Professional Documents
Culture Documents
DOI 10.1007/s10064-015-0721-1
ORIGINAL PAPER
Abstract The strength conversion factor (k) is the ratio Keywords Uniaxial compressive strength Point load
between the uniaxial compressive strength (UCS) and the index Strength conversion factor Statistical methods
point load index (PLI). It has been used to estimate the
UCS from the PLI since the 1960s. Many researchers have
investigated the relationship between UCS and PLI for Introduction
various rock types of different geological origins, such as
igneous, sedimentary, and metamorphic rocks. In this The uniaxial compressive strength (UCS) of a rock ma-
study, the k values for subclasses of igneous (pyroclastic, terial is a crucial parameter, as it is employed in rock
volcanic, and plutonic), sedimentary (chemical and clas- mass classification systems and in various aspects of rock
tic), and metamorphic (foliated and nonfoliated) rocks were engineering design, such as in tunneling and when
evaluated. For this purpose, UCS and PLI data for a total of evaluating slope stability. However, the direct determi-
410 rock samples extracted from literature published nation of this parameter during the preliminary studies
around the world as well as UCS and PLI data obtained in performed during an engineering project is a relatively
this work for 80 rock samples taken from the Eastern Black expensive, troublesome, and time-consuming task. It also
Sea Region in Turkey were evaluated together to determine requires high-quality core samples, and it is often difficult
the k values of different rock classes. Strength conversion to obtain recommended NX-sized core samples, par-
factors were obtained using zero-intercept regression ana- ticularly when there are problematic ground conditions
lysis, formulation, and a graphical approach. This study (thinly bedded, weak, highly fractured rocks, etc.). For
confirmed that there is no single k value that is applicable this reason, UCS has largely been replaced by indirect,
to all rock classes. According to statistical analyses, k var- simpler, faster, and more economic tests. Since the 1960s,
ied between 12.98 and 18.55 for the rocks studied. These the point load index (PLI) test—which is easy to apply
findings demonstrate that the k values derived in this work because the sample preparation process is relatively sim-
can be reliably used to estimate the strengths of rock ple, the test is easy to perform, and it can be used in the
samples with specific lithologies. field—has become the most common indirect method for
determining UCSs (D’Andrea et al. 1964; Deere and
Miller 1966; Broch and Franklin 1972; Bieniawski 1975;
Hassani et al. 1980; Read et al. 1980; Brook 1985; ISRM
1985; Turk and Dearman 1986; Turk 1989; Vallejo et al.
A. Kaya (&)
1989; Cargill and Shakoor 1990; Singh and Singh 1993;
Department of Geological Engineering, Faculty of Engineering,
Recep Tayyip Erdogan University, 53100 Rize, Turkey Chau and Wong 1996; Smith 1997; Hawkins 1998; Ro-
e-mail: ayberk.kaya@erdogan.edu.tr mana 1999; Thuro and Plinninger 2001; Kahraman 2001;
Palchik and Hatzor 2004; Fener et al. 2005; Kahraman
K. Karaman
et al. 2005; Kiliç and Teymen 2008; Diamantis et al.
Department of Mining Engineering, Faculty of Engineering,
Karadeniz Technical University, 61080 Trabzon, Turkey 2009; Singh et al. 2012; Mishra and Basu 2012, 2013;
e-mail: kadirkaraman@ktu.edu.tr Karaman et al. 2014).
123
A. Kaya, K. Karaman
Fig. 1 Map showing the locations of the block samples taken in the Eastern Black Sea Region (EBSR)
Table 1 Properties of rocks obtained in the Eastern Black Sea Region (EBSR) for statistical analyses
Rock class Rock type Lithology Location Number PLI(50) EBSR (MPa) UCSEBSR (MPa)
of
location Min. Max. Ave. SD Min. Max. Ave. SD
Igneous Pyroclastic Agglomerate, volcanic Trabzon 10 0.8 8.1 3.2 2.1 10.4 157.6 55.5 47.2
breccia, tuff
Volcanic Andesite, basalt, dacite, Artvin, 35 2.7 10.0 5.2 1.8 34.0 197.0 102.2 43.6
rhyodacite Trabzon,
Gumushane
Subvolcanic Diabase, microdiorite Artvin, 2 8.2 15.2 11.7 4.9 168.8 203.3 186.0 24.4
Giresun
Plutonic Granite, granodiorite Artvin, 6 1.8 12.9 8.4 3.8 52.7 221.0 148.9 55.8
Giresun,
Gumushane,
Rize
Sedimentary Chemical Limestone, biomicritic Trabzon, 15 0.9 5.7 3.2 1.8 7.7 120.0 59.1 37.6
limestone, sandy Gumushane
limestone, travertine
Clastic Conglomerate Trabzon 1 – – 1.1 – – – 8.3 –
Evaporites – – – – – – – – – – –
Metamorphic Foliated Schist Giresun 1 – – 17.8 – – – 262.5 –
Nonfoliated Marble, metabasalt Giresun, 11 3.8 17.8 7.9 3.9 66.0 262.5 136.8 68.6
Trabzon
Min. minimum, Max. maximum, Ave. average, SD standard deviation
Many researchers have investigated the relation between (50-mm) core. Bieniawski (1975) showed that the UCS
UCS and PLI for different rock types using the UCS/PLI was nearly 23 times the PLI. Pells (1975) stated that using
ratio, which is termed the strength conversion factor (k). an index to strength conversion factor of 24 yielded a 20 %
Broch and Franklin (1972) indicated that the UCS was error in the predicted value of the UCS for rocks such as
approximately 24 times the PLI for a standard-size dolerite, norite, and pyroxenite. Read et al. (1980) showed
123
Estimation of uniaxial compressive strength
that the UCS/PLI ratio varies with both rock type and strong relationship between UCS and PLI for granitic rocks
weathering grade. Greminger (1982) indicated that a using zero-intercept equations. Rusnak and Mark (2000)
k value of 24 was not adequate for anisotropic rocks. evaluated the strength test results of sedimentary rocks
Forster (1983) showed that the UCS/PLI ratio varied from (shale, siltstone, sandstone, and limestone) from six states
11.8 to 17.6 for samples of dolerite and sandstone. ISRM in the USA. It was shown that the UCS/PLI ratio lies be-
(1985) assigned k values of 20–25 based on the UCS/PLI tween 20 and 22, regardless of rock type or geological
ratio. Vallejo et al. (1989) indicated that the UCS/PLI ratio origin. Singh et al. (2012) stated that applying a single
was 12.5 for shales and 17.4 for sandstones. Ghosh and k value to all rock types would give erroneous results.
Srivastava (1991) evaluated the test results for some Strength conversion factors of 21–24 should be used for
granitic rocks from Western Himalaya and derived a UCS/ harder rocks and 14–16 for softer rocks. Karaman et al.
PLI ratio of 16. Smith (1997) investigated the applicability (2014) obtained k values using zero-intercept equations;
of the PLI test to the weak rock materials typical of many these strength conversion factors were 18.2, 16.6, and 18.2
coastal deposits. It was shown that the average UCS/PLI for volcanic, metamorphic, and sedimentary rocks,
ratio for three limestone sites was 14.3, which was low respectively.
compared to the expected value of 24 derived from hard Some researchers have used the regression method to
rock tests. Tugrul and Zarif (1999) found there to be a very elucidate linear relationships between UCS and PLI
123
Table 2 Rock properties compiled from the literature and used in the statistical analyses
Rock class Rock type Lithology Location Researchers Number of PLI(50) (MPa) UCS (MPa)
location
123
Min. Max. Ave. SD Min. Max. Ave. SD
Igneous Pyroclastic Agglomerate, volcanic breccia, Japan This study, Kahraman (2014), 85 0.1 13.0 1.6 1.7 2.1 170.8 21.3 28.0
volcanic bomb, tuff Nigeria Gul and Ceylanoglu (2013),
Volcanic Andesite, basalt, dacite, rhyodacite, Gurocak et al. (2008), Kohno 52 1.2 13.2 5.6 2.8 23.7 202.9 101.3 46.4
Turkey
trachyte, trachyandesite and Maeda (2012), Mishra and
USA Basu (2012), Singh et al.
Subvolcanic Diabase, microdiorite 5 5.4 15.2 10.9 4.0 93.9 203.3 136.5 46.8
(2012), Alemdag et al. (2011),
Plutonic Granite, granodiorite, diorite, Adebayo et al. (2010), 88 1.2 14.6 8.4 3.3 20.5 239.0 123.1 47.2
syenite, gabbro, quartz diorite, Kahraman and Gunaydın
quartz monzodiorite, quartz (2009), Kiliç and Teymen
monzonite, quartz syenite, (2008), Karakus and Tutmez
monzonite, tonalite, granite (2006), Kahraman et al. (2005),
monzogranite, monzogranite Quane and Russell (2005),
Sulukcu and Ulusay (2001),
Kahraman (2001), Topal
(2000), Tugrul and Zarif
(1999), Ghosh and Srivastava
(1991), Catallini (1986)
Sedimentary Chemical Limestone, dolomitic limestone, India This study, Gul and Ceylanoglu 85 0.9 9.8 4.4 1.7 7.7 175.0 80.3 37.5
clayey limestone, gravelly Japan (2013), Kohno and Maeda
limestone, biosparitic limestone, (2012), Kahraman and
Turkey
sparitic crystallised limestone, Gunaydın (2009), Mishra and
biomicritic limestone, sandy USA Basu (2012), Singh et al.
sparitic limestone, high porous (2012), Kiliç and Teymen
limestone, biomicritic limestone, (2008), Karakus and Tutmez
sandy limestone, travertine, (2006), Kahraman et al. (2005),
biomicrite, biosparite limestone, Zarif and Tugrul (2003), Grene
pelmicrite limestone, biopelsparite (2001), Kahraman (2001),
limestone, pelsparite limestone, Sulukcu and Ulusay (2001)
micrite limestone, clayey
biomicrite limestone, biomicrite
with quartz clast limestone, clayey
biomicrite with quartz clast
limestone
Clastic Sandstone, altered sandstone, 57 0.2 13.8 3.8 3.3 2.0 172.0 53.7 44.0
greywacke, mudstone,
conglomerate, marl, siltstone,
shale, tuffaceous conglomerate,
tuffaceous sandstone, siltshale,
mudshale, clayshale, claystone
Evaporites Anhydrite, dolomite, rock salt, 11 1.3 12.0 4.2 3.3 14.6 159.4 67.0 50.2
gypsium
A. Kaya, K. Karaman
Estimation of uniaxial compressive strength
43.5
49.9
(D’Andrea et al. 1964; Deere and Miller 1966; Gunsallus
SD
and Kulhawy 1984; O’Rourke 1989; Cargill and Shakoor
Ave. 1990; Kahraman 2001; Fener et al. 2005; Cobanoglu and
64.3
78.9
Celik 2008; Basu and Kamran 2010). Azimian et al.
(2014) and Kiliç and Teymen (2008) obtained a strong
262.5
243.0
Max.
UCS (MPa)
14.6
22.2
uncovered power relationships between related pa-
rameters (Grasso et al. 1992; Tsiambaos and Sabatakakis
SD
2.9
2.2
2004; Santi 2006). Recently, Azimian et al. (2014) and
Kahraman (2014) published a relatively detailed list
Ave.
4.1
4.8
17.8
9.2
literature.
According to the literature, no single k value is appli-
Min.
1.6
37
paper was to estimate the UCS from the PLI for various
Researchers
rock classes. For this purpose, the UCS and PLI data
gathered from 410 locations and reported in literature from
around the world, as well as UCS and PLI data obtained in
this work from rock samples taken from 80 locations in the
Location
Experimental studies
Non-foliated
Sampling location
Rock type
Foliated
123
A. Kaya, K. Karaman
UCS (MPa)
Eastern Black Sea Region for
this study 200 EBSR
150 Doğrusal
Zero -intercept
Linear
100 (Literature)
UCS = 12.93 PLI (50) + 13.43 95 % confidence
r : 0.83 limit
50
0
0 2 4 6 8 10 12 14 16 18 20
PLI (50) (MPa)
b 350
300 UCS EBSR = 17.92 PLI (50) EBSR
r : 0.89
250
UCSEBSR (MPa)
EBSR
200
Linear
150
Zero -intercept
UCS EBSR = 16.4 8 PLI (50) EBSR + 10.27 95 % confidence
100
r : 90 limit
50
0
0 2 4 6 8 10 12 14 16 18 20
PLI (50) EBSR (MPa)
experimental studies were taken from blocks of the re- obtained by averaging the strength values of five tests
quired dimensions obtained using a core drilling machine. performed at the same lithotype/sampling point.
The ends of the samples were machined flat, ground, and
made parallel to each other. The cut end faces of the cores Point load index (PLI) test
were smoothened to maintain a precision of within
±0.02 mm and made perpendicular to within 0.05 mm In this study, the axial method of PLI testing was applied to
with respect to the core axis using a comparator. Following NX-sized core samples in accordance with the ISRM
a macroscopic inspection, only samples that were un- (Ulusay and Hudson 2007) procedure. A digital test appa-
weathered and free of visible joints were retained. In order ratus was used for PLI testing of the rock samples. At least
to obtain precise results and allow accurate comparisons, ten samples were tested for each rock type, and the mean
the experiments were carried out under the same (natural) value was obtained by discarding the two lowest and highest
conditions. The experimental program included UCS and values and then calculating the mean of the remaining
PLI tests (Fig. 2a–d). values. Tests were carried out along the direction perpen-
dicular to the planes of anisotropy for the foliated meta-
Uniaxial compressive strength (UCS) test morphic rocks. This test allowed the uncorrected PLI to be
determined, so the results were corrected to the standard
The UCS tests were carried out on fresh rock samples with equivalent diameter (De) of 50 mm. The value of the
length to diameter ratios of 2.5–3 following the recom- PLI(50) (MPa) was determined using the following equation:
mendations of the ISRM (Ulusay and Hudson 2007). The
P
tests were performed perpendicular to the planes of ani- PLIð50Þ ¼ F ; ð1Þ
De2
sotropy for the foliated metamorphic rocks collected from
the EBSR. A servo-controlled testing machine with a load where P is the failure load in kN, F is the size correction
capacity of 300 tons was used. A stress rate of 0.75 MPa/s factor ((De/50)0.45), and De is the equivalent core diameter
was applied during the tests. The mean UCS value was in mm2 (De2 = 4A/p).
123
Estimation of uniaxial compressive strength
a 300
UCS = 14.58 PLI (50) b 300 UCS EBSR = 18.2 2 PLI (50) EBSR
r : 0.88
r : 0.89
250 250
UCSEBSR (MPa)
200 200 EBSR
UCS (MPa)
Literature
Linear
150 EBSR 150 Zero -intercept
Zero -intercept UCS EBSR = 16.34 PLI (50) EBSR + 13.0 6
r : 0.89 95 % confidence
100 Linear 100 limit
95 % confidence
UCS = 13.53 PLI (50) + 8.84
50 r : 0.8 9
limit 50
0 0
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
PLI (50) (MPa) PLI (50) EBSR (MPa)
c 240
UCS = 15.1 4 PLI (50) d 120
140
r : 0.61
200
UCSEBSR = 19.94 PLI(50) EBSR - 4.84
100 r : 0.92
UCSEBSR (MPa)
160 Literature
UCS (MPa)
EBSR 80 EBSR
120 Linear Linear
60
Zero -intercept Zero-intercept
80
UCS = 11.11 PL I (50) + 23.0 6 95 % confidence 40 95 % confidence
r : 0. 67 limit limit
40 20 UCSEBSR = 18.74 PLI(50) EBSR
r : 0.92
0 0
0 2 4 6 8 10 12 14 16 0 1 2 3 4 5 6 7
PLI (50) (MPa) PLI(50) EBSR (MPa)
e 300
UCS = 15.16 PLI (50) f
300 UCSEBSR = 16.96 PLI(50) EBSR
r : 85
r : 0.74
250 250
Literature
UCS EBSR (MPa)
EBSR
Zero-intercept
150 Zero-intercept 150 UCSEBSR = 15.10 PLI(50) EBSR + 17.30 Linear
Linear r : 86
95 % confidence
100 UCS = 12.84 PLI (50) + 13.93 95 % confidence 100 limit
r : 0.76 limit
50 50
0 0
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18 20
PLI (50) (MPa) PLI(50) EBSR (MPa)
Fig. 4 Linear and zero-intercept regressions between UCS and PLI values for a, b igneous, c, d sedimentary, and e, f metamorphic rocks, as
compiled from the literature and samples taken in the EBSR
Analysis of collected data evaporites. However, the subvolcanic rock samples (a total
of 5 locations) and evaporites (a total of 11 locations) were
Paired UCS and PLI data collected from literature pub- too scarce to allow separate statistical analyses to be per-
lished around the world and obtained during this study formed for those rock types. Although subvolcanic rocks
were used for the statistical analyses. The results of UCS and evaporites were not taken into consideration in sepa-
and PLI tests performed on samples taken from 490 dif- rate statistical analyses, they were included in the regres-
ferent locations were evaluated (Tables 1, 2). Studies sion analyses of all rocks and of the igneous and
conducted after 1985 were favored for selection because of sedimentary rock groups (Figs. 3a, 4a, c). Metamorphic
the fact that the ISRM (1985) published a suggested rocks were divided into two subclasses: foliated and non-
method for the PLI test in that year, so studies performed foliated rocks.
after that date employed similar test conditions. Initially, In this study, k values were determined based on three
all data from the 490 locations were evaluated together, methods: zero-intercept regression, formulation, and a
without focusing on any particular geological origin (ig- graphical approach. The procedures followed for these
neous, metamorphic, or sedimentary), as in the common methods are described below.
literature. The data were then grouped into rock types In the first method, regression analyses were done to
(igneous, metamorphic, and sedimentary). Igneous rocks investigate the relationship between UCS and PLI (Figs. 3,
were further subdivided into four subclasses: pyroclastic, 4, 5, 6, 7). The relationship between the UCS and PLI
volcanic, subvolcanic, and plutonic. Sedimentary rocks values obtained from the EBSR samples are given in
were individually categorized into chemical, clastic, and Figs. 3b, 4b, d, f. Further, those test data (obtained in this
123
A. Kaya, K. Karaman
UCS (MPa)
120
EBSR
UCS = 14.16 PLI(50)
r : 0.92 Linear
80
Zero-intercept
95 % confidence
40 limit
0
0 2 4 6 8 10 12 14 16
PLI(50) (MPa)
250
b
UCS = 17.15 PLI(50)
r : 0.77
200
Literature
UCS (MPa)
EBSR
150
Linear
Zero-intercept
100
UCS = 13.42 PLI(50) + 26.09 95 % confidence
r : 0.81 limit
50
0
0 2 4 6 8 10 12 14 16
PLI(50) (MPa)
250
c
200
UCS = 10.49 PLI(50) + 35.34 Literature
r : 0.75
EBSR
UCS (MPa)
150
Linear
100 Zero-intercept
95 % confidence
limit
50
UCS = 14.14 PLI(50)
r : 0.70
0
0 2 4 6 8 10 12 14 16
PLI(50) (MPa)
study) were combined with literature data in order to per- the correlation coefficients and the statistically derived
form the regression analyses shown in Figs. 3a, 4a, c, e, zero-intercept equations were confirmed to within the 95 %
5a–c, 6a, b, and 7a, b. In those figures, the test and the confidence level (a: 0.05 significance level) using the SPSS
literature data are distinguished by using symbols with v.15.0 (2006) program. First, the zero-intercept correlation
different colors. The equation for the line of best fit and the coefficients (rcal.) of the rock groups were tested. For this
correlation coefficient (r) were determined for each re- purpose, the calculated correlation coefficients (rcal.) were
gression. The best relationship between the UCS and PLI compared with the critical Pearson’s correlation coeffi-
was linear for each rock group analyzed. The validities of cients (rtable) using the null hypothesis. When the value of
123
Estimation of uniaxial compressive strength
UCS (MPa)
120 EBSR
Linear
UCS = 17.20 PLI (50) Zero -intercept
80 r : 0.39
95 % confidence
limit
40
0
0 2 4 6 8 10 12
PLI (50) (MPa)
b 210
UCS = 12.5 9 PLI (50)
180 r : 0.77
r : 0.79
120 EBSR
Linear
90
Zero -intercept
60 95 % confidence
limit
30
0
0 2 4 6 8 10 12 14 16
PLI (50) (MPa)
rcal. was greater than that of rtable, the null hypothesis was number of k values in the data set. The mean k value of the
rejected, which meant that r was significant. As shown in grouped data sets was found using
Table 3, the rcal. values are greater than the rtable values for " # " #
X
n X
n
all of the derived equations, suggesting that they are valid x ¼ fi :xi = fi ; ð3Þ
according to the correlation coefficient test. Second, the i¼1 i¼1
significance of each regression was determined by per-
where x is the mean k value for the grouped data, fi is the
forming an analysis of variance. When the calculated
frequency of the interval, and xi is the midpoint of the
F value (Fcal.) was greater than the tabulated F value
interval.
(Ftable), the null hypothesis—that there was no relationship
In the third method, k values were determined using a
between the dependent and independent variables—was
graphical approach. Cumulative frequency (%) and his-
rejected. Since the Fcal. values were greater than the
togram plots were plotted for all of the data sets based on
tabulated Ftable values for all of the statistically derived
the frequency values derived via Eq. 2. When all of the
equations, the null hypothesis was rejected in each case
rock groups were combined, the data set showed a normal
(Table 4). Therefore, it was concluded that all of the
distribution (Fig. 8a, b). However, logarithmic distribu-
derived equations are valid according to the F test.
tions were obtained for igneous, sedimentary, and meta-
In the second method, k values were obtained from the
morphic rocks when they were analyzed separately
UCS/PLI ratio and classified using the following equation:
(Fig. 9a–f). The data sets exhibited normal distributions
xmax xmin
h¼ ð2Þ when igneous, sedimentary, and metamorphic rocks were
1 þ 3:322 log N subclassified into pyroclastic, volcanic, plutonic, chemical,
where h is the amount of space, xmax - xmin is the differ- clastic, foliated, and nonfoliated classes (Figs. 10, 11, 12).
ence between the highest and lowest k values, and N is the This may be because the UCS and PLI values vary
123
A. Kaya, K. Karaman
UCS (MPa)
160 Linear
Zero -intercept
120 UCS = 11.62 PLI (50) + 16.99
95 % confidence
r : 0.79
limit
80
40
0
0 2 4 6 8 10 12 14 16 18 20
PLI (50) (MPa)
b 280
240
200
UCS = 16.74 PLI (50) - 0.7 3 Literature
UCS (MPa)
r : 0.74
160 EBSR
Linear
120
Zero -intercept
80 95 % confidence
limit
40 UCS = 16.6 2 PLI (50)
r : 0.74
0
0 1 2 3 4 5 6 7 8 9 10
PLI (50) (MPa)
according to the geological origin of the rock. Cumulative present study (Fig. 3b). No difference between the zero-
frequency (%) and k values were overlaid on normal intercept and linear regression analyses was found in terms
probability paper in order to calculate the mean k values for of their r values. As shown in Fig. 3a, b, there are two
the normal distributions. The data sets exhibiting separate trends: one for all of the rock groups combined
logarithmic distributions were converted into normal dis- and the other for the rocks from the EBSR, with the EBSR
tributions by taking the logarithm in each case. The mean rock data yielding a more steeply sloped line than the rock
k values for igneous, sedimentary, and metamorphic rocks data taken from the literature does. Similar analyses were
were determined using logarithmic probability paper. This executed for igneous, sedimentary, and metamorphic rocks
study therefore indicates that it is very important to know (Fig. 4a–f). The linear and zero-intercept lines were very
the detailed rock type (i.e., volcanic or plutonic rather than close to each other for the literature data and the EBSR
simply ‘‘igneous;’’ chemical or clastic rather than simply data on igneous rocks (Fig. 4a, b). There were reasonable
‘‘sedimentary,’’ etc.) when attempting to estimate rock linear relationships between UCS and PLI for igneous,
strength. sedimentary, and metamorphic rocks but, for sedimentary
and metamorphic rocks, the EBSR rocks showed higher
coefficients of determination (r: 0.85–0.92) than the rocks
Results and discussion described in the literature did (r: 0.61–0.76) (Fig. 4c–f).
Igneous rocks were divided into three subclasses for
In order to be able to examine the relationship between estimating UCS from PLI, based on the lithological simi-
UCS and PLI, regression lines were drawn in Fig. 3 for all larity of the rocks (Fig. 5a–c). Pyroclastic rock data from
rock types. As can be seen in Fig. 3, zero-intercept and the literature gave the highest coefficients of determination
linear regression analyses were implemented for rock data (r [ 0.92) and a steeply sloped line (Fig. 5a). Volcanic and
obtained from literature published around the world plutonic rocks presented similar relationships in terms of
(Fig. 3a) and for rock data from the EBSR obtained in the the slopes of the regression lines. However, the coefficients
123
Estimation of uniaxial compressive strength
Table 3 Test of the correlation coefficients obtained for each rock class at the significance level of a = 0.05
Rock group Rock class Calculated correlation Critical Pearson correlation Test of the correlation coefficient
coefficient (rcal.) coefficient (rtable)
Table 4 Test of the regression equations obtained for each rock class at the significance level of a = 0.05 using analysis of variance (F test)
Rock group Rock class Regression equation Fcal. Ftable Test of the regression equation
All rock groups – UCS = 14.81 PLI(50) 14.37 1.00 H0 rejected Meaningful regression
Igneous – UCS = 14.58 PLI(50) 24.98 1.00 H0 rejected Meaningful regression
Sedimentary – UCS = 15.14 PLI(50) 3.75 1.00 H0 rejected Meaningful regression
Metamorphic – UCS = 15.16 PLI(50) 5.60 1.00 H0 rejected Meaningful regression
Igneous Pyroclastic UCS = 14.16 PLI(50) 53.32 1.46 H0 rejected Meaningful regression
Igneous Volcanic UCS = 17.15 PLI(50) 6.20 1.66 H0 rejected Meaningful regression
Igneous Plutonic UCS = 14.14 PLI(50) 7.66 1.45 H0 rejected Meaningful regression
Sedimentary Chemical UCS = 17.20 PLI(50) 1.80 1.46 H0 rejected Meaningful regression
Sedimentary Clastic UCS = 12.59 PLI(50) 5.76 1.65 H0 rejected Meaningful regression
Metamorphic Foliated UCS = 14.38 PLI(50) 6.59 1.50 H0 rejected Meaningful regression
Metamorphic Nonfoliated UCS = 16.62 PLI(50) 1.83 1.75 H0 rejected Meaningful regression
H0 hypothesis: there is no zero-intercept relation between UCS and PLI(50)
H1 hypothesis: there is zero-intercept relation between UCS and PLI(50)
of determination were slightly higher for the volcanic rocks Table 5 and Fig. 13. The zero-intercept regression method
than for the plutonic rocks (Fig. 5b, c). Figure 6a and b usually gave lower k values than the graphical and formu-
present the relationship between UCS and PLI for chemical lation methods. The k values obtained from the three
and clastic sedimentary rocks, respectively. Relatively poor methods were similar for volcanic rocks. Pyroclastic and
and moderate relationships were obtained for chemical clastic sedimentary rocks had lower average k values than
rocks using zero-intercept and linear regression analyses, those of the other rock types. As shown in Table 5, the
respectively (Fig. 6a). Fairly good relationships (r: average k values were lower than 20 for all rock types except
0.77–0.79) were obtained for clastic sedimentary rocks for metamorphic rocks. According to the literature, reported
(Fig. 6b). Similar relationships (r: 0.74–0.79) and line k values were generally higher than 20 before 1985, whereas
slopes were noted for foliated and nonfoliated metamor- the values have generally been lower than 20 since 1985. It
phic rocks (Fig. 7a, b). is presumed that these deviations are due to variations
Average k values were proposed for the various litholo- among rock types (e.g., in terms of their strengths) and—
gies by averaging the results of the three analyses given in especially—differences in the testing methods used by
123
A. Kaya, K. Karaman
195
200 100
a b
% Cumulative frequency
160 80
116 114
Frequency
120 60
80 40
40 31 20
16
5 6 6
1
0 0
6 12 18 24 30 36 42 48 54 60 6 12 18 24 30 36 42 48 54
k value k value
Fig. 8 a Histogram and b cumulative frequency plot for all rock groups combined, showing a normal distribution (
x indicates the mean k value)
58
a 60 b 100
% Cumulative frequency
50 48
44 80
40 35
60
Frequency
30
24
40
20
14
20
10
2
0
0 0
2 4 6 8 10 12 14 16 18 2 4 6 8 10 12 14 16
k value k value
50 100
c d
42
% Cumulative frequency
40 80
33
60
Frequency
30 27 26
20 40
13
11
10 20
0 0
16 17 18 19 20 21 22 16 17 18 19 20 21
k value k value
e 70 f 100
63
60
% Cumulative frequency
80
50
60
Frequency
40
30 25 40
20
20
10 8
3 5 5
2 1
0 0
24 28 32 36 40 44 48 52 56 24 28 32 36 40 44 48 52
k value k value
Fig. 9 Histograms and cumulative frequency plots for a, b igneous, c, d sedimentary, and e, f metamorphic rocks, showing logarithmic
distributions (
x indicates the mean k value)
123
Estimation of uniaxial compressive strength
30 100
a b
24
% Cumulative frequency
25
80
20 19 19
Frequency
60
15 14
40
10
6
5 20
1 1 1
0 0
4 8 12 16 20 24 28 32 36 4 8 12 16 20 24 28 32
k value k value
24 100
c d
% Cumulative frequency
20 19
80
16
Frequency
60
12
12 10
40
8
5
4 20
1 1 1 1
0 0
0 0
5 10 15 20 25 30 35 40 45 50 55 5 10 15 20 25 30 35 40 45 50
k value k value
35 100
e 32 f
30
% Cumulative frequency
80
25 22
60
Frequency
20 17
15
40
9
10
20
5 2
1 1 1
0 0
0 0
4 8 12 16 20 24 28 32 36 40 44 4 8 12 16 20 24 28 32 36 40
k value k value
Fig. 10 Histograms and cumulative frequency plots of a, b pyroclastic, c, d volcanic, and e, f plutonic rocks, showing normal distributions (
x
indicates the mean k value)
different researchers. In 1985, the ISRM (1985) defined the k is lower for softer rocks and higher for harder rocks.
standard PLI test conditions for researchers. Indeed, the k values listed in Table 5 are consistent with
A strength conversion factor of 24 was proposed by those found in the literature. Hassani et al. (1980) and
Broch and Franklin (1972), and this value has often been Read et al. (1980) reported k values of 29 and 20, re-
used for the practical estimation of UCS in the literature. spectively, for sedimentary rocks. The present study
However, many researchers (Pells 1975; Greminger confirmed that k varies with rock type, so we evaluated k
1982; Smith 1997; Topal 2000) have stated that a separately for rocks with different geological origins.
strength conversion factor of 24 would give erroneous Since this study has investigated the relationship between
results. Kahraman (2014) pointed out that in tests of UCS and PLI for a great number of rocks and rock
many different rock types, k varied between 15 and 50, types, the k values proposed here for each rock type can
especially for anisotropic rocks. Recent studies (Sa- be reliably used in the preliminary studies performed for
batakakis et al. 2008; Singh et al. 2012) revealed that engineering projects.
123
A. Kaya, K. Karaman
a 30 b 100
25 24
% Cumulative frequency
80
21
20 18
Frequency
60
15
12
40
10
5 20
3
2 2 2
1
0 0
6 12 18 24 30 36 42 48 54 60 6 12 18 24 30 36 42 48 54
k value k value
c 25 d 100
21
20
% Cumulative frequency
80
Frequency
15 13
60
10
10
6 40
5 3
2
1 1
0 20
0
5 10 15 20 25 30 35 40 45 50
k value 0
5 10 15 20 25 30 35 40 45
k value
Fig. 11 Histograms and cumulative frequency plots of a, b chemical and c, d clastic sedimentary rocks, showing normal distributions (
x
indicates the mean k value)
a 24 b 100
20
20
% Cumulative frequency
80
16
Frequency
13 60
12 12
12
9
40
8 6
4 2 20
1 1
0
0
3 6 9 12 15 18 21 24 27 30 33 0
3 6 9 12 15 18 21 24 27 30
k value
k value
c 12 d 100
10
10
% Cumulative frequency
9 9 80
8
Frequency
60
6 5
4 40
2
2 1 1 20
0 0
0
5 10 15 20 25 30 35 40 45 50 0
k value 5 10 15 20 25 30 35 40 45
k value
Fig. 12 Histograms and cumulative frequency plots of a, b foliated and c, d nonfoliated metamorphic rocks, showing normal distributions (
x
indicates the mean k value)
123
Estimation of uniaxial compressive strength
Nonfoliated
25
Metamorphic rocks
16.62
15.81
17.50
16.64
20 18.55
17.68
16.64
15.32
14.26
k value
15
Foliated
12.98 13.51
14.38
11.61
16.80
14.26
NON-FOLIATED
PYROCLASTIC
10
CHEMICAL
FOLIATED
VOLCANIC
PLUTONIC
CLASTIC
Clastic
Sedimentary rocks
12.59
12.85
15.10
13.51
5
0
Chemical
17.20
17.36
21.10
18.55
Fig. 13 Histogram showing the mean k values for all rock classes
Plutonic
14.14
14.71
17.10
15.32
Conclusion
Volcanic
phical approach.
The following conclusions can drawn from the results
obtained in this study. The geological origin of the rock has
15.14
17.36
17.30
11.60
10.80
11.30
123
A. Kaya, K. Karaman
123
Estimation of uniaxial compressive strength
Sabatakakis N, Koukis G, Tsiambaos G, Papanakli S (2008) Index Topal T (2000) Problems faced in the applications of the point load
properties and strength variation controlled by microstructure for index test (in Turkish). J Geol Eng 24(1):73–86
sedimentary rocks. Eng Geol 97:80–90 Tsiambaos G, Sabatakakis N (2004) Considerations on strength of
Santi PM (2006) Field methods for characterizing weak rock for intact sedimentary rocks. Eng Geol 72:261–273
engineering. Environ Eng Geosci XII 1:1–11 Tugrul A, Zarif IH (1999) Correlation of mineralogical and textural
Singh VK, Singh DP (1993) Correlation between point load index and characteristics with engineering properties of selected granitic
compressive strength for quartzite rocks. Geotech Geol Eng rocks from Turkey. Eng Geol 51:303–317
11(4):269–272 Turk N (1989) A new procedure for determination of point load
Singh TN, Kainthola A, Venkatesh A (2012) Correlation between strength of rocks. Bull Eng Geol 10:25–31
point load index and uniaxial compressive strength for different Turk N, Dearman WR (1986) A new procedure for determination of point
rock types. Rock Mech Rock Eng 45(2):259–264 load strength in site investigation. Eng Geol Spec Pub 6:405–411
Smith HJ (1997) The point load test for weak rock in dredging Ulusay R, Hudson JA (eds) (2007) The complete ISRM Suggested
applications. Int J Rock Mech Min Sci 34(3–4), Paper 295 Methods for rock characterization, testing and monitoring.
SPSS (2006) Statistical Package for the Social Sciences, SPSS v.15.0 Suggested Methods prepared by the Commission on Testing
for Windows. SPSS Inc., Chicago Methods, International Society for Rock Mechanics. ISRM
Sulukcu S, Ulusay R (2001) Evaluation of the block punch index test Turkish National Group, Ankara, p 628
with particular reference to the size effect, failure mechanism Vallejo LE, Welsh RA, Robinson MK (1989) Correlation between
and its effectiveness in predicting rock strength. Int J Rock Mech unconfined compressive and point load strength for Appalachian
Min Sci 38:1091–1111 rocks. In: Proc 30th US Symp Rock Mech, Morgantown, WV,
Thuro K, Plinninger RJ (2001) Scale effects in rock strength USA, 19–21 June 1989, pp 461–468
properties. Part 2: Point load test and point load strength index. Zarif IH, Tugrul A (2003) Aggregate properties of Devonian
In: Proc ISRM Reg Symp Eurock 2001, Espoo, Finland, 4–7 limestones for use in concrete in Istanbul, Turkey. Bull Eng
June 2001, pp 175–180 Geol Env 62:379–388
123