS = 205.9 + 1.3 0.579 12 (Güneş Yilmaz
7 et al., 2011)
Feldspar Compressive AK%
CE

Modal Strength - UCS = 127.8 + 0.9 0.075 45 Various


Percentage UCS (MPa) AK%
- AK% (%)
AC

3.3 New correlations


After an examination of all bi-variate relationships, two criteria were used to establish which
relationships were most significant: the number of data points was required to be at least 10, and the r 2
value was required to be at least 0.6. Note that all relationships involving E, ν, and mi did not meet one,
or both, of the requirements to be identified as a statistically significant relationship.

Six such relationships were found, and all of these involve brittle strength thresholds (CI, CD, and BTS).
ACCEPTED MANUSCRIPT

These relationships are based on data from multiple studies involving rock samples from across the globe,
thus minimizing the risk of any confounding variable influence. Figure 8 shows each of these
relationships, and Table 15 summarizes the results of the associated regression analyses. In considering
only monzo-granites, only BTS versus Mica % produced a stronger correlation relative to the case
considering all granitic rock types. All other possible combinations produced an r 2 value less than 0.6,
had less than 10 observations, or did not significantly strengthen the existing correlation.

The proposed power relationship between mica modal percentage (Mica %) and BTS is shown in Fig 8a.

T
Linear and log base 10 regressions also produced high r2 values, but power based regression produced the

IP
strongest of the three. Not only is this relationship associated with a fairly strong correlation (r 2 = 0.738),

CR
but the potential for a causal relationship also makes geological sense. This relationship is most likely a
consequence of the unique mechanical properties of mica, which is much softer and has less frictional
strength than the other minerals commonly found within granitic rocks. Mica content is similarly

US
correlated with CI and CD (Fig 8b and 8c), which is consistent with the strong theoretical (Griffith, 1921;
Murrell, 1963) and empirical (Ghazvinian et al., 2013; Nicksiar and Martin, 2014; Perras and Diederichs,
AN
2014) relationship between these damage thresholds and tensile strength. When isolating monzo-
granites, this relationship produced a stronger power based correlation (r 2 = 0.817) and had an acceptable
M

number of observations (N = 14). The shape of the monzo-granite regression fit is very similar to the fit
to the full data set, however, and so it is suggested that the value of considering distinct granitic
ED

classifications (such as monzo-granite) is limited.

Overall minimum grain size was found to have a strong positive relationship with the CD : CI ratio, as
PT

shown in Fig 8d. Previous studies (Eberhardt et al., 1999; Nicksiar and Martin, 2014) determined that the
CI appears independent of average grain size for a rock with a uniform grain size distribution and
CE

constant mineralogy; another study (Hofmann et al., 2015) found that grain size had minimal effects in
terms of when intergranular cracking initiates. The development of the cracks was found to be more
AC

dependent on the strength of the minerals rather than the grain size. In the same study, the behaviour of
cracking during propagation (i.e. at stresses near CD) seemed to be highly influenced by the grain size.
These findings aid in the development of a physical explanation for the observed correlation between the
CD : CI ratio and overall minimum grain size, because as the minimum grain size affects CD but not CI,
it should affect the ratio between these two quantities.

Average Mohs Hardness (Mohs) versus CI and Mohs versus CD (Figure 8e,f) both have statistically
significant positive correlations, although these relationships are likely purely a reflection of the impact
of mica content, since mica content directly affects the average Mohs Hardness based on its relatively
ACCEPTED MANUSCRIPT

low value on the Mohs Hardness scale (Tabor, 1954). Figure 9 shows predicted strength values (using
the proposed equations) compared to actual strength values for most statistically significant relationships
shown in Table 15 and Figure 8.

T
IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

Table 15: Newly proposed relationships.

Model Predictor Variable Response Variable Regression Equation r2 No.


No. Obs.
1 Mica Modal Percentage – Brazilian Tensile Strength – BTS = 37.5 (Mica%)-0.7 0.738 41
Mica% (%) BTS (MPa)
2 Mica Modal Percentage – Crack Initiation Stress – CI CI = 94.4 – 3.0 Mica% 0.739 12
Mica% (%) (MPa)
3 Mica Modal Percentage – Crack Damage Stress – CD CD = 162.3 – 4.7 0.772 12
Mica% (%) (MPa) Mica%

T
4 Overall Minimum Grain Size Crack Damage : Crack Initiation CD/CI = 1.5 + 0.5 0.700 10
– GSMin (mm) Stress – CD/CI GSMin

IP
5 Average Mohs Hardness – Crack Initiation Stress – CI CI = -162.6 + 38.0 0.632 13
Mohs (MPa) Mohs

CR
6 Average Mohs Hardness – Crack Damage Stress – CD CD = -280.6 + 66.6 0.770 13
Mohs (MPa) Mohs
4. MULTIVARIATE ANALYSIS

US
In the case of the mechanical properties considered in this study, there are almost certainly multiple
mineralogical parameters that have coherent influences on rock mechanical properties. With that in mind,
attempting to relate each mechanical parameter to all 17 mineralogical parameters is both impractical and
AN
risks over-fitting. Part of the potential for over-fitting stems from the fact that several mineralogical
parameters are, by definition, interdependent. For example, if there is a large increase in quartz content,
M

there must (by definition) be an accompanying decrease in alkali feldspar, plagioclase, and/or mica
content. Other correlations between mineralogical parameters may exist as a consequence of their genesis;
ED

for example, granites with higher mica contents may also tend to have larger mica grains.

To understand the potential number of underlying factors influencing the observed mineralogical
PT

parameters, a principal component analysis was performed on the Z-scores of each characteristic (the
normalization associated with the Z-transformation was required to allow for comparison of
CE

characteristics measured in different units). The principal component analysis showed that 97.4% of the
variability in the 17 mineralogical parameters considered can be explained by four principal components.
AC

As such, it was decided that the number of mineralogical parameters considered as independent variables
for multivariate regression should not greatly exceed four.

Ultimately, five characteristics were selected as independent variables to be considered in multivariate


regression: overall minimum grain size (GSMin ), overall average grain size (GSAvg ), overall standard
deviation of grain size (GSDev ), quartz modal percentage (Qtz%), and mica modal percentage (Mica %).
Because all of the statistically significant bi-variate regression relationships that were identified involved
CI, CD, or BTS, these are the response variables that were considered in the multivariate analysis. In
order to quantify variables that contribute to the regression models the most, multivariate analyses with
ACCEPTED MANUSCRIPT

two, three, and four predictor variables were performed. r2 -adjusted was utilized to assess whether the
addition of a predictor variable actually strengthened the model relationship, or if the observed increase in
r2 was simply due to the introduction of additional model degrees of freedom. r2 -adjusted, which is used
commonly in multivariate regression analyses, modifies r2 by accounting for the number of predictor
variables in the model; r2 -adjusted increases only if the new variable improves the model more than what
would be expected by chance when a new predictor variable is added (Helland, 1987).

Combinations of GSMin , GSAvg , GSDev , Qtz%, and Mica% were tested against CI, CD, and BTS to produce

T
nine models for the multivariate analysis using two predictor variables. For each response variable, the

IP
three models with the highest r2 -adjusted were chosen to be reported in Table 16; all of these models have

CR
r2 -adjusted greater than 0.65. For each response variable (CI, CD, and BTS), the model with the highest
adjusted r2 is shown graphically in 3-dimensional space in Figure 10. Figure 10a (Mica% and GSMin versus
CI), 10b (Mica% and Qtz% versus CD), and 10c (Mica% and GSAvg versus BTS) each show two different

US
perspectives of the same multivariate regression model. Multivariate regression models with three and
four predictor variables were also produced, and the results are shown in Tables B1 and B2 in Appendix
AN
B. Both of these tables contain twelve models with each model being produce using various combinations
of the five geological predictor variables considered.
M

As mentioned previously, r2 -adjusted was the main indicator of model quality considered. For example,
the r2 -adjusted for Model 1 (GSMin and Mica% versus CI) in Table 7 is 0.91; when GSAvg is added as a
ED

predictor variable to the model, the r2 -adjusted increases to 0.941; however, when GSDev is added as the
fourth predictor variable, the r2 -adjusted decreases to 0.936. Therefore, GSDev does not improve the
PT

model. In most cases, the change in r2 -adjusted when adding a third or fourth predictor variable was
negative or very small.
CE
AC
ACCEPTED MANUSCRIPT

Table 16: Multivariate analysis results considering two predictor variables. Models are ordered by
descending r2 -adjusted value.
r2-
M odel 2 No.
Predictor Variables Response Variable Regression Equation r Adjuste
No. Obs.
d
Overall M inimum Grain Size
– GSMin (mm) Crack Initiation Stress – CI CI = 107.7 – 30.1 GSMin – 0.9
1 0.91 9
M ica M odal Percentage – (M Pa) 2.4 M ica% 32
M ica% (%)
Quartz M odal Percentage –
Qtz % (%) Crack Damage Stress – CI CD = 129.9 + 1.0 Qtz % - 0.8

T
2 0.788 12
M ica M odal Percentage – (M Pa) 4.4 M ica% 27
M ica% (%)

IP
Overall Average Grain Size –
GSAvg (mm) Crack Initiation Stress – CI CI = 102.2 – 3.3 GSAvg – 0.8
3 0.768 12
M ica M odal Percentage – (M Pa) 3.1 M ica% 1

CR
M ica% (%)
Quartz M odal Percentage –
Qtz % (%) Crack Initiation Stress – CI CI = 74.6 + 0.6 Qtz % - 2.8 0.7
4 0.74 12
M ica M odal Percentage – (M Pa) M ica% 87

US
M ica% (%)
Overall M inimum Grain Size
– GSMin (mm) Crack Damage Stress – CI CD = 167.8 – 19.9 GSMin 0.8
5 0.736 9
M ica M odal Percentage – (M Pa) – 4.1 M ica% 02
AN
M ica% (%)
Overall Average Grain Size –
GSAvg (mm) Crack Damage Stress – CI CD = 166.3 – 1.7 GSAvg – 0.7
6 0.732 12
M ica M odal Percentage – (M Pa) 4.7 M ica% 8
M

M ica% (%)
Overall Average Grain Size –
GSAvg (mm) Brazilian Tensile Strength - BTS = – 1.4 GSAvg + 36.6 0.7
7 0.735 30
M ica M odal Percentage – M ica%-0.4
ED

BTS (M Pa) 29
M ica% (%)
Quartz M odal Percentage –
Qtz % (%) Brazilian Tensile Strength - BTS = 19.3 + 0.2 Qtz % - 0.7
8 0.706 31
M ica M odal Percentage – BTS (M Pa) 1.1 M ica% 26
PT

M ica% (%)
Overall M inimum Grain Size
– GSMin (mm) Brazilian Tensile Strength - BTS = 24.3 – 4.6 GSMin – 0.6
9 0.664 26
M ica M odal Percentage –
CE

BTS (M Pa) 0.8 M ica% 91


M ica% (%)
5. CONCLUSIONS
AC

Quantitative geological and mechanical data were compiled from 58 different granitic rock types, and the
resultant database was used to scrutinize previously proposed relationships and to identify new ones.

Many of the correlations proposed previously in the literature (e.g. Mica GS versus BTS, QFR versus BTS)
appear to be spurious when considered in the context of a larger data set which considers a wider variety
of rock types with more varied origins. It is probable that in cases such as these, the previously proposed
relationships are the result of an unknown third variable (a confounding variable) that is correlated to
both the independent and dependent variable. In addition, this behaviour could be consistently occurring
within rocks that share a set of characteristics (e.g. collected from a single region, very similar
ACCEPTED MANUSCRIPT

mineralogy, etc.).

According to several sources, overall tensile strength is directly related to crack initiation and
propagation, because the first cracking that occurs in intact rock is locally tensile in nature, even under
globally compressive loading conditions (Murrell, 1963; Eberhardt et al., 1998; Diederichs, 2007; ). This
relationship suggests that these mechanical properties should be influenced by similar geological factors.
Through study of the database that was compiled, it was found that the abundance of mica minerals in
granitic rocks is such a factor, showing strong correlations with CI, CD, and BTS.

T
IP
Principal component analysis determined that 97.4% of the variability in the mineralogical parameters
considered can be explained by four principal components. This information was applied to guide

CR
decisions on how many geological factors were considered for the purpose of multivariate regression.
Multivariate analysis showed that the use of two geological predictor variables typically improved

US
models for CI, CD, and BTS, but the addition of further predictor variables produced minimal
improvement. The predictor variables that produced the strongest statistical relationships with BTS, CI,
and CD are GSAvg and Mica%, GSMin and Mica%, and Qtz% and Mica%, respectively.
AN
The conclusions of this study must be considered in the context of their limitations. The findings of this
M

study apply only to granitic rocks. Additionally, although the breadth of the data considered in this study
mitigates against the risk of confounding variable influences on the apparent correlations shown,
ED

additional data should be considered as it becomes available for the purpose of validating these
correlations.
PT

6. ACKNOWLEDMENTS
The authors would like to gratefully acknowledge Denis Labrie of CanmetMINING for providing testing
CE

data for the Laurentian Pink, Pinawa Pink, Westerly Grey, and Stantead granites. The authors would also
like to acknowledge “30micron Geological Thin Sections” for offering their ternary plots generation tool
AC

for igneous rocks as an Excel plugin online.


ACCEPTED MANUSCRIPT

REFERENCES
Arzúa, J., Alejano, L.R., 2013. Dilation in granite during servo-controlled triaxial strength tests. Int. J.
Rock Mech. Min. Sci. 61, 43–56. https://doi.org/10.1016/j.ijrmms.2013.02.007

Bieniawski, Z.T., 1967. Mechanism of brittle fracture of rock; Part I-theory of the fracture process. Int. J.
Rock Mech. Min. Sci. 4, 395–406. https://doi.org/http://dx.doi.org/10.1016/0148-9062(67)90030-7

Bieniawski, Z.T., Bernede, M.J., 1979. Suggested methods for determining the uniaxial compressive
strength and deformability of rock materials. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 16, 137–

T
140. https://doi.org/10.1016/0148-9062(79)91450-5

IP
Blanks, R.F., McNamara, C.C., 1935. Mass concrete tests in large cylinders. ACI J. Proc. 31, 280–303.

CR
Brace, W.F., Paulding, B.W., Scholz, C., 1966. Dilatancy in the fracture of crystalline rocks. J. Geophys.
Res. 71, 3939–3953. https://doi.org/10.1029/JZ071i016p03939

US
Brown, E.T., 1981. Rock characterisation, testing and monitoring. ISRM suggested methods: Oxford:
Pergamon Press, 1981, 211P. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 18, 109.
AN
https://doi.org/10.1016/0148-9062(81)90524-6

Diederichs, M.S., 2007. Mechanistic interpretation and practical application of damage and spalling
M

prediction criteria for deep tunnelling. Can. Geotech. J. 44, 1082–1116.

Diederichs, M.S., Kaiser, P.K., Eberhardt, E., 2004. Damage initiation and propagation in hard rock
ED

during tunnelling and the influence of near-face stress rotation. Int. J. Rock Mech. Min. Sci. 41,
785–812. https://doi.org/10.1016/j.ijrmms.2004.02.003
PT

Duevel, B., Haimson, B., 1997. Mechanical characterization of pink Lac du Bonnet granite: evidence on
nonlinearity and anisotropy. Int. J. rock Mech. Min. Sci. Geomech. Abstr. 34, 543.
CE

https://doi.org/10.1016/S1365-1609(97)00241-4

Eberhardt, E., Stead, D., Stimpson, B., Read, R.S., 1998. Identifying crack initiation and propagation
AC

thresholds in brittle rock. Can. Geotech. J. 35, 222–233. https://doi.org/10.1139/t97-091

Eberhardt, E., Stimpson, B., Stead, D., 1999. Effects of Grain Size on the Initiation and Propagation
Thresholds of Stress-induced Brittle Fractures. Rock Mech. Rock Eng. 32, 81–99.
https://doi.org/10.1007/s006030050026

Eloranta, P., Hakala, M., 1999. Laboratory testing of Hiistholmen equigranular rapakivi granite in
borehole HH-KR6. Working Report 98-26. Posiva Oy, Helsinki.

Fairhurst, C.E., Hudson, J. a, 1999. Draft ISRM suggested method for the complete stress-strain curve for
intact rock in uniaxial compression. Int. J. Rock Mech. Min. Sci. 36, 281–289.
ACCEPTED MANUSCRIPT

https://doi.org/10.1016/0148-9062(87)91231-9

Ghazvinian, E., 2015. Fracture Initiation And Propagation In Low Porosity Crystalline Rocks :
Implications For Excavation Damage Zone (EDZ) Mechanics. Ph.D. Thesis. Queen's University.

Ghazvinian, E., Diederichs, M., Martin, D., 2012. Identification of Crack Damage Thresholds in
Crystalline Rock. In Proceedings of Eurock 2012, Stockholm, Sweden.

Ghazvinian, E., Perras, M.A., Diederichs, M., 2013. The Effect of Anisotropy on Crack Damage
Thresholds in Brittle Rocks, in: 47th US Rock Mechanics/Geomechanics Symposium.

T
IP
Griffith, A., 1921. Theory of rupture, in: Proceedings of the 1st International Congress on Applied
Mechanics. Delft, pp. 55–63.

CR
Güneş Yilmaz, N., Mete Goktan, R., Kibici, Y., 2011. Relations between some quantitative petrographic
characteristics and mechanical strength properties of granitic building stones. Int. J. Rock Mech.

US
Min. Sci. 48, 506–513. https://doi.org/10.1016/j.ijrmms.2010.09.003

Heikkila, E., Hakala, M., 1998. Laboratory testing of Kivetty granite in borehole KI-KR 1 0. Working
AN
Report 98-21e. Posiva Oy, Helsinki.

Helland, I.S., 1987. On the Interpretation and Use of R2 in Regression Analysis. Biometrics 43, 61–69.
M

https://doi.org/10.2307/2531949

Hoek, E., Martin, C.D., 2014. Fracture initiation and propagation in intact rock - A review. J. Rock Mech.
ED

Geotech. Eng. https://doi.org/10.1016/j.jrmge.2014.06.001

Hofmann, H., Babadagli, T., Yoon, J.S., Zang, A., Zimmermann, G., 2015. A grain based modeling study
PT

of mineralogical factors affecting strength, elastic behavior and micro fracture development during
compression tests in granites. Eng. Fract. Mech. 147, 261–275.
CE

https://doi.org/10.1016/j.engfracmech.2015.09.008

Keikha, T., Keykha, H.A., 2013. Correlation between mineralogical characteristics and engineering
AC

properties of granitic rocks. Electron. J. Geotech. Eng. 18 S, 4055–4065.

Lajtai, E.Z., 1974. Brittle fracture in compression. Int. J. Fract. 10, 525–536.
https://doi.org/10.1007/BF00155255

Liang, C., Wu, S., Li, X., Xin, P., 2015. International Journal of Rock Mechanics & Mining Sciences
Effects of strain rate on fracture characteristics and mesoscopic failure mechanisms of granite. Int. J.
Rock Mech. Min. Sci. 76, 146–154. https://doi.org/10.1016/j.ijrmms.2015.03.010

Martin, C.D., Chandler, N.A., 1994. The progressive fracture of Lac du Bonnet granite. Int. J. Rock
Mech. Min. Sci. 31, 643–659. https://doi.org/10.1016/0148-9062(94)90005-1
ACCEPTED MANUSCRIPT

Murrell, S.A.F., 1963. A criterion for brittle fracture of rocks and concrete under triaxial stress, and the
effect of pore pressure on the criterion, in: Fifth Symp. on Rock Mechanics. Pergamon Press, pp.
563–577.

Nasseri, M.H.B., Mohanty, B., 2008. Fracture toughness anisotropy in granitic rocks. Int. J. Rock Mech.
Min. Sci. 45, 167–193. https://doi.org/10.1016/j.ijrmms.2007.04.005

Nicksiar, M., Martin, C.D., 2014. Factors affecting crack initiation in low porosity crystalline rocks. Rock
Mech. Rock Eng. 47, 1165–1181. https://doi.org/10.1007/s00603-013-0451-2

T
Perras, M.A., Diederichs, M.S., 2014. A Review of the Tensile Strength of Rock: Concepts and Testing.

IP
Geotech. Geol. Eng. 32, 525–546. https://doi.org/10.1007/s10706-014-9732-0

CR
Stacey, T.R., 1981. A simple extension strain criterion for fracture of brittle rock. Int. J. Rock Mech. Min.
Sci. 18, 469–474. https://doi.org/10.1016/0148-9062(81)90511-8

US
Streckeisen, A., 1976. To each plutonic rock its proper name. Earth Sci. Rev. 12, 1–33.
https://doi.org/10.1016/0012-8252(76)90052-0
AN
Tabor, D., 1954. Mohs’s Hardness Scale -A Physical Interpretation Mohs’s Hardness Scale-A Physical
Interpretation. Proc. Phys. Soc. B 67.
M

Takarli, M., Prince, W., Siddique, R., 2008. Damage in granite under heating/cooling cycles and water
freeze-thaw condition. Int. J. Rock Mech. Min. Sci. 45, 1164–1175.
ED

https://doi.org/10.1016/j.ijrmms.2008.01.002

The MathWorks Inc., 2016. MATLAB R2016a.


PT

Tuǧrul, A., Zarif, I.H., 1999. Correlation of mineralogical and textural characteristics with engineering
properties of selected granitic rocks from Turkey. Eng. Geol. 51, 303–317.
CE

https://doi.org/10.1016/S0013-7952(98)00071-4

Turkish Standard TS 699, Natural stone test methods - determination of splitting tensile strength, 1987.
AC

Turkish Standard TS EN 1926, Natural stone test methods - determination of compressive strength, 2000.

Ulusay, R., Hudson, J.A., 2007. The ISRM Suggested Methods for Rock Characterization, Testing and
Monitoring: 1974-2006. Ankara, Turkey. https://doi.org/10.1007/978-3-319-07713-0

Yesiloglu-Gultekin, N., Sezer, E.A., Gokceoglu, C., Bayhan, H., 2013. An application of adaptive neuro
fuzzy inference system for estimating the uniaxial compressive strength of certain granitic rocks
from their mineral contents, in: Expert Systems with Applications. pp. 921–928.
https://doi.org/10.1016/j.eswa.2012.05.048

Zhao, J., Zhou, Y., Sun, J., Low, B.K., Choa, V., 1995. Engineering geology of the Bukit Timah Granite
ACCEPTED MANUSCRIPT

for cavern construction in Singapore. Q. J. Eng. Geol. 28, 153–162.

Zhao, X.G., Cai, M., Wang, J., Li, P.F., 2015. Strength comparison between cylindrical and prism
specimens of Beishan granite under uniaxial compression. Int. J. Rock Mech. Min. Sci. 76, 10–17.
https://doi.org/10.1016/j.ijrmms.2015.02.009

Zhao, X.G., Cai, M., Wang, J., Ma, L.K., 2013. Damage stress and acoustic emission characteristics of
the Beishan granite. Int. J. Rock Mech. Min. Sci. 64, 258–269.
https://doi.org/10.1016/j.ijrmms.2013.09.003

T
IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

APPENDICES
APPENDIX A MINERALOGICAL PARAMETERS AND MECHANICAL PROPERTIES
DATABASE FOR GRANITIC ROCK

T
IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

Table A1: Mineralogy, average Mohs Hardness, and reference for each rock type in the database. Blank
entries represent values that were not provided by the author(s) of the study in question and could not be
estimated or back-calculated.

Average
Granite Name Alkali Plagioclase Micas Mohs Reference
Quartz (%)
Feldspar (%) (%) (%) Hardness
Amarelo Pais 26.0 25.0 22.0 22.0 5.36 (Arzúa and Alejano, 2013)

Blanco Mera 20.0 27.0 35.0 12.0 5.61 (Arzúa and Alejano, 2013)

T
Vilachan Granite 41.0 15.0 12.0 20.0 5.11 (Arzúa and Alejano, 2013)

IP
(Zhao et al., 2013)
Beishan 17.0 15.0 52.0 12.0 5.71
(Eberhard t et al., 1998)
Lac du Bonnet Lvl 130 31.0 27.0 38.0 4.0 6.34
(Duevel and Haimson, 1997)
Lac du Bonnet 30.0 40.0 20.0 10.0 6.13

CR
(Liang et al., 2015)
Liaoning Province 40.0 20.0 35.0 3.0 6.32
Smaland Granite (Ghazvinian et al., 2012)

Bukit Timah Granite 30.0 60.0 0.0 10.0 6.13 (Zhao et al., 1995)
(Takarli et al., 2008)

US
French Pyrenees G-1 40.0 23.0 31.0 6.0 6.34
(Takarli et al., 2008)
French Pyrenees G-2 30.0 10.0 56.0 4.0 6.34
(Yesiloglu -G ult ek in et al., 2013)
Bergama 26.7 20.1 35.3 5.33
(Yesiloglu -G ult ek in et al., 2013)
Camili 31.4 12.6 45.0 5.79
AN
Yaylak 32.4 13.5 47.9 6.10 (Yesiloglu -G ult ek in et al., 2013)

Koprukoy 28.8 20.7 36.5 5.59 (Yesiloglu -G ult ek in et al., 2013)


(Yesiloglu -G ult ek in et al., 2013)
Keslon 25.2 16.9 44.2 5.58
(Yesiloglu -G ult ek in et al., 2013)
Sivriliisar 34.5 4.6 50.7 5.87
M

(Güneş Yilmaz et al., 2011)


Alkali Granite 36.0 48.0 9.0 6.0 6.25
Monzonite 27.0 28.0 32.0 7.0 5.83 (Güneş Yilmaz et al., 2011)

Monzonite 24.0 30.0 29.0 5.0 5.51 (Güneş Yilmaz et al., 2011)
ED

(Güneş Yilmaz et al., 2011)


Granodiorite 25.0 22.0 39.0 12.0 5.89
(Güneş Yilmaz et al., 2011)
Porphiroid Granite 20.0 56.0 14.0 2.0 5.83
(Güneş Yilmaz et al., 2011)
Calc-Alkali Granite 12.0 62.0 17.0 7.0 5.97
(Güneş Yilmaz et al., 2011)
PT

Monzo-Granite 11.0 30.0 41.0 4.0 5.32


Monzo-Granite 18.0 36.0 25.0 5.07 (Güneş Yilmaz et al., 2011)

Granodiorite 22.0 24.0 40.0 8.0 5.76 (Güneş Yilmaz et al., 2011)
(Güneş Yilmaz et al., 2011)
Calc-Alkali Granite 40.0 41.0 12.0 6.0 6.28
CE

(Güneş Yilmaz et al., 2011)


Porphiroid Granite 21.0 55.0 15.0 7.0 6.04
(Güneş Yilmaz et al., 2011)
Granodiorite 15.0 16.0 54.0 6.0 5.59
(Keikha and Keykha, 2013)
Bouk 26.8 38.6 10.0 20.0 5.46
(Keikha and Keykha, 2013)
AC

Luchan 13.6 17.8 51.0 17.6 5.74


(Nasseri and Mohanty, 2008)
Laurentian Pink 32.3 30.5 30.5 3.7 6.18
(CANMET)
Pinawa Pink 25.8 28.1 33.9 7.7 5.89 (CANMET)
(Ghazvinian, 2015) (CANMET)
Westerly Grey 30.0 30.0 30.0 10.0 6.13
(Nasseri and Mohanty, 2008)
Stanstead Granite 24.7 33.3 33.3 8.7 6.13
(CANMET)
(Nicksiar and Martin, 2014)
Aspo Diorite 19.0 8.0 73.0 0.0 6.39
Beishan Granite 31.1 13.6 33.7 20.1 5.68 (Zhao et al., 2015)

Table A1 (cont.): Mineralogy, average Mohs Hardness, and reference for each rock type in the database.
Blank entries represent values that were not provided by the author(s) of the study in question and could
ACCEPTED MANUSCRIPT

not be estimated or back-calculated.

Average
Alkali Plagioclase Micas References
Granite Name Quartz (%) Mohs
Feldspar (%) (%) (%)
Hardness
6.28 (Heikkila and Hakala, 1998)
Kivetty Granite 27.0 34.0 34.0 5.0
6.24 (Eloranta and Hakala, 1999)
Hastholmen 35.0 41.0 17.0 6.0
6.11 (Tuǧrul and Zarif, 1999)
Quartz monzonite- 15.6 39.3 40.3 1.5
monzonite (Gr1)
(Tuǧrul and Zarif, 1999)
Quartz monzodiorite (Gr2) 15.6 18.4 59.5 3.3 6.05
Quartz monzonite (Gr3) 16.5 38.4 39.1 2.5 6.07 (Tuǧrul and Zarif, 1999)

T
(Tuǧrul and Zarif, 1999)
Quartz diorite (Gr4) 12.0 75.3 5.5 2.4 5.96
(Tuǧrul and Zarif, 1999)

IP
Quartz monzodiorite (Gr5) 9.5 5.5 75.5 3.8 5.83
(Tuǧrul and Zarif, 1999)
Quartz monzodiorite (Gr6) 14.3 18.5 59.5 2.9 5.96
(Tuǧrul and Zarif, 1999)
Granite (Gr7) 32.8 31.5 29.5 2.5 6.18

CR
(Tuǧrul and Zarif, 1999)
Granite (Gr8) 32.2 34.0 28.2 2.5 6.21
Monzogranite (Gr9) 25.3 45.3 21.3 3.7 6.04 (Tuǧrul and Zarif, 1999)

Granite monzogranite 26.4 43.3 22.8 3.6 6.08 (Tuǧrul and Zarif, 1999)

(Gr10)

US
(Tuǧrul and Zarif, 1999)
Granite (Gr11) 31.9 36.3 25.5 3.1 6.18
(Tuǧrul and Zarif, 1999)
Quartz monzonite (Gr12) 14.6 37.2 40.5 2.7 5.95
(Tuǧrul and Zarif, 1999)
Quartz monzonite (Gr13) 17.7 35.3 39.2 2.4 5.96
(Tuǧrul and Zarif, 1999)
Granodiorite (Gr14) 25.3 18.8 45.3 3.0 5.86
AN
(Tuǧrul and Zarif, 1999)
Granite (Gr15) 32.3 34.2 28.5 2.3 6.24
(Tuǧrul and Zarif, 1999)
Tonalite (Gr16) 10.1 5.3 74.4 4.2 5.80
(Tuǧrul and Zarif, 1999)
Granodiorite (Gr17) 19.0 17.0 55.2 3.8 5.95
M

(Tuǧrul and Zarif, 1999)


Granodiorite (Gr18) 30.2 17.3 46.3 2.4 6.16
(Tuǧrul and Zarif, 1999)
Granodiorite (Gr19) 31.4 12.1 50.3 2.5 6.17
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

Table A2: Mineral specific and overall average grain sizes and standard deviation information
for each rock type. Blank entries in Table A2 represent values that were not provided by the
author(s) of the study in question and could not be estimated based on thin section images.
Quartz Alkali Plagioclase Micas (mm)
Overall (mm)
(mm) Feldspar (mm) (mm)
Granite Av Std Avg. Std Av Std Av Std Mi M Av Std.
Name g. . . g. . g. . n ax g. Dev
De De De De .
v. v. v. v.
Amarelo 1. 3.0 2.0 0.50
Pais 0
Blanco 1. 6.0 3.5 1.25

T
Mera 0

IP
Vilachan 0. 1.0 0.8 0.13
Granite 5
Beishan 0.5 0.1 0. 0.1 1.1 0.4 0.4 0.2 0. 1.4 0.9 0.28

CR
5 6 9 5 2 3
Lac du 3.5 0.50
Bonnet
Lvl 130
Lac du 0. 1.0 0.8 0.13

US
Bonnet 5
Liaoning 0.5 0.1 0. 0.3 0.2 0. 0.7 0.4 0.13
Province 6 3 4 2
Smaland 0. 1.0 0.8 0.13
AN
Granite 5
Bukit 2. 5.0 3.5 0.75
Timah 0
Granite
M

French 0. 0.8 0.5 0.15


Pyrenees 2
G-1
ED

French 0. 1.5 0.9 0.30


Pyrenees 3
G-2
Bergama
PT

Camili
Yaylak
Koprukoy
Keslon
CE

Sivriliisar
Alkali 7.5 0.6 15.0 1.6 1.6 0.1 0.6 10. 5.22
Granite 6 0 7 1
AC

Monzonite 5.0 0.7 8. 4.0 1.7 0.3 4.9 2.41


1 0 5
Monzonite 1.0 0.1 0. 0.1 1.0 0.4 0.2 0.7 0.41
4 5 7 7
Granodiori 2.5 0.3 10.0 3.5 0.6 0.4 4.2 3.22
te 4 3
Porphiroid 5.0 25.0 1.4 0.4 0.4 15. 10.8
Granite 0 1 6
Calc- 6.0 10.0 0.7 5.0 1.4 1.3 7.9 2.93
Alkali 1 8
Granite
Monzo- 2.0 0.2 2. 2.1 5.0 0.8 2.9 2.00
Granite 2 0 2
Monzo- 4.5 1. 0.8 0.9 1.5 1.53
Granite 3 5
Granodiori 4.0 7. 1.1 4.5 0.4 0.6 4.5 2.06
ACCEPTED MANUSCRIPT

te 5 9 2
Calc- 3.5 0.4 12.0 2.3 1.9 1.5 6.6 4.81
Alkali 7 3
Granite
Porphiroid 3.5 0.3 20.0 0.6 8.0 0.5 0.5 0.4 13. 7.88
Granite 8 5 7 1 0
Granodiori 2.5 0.5 9. 0.1 3.5 0.4 1.6 3.8 2.32
te 1 0 7 2
Bouk 4.7 0.8 4. 0.4 3.4 0.8 2.2 0.5 3.6 1.13
5 0 4 0 2
Luchan 2.7 0.7 1. 0.5 2.4 0.5 1.2 0.5 2.1 0.86
0 3 5 7 7

T
Laurentian 0.5 0.0 0. 0.0 0.4 0.0 0.3 0.0 0.4 0.09
Pink 9 4 7 7 9

IP
Pinawa
Pink
Westerly 0. 2.2 1.4 0.43

CR
Grey 5
Stanstead 0.9 0.0 1. 0.1 1.5 0.1 0.6 0.0 1.3 0.33
Granite 1 5 2 2 1
Aspo

US
Diorite
Another
Beishan
Kivetty 6.5 1.75
AN
Granite

Table A2 (cont.): Mineral specific and overall average grain sizes and standard deviation information
M

for each rock type. Blank entries in Table A2 represent values that were not provided by the author(s)
of the study in question and could not be estimated based on thin section images.
ED

Alkali
Quartz Micas
Feldspar Plagioclas Overall (mm)
(mm) (mm)
Granite (mm) e (mm)
Av Std Std Av Std Std Mi M Av Std
PT

Name
g. . . g. . Av . n ax g. .
De Avg. De De g. De De
v. v. v. v. v.
Hastholmen 1.0 3.0 2.0 0.5
CE

0
Quartz 0.9 0.2 2. 1.1 0.8 0.2 0.8 0.2 0.2 4.9 1.5 1.1
monzonite- 8 6 5 8 3 8
monzonite
AC

(Gr1)
Quartz 0.6 0.2 4. 1.8 1.3 0.4 1.3 0.3 0.2 8.0 1.7 1.9
monzodiori 0 4 3 3 5 5
te (Gr2)
Quartz 2.2 0.8 4. 2.0 1.6 0.6 0.6 0.2 0.2 8.8 2.8 2.1
monzonite 0 7 8 0 0 5
(Gr3)
Quartz 0.7 0.1 3. 1.4 0.3 6.7 2.9 1.6
diorite 8 8 8 0
(Gr4)
Quartz 0.7 0.2 2. 0.9 0.9 0.2 0.5 0.1 0.1 4.0 0.8 0.9
monzodiori 5 2 0 8 8 8
te (Gr5)
Quartz 0.7 0.1 3. 1.3 1.1 0.2 1.0 0.4 0.2 6.2 1.4 1.5
monzodiori 8 5 5 8 0 0
ACCEPTED MANUSCRIPT

te (Gr6)
Granite 1.1 0.3 2. 0.8 0.9 0.2 0.7 0.1 0.3 4.0 1.3 0.9
(Gr7) 8 3 5 8 8 3
Granite 0.5 0.1 1. 0.4 1.3 0.4 0.2 0.0 0.1 2.2 0.9 0.5
(Gr8) 5 2 8 8 5 3
Monzogran 0.8 0.3 1. 0.3 0.8 0.2 0.2 0.0 0.1 1.6 0.8 0.3
ite (Gr9) 3 0 3 8 3 8
Granite 0.7 0.1 2. 1.0 0.4 0.1 0.5 0.1 0.1 4.8 1.4 1.1
monzograni 8 7 8 5 8 8
te (Gr10)
Granite 0.8 0.3 1. 0.3 0.5 0.1 0.5 0.2 0.1 1.8 0.8 0.4
(Gr11) 3 1 8 0 0 3

T
Quartz 0.6 0.1 2. 0.7 1.0 0.4 0.4 0.1 0.1 3.6 1.3 0.8
monzonite 8 2 3 3 0 8

IP
(Gr12)
Quartz 0.5 0.1 2. 0.7 0.8 0.2 0.5 0.1 0.2 3.8 1.2 0.9
monzonite 5 3 5 8 0 0

CR
(Gr13)
Granodiorit 0.4 0.1 2. 0.8 1.2 0.4 0.6 0.1 0.1 4.0 1.1 0.9
e (Gr14) 5 3 8 0 3 8
Granite 0.6 0.1 1. 0.3 0.4 0.1 0.7 0.1 0.1 1.6 0.6 0.3

US
(Gr15) 8 0 0 5 8 8
Tonalite 1.6 0.6 1. 0.5 2.5 0.9 0.6 0.1 0.3 4.4 2.1 1.0
(Gr16) 3 5 3 8 5 3
Granodiorit 0.9 0.2 2. 0.7 1.6 0.6 0.7 0.2 0.2 3.4 1.4 0.8
AN
e (Gr17) 3 0 3 3 5 0
Granodiorit 1.2 0.4 1. 0.4 2.7 1.1 0.4 0.1 0.2 5.0 1.9 1.2
e (Gr18) 0 6 8 5 0 0
Granodiorit 1.3 0.5 1. 0.6 2.9 1.2 0.3 0.0 0.1 5.4 2.1 1.3
M

e (Gr19) 8 9 5 8 3 3
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

Table A3: Mechanical properties of rock types in the database. Blank entries correspond to data not
provided by authors.
Uncon Cra Bra
Elasti Crack
No. fined Poi ck zili
c Damag m
Sam Compre sso Init an
Mod e i
ples ssive n's iati Te
ulus, Stress,
Strengt Rat on nsil
E CD
h, UCS io, Str e
(MPa (MPa)
(MPa) ν ess Str
)
, eng

T
Granite CI th,
Name (MPa) BTS (MPa)

IP
S S S S S S
Av Avg. Avg. Av A Avg. Avg.
t td t t t td
g. g. vg

CR
d. . d d. d. .
.
D D . D D D
e e D e e e
v. v. e v. v. v.

US
v.
Amarelo 12 76. 3.87 0.17 0.02 18.97 2.43 26. 2.38 51.0 6.31 6.65 40.96
Pais 93 75 8
Blanco 12 108 15.37 0.14 0.03 30.64 3.23 38. 8.50 78.7 12.84 6.12 37.82
Mera .96 59 2
AN
Vilacha 12 116 7.62 0.17 0.05 23.48 1.92 42. 7.06 84.5 13.62 6.93 28.23
n .07 51 5
Granite
Beishan 6 145 13.69 0.14 0.02 63.49 1.73 74. 9.97 113. 13.08
M

.99 19 73
Lac du 20 206 10.00 0.31 0.04 66.10 2.50 81. 3.70 157. 9.90
Bonnet .50 50 30
ED

Lvl 130
Lac du 46 219 9.00 0.25 0.03 55.60 5.90 11.40 8.80 37.20
Bonnet .00
Liaonin 11 155 31.66 0.14 51.76 92. 29.02 135.
PT

g .44 42 00
Province
Smaland 15 240 28.82 0.27 0.04 70.64 9.79 121 18.29 202. 20.11
Granite .06 .52 10
CE

Bukit 186 18.25 0.25 0.04 84.00 16.00 11.40 0.45


Timah .00
Granite
French 12 247 11.20 71.50 7.50 92. 157.
AC

Pyrenee .30 74 04
s G-1
French 12 211 1.20 64.00 2.00 80. 130.
Pyrenee .80 91 68
s G-2
Bergam 25 133 17.83
a .10
Camili 3 73. 8.28
46
Yaylak 10 87. 15.88
85
Kopruko 8 106 25.61
y .40
Keslon 12 103 22.44
.50
ACCEPTED MANUSCRIPT

Sivriliis 17 120 14.67


ar .70
Alkali 10 143 5.78 9.06 8.39
Granite .19
Monzon 10 145 14.72 8.20 5.00
ite .20
Monzon 10 185 9.38 11.72 5.20
ite .96
Granodi 10 196 17.98 9.22 4.55
orite .74
Porphiro 10 152 16.05 9.66 18.01
id .77

T
Granite
Calc- 10 121 14.10 6.88 26.89

IP
Alkali .37
Granite
Monzo- 10 172 12.32 11.02 14.06

CR
Granite .28
Monzo- 10 164 8.19 14.15 10.53
Granite .25
Granodi 10 193 10.65 8.21 18.03

US
orite .51
Calc- 10 138 15.22 3.13 22.68
Alkali .82
Granite
AN
Porphiro 10 128 12.38 10.16 9.94
id .87
Granite
Granodi 10 162 14.03 7.88 15.36
M

orite .46
Bouk 5 86. 18.51 3.90
70
Luchan 5 64. 16.03 3.43
ED

00
Laurenti 3 248 4.04 0.35 0.02 63.23 0.26 33.19
an Pink .56
Pinawa 3 259 6.11 0.29 0.03 71.44 2.17 28.29
PT

Pink .19
Westerl 4 160 2.03 0.41 0.04 49.82 0.40 70. 7.20 127. 10.20 8.07 0.30 40.06
y Grey .79 20 00
Stanstea 4 137 1.55 0.17 0.02 45.66 1.30 53. 1.22 105. 3.11 7.68 0.59
CE

d .11 94 07
Granite
AC

Table A3 (cont.): Mechanical properties of rock types in the database. Blank entries correspond to data
not provided by authors.
Unco Cra Bra
Elast Crack
No. nfined Poi ck zili
ic Damag m
Sam Compr sso Init an
Mod e i
ples essive n's iati Te
ulus, Stress,
Strengt Rat on nsil
Granite E CD
h, UCS io, Str e
Name (MP (MPa)
(MPa) ν ess Str
a)
, eng
CI th,
(MPa) BTS (MPa)
ACCEPTED MANUSCRIPT

S S
Avg. A Avg.
td t Std
Std vg
Av Std. Std.
.
.
.
d. .
g. Dev. Dev. Avg. Avg. D D Avg. De
De
e e v.
v.
v. v.
Aspo 169.0 146.
10 14.00 0.22 0.04 61.00 2.00 64.00 6.90 3.00
Diorite 0 00
Another
4 79.75 13.02 0.18 0.04 52.10 4.01 0.66
Beishan 5.05
Kivetty 178.0 136. 1.91
14 18.11 0.25 0.01 65.00 2.81 74.00 5.61 13.78
Granite 0 00 9.40

T
Hastholm 194.0 146.
9 15.31 0.25 0.01 71.00 2.04 72.00 2.04 12.76 1.68
en 0 00 11.30

IP
Australia 238.6
n Granite 5 15.80 0.22 0.03 79.60 1.30 1.57
0 13.46

CR
Quartz
monzonit
158.1
e- 11.26 44.80 1.23 1.01
6 20.45
monzonit

US
e (Gr1)
Quartz
monzodi 125.2
8.03 30.10 0.21 0.36
orite 8 16.40
(Gr2)
AN
Quartz
129.0
monzonit 5.68 29.80 0.78 0.46
4 19.88
e (Gr3)
Quartz
M

142.4
diorite 3.67 38.70 2.27 0.89
0 19.60
(Gr4)
Quartz
ED

monzodi 109.1 1.06


8.42 30.31 4.80
orite 7 14.89
(Gr5)
Quartz
PT

monzodi 121.2
8.94 34.00 1.84 1.21
orite 8 15.60
(Gr6)
Granite 165.7
CE

(Gr7) 7.17 2.07 0.09


9 45.09 23.15
Granite 193.3
(Gr8) 21.92 3.00 1.12
3 58.50 28.03
AC

Monzogr
173.4 1.56 2.49
anite 14.63
0 45.31 21.25
(Gr9)
Granite
monzogr 170.5
6.10 1.38 1.50
anite 6 52.85 25.88
(Gr10)
Granite 175.4
(Gr11) 24.12 2.95 0.62
9 51.20 27.28
Quartz
159.5
monzonit 2 10.00 3.69 2.45
e (Gr12) 43.12 21.80
Quartz
129.6
monzonit 9.76 2.24 18.1 1.60
0 35.39
e (Gr13) 8
Granodio 150.0 10.98 2.88 2.33
ACCEPTED MANUSCRIPT

rite 8 38.80 22.2


(Gr14) 3
Granite 175.8
(Gr15) 7.63 3.09 0.52
5 51.80 24.55
Tonalite
157.6 1.11
(Gr16) 15.35 22.1 0.69
2 48.61
5
Granodio
136.3
rite 13.09 0.97 21.6 1.34
2 31.18
(Gr17) 0
Granodio
177.5
rite 16.04 1.90 23.4 1.30
2 47.20

T
(Gr18) 8
Granodio
178.5

IP
rite 6.48 1.41 0.29
6 51.30 26.05
(Gr19)

CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

APPENDIX B – MULTIVARIATE REGRESSION RESULTS FOR THREE AND FOUR


PREDICTOR VARIABLES

T
IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

Table B1: Multivariate analysis results considering three predictor variables. Models are ordered by
descending r2 -adjusted value.

r2 - No.
Model Predictor Variables Response Regression Equation r2
Adjusted Obs.
No. Variable
Overall Minimum Grain Size -
GSMin (mm) Crack Initiation CI = 108.8 - 4.8
1 Overall Average Grain Size - Stress - CI GSMin - 9.3 0.963 0.941 9
GSAvg (mm) (MPa) GSAvg - 2.6 Mica%
Mica Modal Percentage - Mica%
(%)

T
Overall Minimum Grain Size -
GSMin (mm) Crack Initiation CI = 108.9 - 14.1

IP
2 Overall Std. Deviation Grain Size Stress - CI GSMin - 18.7 0.963 0.941 9
- GSDev (mm) (MPa) GSDev - 2.6 Mica%
Mica Modal Percentage - Mica%

CR
(%)
Overall Minimum Grain Size -
GSMin (mm) Crack Initiation CI = 101.4 - 28.7
3 Quartz Modal Percentage - Qtz% Stress - CI GSMin + 0.936 0.897 9

US
(%) (MPa) 0.2 Qtz% - 2.4 Mica%
Mica Modal Percentage - Mica%
(%)
Overall Minimum Grain Size -
AN
GSMin (mm) Crack Damage CD = 170.3 + 14.0
4 Overall Std. Deviation Grain Size Stress - CD GSMin - 0.868 0.789 9
- GSDev (mm) (MPa) 40.0 GSDev - 4.6
Mica Modal Percentage - Mica% Mica%
M

(%)
Overall Average Grain Size -
GSAvg (mm) Crack Damage CD = 164.6 + 11.9
Stress - CD GSAvg -
ED

5 Overall Std. Deviation Grain Size 0.846 0.789 12


- GSDev (mm) (MPa) 51.5 GSDev - 4.5
Mica Modal Percentage - Mica% Mica%
(%)
PT

Overall Minimum Grain Size -


GSMin (mm) Crack Damage CD = 170.3 + 33.7
6 Overall Average Grain Size - Stress - CD GSMin - 0.867 0.788 9
GSAvg (mm) (MPa) 19.9 GSAvg - 4.6
CE

Mica Modal Percentage - Mica% Mica%


(%)
Overall Average Grain Size -
GSAvg (mm) Crack Initiation CI = 101.6 + 1.4
Stress - CI
AC

7 Overall Std. Deviation Grain Size GSAvg - 0.828 0.764 12


- GSDev (mm) (MPa) 17.7 GSDev - 3.0
Mica Modal Percentage - Mica% Mica%
(%)
Overall Minimum Grain Size -
GSMin (mm) Crack Damage CD = 138.5 -
8 Quartz Modal Percentage - Qtz% Stress - CD 14.2 GSMin + 0.838 0.740 9
(%) (MPa) 0.8 Qtz% - 3.9
Mica Modal Percentage - Mica% Mica%
(%)
Overall Minimum Grain Size -
GSMin (mm) Brazilian Tensile BTS = 19.8 - 4.8
9 Quartz Modal Percentage - Qtz% Strength GSMin + 0.758 0.725 26
(%) - BTS (MPa) 0.2 Qtz% - 0.9 Mica%
Mica Modal Percentage - Mica%
(%)
ACCEPTED MANUSCRIPT

Overall Average Grain Size -


GSAvg (mm) Brazilian Tensile BTS = 25.8 - 1.6
10 Overall Std. Deviation Grain Size Strength GSAvg + 0.729 0.698 30
- GSDev (mm) - BTS (MPa) 0.03 GSDev – 1.0
Mica Modal Percentage - Mica% Mica%
(%)
Overall Minimum Grain Size -
GSMin (mm) Brazilian Tensile BTS = 26.8 - 4.3
11 Overall Std. Deviation Grain Size Strength GSMin - 0.714 0.675 26
- GSDev (mm) - BTS (MPa) 2.2 GSDev - 0.9 Mica%
Mica Modal Percentage - Mica%
(%)

T
Overall Minimum Grain Size -
GSMin (mm) Brazilian Tensile BTS = 24.9 - 3.9

IP
12 Overall Average Grain Size - Strength GSMin - 0.693 0.651 26
GSAvg (mm) - BTS (MPa) 0.4 GSAvg - 0.8 Mica%
Mica Modal Percentage - Mica%

CR
(%)

US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

Table B2: Multivariable analysis results considering four predictor variables. Models are ordered by
descending r2 -adjusted value.

r2 - No.
Model Predictor Variables Response Regression r2
Adjusted Obs.
No. Variable Equation
Overall Minimum Grain Size -
GSMin (mm) CI = 118.7 - 10.8
Overall Std. Deviation Grain Crack GSMin -
1 Size - GSDev (mm) 24.8 GSDev - 0.968 0.936 9
Initiation
Quartz Modal Percentage - Stress - CI 0.3 Qtz% - 2.8

T
Qtz% (%) Mica%
(MPa)

IP
Mica Modal Percentage -
Mica% (%)
Overall Minimum Grain Size -

CR
GSMin (mm) CI = 118.7 + 1.5
Overall Average Grain Size - Crack GSMin -
2 GSAvg (mm) 12.4 GSAvg - 0.968 0.935 9
Initiation
Quartz Modal Percentage - Stress - CI 0.03 Qtz% -

US
Qtz% (%) 2.8 Mica%
(MPa)
Mica Modal Percentage -
Mica% (%)
Overall Minimum Grain Size -
AN
GSMin (mm) CI = 110.9 - 511.5
Overall Average Grain Size - Crack GSMin +
3 GSAvg (mm) 497.1 GSAvg - 0.967 0.933 9
Initiation
Overall Std. Deviation Grain Stress - CI 1014 GSDev -
M

Size - GSDev (mm) 2.7 Mica%


(MPa)
Mica Modal Percentage -
Mica% (%)
ED

Overall Minimum Grain Size -


GSMin (mm) CD = 182.0 -
Overall Average Grain Size - Crack 2845 GSMin +
4 GSAvg (mm) 2857 GSAvg - 0.925 0.849 9
Damage 5759 GSDev -
PT

Overall Std. Deviation Grain Stress - CD


Size - GSDev (mm) 5.1 Mica%
(MPa)
Mica Modal Percentage -
Mica% (%)
CE

Overall Average Grain Size -


GSAvg (mm) CD = 141.0 +
Overall Std. Deviation Grain Crack 10.3 GSAvg -
5 Size - GSDev (mm) 41.2 GSDev + 0.7 0.864 0.786 12
Damage
AC

Quartz Modal Percentage - Stress - CD Qtz% -


Qtz% (%) (MPa) 4.4 Mica%
Mica Modal Percentage -
Mica% (%)
Overall Average Grain Size -
GSAvg (mm) CI = 91.9 + 0.7
Overall Std. Deviation Grain Crack GSAvg -
6 Size - GSDev (mm) 13.5 GSDev + 0.835 0.741 12
Initiation
Quartz Modal Percentage - Stress - CI 0.3 Qtz% - 3.0
Qtz% (%) Mica%
(MPa)
Mica Modal Percentage -
Mica% (%)
Overall Minimum Grain Size -
GSMin (mm) CD = 163.6 +
ACCEPTED MANUSCRIPT

Overall Std. Deviation Grain Crack 11.8 GSMin -


7 Size - GSDev (mm) 0.869 0.738 9
Damage 35.9 GSDev + 0.2
Quartz Modal Percentage - Stress - CD Qtz% -
Qtz% (%) (MPa) 4.5 Mica%
Mica Modal Percentage -
Mica% (%)
Overall Minimum Grain Size -
GSMin (mm) CD = 163.3 +
Overall Average Grain Size - Crack 29.2 GSMin -
8 GSAvg (mm) 17.7 GSAvg + 0.2 0.868 0.737 9
Damage
Quartz Modal Percentage - Stress - CD Qtz% -
Qtz% (%) (MPa) 4.5 Mica%

T
Mica Modal Percentage -

IP
Mica% (%)
Overall Average Grain Size -
GSAvg (mm) BTS = 18.9 -
2.1 GSAvg + 2.7

CR
Overall Std. Deviation Grain Brazilian
9 Size - GSDev (mm) GSDev + 0.2 0.766 0.729 30
Tensile Qtz% -
Quartz Modal Percentage - Strength -
Qtz% (%) 0.9 Mica%
BTS (MPa)

US
Mica Modal Percentage -
Mica% (%) AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

Table B2 (cont.): Multivariable analysis results considering four predictor variables. Models are ordered
by descending r2 -adjusted value.
Model r2 - No.
Predictor Variables Response Regression r2
No. Adjusted Obs.
Variable Equation
Overall Minimum Grain Size
- GSMin (mm) BTS = 17.4 -
Overall Average Grain Size - Brazilian 6.7 GSMin +
10 GSAvg (mm) 1.2 GSAvg + 0.767 0.726 26
Tensile
Quartz Modal Percentage - Strength - BTS 0.3 Qtz% -
Qtz% (%) 0.9 Mica%
(MPa)
Mica Modal Percentage -

T
Mica% (%)
Overall Minimum Grain Size

IP
- GSMin (mm) BTS = 20.1 -
Overall Std. Deviation Grain Brazilian 4.5 GSMin -
11 0.1 GSDev + 0.758 0.712 26

CR
Size - GSDev (mm) Tensile
Quartz Modal Percentage - Strength - BTS 0.2 Qtz% -
Qtz% (%) 0.9 Mica%
(MPa)
Mica Modal Percentage -

US
Mica% (%)
Overall Minimum Grain Size
- GSMin (mm) BTS = 26.5 -
Overall Average Grain Size - Brazilian 8.8 GSMin +
12 3.1 GSAvg - 0.738 0.688 26
AN
GSAvg (mm) Tensile
Overall Std. Deviation Grain Strength - BTS 5.5 GSDev -
Size - GSDev (mm) 0.9 Mica%
(MPa)
Mica Modal Percentage -
M

Mica% (%)
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

LIST OF FIGURES:

Figure 1: Quartz - Alkali Feldspar - Plagioclase Feldspar (QAP) plot of coarse grained igneous rocks.
Granitic rocks are classified according to their mineralogical composition using Streckeisen’s
classification system (Streckeisen, 1976), which follows the International Union of Geological Sciences
(IUGS) classification system for plutonic rocks. Granitic rock types considered in this study are shown as
red circles.

T
IP
Figure 2: Geographic origin of granitic rocks considered in this study. Associated references are
included within the figure as well as how many rock types each reference includes if there are more than

CR
one (which is referenced in parentheses next to the rock name).

US
Figure 3: Hand samples of selected rock types considered in this study. The rocks shown illustrate the
variety in grain size and mineralogy of the granitic rock types considered. A – Amarelo Pais; Monzo-
AN
granite; average grain size: 2.0 mm (Arzúa and Alejano, 2013); B – Stanstead; Monzo-granite; average
grain size: 1.3 mm (CANMET); C – Beishan; Quartz Monzodiorite/Monzogabbro; average grain size:
M

0.9 mm (Zhao et al., 2013); D – Aspo Diorite; Quartz Diorite/Gabbro/Anorthosite; no grain size reported
(Nicksiar and Martin, 2014); E – Smaland; Granite type not specified; average grain size: 0.8 mm
ED

(Ghazvinian et al., 2012); F – Westerly; Monzo-granite; average grain size: 1.4 mm (CANMET); G –
Vilachan; Monzo-granite; average grain size: 0.8 mm (Arzúa and Alejano, 2013).
PT

Figure 4: Thin sections of selected rocks to illustrate variety in grain size and mineralogy of the granitic
CE

rock types considered in this study. A – Bouk; Syeno-granite; average grain size: 3.6 mm (Keikha and
Keykha, 2013); B – Beishan; Quartz Monzodiorite/Monzogabbro; average grain size: 0.9 mm (Zhao et
al., 2013); C – Blanco Mera; Monzo-granite; average grain size: 3.5 mm (Arzúa and Alejano, 2013); D -
AC

Liaoning Province Granite; Monzo-granite; average grain size: 0.4 mm (Liang et al., 2015); E –
Vilachan; Monzo-granite; average grain size: 0.8 mm (Arzúa and Alejano, 2013); F - Quartz Monzonite
(Gr12); Quartz Monzonite; average grain size: 1.3 mm (Tuǧrul and Zarif, 1999); G – Quartz
Monzonite/Monzonite (Gr1); Quartz Monzonite; average grain size: 1.5 mm (Tuǧrul and Zarif, 1999); H
– Granite Monzo-granite (Gr10); Syeno-granite; average grain size: 1.4 mm (Tuǧrul and Zarif, 1999); I -
Granite (Gr7); Monzo-granite; average grain size: 1.3 mm (Tuǧrul and Zarif, 1999).

Figure 5: Stages in the progressive brittle failure of intact rock specimens subjected to compressive
ACCEPTED MANUSCRIPT

loading (after Hoek and Martin, 2014).

Figure 6: Graphs comparing relationships previously proposed by Tuǧrul and Zarif (1999) with the
database compiled for this study. (a) Alkali Feldspar Average Grain Size (AKGS) versus Unconfined
Compressive Strength (UCS); (b) Plagioclase Average Grain Size (PlagGS) versus UCS; (c) Quartz :
Feldspar Modal Percentage (QFR) versus UCS; (d) QFR versus Brazilian Tensile Strength (BTS).

T
Notes: 1 Blue circles and dashed blue lines indicate regression data from Tuǧrul and Zarif (1999), and

IP
black squares and solid black lines indicate regression data from all other available data. 2 ‘All other data’
considers all data, other than data from Tuǧrul and Zarif (1999), that could possibly be used with the

CR
variables for a specific regression. 3 When considering QFR, ‘Feldspar’ is inclusive of Alkali and
Plagioclase Feldspars.

US
Figure 7: Graphs comparing relationships previously proposed by Güneş Yilmaz et al. (2011) and Tuǧrul
AN
and Zarif (1999) with the database compiled for this study. (a) Quartz Average Grain Size (QtzGS) versus
Unconfined Compressive Strength (UCS); (b) Alkali Feldspar Modal Percentage (AK%) versus UCS; (c)
M

Mica Average Grain Size (Mica GS) versus Brazilian Tensile Strength (BTS).
Notes: 1 Blue circles and dashed blue lines indicate regression data from Tuǧrul and Zarif (1999), red
ED

diamonds and dotted red lines indicate regression data from Güneş Yilmaz et al. (2011), and black
squares and solid black lines indicate regression data from all other available data. 2 ‘All other data’
considers all data, other than data from Tuǧrul and Zarif (1999) and Güneş Yilmaz et al. (2011), that
PT

could possibly be used with the variables for a specific regression. The only exception is in (c), where
Güneş Yilmaz et al. (2011) data is grouped with ‘All other data’ in order to create a sample size large
CE

enough for analysis.


AC

Figure 8: Bivariate regression graphs, which highlight the most statistically significant six relationships
identified for the prediction of CI, CD, and BTS. (a) Mica Modal Percentage (Mica %) versus Brazilian
Tensile Strength (BTS); (b) Mica% versus Crack Initiation Stress (CI); (c) Mica% versus Crack Damage
Stress (CD); (d) Overall Minimum Grain Size (GSMin ) versus Crack Damage : Crack Initiation Stress
(CD/CI); (e) Average Mohs Hardness (Mohs) versus CI; (f) Mohs versus CD.

Figure 9: Predictive strength equations compared to actual strength values for the statistically significant
relationships shown in Figure 8. The dashed line on each sub-figure indicates a line with a slope of unity.
ACCEPTED MANUSCRIPT

(a) Actual Brazilian Tensile Stress (BTS) versus Predicted Brazilian Tensile Stress; (b) Actual Crack
Initiation Stress versus Predicted Crack Initiation Stress; (c) Actual Crack Damage Stress versus
Predicted Crack Damage Stress; (d) Actual Crack Damage : Crack Initiation Stress versus Predicted
Crack Damage : Crack Initiation Stress; (e) Actual Crack Initiation Stress versus Predicted Crack
Initiation Stress; (f) Actual Crack Damage Stress versus Predicted Crack Damage Stress.

Figure 10: 3-Dimensional graphs of selected multivariate analysis models. Model 7 (see Table 7) has the

T
largest r2 -adjusted value out of the Brazilian Tensile Strength (BTS) models, and is shown from two

IP
perspectives in (a). Model 1 has the largest r2 -adjusted value out of the Crack Initiation (CI) models, and
is shown from two perspectives in (b). Model 2 has the largest r2 -adjusted value out of the Crack Damage

CR
(CD) models, and is shown from two perspectives in (c).

US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
Fig 1
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M

Fig 2
ED
PT
CE
AC
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

Fig 3
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED

Fig 4
PT
CE
AC
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

Fig 5
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

Fig 6
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

Fig 7
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

Fig 8
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

Fig 9
ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

Fig 10
ACCEPTED MANUSCRIPT

HIGHLIGHTS:

 Mineralogical and mechanical properties of granitic rocks are compiled in a database

 Statistical and multivariate analyses are run to identify relationships between parameters

 Previously proposed relationships are scrutinized

 Novel relationships for prediction of crack initiation stress, crack damage stress, and

T
tensile strength are proposed

IP
CR
US
AN
M
ED
PT
CE
AC

You might also like