Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Georisk: Assessment and Management of Risk for

Engineered Systems and Geohazards

ISSN: 1749-9518 (Print) 1749-9526 (Online) Journal homepage: https://www.tandfonline.com/loi/ngrk20

Reliability of empirical equations to predict


uniaxial compressive strength of rocks using
Schmidt hammer

Diyari Abdulalrahman Mohammed, Y. M. Alshkane & Y. A. Hamaamin

To cite this article: Diyari Abdulalrahman Mohammed, Y. M. Alshkane & Y. A. Hamaamin (2019):
Reliability of empirical equations to predict uniaxial compressive strength of rocks using Schmidt
hammer, Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards,
DOI: 10.1080/17499518.2019.1658881

To link to this article: https://doi.org/10.1080/17499518.2019.1658881

Published online: 28 Aug 2019.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=ngrk20
GEORISK
https://doi.org/10.1080/17499518.2019.1658881

Reliability of empirical equations to predict uniaxial compressive strength of rocks


using Schmidt hammer
Diyari Abdulalrahman Mohammed , Y. M. Alshkane and Y. A. Hamaamin
Department of Civil Engineering, University of Sulaimani, Sulaimani, Iraq

ABSTRACT ARTICLE HISTORY


While the direct measurement of uniaxial compressive strength (UCS) of rocks is expensive and Received 26 March 2019
time-consuming, many empirical equations were developed to predict UCS of rocks by the Accepted 24 July 2019
indirect method such as Schmidt hammer rebound number (RL). However, to the best of our
KEYWORDS
knowledge, the reliability of these equations is not investigated. The aim of this paper is to Reliability; limestone rocks;
check the reliability of the available empirical equations to predict UCS for limestone rocks using empirical equations; uniaxial
L-type Schmidt hammer (RL) test. A total of 112 core sample test results of limestone rocks, compressive strength;
which were collected from four different borrow areas in Iraq, were measured. Results of this Schmidt hammer test
study showed that the equation developed by Arslan et al., 2015 [“Prediction of Durability and
Strength from Schmidt Rebound Hammer Number for Limestone Rocks from Salt Range,
Pakistan.” Journal of Himalayan Earth Science 48: 1] is the more reliable regression model to
predict UCS from RL. Furthermore, the performance of the selected equation is enhanced using
the measured values of UCS from this study.

1. Introduction
Schmidt hammer test has been used as a non-
Rocks are one of the abundant materials used in the destructive test to find UCS of rocks since early
construction of engineering projects such as buildings, 1960s (Aydin and Basu 2005; Kallu and Roghanchi
monuments, historical buildings and wall gardens 2015). There are three types of Schmidt hammer: N-
(Karaca, Hacimustafaoʇlu, and Gökçe 2015; La Russa type hammer used for UCS range of weak to very
et al. 2015; Sirdesai et al. 2018; Zhou et al. 2018). Uni- strong, L-type hammer used when the hardness of
axial compressive strength (UCS) can be measured rock is lower than that of N-type hammer and P-
according to both the American Society for Testing type hammer is a pendulum hammer used for very
and Materials (ASTM D7012 2010) and the Inter- low hardness rock (Saptono, Kramadibrata, and Sulis-
national Society for Rock Mechanics (ISRM 1981; Car- tianto 2013). Schmidt hammer test can be used by
gill and Shakoor 1990; Kahraman 2001). To investigate two different laboratory methods, vertically (measure
the mechanical behaviour of various rock types, UCS position 90°) and horizontally (measure position 0°)
value is the critical parameter (Yurdakul and Akdas (Yaşar and Erdog an 2004). Many researchers investi-
2013). UCS of rock is the most important mechanical gated the correlation between UCS and RL for various
property that affects the design and construction on/in rock types in several countries and they developed the
rock foundations such as building structure, tunnel model equations by using advance statistics such as
and dam (Jabbar 2011; Mishra and Basu 2013; Barone fuzzy inference system and Artificial Neural Network
et al. 2015; Kurtuluş, Sertçelik, and Sertçelik 2016; (ANN) (Gökceoglu and Aksoy 2002; Yilmaz and Yuk-
Daoud, Alshkane, and Rashid 2018). For poor quality sek 2008; Mishra and Basu 2013). Nevertheless, the
rocks, drilling core samples are difficult and the prep- reliability of these empirical equations is not evaluated
aration of the samples for UCS test is costly, so, and investigated yet according to the authors’ knowl-
Schmidt hammer test can be used to predict UCS, edge. Also, another method such as multivariate adap-
which is a simple, quick test and does not need sample tive regression splines (MARS) and back-propagation
preparation (Yaşar and Erdog an 2004; Çobanoglu and algorithm (BPNN) can be used for creating a model
Çelik 2008; Sengun et al. 2011; Kallu and Roghanchi equation (Zhang and Goh 2013; Zhang and Goh
2015; Çobanoğlu and Çelik 2017; Daoud, Rashed, 2016), but more importantly the reliability of the
and Alshkane 2018). regression models should be studied (Alshkane 2017).

CONTACT Y. M. Alshkane Younis.ali@univsul.edu.iq Department of Civil Engineering, University of Sulaimani, Sulaimani, Kurdistan Region, Iraq
© 2019 Informa UK Limited, trading as Taylor & Francis Group
2 D. A. MOHAMAD ET AL.

ISRM (1981) suggested L-type hammer test as a stan- Kometan formation at Qlyasan region, Pila spy for-
dard method for the measurement of compressive mation at Qaradax region and Sinjar formation at
strength of rock materials, whereas, N-type hammer is Tasluja region with one formation of Mosul province
appropriate for concrete materials. The main purpose in the north of Iraq which is known as Lower Fars for-
of this study is to investigate the reliability of L-type mation at Al-Shora and Ayn al-Jahesh, as shown in
Schmidt hammer (RL) to predict UCS of limestone Figure 1. UCS and RL tests were conducted on the rock
rocks from the empirical equations developed from the samples. Single core barrel with 63 mm outside diameter,
literature. In this study, 112 measured data of UCS and 54.7 mm inside diameter and 400 mm length according
RL were obtained from the test results of core samples to ISRM (1981) were used for drilling the intact rock,
and used to check the reliability of these empirical then the limestone core samples were obtained. The
equations. In addition, the statistical analysis of each number of samples drilled in each location is presented
specific site and the total 112 data of UCS and RL includ- in Table 1 from different points with different layers of
ing standard deviation, mean, variance, maximum and rock masses. It should be noted that the samples were
minimum values of UCS and RL were presented. For- taken in two directions: vertically (Figure 2(A)) and lat-
merly, the more reliable empirical equation was selected erally (Figure 2(B)) on the rock masses to obtain the
through the study. Finally, a modification of the selected different strength and representative samples of the
best equation is presented based on the 112 measured selected site locations.
data points, to be used with more consistency to predict
UCS of limestone rocks.
2.2. Experimental tests
2.2.1. Sample preparation
2. Methodology For UCS test, the samples were prepared according to
2.1. Measured data and rock sampling ASTM D7012 (2010) with a core size diameter of NX
(54.7 mm) and trimmed (Figure 3(A)) for length to
Limestone rocks can be considered as one of the abun- diameter ratio 2:1. Sample ends were further smooth-
dant rocks used in the construction of building and dec- ened using a surface grinder machine (Figure 3(B))
oration in Sulaymaniyah, Kurdistan Region of Iraq. and polished to minimise the contact friction between
Generally, in Iraq, limestone rocks are available in platens and sample ends during loading. For RL test,
three main formations which include Kometan, Pila the samples were prepared according to ISRM (1981)
Spy and Sinjar formations and one formation from with a core size diameter of NX (54.7 mm) with a length
Mosul province includes the Lower Fars formation. of 100 mm. The prepared samples were divided into two
Kometan formation is a limestone formation that was groups to perform UCS and RL tests, half of the samples
first described by Dunnington (1958) at Kometan village were tested in dry case, while the other half were tested in
of the High Folded Zones (Al-Barzinjy 2008). Pila Spi saturated case. Samples were saturated in a special
formation is the limestone formation that was first device, as shown in Figure 3(C).
reported by Lees in 1930 from the surrounding of Pila
Spi village of the High Folded Zones those crop out in
anticlinal limbs and cores in the foreland folds of the 2.2.2. Uniaxial compressive strength test
northern Iraq (Numan, Hammoudi, and Chorowicz UCS tests were conducted on the prepared NX size
1998). Sinjar Formation is a limestone formation that samples based on ISRM (1981) with a compressive test-
was first described by Keller (1941) from the Jabal Sinjar ing machine manufactured by Control Company (Figure
area near Mantissa village (Buday 1980; Daoud 2009). 3(F)). In order to select the more reliable model that can
Lower Fars formation also called Fatha formation is be used for dry and saturated condition, samples were
the limestone formation that was originally described divided into two groups to perform UCS test, dry and
in the Fars province of Iran by Busk and Mayo in 1918 saturated samples.
(Al-Juboury and McCann 2008; Busk and Mayo 1918).
The Middle Miocene Lower Fars formation in the north- 2.2.3. Schmidt hammer test
ern Iraq was particularly in the area surrounding the RL tests were conducted based on ISRM (1981). Initially,
Mosul city (Jassim, Jibril, and Numan 1997; Al-Juboury the hammer was calibrated by using a calibration test
and McCann 2008). anvil (Figure 3(D)). L-type Schmidt hammer instrument
In this study, 112 rock samples using NX core size with a steel base of about 20 kg weight (Figure 3(E)) was
(54.7 mm in diameter) were drilled from three different used. Also, the perpendicularity of the hammer was con-
formations of Sulaymaniyah province including sidered during RL tests.
GEORISK: ASSESSMENT AND MANAGEMENT OF RISK FOR ENGINEERED SYSTEMS AND GEOHAZARDS 3

Figure 1. Study area of limestone rocks. (https://www.google.com/maps/place/Iraq/).

2.3. Literature equations UCS. To compare the predicted data points with the
measured data points, the Root Mean Square Error
In this study, 40 empirical equations were collected, as
(RMSE), the Mean Absolute Error (MAE), the coefficient
summarised in Table 2, for various rock types from the
of determination (R 2) and p-value from Mann–Whitney
literature.
test were used. Formerly, the more reliable equation of
limestone UCS estimation was selected.
2.4. Assessment of equations
2.5. Modification of the best empirical equation
The assessment of 40 empirical equations (Table 2) was
performed by applying the equations to the total 112 RL After the selection of the best empirical equation, based
collected data points for limestone rocks to predict the on the correlation equation of the predicted values
4 D. A. MOHAMAD ET AL.

Table 1. Number of drilled samples in each location. of the measured and predicted UCS data was checked
Number of using Minitab 17 statistical software program (Minitab
drilled Limestone
Cities Formations Locations samples types 2017). According to the normality test of the measured
Sulaymaniyah Kometan Qlyasan 20 White and predicted data of UCS, the statistical test was
limestone selected. For non-normal datasets (predicted and
Pila Spy Qaradax 27 Dolomitic
limestone
measured), the p-value of each empirical equation was
Sinjar Tasluja 25 Milky limestone considered using nonparametric Mann–Whitney test
Mosul Lower Fars Al-Shora 16 Fossiliferous (Ott and Longnecker 2015). Figure 4 shows the statistical
limestone
Lower Fars Ayn al- 24 Marley analysis flowchart. Wang and Aladejare (2015) tested the
Jahesh limestone reliability of empirical equations based on the trend line
of each empirical equation prediction compared with
that of measured test results. The most reliable equation
from the best equation and 112 measured data values was selected depending on the best-fitted line.
from this study, a modified equation was obtained. In Figure 4, a two-tail test with 95% confident interval
was used with test hypothesis assumptions, null hypoth-
3. Statistical analysis and tests eses (Ho: measured and predicted data are identical) and
research hypotheses (Ha: measured and predicted data
3.1. Statistical analysis are not identical).
The descriptive statistics of measured UCS with RL data During statistical analysis, Ho was failed to be rejected
points of limestone rocks for any site specific from any (accepted) for p-value ≥ 0.05, and Ho was rejected for p-
rock formation are shown in Table 3. Aladejare and value <0.05 for all tests.
Wang 2017 performed this procedure for three main 
n 2
rock types including: igneous, sedimentary and meta- i=1 (Yp − Yi)
RMSE = (41)
morphic rocks in many countries. N

n
3.2. Statistical analysis and test hypothesis i=1 /(Yp − Yi)/
MAE = (42)
N
The values of UCS were predicted for all 40 empirical
equations separately using 112 measured limestone where: RMSE = root mean square error MAE = mean
data of RL. Subsequently, the RMSE (Equation (41)), absolute error Yp= predicted value from empirical
MAE (Equation (42)), and R 2 were computed for equations Yi = actual value of collected data N = no. of
measured and predicted data points. Also, the normality measured or predicted data Yp- Yi = error

Figure 2. Rock sampling in the field. A. Vertically drilling, B. Laterally drilling.


GEORISK: ASSESSMENT AND MANAGEMENT OF RISK FOR ENGINEERED SYSTEMS AND GEOHAZARDS 5

Figure 3. Apparatus used to perform tests in this study. A. Rock trimmer, B. Rock grinding, C. Saturation machine, D. Schmidt hammer
calibration, E. Schmidt hammer machine with base and F. Rock compressive testing machine.

4. Results and discussion


UCS was 6.8 MPa to 184.95 MPa, while the minimum
The reliability of empirical equations was conducted by and maximum values of measured RL were 10.462 and
using 112 measured data points of UCS and RL to com- 43.347, respectively. From Table 4, the minimum values
pute RMSE, MAE and R 2 between measured and pre- of RMSE and MAE were 35.372 MPa and 28.345,
dicted data points (Table 4). The range of measured respectively, for Equation (5). The maximum value of
6 D. A. MOHAMAD ET AL.

Table 2. Empirical correlation equations from the literature to predict UCS from RL.
Eq. No. References Equations R2 Rock types
1 Deere and Miller (1966) UCS = 0.0086 *RL – 0.241 0.77 Various rock types
2 Singh, Hassani, and Elkington (1983) UCS = 2*RL 0.74 Sedimentary (sandstone, siltstone, mudstone and seat earth)
3 O’Rourke (1989) UCS = 4.84*RL – 76.118 0.59 Sedimentary
4 Sachpazis (1990) UCS = 4.2937*RL – 67.5157 0.92 Marble, dolomite and limestone
5 Tug rul and Zarif (1999) UCS = 8.36*RL – 416 0.88 Granite
6 Gökceoglu and Aksoy (2002) UCS = 0.7291*RL – 12.774 0.67 Marlstone
7 UCS = 18.294* Ln(RL) –53.301 0.66 Marlstone
8 UCS = 0.0001*(RL)3.2658 0.71 Marlstone
9 UCS = 0.1969*exp(0.1285*RL) 0.69 Marlstone
10 Yilmaz and Sendr (2002) UCS = exp(0.818+0.059*RL) 0.96 Gypsum
11 Yaşar and Erdog an (2004) UCS = 0.000004*(RL)4.2917 0.80 Marlstone, limestone, sandstone and basalt
12 Dinçer et al. (2004) UCS = 2.75*RL – 36.83 0.95 Andesite, basalt and tuffs
13 Aydin and Basu (2005) UCS = 1.4459*exp(0.0706*RL) 0.85 Granite
14 Haramy and DeMarco (2005) UCS = 0.994*RL – 0.383 0.71 Coal measure rock
15 UCS = 0.287*(RL)1.3252 0.72
16 Grasso, Xu, and Mahtab (2005) UCS = exp(0.045*RL + 2.27) 0.56 Mudstone
17 Buyuksagis and Goktan (2007) UCS = 2.4817*exp(0.0725*RL) 0.88 Various rock types (granite, marble, limestone and travertine)
18 UCS = 2.6363*exp(0.0659*RL) 0.89
19 UCS = 3.2123*exp(0.0706*RL) 0.86
20 UCS = 3.6834*exp(0.0679*RL) 0.86
21 UCS = 2.4736*exp(0.0691*RL) 0.88
22 Shalabi, Cording, and Al-Hattamleh (2007) UCS = 3.201* RL – 46.59 0.76 Various rock types
23 Sabatakakis et al. (2008) UCS = 3.1*exp(0.09*RL) 0.79 Sedimentary (marlstone, sandstone and limestone)
24 Çobanog lu and Çelik (2008) UCS = 6.59*RL – 212.63 0.65 Sedimentary (cement mortar, sandstone and limestone)
25 Yilmaz and Yuksek (2008) UCS = 1.2483*RL – 24.723 0.96 Gypsum
26 Yagiz (2009) UCS = 0.0028*(RL)2.584 0.92 Travertine, limestone, dolomitic limestone and schist
27 Mishra and Basu (2013) UCS = 3.79*exp(0.055*RL) 0.85 Sandstone
28 UCS = 5.19*RL – 168.1 0.75 Granite
29 UCS = 2.46*exp(0.06*RL) 0.78 Schist
30 UCS = 2.38*exp(0.065*RL) 0.87 Sandstone, granite and schist
31 Minaeian and Ahangari (2013) UCS = 0.678* RL 0.98 Weak, conglomeratic rock, tuff, limestone, and marl
32 UCS = 6.038*Ln(RL) 0.88
33 UCS = (RL)0.885 0.87
34 UCS = exp(0.092* RL) 0.87
35 Kallu and Roghanchi (2015) UCS = 0.25*(RL)1.77 0.88 Rhyolite and basalt
36 Arslan, Khan, and Yaqub (2015) UCS = 23.8*exp(0.032* RL) 0.78 Limestone
37 Daoud, Alshkane, and Rashid (2018) UCS = 0.00004*(RL)4.1643 0.88 Limestone and sandstone
38 Yilmaz and Goktan (2018) UCS = 0.0001*(RL)3.5486 0.84 Sedimentary
39 UCS = 0.00007*(RL)3.6023 0.88 Sedimentary
40 UCS = 0.0001*(RL)3.4798 0.85 Sedimentary
Note: UCS in MPa.

Table 3. Summary of statistical analysis of limestone rocks data for all sites.
Parameter Locations No. of data Mean Minimum Maximum Variance Standard deviation
UCS (MPa) Qlyasan 20 117.56 64.79 178.03 892.84 29.88
Qaradax 27 102.53 51.46 184.95 1353.51 36.79
Tasluja 25 69.29 23.48 164.86 1319.56 36.33
Al-Shora 16 13.114 6.803 18.434 10.467 3.235
Ayn al-Jahesh 24 18.81 9.71 30.69 62.20 7.89
Over all 112 67.08 6.8 184.95 2458.93 49.59
RL Qlyasan 20 36.269 32.894 43.347 5.634 2.374
Qaradax 27 29.454 16.206 35.567 17.020 4.126
Tasluja 25 31.975 26.269 36.341 6.340 2.518
Al-Shora 16 13.748 10.462 18.082 6.326 2.515
Ayn al-Jahesh 24 15.440 11.483 20.556 10.413 3.227
Over all 112 25.987 10.462 43.347 85.023 9.221

R 2 was 0.624 for Equation (4) whereas the maximum values of two models have the same values, and the
values of RMSE and MAE were 263.189 and MAE values are different, MAE can be used for compari-
256.894 MPa, respectively, for Equation (12) and the sons of the two model equations (Chai and Draxler
minimum value of R 2 was 0.211 for Equation (24). 2014).
To find out the accuracy of a model, RMSE is more In this study, the error distributions (difference
suitable than MAE when the error distribution is normal between measured and predicted values) of all empirical
(Chai and Draxler 2014). However, when the RMSE equations except the Equation (24) were non-normally
GEORISK: ASSESSMENT AND MANAGEMENT OF RISK FOR ENGINEERED SYSTEMS AND GEOHAZARDS 7

Figure 4. Flowchart of empirical equation performance evaluation.

distributed (p-value for error distribution <0.05);there- maximum p-values were 0 and 0.522, respectively.
fore, RMSE was not considered (Chai and Draxler Based on the p-value from this test, only one empirical
2014). Also, MAE was not considered for the evaluation equation, Equation (36) was more reliable to predict
of empirical equations because MAE correlates to RMSE. UCS of limestone rocks that was developed by Arslan,
Furthermore, R 2 was not considered because R 2 is a weak Khan, and Yaqub (2015). The p-value of this equation
indicator (Willmott and Matsuura 2005). is greater than 0.05; therefore, this equation can be
In addition, while the error distribution of Equation used with more confidence compared to the other
(24) was normally distributed, RMSE and MAE of this empirical equations to predict the UCS of limestone
equation were higher than the other equations and R 2 rocks. However, Arslan, Khan, and Yaqub (2015) used
was lower than the other equations; therefore, there 18 data of limestone rocks to develop this equation
was insufficient evidence to accept this equation. To (Equation (36)).
further evaluate the performance of the 40 empirical Then correlation between measured and predicted
equations, the nonparametric two sample test (Mann– UCS data points for Equation (36) is presented
Whitney test) was performed. Initially the normality of (Figure 6). It can be realised from Figure 6 that Equation
the output from the empirical equations was tested. (36) underestimates the UCS value, but it produces a
Then depending on the frequency distribution of the safe value. In order to enhance the under estimation of
data (normality of the measured and predicted data) of the model, the correlation between Measured UCS and
UCS, the statistical tests to check the reliability of empiri- Predicted UCS was determined as in the following
cal equations were selected (Figure 4). Equation (43):
Since the statistical results (the measured and pre-
dicted UCS) were non-normally distributed; the UCS(measured) = 2.4786∗UCS(predicted) −74.128 (43)
Mann–Whitney test was used to check the reliability of
empirical equations. The normal probability plot of the
measured UCS data is shown in Figure 5 and the normal Combining Equation (36) and Equation (43), a
probability for 40 predicted dataset from 40 empirical modified regression model can be obtained as presented
equations is presented in Table 5. in Equation (44):
Table 5 shows the p-value from the Mann–Whitney
test results for empirical equations, the minimum and UCS(modified) = 58.991∗exp(0.032∗RL) −74.128 (44)
8 D. A. MOHAMAD ET AL.

Table 4. Results of evaluation criteria for empirical equations of It is worth to mention that Equation (36) was created
UCS compared to measured data. based on 18 data points, whereas Equation (44) is devel-
Evaluation criteria oped based on 112 measured data points.
Eq. No. Equations RMSE (MPa) MAE (MPa) R2
In addition, to show the performance of the modified
1 UCS = 0.0086 *RL – 0.241 83.264 67.097 0.623
2 UCS = 2*RL 39.638 29.652 0.623
equation (Equation (44)), predictions from this equation
3 UCS = 4.84*RL – 76.118 35.372 28.345 0.623 against the measured data are plotted in Figure 7. It can be
4 UCS = 4.2937*RL – 67.5157 38.051 30.400 0.623 observed that the trend line of Equation (44) is almost
5 UCS = 8.36*RL – 416 263.189 256.894 0.465
6 UCS = 0.7291*RL – 12.774 75.297 60.906 0.623 along the equality 45° line, while the trend line of the
7 UCS = 18.294* Ln(RL) –53.301 76.041 62.193 0.603 Equation (36) (Figure 6) is more deviated from the equal-
8 UCS = 0.0001*(RL)3.2658 76.351 61.073 0.579
9 UCS = 0.1969*exp(0.1285*RL) 72.272 57.506 0.494 ity line. Therefore, the trend line of the modified equation
10 UCS = exp(0.818+0.059*RL) 59.765 46.147 0.384 (Equation (44)) is the best fit with the equality line.
11 UCS = 0.000004*(RL)4.2917 62.121 49.462 0.281
12 UCS = 2.75*RL – 36.83 46.467 35.218 0.623
13 UCS = 1.4459*exp(0.0706*RL) 71.909 56.165 0.590
14 UCS = 0.994*RL – 0.383 59.516 42.243 0.623
15 UCS = 0.287*(RL)1.3252 61.568 44.995 0.624 5. Conclusions and recommendations
16 UCS = exp(0.045*RL + 2.27) 52.242 37.110 0.613
17 UCS = 2.4817*exp(0.0725*RL) 62.794 47.211 0.588 In this study, statistical parameters of RMSE, MAE, R 2
18 UCS = 2.6363*exp(0.0659*RL) 65.724 49.858 0.595
19 UCS = 3.2123*exp(0.0706*RL) 58.613 42.881 0.590 and nonparametric Mann–Whitney’s p-value were
20 UCS = 3.6834*exp(0.0679*RL) 57.499 41.617 0.593 used to check the reliability of empirical equations to
21 UCS = 2.4736*exp(0.0691*RL) 64.976 49.251 0.592
22 UCS = 3.201* RL – 46.59 44.038 34.122 0.623 predict UCS from RL for limestone rocks. A total of
23 UCS = 3.1*exp(0.09*RL) 41.331 28.795 0.564 112 field data points were used for the evaluation
24 UCS = 6.59*RL – 212.63 112.023 101.142 0.211
25 UCS = 1.2483*RL – 24.723 72.104 59.363 0.623
process.
26 UCS = 0.0028*(RL)2.584 65.883 51.265 0.603 Depending on the results from this study, the follow-
27 UCS = 3.79*exp(0.055*RL) 65.718 49.299 0.606 ing conclusions are drawn:
28 UCS = 5.19*RL – 168.1 105.141 100.307 0.623
29 UCS = 2.46*exp(0.06*RL) 69.732 53.654 0.601
30 UCS = 2.38*exp(0.065*RL) 67.845 51.955 0.596 . Probability Distribution of data points is the key in
31 UCS = 0.678* RL 66.610 49.483 0.623
32 UCS = 6.038*Ln(RL) 67.408 49.499 0.603 selecting the more reliable empirical equations to pre-
33 UCS = (RL)0.885 66.802 49.387 0.622 dict UCS from RL. When the distribution is non-nor-
34 UCS = exp(0.092* RL) 67.274 52.294 0.561
35 UCS = 0.25*(RL)1.77 37.151 28.500 0.621 mal, the following statistical parameters RMSE, MAE
36 UCS = 23.8*exp(0.032* RL) 39.580 31.641 0.620 and R2 may not be a good indicator. Therefore, the p-
37 UCS = 0.00004*(RL)4.1643 38.535 29.134 0.535
38 UCS = 0.0001*(RL)3.5486 64.936 50.983 0.566
value of nonparametric Mann–Whitney test is the
39 UCS = 0.00007*(RL)3.6023 67.668 53.489 0.564 best indicator for the evaluation of empirical
40 UCS = 0.0001*(RL)3.4798 68.754 54.419 0.569 equations.

Figure 5. The normal probability plot of measured UCS data.


GEORISK: ASSESSMENT AND MANAGEMENT OF RISK FOR ENGINEERED SYSTEMS AND GEOHAZARDS 9

Table 5. Results of Mann–Whitney test for empirical equations of UCS compared to measured data.
Normality test
Eq. No. Equations Measured UCS Predicted UCS Normality test for error distribution p-value from Mann–Whitney test
1 UCS = 0.0086 *RL – 0.241 <0.005 <0.005 <0.005 0.00000
2 UCS = 2*RL <0.005 <0.005 <0.005 0.00000
3 UCS = 4.84*RL – 76.118 <0.005 <0.005 <0.005 0.01206
4 UCS = 4.2937*RL – 67.5157 <0.005 <0.005 <0.005 0.00037
5 UCS = 8.36*RL – 416 <0.005 <0.005 <0.005 0.00000
6 UCS = 0.7291*RL – 12.774 <0.005 <0.005 <0.005 0.00000
7 UCS = 18.294* Ln(RL) –53.301 <0.005 <0.005 <0.005 0.00000
8 UCS = 0.0001*(RL)3.2658 <0.005 <0.005 <0.005 0.00000
9 UCS = 0.1969*e(0.1285*RL) <0.005 <0.005 <0.005 0.00000
10 UCS = exp(0.818+0.059*RL) <0.005 <0.005 <0.005 0.00000
11 UCS = 0.000004*(RL)4.2917 <0.005 <0.005 <0.005 0.00000
12 UCS = 2.75*RL – 36.83 <0.005 <0.005 <0.005 0.00000
13 UCS = 1.4459*exp(0.0706*RL) <0.005 <0.005 <0.005 0.00000
14 UCS = 0.994*RL – 0.383 <0.005 <0.005 <0.005 0.00000
15 UCS = 0.287*(RL)1.3252 <0.005 <0.005 <0.005 0.00000
16 UCS = exp(0.045*RL + 2.27) <0.005 <0.005 <0.005 0.00018
17 UCS = 2.4817*exp(0.0725*RL) <0.005 <0.005 <0.005 0.00000
18 UCS = 2.6363*exp(0.0659*RL) <0.005 <0.005 <0.005 0.00000
19 UCS = 3.2123*exp(0.0706*RL) <0.005 <0.005 <0.005 0.00000
20 UCS = 3.6834*exp(0.0679*RL) <0.005 <0.005 <0.005 0.00000
21 UCS = 2.4736*exp(0.0691*RL) <0.005 <0.005 <0.005 0.00000
22 UCS = 3.201* RL –46.59 <0.005 <0.005 <0.005 0.00000
23 UCS = 3.1*exp(0.09*RL) <0.005 <0.005 <0.005 0.00014
24 UCS = 6.59*RL – 212.63 <0.005 <0.005 0.126 0.00000
25 UCS = 1.2483*RL – 24.723 <0.005 <0.005 <0.005 0.00000
26 UCS = 0.0028*(RL)2.584 <0.005 <0.005 <0.005 0.00000
27 UCS = 3.79*exp(0.055*RL) <0.005 <0.005 <0.005 0.00000
28 UCS = 5.19*RL – 168.1 <0.005 <0.005 <0.005 0.00000
29 UCS = 2.46*exp(0.06*RL) <0.005 <0.005 <0.005 0.00000
30 UCS = 2.38*exp(0.065*RL) <0.005 <0.005 <0.005 0.00000
31 UCS = 0.678* RL <0.005 <0.005 <0.005 0.00000
32 UCS = 6.038*Ln(RL) <0.005 <0.005 <0.005 0.00000
33 UCS = (RL)0.885 <0.005 <0.005 <0.005 0.00000
34 UCS = exp(0.092* RL) <0.005 <0.005 <0.005 0.00000
35 UCS = 0.25*(RL)1.77 <0.005 <0.005 <0.005 0.00024
36 UCS = 23.8*exp(0.032* RL) <0.005 <0.005 <0.005 0.52202
37 UCS = 0.00004*(RL)4.1643 <0.005 <0.005 <0.005 0.00992
38 UCS = 0.0001*(RL)3.5486 <0.005 <0.005 <0.005 0.00000
39 UCS = 0.00007*(RL)3.6023 <0.005 <0.005 <0.005 0.00000
40 UCS = 0.0001*(RL)3.4798 <0.005 <0.005 <0.005 0.00000

Figure 6. Relationship between measured and predicted UCS for Equation (36).
10 D. A. MOHAMAD ET AL.

Figure 7. Relationship between measured UCS from test results and modified UCS from Equation (44).

. The recommended equation to predict UCS from RL Al-Juboury, A. I., and T. McCann. 2008. “The Middle Miocene
of limestone rocks is Equation (36) which has a higher Fatha (Lower Fars) Formation, Iraq.” Geoarabia-Manama
p-value compared with other equations. 13 (3): 141.
Aladejare, A. E., and Y. Wang. 2017. “Evaluation of Rock
. While the range of the total of the 112 RL data was Property Variability.” Georisk: Assessment and Management
between 10.462 and 43.347 it is recommended to of Risk for Engineered Systems and Geohazards 11 (1): 22–
use the selected best equation to predict UCS of 41. doi:10.1080/17499518.2016.1207784.
rocks with the same range of RL. Alshkane, Y. M. 2017. “Discussion on “Property
. A modified regression model (Equation (44)) was Correlations and Statistical Variations in the
Geotechnical Properties of (CH) Clay Soils.”
obtained to predict UCS for limestone rocks
Geotechnical and Geological Engineering, doi:10.1007/
based on Equation (36) and the test results of s10706-017-0416-4.
this study. Arslan, M., M. S. Khan, and M. Yaqub. 2015. “Prediction
of Durability and Strength from Schmidt Rebound
Hammer Number for Limestone Rocks from Salt
Disclosure statement Range, Pakistan.” Journal of Himalayan Earth Science
48: 1.
No potential conflict of interest was reported by the authors. ASTM D7012-10. 2010. “Standard Test Method for
Compressive Strength and Elastic Moduli of Intact Rock
Core Specimens Under Varying States of Stress and
ORCID Temperatures.” ASTM International, West
Diyari Abdulalrahman Mohammed http://orcid.org/0000- Conshohocken, PA, (C): 1–8. doi:10.1520/D7012-13.1.
0002-5046-3420 Aydin, A., and A. Basu. 2005. “The Schmidt Hammer in Rock
Y. M. Alshkane http://orcid.org/0000-0003-4699-0129 Material Characterization.” Engineering Geology 81 (1):
Y. A. Hamaamin http://orcid.org/0000-0003-2621-7489 1–14. doi:10.1016/j.enggeo.2005.06.006.
Barone, G., P. Mazzoleni, G. Pappalardo, and S. Raneri. 2015.
“Microtextural and Microstructural Influence on the
Changes of Physical and Mechanical Proprieties Related
References
to Salts Crystallization Weathering in Natural Building
Al-Barzinjy, S. T. 2008. “Origins of Chert Nodules in Kometan Stones. The Example of Sabucina Stone (Sicily).”
Formation from Dokan Area, NE-Iraq.” Iraqi Bulletin of Construction and Building Materials 95: 355–365. doi:10.
Geology and Mining 4 (1): 95–104. 1016/j.conbuildmat.2015.07.131.
GEORISK: ASSESSMENT AND MANAGEMENT OF RISK FOR ENGINEERED SYSTEMS AND GEOHAZARDS 11

Brown, E. T., and ISRM. 1981. Rock Characterization, Testing Geology and Productivity, AAPG Foreign Reprint Series
and Monitoring. ISRM Suggested Methods. Oxford: 125: 1194–1251.
Pergamon Press. Gökceoglu, C., and H. Aksoy. 2002. “Landslide Susceptibility
Buday, T. 1980. “The Regional Geology of Iraq,” Vol. 1, Mapping of the Slopes in the Residual Soils of the
Stratigraphy and Paleogeography. GEOSURV. Baghdad, Mengen Region (Turkey) by Deterministic Stability
Iraq. 445. Analyses and Image Processing Techniques.” Engineering
Busk, H. G., and H. T. Mayo. 1918. “Some notes on the Geology 44 (1-4): 147–161. doi:10.1016/s0013-7952
Geology of Persian Oilfields.” The Journal of Petroleum (97)81260-4.
Technology 5: 5–26. Grasso, P., S. Xu, and A. Mahtab. 2005. “Problems and
Buyuksagis, I. S., and R. M. Goktan. 2007. “The Effect of Promises of Index Testing of Rocks.” International
Schmidt Hammer Type on Uniaxial Compressive Strength Journal of Rock Mechanics and Mining Sciences &
Prediction of Rock.” International Journal of Rock Geomechanics Abstracts 30 (5): 278. doi:10.1016/0148-
Mechanics and Mining Sciences 44 (2): 299–307. doi:10. 9062(93)92299-6.
1016/j.ijrmms.2006.07.008. Haramy, K. Y., and M. J. DeMarco. 2005. “Use of the Schmidt
Cargill, J. S., and A. Shakoor. 1990. “Evaluation of Empirical Hammer for Rock and Coal Testing.” International Journal
Methods for Measuring the Uniaxial Compressive of Rock Mechanics and Mining Sciences & Geomechanics
Strength of Rock.” International Journal of Rock Abstracts 23 (4): 132. doi:10.1016/0148-9062(86)90704-7.
Mechanics and Mining Sciences 27 (6): 495–503. doi:10. Jabbar, M. A. 2011. “Correlations of Point Load Index and
1016/0148-9062(90)91001-N. Pulse Velocity with the Uniaxial Compressive Strength for
Chai, T., and R. R. Draxler. 2014. “Root Mean Square Error Rocks.” Journal of Engineering 17 (4): 992–1006.
(RMSE) or Mean Absolute Error (MAE)? -Arguments Jassim, S. Z., A. S. Jibril, and N. M. S. Numan. 1997. “Gypsum
Against Avoiding RMSE in the Literature.” Geoscientific Karstification in the Middle Miocene Fatha Formation,
Model Development 7 (3): 1247–1250. doi:10.5194/gmd-7- Mosul Area, Northern Iraq.” Geomorphology 18 (2): 137–
1247-2014. 149. doi:10.1016/S0169-555X(96)00018-9.
Çobanoglu, I., and S. B. Çelik. 2008. “Estimation of Uniaxial Kahraman, S. 2001. “Evaluation of Simple Methods for
Compressive Strength From Point Load Strength, Schmidt Assessing the Uniaxial Compressive Strength of Rock.”
Hardness and P-Wave Velocity.” Bulletin of Engineering International Journal of Rock Mechanics and Mining
Geology and the Environment 67 (4): 491–498. doi:10. Sciences 38 (7): 981–994. doi:10.1016/S1365-1609
1007/s10064-008-0158-x. (01)00039-9.
Çobanoğlu, İ, and S. B. Çelik. 2017. “Assessments on the Kallu, R., and P. Roghanchi. 2015. “Correlations Between
Usability of Wide Wheel (Capon) Test as Reference Direct and Indirect Strength Test Methods.” International
Abrasion Test Method for Building Stones.” Construction Journal of Mining Science and Technology 25 (3): 355–360.
and Building Materials 151: 319–330. doi:10.1016/j. doi:10.1016/j.ijmst.2015.03.005.
conbuildmat.2017.06.045. Karaca, Z., R. Hacimustafaoʇlu, and M. V. Gökçe. 2015. “Grain
Daoud, H. 2009. “Carbonate Microfacies Analysis of Sinjar Properties, Grain-Boundary Interactions and Their Effects
Formation from Qara Dagh Mountain. Southwest of on the Characteristics of Marbles Used as Building
Sulaimaniyah City, Kurdistan Region, Iraq.” World Stones.” Construction and Building Materials 93: 166–171.
Academy of Science, Engineering and Technology 58: doi:10.1016/j.conbuildmat.2015.05.023.
1227–1237. Keller, F. S. 1941. “Light-aversion in the White rat.” The
Daoud, H., Y. Alshkane, and K. A. Rashid. 2018. “Prediction of Psychological Record 4 (17): 235–250. doi:10.1007/
Uniaxial Compressive Strength and Modulus of Elasticity BF03393248.
for Some Sedimentary Rocks in Kurdistan Region- Iraq Kurtuluş, C., F. Sertçelik, and I. Sertçelik. 2016. “Correlating
Using Schmidt Hammer.” Kirkuk University Journal for Physico-Mechanical Properties of Intact Rocks with P-
Scientific Studies 13 (1): 52–67. Wave Velocity.” Acta Geodaetica et Geophysica 51 (3):
Daoud, H. S. D., K. A. R. Rashed, and Y. M. A. Alshkane. 2018. 571–582. doi:10.1007/s40328-015-0145-1.
“Correlations of Uniaxial Compressive Strength and La Russa, M. F., C. M. Belfiore, G. V. Fichera, R. Maniscalco, C.
Modulus of Elasticity with Point Load Strength Index, Calabrò, S. A. Ruffolo, and A. Pezzino. 2015. “The
Pulse Velocity and Dry Density of Limestone and Behaviour to Weathering of the Hyblean Limestone in the
Sandstone Rocks in Sulaimani Governorate, Kurdistan Baroque Architecture of the Val di Noto (SE Sicily): An
Region, Iraq.” Journal of Zankoy Sulaimani - Part A 19 Experimental Study on the “Calcare a Lumachella” Stone.”
(3&4): 57–72. doi:10.17656/jzs.10632. Construction and Building Materials 77: 7–19. doi:10.
Deere, D. U., and R. P. Miller. 1966. “Engineering Classification 1016/j.conbuildmat.2014.11.073.
and Index Properties for Intact Rock.” Illinois Univ at Minaeian, B., and K. Ahangari. 2013. “Estimation of Uniaxial
Urbana Dept of Civil Engineering Chicago. 1–327. Compressive Strength Based on P-Wave and Schmidt
Dinçer, I., A. Acar, I. Çobanog lu, and Y. Uras. 2004. Hammer Rebound Using Statistical Method.” Arabian
“Correlation Between Schmidt Hardness, Uniaxial Journal of Geosciences 6 (6): 1925–1931. doi:10.1007/
Compressive Strength and Young’s Modulus for s12517-011-0460-y.
Andesites, Basalts and Tuffs.”.” Bulletin of Engineering Minitab, I. N. C. 2017. “MINITAB Statistical Software.”
Geology and the Environment 63 (2): 141–148. doi:10. Minitab Release. 13.
1007/s10064-004-0230-0. Mishra, D. A., and A. Basu. 2013. “Estimation of Uniaxial
Dunnington, H. V. 1958. “Generation, Migration and Compressive Strength of Rock Materials by Index Tests
Dissipation of Oil in Northern Iraq.” In: Arabian Gulf, Using Regression Analysis and Fuzzy Inference System.”
12 D. A. MOHAMAD ET AL.

Engineering Geology 160: 54–68. doi:10.1016/j.enggeo.2013. Turkey.” Engineering Geology 51 (4): 303–317. doi:10.
04.004. 1016/S0013-7952(98)00071-4.
Numan, N. M. S., R. A. Hammoudi, and J. Chorowicz. 1998. Wang, Y., and A. E. Aladejare. 2015. “Selection of Site-Specific
“Synsedimentary Tectonics in the Eocene Pila Spi Regression Model for Characterization of Uniaxial
Limestone Formation in Iraq and its Geodynamic Compressive Strength of Rock.” International Journal of
Implications.” Journal of African Earth Sciences 27 (1): Rock Mechanics and Mining Sciences 75: 73–81. doi:10.
141–148. doi:10.1016/S0899-5362(98)00052-9. 1016/j.ijrmms.2015.01.008.
O’Rourke, J. E. 1989. “Rock Index Properties for Willmott, C. J., and K. Matsuura. 2005. “Advantages of the
Geoengineering in Underground Development.” Mining Mean Absolute Error (MAE) Over the Root Mean Square
Engineering (Littleton, Colorado 41 (2): 106–109. Error (RMSE) in Assessing Average Model Performance.”
Ott, R. L., and M. T. Longnecker. 2015. “An Introduction to Climate Research 30 (1): 79–82. doi:10.3354/cr030079.
Statistical Methods and Data Analysis.” Nelson Education, Yagiz, S. 2009. “Predicting Uniaxial Compressive Strength,
doi:10.5194/gmd-7-1247-2014,. 2014. Modulus of Elasticity and Index Properties of Rocks
Sabatakakis, N., G. Koukis, G. Tsiambaos, and S. Papanakli. Using the Schmidt Hammer.” Bulletin of Engineering
2008. “Index Properties and Strength Variation Geology and the Environment 68 (1): 55–63. doi:10.1007/
Controlled by Microstructure for Sedimentary Rocks.” s10064-008-0172-z.
Engineering Geology 97 (1-2): 80–90. doi:10.1016/j. Yaşar, E., and Y. Erdogan. 2004. “Estimation of Rock
enggeo.2007.12.004. Physicomechanical Properties Using Hardness Methods.”
Sachpazis, C. I. 1990. “Correlating Schmidt Hardness with Engineering Geology 71 (3-4): 281–288. doi:10.1016/S0013-
Compressive Strength and Young’s Modulus of Carbonate 7952(03)00141-8.
Rocks.” Bulletin of the International Association of Yilmaz, N. G., and R. M. Goktan. 2018. “Comparison and
Engineering Geology-Bulletin de L’Association Combination of two NDT Methods with Implications for
Internationale de Géologie de L’Ingénieur 42 (1): 75–83. Compressive Strength Evaluation of Selected Masonry and
Saptono, S., S. Kramadibrata, and B. Sulistianto. 2013. “Using Building Stones.” Bulletin of Engineering Geology and the
the Schmidt Hammer on Rock Mass Characteristic in Environment, 1–11. doi:10.1007/s10064-018-1382-7.
Sedimentary Rock at Tutupan Coal Mine.” Procedia Earth Yilmaz, I., and H. Sendr. 2002. “Correlation of Schmidt Hardness
and Planetary Science 6: 390–395. doi:10.1016/j.proeps. with Unconfined Compressive Strength and Young’s Modulus
2013.01.051. in Gypsum from Sivas (Turkey).” Engineering Geology 66 (3-
Sengun, N., R. Altindag, S. Demirdag, and H. Yavuz. 2011. “P- 4): 211–219. doi:10.1016/S0013-7952(02)00041-8.
wave Velocity and Schmidt Rebound Hardness Value of Yilmaz, I., and A. G. Yuksek. 2008. “An Example of Artificial
Rocks Under Uniaxial Compressional Loading.” Neural Network (ANN) Application for Indirect Estimation
International Journal of Rock Mechanics and Mining of Rock Parameters.” Rock Mechanics and Rock Engineering
Sciences 48 (4): 693–696. doi:10.1016/j.ijrmms.2011.02.007. 41 (5): 781–795. doi:10.1007/s00603-007-0138-7.
Shalabi, F. I., E. J. Cording, and O. H. Al-Hattamleh. 2007. Yurdakul, M., and H. Akdas. 2013. “Modeling Uniaxial
“Estimation of Rock Engineering Properties Using Compressive Strength of Building Stones Using non-
Hardness Tests.” Engineering Geology 90 (3–4): 138–147. Destructive Test Results as Neural Networks Input
doi:10.1016/j.enggeo.2006.12.006. Parameters.” Construction and Building Materials 47:
Singh, R. N., F. P. Hassani, and P. A. S. Elkington. 1983. 1010–1019. doi:10.1016/j.conbuildmat.2013.05.109.
“Application of Strength and Deformation Index Testing Zhang, W. G., and A. T. C. Goh. 2013. “Multivariate Adaptive
to the Stability Assessment of Coal Measures Regression Splines for Analysis of Geotechnical Engineering
Excavations.” International Journal of Rock Mechanics and Systems.” Computers and Geotechnics 48: 82–95. doi:10.
Mining Sciences & Geomechanics Abstracts 21 (2): 48. 1016/j.compgeo.2012.09.016.
doi:10.1016/0148-9062(84)91244-0. Zhang, W., and A. T. Goh. 2016. “Multivariate Adaptive
Sirdesai, N. N., T. Gupta, T. N. Singh, and P. G. Ranjith. 2018. Regression Splines and Neural Network Models for
“Studying the Acoustic Emission Response of an Indian Prediction of Pile Drivability.” Geoscience Frontiers 7 (1):
Monumental Sandstone Under Varying Temperatures and 45–52. doi:10.1016/j.gsf.2014.10.003.
Strains.” Construction and Building Materials 168: 346– Zhou, Z., X. Cai, D. Ma, L. Chen, S. Wang, and L. Tan. 2018.
361. doi:10.1016/j.conbuildmat.2018.02.180. “Dynamic Tensile Properties of Sandstone Subjected to
Tug rul, A., and I. H. Zarif. 1999. “Correlation of Wetting and Drying Cycles.” Construction and Building
Mineralogical and Textural Characteristics with Materials 182: 215–232. doi:10.1016/j.conbuildmat.2018.
Engineering Properties of Selected Granitic Rocks from 06.056.

You might also like