Professional Documents
Culture Documents
PII: S1674-7755(19)30722-X
DOI: https://doi.org/10.1016/j.jrmge.2019.08.001
Reference: JRMGE 606
Please cite this article as: Aladejare AE, Evaluation of empirical estimation of uniaxial compressive
strength of rock using measurements from index and physical tests, Journal of Rock Mechanics and
Geotechnical Engineering, https://doi.org/10.1016/j.jrmge.2019.08.001.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. 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.
© 2019 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Elsevier B.V. All rights reserved.
Full Length Article
Oulu Mining School, University of Oulu, Pentti Kaiteran katu 1, Oulu, 90014, Finland
Received 15 March 2019; received in revised form 5 June 2019; accepted 6 August 2019
Abstract: The uniaxial compressive strength (UCS) of rock is an important parameter required for design and analysis of rock structures, and rock mass
classification. Uniaxial compression tests are a direct means to obtain the UCS values. However, these tests are generally tedious, time-consuming,
expensive, and sometimes impossible to perform due to difficult rock conditions. Therefore, several empirical equations have been developed to
estimate the UCS from results of index and physical tests of rock. Nevertheless, numerous empirical models available in the literature often make it
difficult for mining engineers to decide which empirical equation provides the most reliable estimate of UCS. This study evaluates estimation of UCS of
rocks from several empirical equations. The study uses data of point load strength (Is(50)), Schmidt rebound hardness (SRH), block punch index (BPI),
effective porosity (n) and density (ρ) as inputs to empirically estimate the UCS. The estimated UCS values from empirical equations are compared with
experimentally obtained or measured UCS values, using statistical analyses. It shows that the reliability of UCS estimated from empirical equations
depends on the quality of data used to develop the equations, type of input data used in the equations, and the quality of input data from index or
physical tests. The results show that the point load strength (Is(50)) is the most reliable index for estimating UCS among the five types of tests evaluated.
Because of type-specific nature of rock, restricting the use of empirical equations to the similar rock types for which they are developed is one of the
measures to ensure satisfactory prediction performance of empirical equations.
Keywords: uniaxial compressive strength (UCS); empirical equation; index test; physical test; uniaxial compression test; statistical analysis
1. Introduction high-quality cores or cut samples from some rocks is rather difficult, even
impossible. For example, it is difficult to extract cores or cut samples to
The uniaxial compressive strength (UCS) of rock is one of the most the required sizes from weak or highly weathered and thin bedded rocks.
frequently used rock properties in civil and mining engineering practices. In addition, direct laboratory test is complicated and time-consuming
Bieniawski (1976) reported that mining engineers request for UCS more (Palchik, 2007; Mishra and Basu, 2013; Wang and Aladejare, 2015, 2016;
often than any other rock material properties. The UCS is required as an Aladejare, 2016). Furthermore, setting up of the equipment required for
input in most rock and mining engineering designs. The UCS is also laboratory determination of UCS is often tedious. For small to
needed in prediction of deformation modulus (Sonmez et al., 2006) and medium-size rock engineering projects, engineers often consider that it is
major rock mass classifications, like the rock mass rating (RMR) system time-consuming to set up such equipment for relatively small number of
(Bieniawski, 1974; Aladejare and Wang, 2019a), rock mass index (RMi) laboratory tests. For this, researchers developed different methods for
(Palmstrøm, 1996), and among others. In addition, the Hoek–Brown indirect estimation of UCS using empirical relationship between UCS and
failure criterion (Hoek et al., 2002) for predicting the strength parameters other index and physical properties of rock which are less difficult to be
of rock masses also requires the UCS of intact rock as one of the inputs measured (Sachpazis, 1990; Tuğrul and Gürpinar, 1997; Chatterjee and
(Aladejare and Wang, 2019b). From surface to underground mining Mukhopadhyay, 2002; Gokceoglu and Zorlu, 2004; Diamantis et al., 2009;
practice, UCS is a critical rock property when considering a variety of Basu and Kamran, 2010). These tests offer estimation of UCS at low cost
issues encountered during drilling, blasting, and support construction of associated with less experimental effort.
open pit slope or underground excavations and tunnels (Adebayo and These advantages have led to an increasing popularity of the use of
Aladejare, 2013). As UCS is important for most mining engineering various indirect tests to estimate UCS. The possibility of little or no
designs, inadequate information about rock’s UCS may lead to unexpected specimen preparation by some of the tests used to indirectly estimate UCS
failure of mining engineering structures. Therefore, determination of UCS through empirical equations makes them more attractive (Gunsallus and
is not only important but also strategic to the overall success of mining Kulhawy, 1984; Ghosh and Srivastava, 1991; Singh and Singh, 1993;
engineering designs and practices. Ulusay et al., 1994; Chau and Wong, 1996; Tuğrul and Zarif, 1999;
The direct laboratory method for determining UCS of rock (i.e. using Chatterjee and Mukhopadhyay, 2002; Gokceoglu and Zorlu, 2004;
uniaxial compression test) was proposed and standardised by the Tsiambaos and Sabatakakis, 2004; Aydin and Basu, 2005; Kahraman and
International Society for Rock Mechanics and Rock Engineering (ISRM) Gunaydin, 2009; Basu and Kamran, 2010; Aliyu et al., 2019). Results of
(Ulusay and Hudson, 2007). The direct laboratory determination of UCS some index and physical tests have been recommended for indirect
requires high-quality cores or cut samples. Nevertheless, obtaining estimation of UCS. Numerous studies of empirical equations developed for
indirect estimation of UCS in the literature generally include those using
Table 1
Simple regression equations for estimating UCS.
Input data Source Equation Rock type
Is(50) Broch and Franklin (1972) UCS = 23.7Is(50) Several
Bieniawski (1975) UCS = 23Is(50) Several
Gunsallus and Kulhawy (1984) UCS = 16.5Is(50) + 51 Sedimentary
Ghosh and Srivastava (1991) UCS = 16Is(50) Igneous
Singh and Singh (1993) UCS = 23.37Is(50) Metamorphic
Ulusay et al. (1994) UCS = 19Is(50) + 12.7 Sedimentary
Chau and Wong (1996) UCS = 12.5Is(50) Igneous
Smith (1997) UCS = 14.3Is(50) Sedimentary
Tuğrul and Zarif (1999) UCS = 15.25Is(50) Igneous
Kahraman (2001) UCS = 8.41Is(50) + 9.51 Several
Sulukcu and Ulusay (2001) UCS = 15.31Is(50) Several
Tsiambaos and Sabatakakis (2004) UCS = 23Is(50) Sedimentary
Fener et al. (2005) UCS = 9.08Is(50) + 39.32 Several
Kahraman et al. (2005) UCS = 10.22Is(50) + 24.31 Several
Agustawijaya (2007) UCS = 13.4Is(50) Several
Kahraman and Gunaydin (2009) UCS = 8.2Is(50) + 36.43 Igneous
UCS = 18.45Is(50) − 13.63 Metamorphic
UCS = 29.77Is(50) − 51.49 Sedimentary
UCS = 10.92Is(50) + 24.24 Several
Basu and Kamran (2010) UCS = 11.103Is(50) + 37.659 Metamorphic
Kohno and Maeda (2012) UCS = 16.4Is(50) Several
Mishra and Basu (2012) UCS = 14.63Is(50) Several
Aliyu et al. (2019) UCS = 17.6Is(50) + 13.5 Sedimentary
SRH Haramy and DeMarco (1985) UCS = 0.99SRH − 0.38 Sedimentary
O’Rourke (1989) UCS = 4.85SRH − 76.18 Sedimentary
Sachpazis (1990) UCS = 4.29SRH − 67.52 Sedimentary
Gökçeoglu (1996) UCS = 0.0001SRH3.2658 Sedimentary
Tuğrul and Zarif (1999) UCS = 8.36SRH − 416 Igneous
Katz et al. (2000) ln(UCS)=0.792 + 0.067SRH ± 0.231 Several
Yılmaz and Sendır (2002) UCS = exp(0.818 + 0.059SRH) Sedimentary
Aydin and Basu (2005) UCS = 1.45e0.07SRH Igneous
Kılıç and Teymen (2008) UCS =0.0137SRH2.2721 Several
Gupta (2009) UCS = 0.64SRH + 37.5 Metamorphic
BPI Van der Schrier (1988) UCS = 6.1BPI − 3.3 Several
Ulusay and Gokceoglu (1997) UCS = 5.5BPI Several
Gokceoglu and Aksoy (2000) UCS = 5.25BPI Several
Sulukcu and Ulusay (2001) UCS = 5.1BPI Several
Gokceoglu and Zorlu (2004) UCS = 2.72BPI + 13.7 Sedimentary
Mishra and Basu (2012) UCS = 4.93BPI Several
n Al-Harthi et al. (1999) UCS = 274 − 8.51n, n < 20%; UCS = 104 − 1.01n, n > 20% Igneous
Chatterjee and Mukhopadhyay (2002) UCS = 34.44e−0.044n Sedimentary
Chatterjee and Mukhopadhyay (2002) UCS = 64.23e−0.085n Sedimentary
ρ Deere and Miller (1966) UCS =(28812.5ρ − 52.586) × 0.0069 Several
Tuğrul and Gürpinar (1997) UCS = 10−5ρ16.7 Igneous
Chatterjee and Mukhopadhyay (2002) UCS = 37.34ρ − 63.11 Sedimentary
Chatterjee and Mukhopadhyay (2002) UCS = 55.57ρ − 100.75 Sedimentary
Table 2
Multiple regression equations for estimating UCS.
Source Technique Input data
Grima and Babuška (1999) FIS ρ, L, n
Meulenkamp and Grima (1999) ANN L, n, ρ, d
Singh et al. (2001) ANN PSV
Gokceoglu and Zorlu (2004) FIS Is(50), BPI, Vp, BTS
Yılmaz and Yuksek (2008) ANFIS Vp, Is(50), SRH, WC
Dehghan et al. (2010) ANN Vp, Is(50), SRH, n
Jahanbakhshi et al. (2011) ANN ρ, n, Vp
Monjezi et al. (2012) ANN-GA SRH, ρ, n
Rabbani et al. (2012) ANN n, BD, Sw
Yagiz et al. (2012) ANN Vp, n, SRH, Id, γd
Beiki et al. (2013) GA ρ, n, Vp
Mishra and Basu (2013) FIS Vp, Is(50), BPI, SRH
Yesiloglu-Gultekin et al. (2013) ANFIS BTS, Vp
Armaghani et al. (2014) ANFIS Vp, ρ, PSV
Rezaei et al. (2014) FIS SRH, ρ, n
Armaghani et al. (2015) ANN-ANFIS Vp, SRH, Is(50)
Momeni et al. (2015a) PSO-ANN SRH, Vp, Is(50), ρ
Mohamad et al. (2015) ANN-PSO Is(50), BD, Vp, BTS
Torabi-Kaveh et al. (2015) ANN Vp, ρ, n
Armaghani et al. (2016) ICA-ANN n, Vp, SRH, Is(50)
Madhubabu et al. (2016) ANN Vp, n, ρ, Is(50), µ
Armaghani et al. (2018) GEP DD, Id, BTS
Note: BD: bulk density; DD: dry density; BTS: Brazilian tensile strength; d: grain size; FIS: fuzzy inference system; GA: genetic algorithm; GEP: gene expression programming;
ICA: imperialist competitive algorithm; Id: slake durability index; L: Equotip value; n: porosity; PSV: petrography study values; Sw: water saturation; Vm: P-wave velocity in solid
part of the sample; Vp: P-wave velocity ; WC: water content; γd: dry unit weight; µ: Poisson’s ratio.
gravity techniques, mercury displacement and Boyle's law techniques 3.2. Performance indicators for the evaluation of empirical equations
(Franklin, 1979). Mishra and Basu (2013) reported that effective porosity
and density are widely used in engineering geological investigations for The technique employed to evaluate estimation of UCS from index
characterising various physico-mechanical parameters of rock materials. and physical tests results using empirical models is to compare the
Empirical equations between UCS and effective porosity and those estimated UCS values with the measured (i.e. experimental) UCS values
between UCS and density have been also reported (Al-Harthi et al., 1999, reported for the same rock type. The trend, statistics and probability
Palchik, 1999; Tuğrul and Zarif, 1999; Chatterjee and Mukhopadhyay, distribution of estimated UCS are compared with measured UCS. Then,
2002). three performance indicators are used in this study to evaluate the
prediction performance of the empirical equations. Absolute average
3. Rock data and performance indicators relative error percentage (AAREP), root mean square error (RMSE) and
variance accounted for (VAF) are adopted as indicators to assess the UCS
To evaluate empirical estimation of UCS of rock, index and physical prediction.
test measurements of three types of rocks are used. Also, three different AAREP is a measure of prediction accuracy of estimation method in
performance indicators are employed to compare the estimated UCS from statistics. Shen and Karakus (2014) explained that AAREP is useful for
different empirical equations with measured UCS. Details of the database evaluating a group of data. They used AAREP in their study to evaluate the
of rock data and performance indicators are discussed in subsequent reliability of Hoek-Brown constant (mi) values calculated by different
subsections. methods. AAREP is calculated as
= ∑
× 100% (1)
3.1. Rock data
=! ∑ " − $%
Shen and Karakus (2014) and Bahaaddini and Moghadam (2019).
(2)
RMSE measures the deviation of predicted or estimated values from
the observed or measured values. RMSE is calculated as
Hyndman and Koehler (2006) explained that RMSE is a measure of measured UCS and underestimates the UCS compared with the measured
accuracy, to compare forecasting errors of different models for dataset. UCS. Note that the estimated UCS has not been presented for igneous and
RMSE is always non-negative, and a value of 0 indicates a perfect fit of a metamorphic rocks in Fig. 3 and metamorphic rock in Figs. 4 and 5. This is
model to the data. In general, low RMSE value indicates high prediction because most of the other models available in the literature are developed
ability of empirical model and input data. Willmott et al. (2006) explained from several rock types and this study is restricted to specific rock types.
that RMSE is the square root of the average of squared errors, and the
effect of each error on RMSE is proportional to the size of the squared 4.2. Statistics and probability distribution of UCS
error. Therefore, larger errors have a disproportionately greater effect on
RMSE. Visual examination is not enough to evaluate the estimated UCS from
VAF is another statistical method used to measure preciseness of the empirical models and input data. Hence, statistics and probability
prediction methods. High VAF denotes high prediction performance for a distribution of estimated UCS are presented and compared with those of
given dataset (Bahaaddini and Moghadam, 2019). VAF is calculated as measured UCS in this section. Table 5 presents the statistics of the
follows: estimated UCS from different models including measured UCS. As rock is
)*+" $
type-specific in nature, Table 5 divides the results into those of igneous,
& ' = (1 −
, × 100% "3$
)*+" $ metamorphic and sedimentary rocks. The minimum values, maximum
values, means and standard deviations of UCS are estimated and
where Var" $ is the variance.
presented. For igneous rock, Models 1 and 3 produce the results close to
The VAF method is often used to verify the correctness of a model, by
the measured UCS in terms of their minimum, maximum and mean values.
comparing the observed or measured values with the predicted or
Many of the models for igneous rock produce estimates with standard
estimated values of the model. If the two values that are the same, the VAF
deviations being relatively close to that of measured UCS, while the mean
is 100%. If not, the VAF will be lower. The higher the difference between
values are significantly different. Therefore, Models 1 and 3 using Is(50)
measured and estimated values, the lower the VAF, and vice versa.
can be taken as satisfactory models in this category. For metamorphic
rock, the estimated UCS values from the models do not provide
4. Evaluation of UCS estimation from empirical models
satisfactory statistics compared with measured UCS values. Although the
estimated UCS from Model 7 using Is(50) produces a mean value relatively
4.1. Trend of UCS
close to the mean of measured UCS, other statistics like minimum,
maximum and standard deviation values are significantly different from
Empirical models 1–27 were used to estimate UCS from their
those of the measured UCS. This again indicates the possibility to
respective input data. The estimated UCS values are plotted with
underestimate or overestimate UCS when UCS is obtained through
measured UCS values in Figs. 1–5, which are used to compare the
empirical models. For sedimentary rock, the estimated UCS values from
measured UCS with estimated UCS from inputs of Is(50), SRH, BPI, n, and
empirical models do not provide satisfactory statistics compared with
ρ, respectively. From Fig. 1, it seems that the Is(50) data and empirical
measured one. The physical tests (i.e. n and ρ) and BPI values lead to
models used provide satisfactory estimation of UCS values. Based on
underestimation of UCS; Is(50) values lead to overestimation of UCS
visual judgement of the performance of the empirical models, the
compared with measured one, whilst SRH values lead to underestimation
estimated UCS from Models 1, 3, 7 and 11 seems to be close to the
and overestimation of UCS.
measured one. In Fig. 1, not all estimated UCS values are close to the
Figs. 6–8 present cumulative distribution functions (CDFs) of UCS for
measured ones, but they have the same trend. From Fig. 2, the estimated
igneous, metamorphic and sedimentary rocks. In Fig. 6, the CDFs of
UCS values from empirical models and SRH data do not provide a
estimated UCS from Models 1 and 3 are close to that of measured UCS.
satisfactory characterisation of UCS at the site. For all the cases plotted in
This fair agreement suggests that the information contained in the
Fig. 2, the estimated UCS plot far away from the measured UCS. In
estimated UCS using Models 1 and 3 is consistent with that obtained from
addition, there are disparities in the trends of the estimated and measured
the measured UCS. The CDFs of the estimated UCS obtained from other
UCS values, unlike those presented in Fig. 1. Fig. 3 presents the plots of
models are not close to that of measured UCS. This further confirms the
estimated UCS values of sedimentary rock from a model using BPI as input
statistical results of the estimated UCS, indicating that Models 1 and 3
data. Although, the empirical model and BPI data underestimate the UCS
provide better estimates than other models. In Fig. 7, the CDFs of the
values, the estimated UCS has a similar trend with the measured UCS. Fig.
estimated UCS generally plot away from that of the measured UCS.
4 presents the estimated UCS values of igneous and sedimentary rocks
Although, the CDF of estimated UCS from Model 7 is slightly close to the
from empirical models using effective porosity (n) as input data. For both
CDF of measured UCS, it is still far away when compared with the
rock types, the estimated UCS values have different trends from the
closeness of the CDFs of UCS from Models 1 and 3 as indicated in Fig. 6.
measured UCS values. Averagely, Model 22 overestimates UCS for igneous
Like Fig. 7, Fig. 8 shows that the CDFs of estimated UCS plot away from the
rock while Models 23 and 24 underestimate UCS for sedimentary rock.
CDF of measured UCS. Although the CDF of estimated UCS from Model 11
Fig. 5 presents the estimated UCS values of igneous and sedimentary rocks
follows the pattern and trend of the CDF of measured UCS, it plots away
from empirical models using density (ρ) as input data. For igneous rock,
from the CDF of estimated UCS. The other CDFs of estimated UCS do not
the estimated UCS has a similar trend with measured one but displays an
follow the pattern or trend of the CDF of measured UCS, and they plot far
overestimation of UCS compared with the measured UCS in the same
away.
figure. For sedimentary rock, the estimated UCS has a different trend from
300 300
UCS (MPa)
180 180
150 150
120 120
90 90
60 60
30 30
0 0
G1
G2
G3
G4
G5
G6
G7
G8
G9
G10
G11
G12
G13
G14
G16
G17
G18
G19
G20
G1
G2
G3
G4
G5
G6
G7
G8
G9
G10
G11
G12
G13
G15
G16
G17
G18
G19
G20
Sample ID Sample ID
(a) (a)
300 300
UCS (MPa)
UCS (MPa)
180 180
150 150
120 120
90 90
60 60
30 30
0 0
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
S12
S13
S14
S15
S16
S17
S18
S19
S20
S1
S3
S4
S5
S6
S7
S8
S9
S10
S11
S12
S13
S14
S15
S16
S17
S18
S19
S20
Sample ID Sample ID
(b) (b)
300 300
Measured UCS Measured UCS
Uniaxial compressive strength,
270 Estimated UCS from Model 8 270 Estimated UCS from Model 16
240 Estimated UCS from Model 9 240 Estimated UCS from Model 17
Estimated UCS from Model 10 Estimated UCS from Model 18
210 Estimated UCS from Model 11 210 Estimated UCS from Model 19
UCS (MPa)
UCS (MPa)
180 Estimated UCS from Model 12 180 Estimated UCS from Model 20
150 150
120 120
90 90
60 60
30 30
0 0
SS1
SS2
SS3
SS4
SS6
SS7
SS9
SS10
SS11
SS12
SS13
SS14
SS15
SS16
SS17
SS18
SS19
SS20
SS1
SS2
SS3
SS4
SS5
SS6
SS7
SS9
SS10
SS11
SS12
SS13
SS14
SS15
SS16
SS17
SS19
SS20
Sample ID Sample ID
(c) (c)
Fig. 1. Comparison of the estimated UCS from different models using Fig. 2. Comparison of the estimated UCS from different models using SRH
Is(50) as input data: (a) Igneous rock; (b) metamorphic rock; and (c) as input data: (a) Igneous rock; (b) metamorphic rock; and (c)
sedimentary rock. sedimentary rock.
180 presents the results of the analyses. Comparison of the results indicates
150 that the prediction performance of empirical equations varies with
120
performance indicators. In this study, the best three prediction
90
60 performances are identified for each of the performance indicators. For
30 AAREP, Models 1, 3 and 2 have the best performances at 7.56%, 10.25%
0 and 23.99%, respectively, because the lower the AAREP, the better the
SS1
SS2
SS4
SS6
SS7
SS10
SS11
SS12
SS13
SS14
SS15
SS16
SS17
SS19
SS20
Sample ID models are empirical models developed for igneous rocks using Is(50) as
input. For RMSE, Models 1, 7 and 3 have the best performances at 15.99
Fig. 3. Comparison of the estimated UCS of sedimentary rock from a MPa, 16.46 MPa and 17.67 MPa, respectively. Like AAREP, the lower the
RMSE, the better the performance of an empirical equation. Like the
model using BPI as input data.
regression models with best performances under AAREP, the three
models with the best performances under RMSE are also using Is(50) as
input. The slight difference from the case of AAREP is that two of three
models
Uniaxial compressive strength, 300 300
UCS (MPa)
UCS (MPa)
180 180
150 150
120 120
90 90
60 Measured UCS 60
30 Estimated UCS from Model 22 30
0 0
SS1
SS2
SS3
SS4
SS5
SS6
SS7
SS8
SS9
SS10
SS11
SS12
SS13
SS14
SS15
SS16
SS17
SS19
SS20
G1
G2
G3
G4
G5
G6
G7
G8
G9
G10
G11
G12
G13
G14
G15
G16
G17
G18
G19
G20
Sample ID Sample ID
(a) (b)
Fig. 4. Comparison of the estimated UCS from different models using effective porosity (n) as data input: (a) Igneous rock; and (b) sedimentary rock.
300 300
UCS (MPa)
UCS (MPa)
180 180
150 150
120 120
90
90
60
60
30
30
0
0
SS1
SS2
SS3
SS4
SS5
SS6
SS7
SS8
SS9
SS10
SS11
SS12
SS13
SS14
SS15
SS16
SS17
SS18
SS19
SS20
G1
G2
G3
G4
G5
G6
G7
G8
G9
G10
G11
G12
G13
G14
G15
G16
G17
G18
G19
G20
Sample ID Sample ID
(a) (b)
Fig. 5. Comparison of the estimated UCS from different models using density (ρ) as data input: (a) Igneous rock; and (b) sedimentary rock.
are for estimating igneous rocks while the other is for metamorphic rocks. 5. Conclusions
For VAF, Models 6, 11 and 2 have the best performances at 83.62%,
82.88% and 78.01%, respectively. Unlike AAREP and RMSE, the higher the The UCS is one of the basic inputs required for designs and analyses
VAF, the better the performance of an empirical equation. In a similar of most mining engineering projects. Uniaxial compression tests are
fashion to the regression models with best performances under AAREP tedious, time-consuming and expensive when they are used to obtain UCS
and RMSE, the three models with the best performances under VAF are of rocks. However, at the early stage of many practical applications, and in
those with Is(50) as input. The difference from the cases of AAREP and many small to medium-sized projects, UCS may not be available. Hence, it
RMSE is that each of igneous, metamorphic and sedimentary rocks has is necessary to estimate UCS without having uniaxial compression test
one model each out of the three models, with best prediction data. In the absence of measured UCS, mining engineers often estimate
performances. rock UCS based on empirical equations and results from index and
From the statistical analyses using AAREP, RMSE and VAF as physical tests. However, using empirical equations comes with the
prediction indicators, no model emerges as the best in all three analyses. possibility of overestimation or underestimation of UCS caused by
However, Models 1, 2 and 3 each has two best performances from the site-specific nature of rock properties, type-specific nature of rock and
three statistical analyses. In addition, Model 1 has the best prediction data that are used to develop empirical equations.
performance in both AAREP and RMSE analyses. From the three statistical Therefore, this study evaluates the prediction performance of 27
analyses performed in this study, the models with the best prediction empirical equations in estimating UCS. There are 60 data of uniaxial
performance are the empirical equations using Is(50) as input. The point compression tests in total from three project sites in India, with 20 data
load strength (Is(50)) appears to be the best proxy for UCS, which is each for igneous, metamorphic and sedimentary rocks. The prediction
consistent with previous studies on estimated UCS (Brook, 1985; Cargill performances of the empirical equations in estimating UCS from index and
and Shakoor, 1990; Ghosh and Srivastava, 1991; Tuğrul and Zarif, 1999; physical test results are evaluated using statistical analyses. It is found
Mishra and Basu, 2013). that the empirical equations can underestimate or overestimate UCS. Only
about 5 out
Table 5
Statistics of estimated UCS of rocks using different empirical models and input data.
Model Rock type Minimum Maximum Mean Standard deviation
Measured UCS of igneous rock 91.48 201.73 150.10 28.31
Model Rock type (igneous)
1 UCS = 16Is(50) 90.56 226.08 144.29 34.41
2 UCS = 12.5Is(50) 70.75 176.63 112.73 26.88
3 UCS = 15.25Is(50) 86.32 215.48 137.53 32.80
4 UCS = 8.2Is(50) + 36.43 82.84 152.30 110.38 17.63
13 UCS = 8.36SRH − 416 26.41 149.47 92.46 37.96
14 UCS = 1.45e(0.07SRH) 58.91 165.07 107.42 34.05
22 UCS = 274 − 8.51n, n < 20%; UCS = 104 − 1.01n, n>20% 103.60 271.49 170.57 84.00
25 UCS = 10−5ρ16.7 159.87 245.14 196.48 23.77
Measured UCS of metamorphic rock 21.36 95.14 46.53 19.01
Model Rock type (metamorphic)
5 UCS = 23.37Is(50) 27.26 175.85 89.85 36.72
6 UCS = 11.103Is(50) + 37.659 50.43 120.04 79.75 17.2
7 UCS = 18.45Is(50) − 13.63 7.59 123.27 56.31 28.58
15 UCS = 0.64SRH + 37.5 57.76 75.00 66.55 4.71
Measured UCS of sedimentary rock 12.8 172.03 57.44 41.57
Model Rock type (sedimentary)
8 UCS = 17.6Is(50) + 13.5 35.5 215.72 102.01 52.89
9 UCS =16.5Is(50) + 51.0 71.63 240.59 133.98 49.58
10 UCS =19Is(50) + 12.7 36.45 231.01 108.25 57.10
11 UCS = 14.3Is(50) 17.88 164.31 71.91 42.97
12 UCS = 23Is(50) 28.75 264.27 115.66 69.12
16 UCS =0.99SRH − 0.38 25.25 57.62 42.17 10.53
17 UCS =4.85SRH − 76.18 49.39 207.98 132.29 51.60
18 UCS = 4.29SRH − 67.52 43.55 183.83 116.88 45.64
19 UCS = 0.0001SRH3.2658 4.12 59.34 26.12 17.62
20 UCS = exp(0.818 + 0.059SRH) 8.27 42.41 22.04 10.73
21 UCS = 2.72BPI + 13.7 20.58 70.25 34.51 14.91
23 UCS = 34.44e−0.044n 17.38 30.33 22.73 4.38
24 UCS = 64.23e−0.085n 17.14 50.24 29.70 11.10
26 UCS = 37.34ρ − 63.11 17.92 33.97 24.30 4.70
27 UCS = 55.57ρ − 100.75 19.84 43.73 29.34 6.99
1
Measured UCS
Estimated UCS from Model 1 (Is(50) data)
0.9 Estimated UCS from Model 2 (Is(50) data)
Estimated UCS from Model 3 (Is(50) data)
Estimated UCS from Model 4 (Is(50) data)
0.8 Estimated UCS from Model 13 (SRH data)
Estimated UCS from Model 14 (SRH data)
Estimated UCS from Model 22 (n data)
0.7 Estimated UCS from Model 25 (ρ data)
Cumulative distribution function
0.6
0.5
0.4
0.3
0.2
0.1
0
0 50 100 150 200 250 300
Uniaxial compressive strength, UCS (MPa)
Fig. 6. Cumulative distribution function (CDF) of UCS of igneous rock. [Please change comma in y-axle number into dot. For example, change 0,1 into 0.1]
1
0.9
0.8
Cumulative distribution function
0.7
0.6
0.5
0.4
0
0 20 40 60 80 100 120 140 160 180 200
Uniaxial compressive strength, UCS (MPa)
1.00
0.90
0.80
Measured UCS
Estimated UCS from Model 8 (Is(50) data)
Cumulative distribution function
0.70
Estimated UCS from Model 9 (Is(50) data)
Estimated UCS from Model 10 (Is(50) data)
0.60
Estimated UCS from Model 11 (Is(50) data)
Estimated UCS from Model 12 (Is(50) data)
0.50
Estimated UCS from Model 16 (SRH data)
Estimated UCS from Model 17 (SRH data)
0.40
Estimated UCS from Model 18 (SRH data)
Estimated UCS from Model 19 (SRH data)
0.30 Estimated UCS from Model 20 (SRH data)
Estimated UCS from Model 21 (PBI data)
0.20 Estimated UCS from Model 23 (n data)
Estimated UCS from Model 24 (n data)
0.10 Estimated UCS from Model 26 (ρ data)
Estimated UCS from Model 27 (ρ data)
0.00
0 50 100 150 200 250 300
Uniaxial compressive strength, UCS (MPa)