Journal of Sustainable Mining 18 (2019) 8–17

HOSTED BY Contents lists available at ScienceDirect

Journal of Sustainable Mining

journal homepage: www.elsevier.com/locate/jsm

Research paper

Sensitivity analysis of fuzzy-analytic hierarchical process (FAHP) decision- T

making model in selection of underground metal mining method
Bhanu Chander Balusa, Amit Kumar Gorai∗
Department of Mining Engineering, National Institute of Technology, Rourkela, 769008, India

ARTICLE INFO ABSTRACT

Keywords: This study aims to analyse the sensitivity in decision-making which results in the selection of the appropriate
Decision-making underground metal mining method using the fuzzy-analytical hierarchy process (FAHP) model. The proposed
Mining methods model considers sixteen criteria for the selection of the most appropriate mining method out of the seven. The
Fuzzy-AHP model consists of three-layer viz. the first layer represents the criteria (factors which influence the mining
Sensitivity analysis
method), the second layer represents the sub-criteria (categorisation of the factors) and the third layer represents
the alternatives (mining methods). The priority of the different mining methods was determined based on global
weights. The global weights of seven mining method were determined using a different fuzzification factor under
different decision-making attitudes (optimistic, pessimistic and unbiased). The sensitivity of the decision-making
results was analysed in order to understand the robustness of the model.

1. Introduction Naghedehi, Mikaeil and Atei (2009) proposed the FAHP decision-
making model for the selection of an appropriate mining method for a
There are many metal mining methods available such as block Bauxite ore deposit in Iran. The decision-making in the selection of best
caving, sub level stoping, sub level caving, room and pillar mining, mining method out of the six was made based on the thirteen influen-
shrinkage stoping, cut and fill stoping, and square set stoping for ex- cing parameters. Gupta and Kumar (2012) suggested an AHP-based
cavating ore reserves from underground. The selection of a particular MCDM model for underground mining method selection. Ataei,
metal mining method depends on multiple factors, and thus the selec- Shahsavany, and Mikaeil (2013) suggested a Monte Carlo based AHP
tion can be made using the multi-criteria decision-making (MCDM) (MAHP) approach for the selection of the mining method for a Bauxite
technique. It is very important to select the suitable mining method for ore deposit in Iran. Yavuz (2015) conducted a study of the underground
excavating an ore deposit for economic reasons and for safety. MCDM coal mining method selection for five alternatives using AHP and Ya-
aims to select the most promising alternative based on the defined ger's multi criteria decision-making techniques. Dehghani, Siami, and
criteria and sub-criteria. In the past, many types of MCDM techniques Haghi (2017) suggested Grey and TODIM approaches for the selection
(AHP, FAHP, TOPSIS, etc.) have been developed and used for the se- of a mining method.
lection of mining methods. Namin, Shahriar, Ataee-Pour, and Dehghani The selection of a mining method for excavating an ore deposit is a
(2008) suggested a TOPSIS and Fuzzy TOPSIS kind of integrated model crucial task for the planners at the decision making stage. The selection
for the selection of coal and metal mining methods. Alpay and Yavuz of a mining method depends on various qualitative and quantitative
(2007) proposed a model using AHP and the Yager's decision making factors of the ore deposit. These qualitative and quantitative parameters
techniques for the selection of the underground mining method. include geometry, geo-mechanical, operational, economical, etc. Mine
Mikaeil, Naghadehi, Ataei, and Khalokakaie (2009) developed the planners face the difficulty while selecting the mining method due to
MCDM models using FAHP and TOPSIS for the selection of the mining more number of interdependent parameters. Once the operation for ore
method for a Bauxite mine in Iran based on thirteen criteria. extraction has begun with a particular mining method, it is not possible

∗
Corresponding author.
E-mail addresses: [email protected] (B.C. Balusa), [email protected] (A.K. Gorai).

https://doi.org/10.1016/j.jsm.2018.10.003
Received 4 September 2018; Received in revised form 9 October 2018; Accepted 29 October 2018
Available online 31 October 2018
2300-3960/ © 2019 Published by Elsevier B.V. on behalf of Central Mining Institute This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
B.C. Balusa, A.K. Gorai Journal of Sustainable Mining 18 (2019) 8–17

or difficult to alter the mining method. Among all the MCDM techni- method was used in many research work problems in the decision-
ques, the most popular technique that solves decision-making problems making of uncertain problems.
is the analytical hierarchy process (AHP). However, AHP is ineffective The proposed study attempts to develop a FAHP model for decision
when applied to ambiguous problems like uncertainty of the criteria making on the selection of the best mining methods out of seven al-
parameters (Tsai, Chang, & Lin, 2010). For accommodating the un- ternatives based on the criteria. The sensitivity of the model output was
certainty of the factors, AHP is integrated with fuzzy logic. Sensitivity analysed by changing the values of the fuzzification factors in different
analysis is an essential component of fuzzy-AHP decision-making decision-making attitudes. The relative priority of each mining method
models. The purpose of the sensitivity analysis is to measure the con- was determined based on the global weights in different condition. The
sistency in selecting the best alternative in different conditions. The flowchart of the working procedure is shown in Fig. 1.
final priorities of the alternatives are heavily dependent on the weights
associated with the main criteria parameters. A small change in the 2.1. Selection of criteria, sub-criteria, and alternatives
weights of the criteria have a significant impact on the final ranking of
the alternatives. Sensitivity analysis provides information about an al- The first step of the proposed model is to select the criteria and sub-
teration in the ranking of the alternatives. The sensitivity analysis of criteria for the prioritization of the alternatives (underground metal
different MCDM models has been performed in different fields, but not mining methods). The selection of a suitable mining method depends on
in the selection of mining method. Chang, Wu, Lin, and Chen (2008) various criteria like ore-geometry, the geo-mechanical conditions of the
conducted sensitivity analysis of the FAHP model for evaluating and ore, the production capacity of the deposit and various operational
controlling silicon wafer slicing quality. The model outputs were ana- parameters (Naghadehi, Mikaeil, & Ataei, 2009). The present study
lysed by increasing each criteria weight by 10%, 20%, and 30%. Hsu considered 16 criteria (dip, shape, thickness, depth, grade distribution,
and Chen (2007) developed FAHP model for the selection of franchisees RMR of the ore, RMR of the hanging wall, RMR of the footwall, RSS of
of a bedding chain retail store. The sensitivity analysis of the model was the ore, RSS of the hanging wall, RSS of the footwall, productivity, re-
conducted by changing the values of the uncertainty factors. Tabari, covery, dilution, flexibility and safety) (shown in Table 1) for selection of
Kaboli, Aryanezhad, Shahanaghi, and Siadat (2008) proposed a FAHP the most suitable mining method for a typical ore deposit. The criteria
model for site selection and sensitivity analysis was performed for the were further classified into 54 sub-criteria, as shown in Table 1, to de-
model they developed by changing the value uncertainty factor from 0 velop the hierarchical model. The number of alternatives (mining
to 1. Tseng and Lin (2008) suggested the selection of competitive ad- methods) considered in the model is seven. These are block caving (BC),
vantages in total quality management implementation using FAHP. The sublevel stoping (SS), sublevel caving (SC), room and pillar mining (RP),
sensitivity analysis of the model was analysed for competitive ad- shrinkage stoping (SH), cut and fill stoping (CF), and square set stoping
vantage by considering the fuzzification factor (α) to be 0.2, 0.4, 0.6 (SQ). The study analyses the sensitivity of the decision-making by con-
and 0.8. Tsai et al. (2010) analysed the sensitivity of the FAHP model sidering a different range of fuzzification factors and the decision maker's
for evaluating hospital organization performance by changing the fuz- attitude. The ranges of the sub-criteria were considered based on pre-
zification factor, in the range of 0.1–1. vious studies (Miller-Tait, Pakalnis, & Poulin, 1995; Tatiya, 2013).
It is clear from the literature that the selection of mining method
depends on multiple factors and thus it is important to analyse the 2.2. Design of the hierarchical structure
degree of sensitivity of decision-making due to either a change in a
factor's uncertainty level or a decision maker's attitude. The literature The hierarchical structure of the proposed FAHP model was de-
study revealed that sensitivity analysis of decision-making on the type signed based on different criteria and sub-criteria. All the identified
of mining has not been conducted by any other researcher to date. Thus, criteria, sub-criteria and the evaluation alternatives (mining methods)
the present study attempts to analyse the decision-making results in were arranged in different levels of the hierarchy (shown in Fig. 2). The
selecting the best mining method using the proposed FAHP model first, second and third level of the hierarchy defines the criteria, sub-
under different fuzzification factors and decision-making attitudes. The criteria and mining methods respectively. The last level defines the goal
model performance was analysed by changing the uncertainty levels of of the decision-making problem, i.e. the best underground metal mining
the factors from minimum to maximum in different decision-making method (UMMS) for the extraction of the ore deposit.
attitudes, i.e. optimistic, pessimistic and unbiased.
2.3. Formation of fuzzy-relative importance matrices for each level
2. Methodology
The next step is to develop relative importance matrices for each
The analytical hierarchy process (AHP) is a decision-making process level using the corresponding parameters. The relative importance
used to select the best choice from multiple alternatives. The AHP matrices for each level were built by using the FAHP scale of 1̄ to 9̄
method was initially proposed by Saaty (1980). The AHP subdivides the (Saaty, 1980). The relative importance scale in normal AHP is 1–9
problem into many levels in a hierarchical structure. The AHP method whereas in FAHP it is 1̄ to 9̄. The relative importances and their cor-
requires designing a pair-wise comparison at each level using the de- responding definitions are listed in Table 2.
cision maker's knowledge. Though AHP solves many decision-making The relative importance values of criteria in selecting the mining
problems; it is inefficient when the influencing parameters in the given method were considered from past studies (Azadeh, Osanloo, & Ataei,
problem are uncertain. AHP method is integrated with fuzzy logic to 2010; Gupta & Kumar, 2012; Naghadehi et al., 2009). The relative
deal with such type of uncertain problems. van Laarhoven and Pedrycz importance matrix of the criteria in the first level of the hierarchy was
(1983) initially proposed fuzzy logic with AHP. After that, FAHP designed as follows:

9
B.C. Balusa, A.K. Gorai Journal of Sustainable Mining 18 (2019) 8–17

DI SH TH DE GD ORMR HRMR FRMR PR RE DL ORSS HRSS FRSS FLE SE

1̄ 3̄ 1̄ 4.¯ 5 3̄ 2̄ 2̄ 4̄ 7̄ 7̄ 7̄ 2̄ 2̄ 4̄ 7̄ 7̄
DI
1/3̄ 1̄ 1/2.¯ 5 2.¯ 5 1̄ 1/3̄ 1/3̄ 1̄ 5̄ 5̄ 5̄ 1/3̄ 1/3̄ 1̄ 5̄ 5̄
SH
TH 1̄ 2.¯ 5 1̄ 4̄ 3̄ 1̄ 1̄ 3̄ 7̄ 7̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄
DE 1/4.¯ 5 1/2.¯ 5 1/4̄ 1̄ 1/3̄ 1/5̄ 1/5̄ 1/3̄ 3̄ 3̄ 3̄ 1/5̄ 1/5̄ 1/3̄ 3̄ 3̄
GD 1/3̄ 1̄ 1/3̄ 3̄ 1̄ 1/7̄ 1/7̄ 1/5̄ 5̄ 5̄ 5̄ 1/7̄ 1/7̄ 1/5̄ 5̄ 5̄
ORMR 1/2̄ 3̄ 1̄ 5̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄
HRMR 1/2̄ 3̄ 1̄ 5̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄
CP = FRMR 1/4̄ 1̄ 1/3̄ 3̄ 5̄ 1/3̄ 1/3̄ 1̄ 5̄ 5̄ 5̄ 1/3̄ 1/3̄ 1̄ 5̄ 5̄
PR 1/7̄ 1/5̄ 1/7̄ 1/3̄ 1/5̄ 1/7̄ 1/7̄ 1/5̄ 1̄ 3̄ 3̄ 1/7̄ 1/7̄ 1/5̄ 3̄ 5̄
RE
1/7̄ 1/5̄ 1/7̄ 1/3̄ 1/5̄ 1/7̄ 1/7̄ 1/5̄ 1/3̄ 1̄ 3̄ 1/7̄ 1/7̄ 1/5̄ 3̄ 5̄
DL
ORSS 1/7̄ 1/5̄ 1/7̄ 1/3̄ 1/5̄ 1/7̄ 1/7̄ 1/5̄ 1/3̄ 1/3̄ 1̄ 1/7̄ 1/7̄ 1/5̄ 3̄ 3̄
HRSS 1/2̄ 3̄ 1̄ 5̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄
FRSS 1/2̄ 3̄ 1̄ 5̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄ 7̄ 1̄ 1̄ 3̄ 7̄ 7̄
FLE 1/4̄ 1̄ 1/3̄ 3̄ 5̄ 1/3̄ 1/3̄ 1̄ 5̄ 5̄ 5̄ 1/3̄ 1/3̄ 1̄ 5̄ 5̄
SE 1/7̄ 1/5̄ 1/7̄ 1/3̄ 1/5̄ 1/7̄ 1/7̄ 1/5̄ 1/3̄ 1/3̄ 1/3̄ 1/7̄ 1/7̄ 1/5̄ 1̄ 3̄
1/7̄ 1/5̄ 1/7̄ 1/3̄ 1/5̄ 1/7̄ 1/7̄ 1/5̄ 1/5̄ 1/5̄ 1/3̄ 1/7̄ 1/7̄ 1/5̄ 1/3̄ 1̄

Similarly, the relative importance matrices for sub-criteria in the (1) (Gorai, Kanchan, Upadhyay, Tuluri, Goyal and Tchounwou, 2015).
second level and mining methods in the third level were developed
based on the UBC technique of the selection of a mining method (Miller- x¯ = [x , x + ];
1
=
1
,
1
Tait et al., 1995) and shown in Appendix A. These relative importance x¯ x+ x (1)
matrices were converted into fuzzy matrices by using the following eqn.
In general, the α value ranges between 0 and 1, and it may be any

Fig. 1. Flowchart of the working methodology.

Table 1
List of criteria, sub-criteria, and alternatives.
Sl. No Criteria Sub-criteria

1 Dip (DI) Flat (FL), Moderate (MO), Steep (ST)

2 Shape (SH) Massive (MA), Tabular (TA), Irregular (IR)
3 Thickness (TH) Narrow (NA), Intermediate (IN), Thick (TI)
4 Depth (DE) Shallow (SA), Moderate (MD), Deep (DP)
5 Grade distribution (GD) Uniform (UN), Gradational (GR), Erratic (ER)
6 RMR of Ore (ORMR) Very weak (OVW), Weak (OW), Moderate (OM), Strong (OS), Very strong (OVS)
7 RMR of hanging wall (HRMR) Very weak (HVW), Weak (HW), Moderate (HM), Strong (HS), Very strong (HVS)
8 RMR of foot wall (FRMR) Very weak (FVW), Weak (FW), Moderate (FM), Strong (FS), Very strong (FVS)
9 Productivity (PR) Low (PL), Medium (PM), High (PH)
10 Recovery (RE) Low (RL), High (RH)
11 Dilution (DL) Low (DW), Medium (DM), High (DH)
12 RSS of ore (ORSS) Very weak (ORVW), Weak (ORW), Moderate (ORM), Strong (ORS)
13 RSS of hanging wall (HRSS) Very weak (HRVW), Weak (HRW), Moderate (HRM), Strong (HRS)
14 RSS of foot wall (FRSS) Very weak (FRVW), Weak (FRW), Moderate (FRM), Strong (FRS)
15 Flexibility (FLE) Low (FLL), High (FLH)
16 Safety (SE) Low (SEL), High (SEH)
Alternatives (Underground Metal Mining Methods)
Block caving (BC), Sub level stoping (SS), Sub level caving (SC), Room and Pillar (RP), Shrinkage stoping (SH), Cut and fill stoping (CF), Square set stoping (SQ)

10
B.C. Balusa, A.K. Gorai Journal of Sustainable Mining 18 (2019) 8–17

Fig. 2. Hierarchical structure of the FAHP model.

11
B.C. Balusa, A.K. Gorai Journal of Sustainable Mining 18 (2019) 8–17

fractional value in between 0 and 1. The higher values of α (i.e., close to the ith row (GMi) of a crisp matrix of a corresponding row parameter
1) represent more uncertainty and the lower values less uncertainty. can be determined using eqn. (4). Where bij in eqn. (4) represents the
Using eqn. (1), the relative importance matrices of criteria sub-criteria value in the ith row and jth column of the crisp comparison matrix. M is
and underground mining methods were converted into fuzzy matrices. the number of parameters in the crisp comparison matrix.
This study used 6 α values (0, 0.2, 0.4, 0.6, 0.8, and 1) to analyse de- 1/ M
M
cision making results.
GMi = bij
j=1 (4)
2.4. Determination of crisp comparison matrices
The local weight of the variable can be determined using eqn. (5)
The fuzzy pair-wise comparison matrices for each level were con- N

verted into crisp comparison matrices using the following eqn. (2) (Lee, wi = GMi / GMi
1995). i=1 (5)
After determining the local weights, determination of the global
aij = aiju + (1 ) aijl (2)
weights at the third and fourth levels needs to be performed. The fuzzy
aiju and aijl in the above eqn. (2) are the upper and lower bound, re- global weights (Gk) can be computed from the local weight of the kth
spectively, of relative importance value aij in the previously developed level and the global weights of the (k-1)th level using eqn. (6).
matrix. The defuzzified value aij returns the crisp value for the relative Gk = wk Gk 1 (6)
importance value aij. In eqn. (2), λ represents the decision making at-
titude. The value of can be any value between 0 and 1. Crisp com- The global weights of each mining method were determined using
parison matrices for the parameters at each level were constructed. the above equation.
After developing the crisp comparison matrices the consistency ratio of
all the crisp comparison matrices at each level was examined using eqn. 2.6. Sensitivity analysis of decision making
(3) (Saaty, 1980). The CR values for all the parameters at each level
were found to be less than 0.1. The consistency ratio of the matrix can The sensitivity analysis of the proposed decision-making model was
be determined as conducted by varying the fuzzification factor (α) in eqn. (1) and deci-
sion-making attitude (λ) in eqn. (2). The decision-making attitude was
CI
CR =
(3) considered for three conditions (the optimistic, pessimistic, and neu-
RI
tral). The λ values for optimistic, pessimistic, and neutral conditions
In the above eqn., the CI and RI are respectively the consistency were chosen as 1, 0, and 0.5 respectively. The model output was also
index and the random index. The CI of a matrix can be determined as analysed for six sets (0, 0.2, 0.4, 0.6, 0.8 and 1) of the fuzzification
n factor (α) in the range of 0–1. The fuzzy pair-wise comparison matrices
max
CI = were formulated using different fuzzification factors (α) for each set of
n 1
criteria and sub-criteria. The crisp comparison matrices corresponding
Where λmax is the maximum eigen value, and n is the size of the crisp to each fuzzy pair-wise comparison matrix were derived for three de-
comparison matrix. cision-making attitudes. In other words, the crisp comparison matrices
The value of RI depends on the size of the matrix. Many researchers were derived for three λ values (λ = 0, 0.5, 1) using eqn. (2). The
determined RI values for various sizes of matrices. In this study, the RI decision-making model output was analysed for each combination of α
values suggested by Alonso and Lamata (2006) were considered for the and λ.
analysis; these RI values for different sizes of matrices are listed in The sensitivity of decision-making in the ranking of seven mining
Table 3. methods was analysed by considering the fuzzification factor in 16-
criteria and 54-sub-criteria. The results indicated that the ranking or
2.5. Determination of the local and global weights for prioritizing the priorities of seven mining methods were not altered by either changing
objectives of the fuzzification factor from 0 to 1 or changing the decision-making
attitude. Therefore, for any value of λ and α, the rank of a particular
The local and global weights of the parameters at each level were mining method remains the same. The rank of a particular mining
determined using the geometric mean concept. The geometric mean of method is decided based on global weights. The higher the global

Table 2
Fuzzy relative importance scale used for making pair-wise comparisons (Saaty, 1980).
Relative importance Fuzzy Scale Definition Explanation

1̄ (1,1,1) Equal importance Two activities contribute equally to the objective.

3̄ (3-α), 3, (3+α) Weak importance Experience and judgement slightly favour one activity over another.
5̄ (5-α), 5, (5+α) Essential or strong importance Experience and judgement strongly favour one activity over another.
7̄ (7-α), 7, (7+α) Demonstrated importance Experience and judgement strongly favour one activity over another.
9̄ (9-α), 9, (9+α) Extreme importance One activity is strongly favoured and demonstrated in practice.
2̄, 4̄, 6̄, 8̄ (x-α), x, (x+α) Intermediate values between two adjacent The evidence favouring one activity over another is of the highest possible order of
judgements affirmation.

Note: α is a fuzzification factor.

Table 3
Random index (RI) values for different matrix size.
Size of matrix 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

RI 0 0 0.52 0.88 1.1 1.24 1.34 1.4 1.44 1.48 1.51 1.53 1.55 1.57 1.58 1.59

12
B.C. Balusa, A.K. Gorai Journal of Sustainable Mining 18 (2019) 8–17

weight of the mining method, the higher is the rank or priority. To

0.0309
0.0083
0.0056
0.0446
0.0027
0.0056
0.0083
α=1
demonstrate the sensitivity of the ranking of various mining methods
under different degrees of uncertainty (α) and different decision-ma-
kers’ attitudes (λ), the results of one criterion (dip) is shown in the text

α = 0.8

0.0316
0.0081
0.0054
0.0461
0.0025
0.0054
0.0081
in order to reduce the manuscript length. The sensitivity of the decision-
making results for all other parameters are shown in Appendix B. The
global weights of different mining methods for a different level of un-
certainty or fuzzification factors (α) and the decision-maker's attitude

α = 0.6

0.0289
0.0075
0.0048
0.0424
0.0023
0.0048
0.0075
(λ) were determined using eqns. (1), (2) and (6), as explained above in
Sections 2.3, 2.4, and 2.5. The global weights of each mining method
correspond to flat-dip, moderate-dip, and steep-dip under different

α = 0.4

0.0314
0.0082
0.0051
0.0464
0.0025
0.0051
0.0082
fuzzification factors (α) and the decision-maker's attitude (λ), all of
which are shown in Tables 4–6 respectively.
Table 4 shows the ranks of seven mining methods for six fuzzifica-

α = 0.2

0.0311
0.0081
0.0050
0.0462
0.0025
0.0050
0.0081
tion factors (α = 0, 0.2, 0.4, 0.6, 0.8 and 1) in three decision-making

λ = 1 (optimistic)
attitudes, these being pessimistic (λ = 0), unbiased (λ = 0.5), and op-
timistic (λ = 1) with the flat dip condition of the ore deposit. The trend
of global weights for different fuzzification factors indicates that the

0.0307
0.0080
0.0049
0.0456
0.0025
0.0049
0.0080
α=0
room and pillar mining method is most appropriate for flat deposit,
irrespective of the fuzzification factors and decision-making attitudes.
The rank of the room and pillar mining method is always at the top and

0.0299
0.0076
0.0051
0.0436
0.0025
0.0051
0.0076
α=1
never alters when changing the values of α and λ. It was also observed
that though the global weights of each mining method were altered due
to changes in the value of α and λ, the rank of the mining methods

α = 0.8

0.0303
0.0078
0.0050
0.0447
0.0025
0.0050
0.0078
never altered.
Similarly, Table 5 shows the ranks of seven mining methods for six
fuzzification factors (α = 0, 0.2, 0.4, 0.6, 0.8 and 1) in three decision-

α = 0.6
making attitudes, these being pessimistic (λ = 0), unbiased (λ = 0.5),

0.0306
0.0079
0.0050
0.0453
0.0025
0.0050
0.0079
Global weights of seven mining methods for different decision-making attitudes (λ) and fuzzification factors (α) for flat-dip.

and optimistic (λ = 1) with the moderate dip condition of the ore de-
posit. The trend of global weights for different fuzzification factors in-
dicates that the cut and fill mining method is most suitable for mod-
α = 0.4

0.0307
0.0079

0.0456
0.0024

0.0079
0.005

0.005
erate-dip ore deposits irrespective of the fuzzification factors and
Dip-Flat

decision-making attitudes. Here, also, the rank of the cut and fill mining
method is always at the top and never alters when changing the values
α = 0.2
λ = 0.5 (unbiased)

0.0307
0.0079
0.0049
0.0457
0.0025
0.0049
0.0079
of α and λ. It was also observed that though the global weights of each
mining method were altered with changes in the value of α and λ, the
rank of the mining methods never altered.
0.0307
0.0079
0.0049
0.0456
0.0025
0.0049
0.0079
In the same way, Table 6 shows the ranks of seven mining methods
α=0

for steeply dipping ore deposits. In this case, the square set stoping
method exhibits the highest global weight. Here, also, the global
weights of each mining method were altered due to changes in the
0.0267
0.0061
0.0040
0.0395
0.0021
0.0040
0.0061
α=1

value of α and λ, but the rank of the mining methods never varied.
Therefore, the square set stoping method was always at the top and did
not alter when changing the values of α and λ.
α = 0.8

0.0292
0.0069
0.0044
0.0434
0.0023
0.0044
0.0069

3. Case study
α = 0.6

0.0297
0.0073
0.0046
0.0444
0.0023
0.0046
0.0073

The validation of the proposed decision-making model was con-

ducted with ore deposit data of the Tummalapalle mine of the Uranium
Corporation of India Limited (UCIL). The latitudes of the deposit ranges
from 14°18‘36.6″N to 14°20′20″N and the longitude from
α = 0.4

0.0302
0.0075
0.0047
0.0451
0.0023
0.0047
0.0075

78°15′16.57″E to 78°18″3.33″E. The deposit is located in the Cuddapah

district of Andhra Pradesh, India as shown in Fig. 3. The direction of the
strike of the deposit is WNW-ESE, and the dip varies from 15° to 17°.
α = 0.2
λ = 0 (pessimistic)

0.0305
0.0077
0.0048
0.0454
0.0024
0.0048
0.0077

The ore body is fairly continuous over the entire strike length of 6.6 km
and uniformly extending to a depth of 275 m. The width of the hang-
wall and footwall are, respectively, 3.2 m and 2.5 m. The ore deposit is
0.0305
0.0080
0.0049
0.0454
0.0025
0.0049
0.0080

tabular in shape with little variation in its grade and thickness along the
α=0

strike and dip direction. The host rock is quite competent. Fig. 3 shows
the location of Tummalapalle Uranium Project of UCIL.
Cut and Fill stoping

In the proposed study, the FAHP model was developed by con-

Shrinkage Stoping

Squareset Stoping
Sublevel Stoping

Room and Pillar

Sublevel Caving

sidering 16-criteria. However, model validation and sensitivity analysis

Block Caving

were conducted based on 8-criteria due to the unavailability of the

remaining data, these being: dip, shape, thickness, grade distribution,
Table 4

depth of the deposit, the RMR of the ore zone, RMR of hanging wall,
and RMR of foot wall. The characteristics of the ore deposit are shown

13
 Table 5
Global weights of seven mining methods for different decision-making attitudes (λ) and fuzzification factors (α) for moderate-dip.
Dip-Moderate

λ = 0 (pessimistic) λ = 0.5 (unbiased) λ = 1 (optimistic)

B.C. Balusa, A.K. Gorai

α=0 α = 0.2 α = 0.4 α = 0.6 α = 0.8 α=1 α=0 α = 0.2 α = 0.4 α = 0.6 α = 0.8 α=1 α=0 α = 0.2 α = 0.4 α = 0.6 α = 0.8 α=1

Block Caving 0.0023 0.0022 0.0020 0.0019 0.0017 0.0016 0.0023 0.0023 0.0023 0.0023 0.0023 0.0023 0.0023 0.0024 0.0025 0.0023 0.0025 0.0029
Sublevel Stoping 0.0017 0.0016 0.0015 0.0014 0.0014 0.0013 0.0017 0.0016 0.0017 0.0017 0.0017 0.0017 0.0017 0.0017 0.0018 0.0017 0.0019 0.0020
Sublevel Caving 0.0017 0.0016 0.0015 0.0014 0.0014 0.0013 0.0017 0.0016 0.0017 0.0017 0.0017 0.0017 0.0017 0.0017 0.0018 0.0017 0.0019 0.0020
Room and Pillar 0.0008 0.0008 0.0008 0.0007 0.0007 0.0007 0.0008 0.0008 0.0008 0.0008 0.0008 0.0009 0.0008 0.0008 0.0009 0.0008 0.0010 0.0010
Shrinkage Stoping 0.0008 0.0008 0.0008 0.0007 0.0007 0.0007 0.0008 0.0008 0.0008 0.0008 0.0008 0.0009 0.0008 0.0008 0.0009 0.0008 0.0010 0.0010
Cut and Fill stoping 0.0092 0.0090 0.0088 0.0085 0.0083 0.0080 0.0092 0.0092 0.0091 0.0091 0.0091 0.0089 0.0091 0.0093 0.0095 0.0088 0.0097 0.0096
Squareset Stoping 0.0092 0.0090 0.0088 0.0085 0.0083 0.0080 0.0092 0.0092 0.0091 0.0091 0.0091 0.0089 0.0091 0.0093 0.0095 0.0088 0.0097 0.0096

14
Table 6
Global weights of seven mining methods for different decision-making attitudes (λ) and fuzzification factors (α) for steep-dip.
Dip-Steep

λ = 0 (pessimistic) λ = 0.5 (unbiased) λ = 1 (optimistic)

α=0 α = 0.2 α = 0.4 α = 0.6 α = 0.8 α=1 α=0 α = 0.2 α = 0.4 α = 0.6 α = 0.8 α=1 α=0 α = 0.2 α = 0.4 α = 0.6 α = 0.8 α=1

Block Caving 0.0020 0.0020 0.0019 0.0018 0.0018 0.0019 0.0020 0.0020 0.0020 0.0020 0.0020 0.0023 0.0020 0.0020 0.0020 0.0018 0.0020 0.0022
Sublevel Stoping 0.0020 0.0020 0.0019 0.0018 0.0018 0.0019 0.0020 0.0020 0.0020 0.0020 0.0020 0.0017 0.0020 0.0020 0.0020 0.0018 0.0020 0.0022
Sublevel Caving 0.0020 0.0020 0.0019 0.0018 0.0018 0.0019 0.0020 0.0020 0.0020 0.0020 0.0020 0.0017 0.0020 0.0020 0.0020 0.0018 0.0020 0.0022
Room and Pillar 0.0002 0.0002 0.0002 0.0002 0.0002 0.0002 0.0002 0.0002 0.0002 0.0002 0.0002 0.0009 0.0002 0.0002 0.0002 0.0002 0.0002 0.0002
Shrinkage Stoping 0.0020 0.0020 0.0019 0.0018 0.0018 0.0019 0.0020 0.0020 0.0020 0.0020 0.0020 0.0009 0.0020 0.0020 0.0020 0.0018 0.0020 0.0022
Cut and Fill stoping 0.0020 0.0020 0.0019 0.0018 0.0018 0.0019 0.0020 0.0020 0.0020 0.0020 0.0020 0.0089 0.0020 0.0020 0.0020 0.0018 0.0020 0.0022
Squareset Stoping 0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0089 0.0003 0.0003 0.0003 0.0003 0.0004 0.0004
Journal of Sustainable Mining 18 (2019) 8–17
B.C. Balusa, A.K. Gorai Journal of Sustainable Mining 18 (2019) 8–17

Fig. 3. Location of the tummalapalle uranium ore deposit.

Table 7 3.1. Sensitivity analysis of the characteristics (criteria parameters) of the

Characteristics of the uranium ore deposit of the Tummalapalle mine. specified mine
Criteria Parameters Field data Characteristics
Sensitivity analysis for the ranking of the mining methods for the
Dip 15 0
Flat specified ore deposit was carried out for six fuzzification factors (α = 0,
Shape tabular Tabular 0.2, 0.4, 0.6, 0.8 and 1) in three decision-making attitudes, these being
Thickness 1.5 m Narrow
pessimistic (λ = 0), unbiased (λ = 0.5), and optimistic (λ = 1). The
Grade distribution Uniform Uniform
Depth 275 m Moderate global weights of each combination of λ and α were determined using
RMR of ore zone 48 Moderate eqns. (1)–(6). At each λ, six fuzzification factors (α) were considered for
RMR of hanging wall 34 Weak analyzing the sensitivity in the ranking of the mining methods. The
RMR of foot wall 48 Moderate
trends of the global weights in each case are shown in Table 9. The
trend of global weights for different fuzzification factors clearly in-
dicates that the room and pillar mining method is most suitable for the
in Table 7.
specified uranium ore deposit, irrespective of the fuzzification factors
The global weights of these characteristics were taken from the
and decision-making attitudes. The rank of room and pillar mining
model developed for the selection of the best mining method for the
method was always ranked top and this did not alter when changing the
specified ore deposit. Global weights corresponding to the flat dip,
values of α and λ. It was also observed that the global weights of each
tabular shape, narrow thickness, moderate depth, uniform grade dis-
mining method were altered with changes in the value of α and λ, but
tribution, the moderate RMR of the ore, the weak RMR of the hanging
the rank of the mining methods rarely changed. The second best mining
wall, and the moderate RMR of the foot wall, with respect to each
method was the block caving method.
mining method, were determined for α value 1 and λ value 0.5 using
Table 9 Global weights of different mining methods for the uranium
eqns. (1)–(6). These global weights were determined using a method
ore deposit (Flat-Dip, Tabular-Shape, Narrow-Thickness, Moderate-
similar to the one explained in Section 2. These global weights are
Depth, Uniform-Grade distribution, the Moderate-RMR of the ore, the
shown in Table 8.
Weak-RMR of the hanging wall, and the Moderate-RMR of the foot-
The total score (shown in Table 8) for a specific mining method was
wall).
determined by summing up the respective weights of all the criteria. All
the scores were ranked from the highest to the lowest value. The
highest score was obtained for the Room and Pillar mining (0.0582) and 4. Conclusions
thus it was assigned first rank. Hence, the best mining method obtained
from the model is the Room and Pillar mining for the excavation of the The study aims to analyse the sensitivity in decision-making for the
ore deposit, and the UCIL adopted the same mining method for ex- selection of a mining method using the FAHP model. The results in-
cavation of the ore deposit. dicate that the proposed FAHP decision-making model could be

Table 8
Relative rankings of seven mining methods for the uranium ore deposit of the Tummalapalle mine.
DI SH TH DE GD ORMR HRMR FRMR Score Rank

BC 0.0299 0.0002 0.0001 0.0007 0.0031 0.0005 0.0006 0.0026 0.0377 2

SS 0.0076 0.0018 0.0002 0.0016 0.0069 0.0054 0.0001 0.0005 0.0241 3
SC 0.0051 0.0018 0.0001 0.0001 0.0031 0.0027 0.0011 0.0026 0.0166 4
RP 0.0437 0.0018 0.0021 0.0007 0.0069 0.0027 0.0001 0.0002 0.0582 1
SH 0.0025 0.0018 0.0021 0.0007 0.0031 0.0027 0.0001 0.0005 0.0135 5
CF 0.0051 0.0018 0.0011 0.0007 0.0007 0.0005 0.0006 0.0005 0.011 7
SQ 0.0076 0.0002 0.0011 0.0001 0.0005 0.0004 0.0011 0.0002 0.0112 6

15
B.C. Balusa, A.K. Gorai Journal of Sustainable Mining 18 (2019) 8–17

Global weights of different mining methods for the uranium ore deposit (Flat-Dip, Tabular-Shape, Narrow-Thickness, Moderate-Depth, Uniform-Grade distribution, Moderate-RMR of ore, Weak-RMR of hanging wall, and
robustly used for the selection of a mining method, as the factor's un-

0.0391
0.0259
0.0175
0.0594
0.0139
0.0120
0.0121
α=1
certainty levels do not influence the final decision. It was observed that
the rank of the highest priority alternative never alters with either
α = 0.8 changes in the fuzzification factor (α) or the decision-making attitude

0.0397
0.0256
0.0171
0.0609
0.0136
0.0114
0.0118
(λ). It can be inferred from the results that the ranking of the most
suitable alternative remains the same irrespective of the fuzzification
factors and decision-making attitudes.
α = 0.6

0.0340
0.0205
0.0138
0.0506
0.0108
0.0120
0.0109
Conflict of interest

Authors state that there is not any conflict of interest.

α = 0.4

0.0393
0.0253
0.0167
0.0611
0.0135
0.0111
0.0118

Ethical statement
α = 0.2

0.0390
0.0250
0.0165
0.0609
0.0135
0.0111
0.0118

Authors state that the research was conducted according to ethical

λ = 1 (optimistic)

standards.

Funding body
0.0385
0.0247
0.0165
0.0602
0.0137
0.0111
0.0117
α=0

There is no special funding for this work.

0.0381
0.0244
0.0170
0.0585
0.0139
0.0114
0.0115
α=1

Acknowledgement

The work has been carried out at the National Institute of

α = 0.8

0.0384
0.0246
0.0168
0.0593
0.0137
0.0112
0.0116

Technology (NIT) Rourkela, Odisha, India. Authors are thankful to

Director, NIT Rourkela for providing the computing facility for ex-
ecuting the work. The authors want to deliver thanks to the authorities
α = 0.6

of Tummalapalle mine of Uranium Corporation of India Limited (UCIL)

0.0384
0.0247
0.0166
0.0599
0.0137
0.0111
0.0117

for providing the data of ore deposit characteristics.

Appendix A and B. Supplementary data

α = 0.4

0.0386
0.0247
0.0166
0.0601
0.0137
0.0110
0.0116

Supplementary data to this article can be found online at https://

doi.org/10.1016/j.jsm.2018.10.003.
α = 0.2
λ = 0.5 (unbiased)

0.0384
0.0247
0.0165
0.0602
0.0137
0.0111
0.0118

References
0.0386
0.0247
0.0165
0.0604
0.0136
0.0110
0.0117

Alonso, J. A., & Lamata, M. T. (2006). Consistency in the analytic hierarchy process: A
α=0

new approach. International Journal of Uncertainty, Fuzziness and Knowledge-based

Systems, 14(4), 445–459. https://doi.org/10.1142/S0218488506004114.
Alpay, S., & Yavuz, M. (2007). A decision support system for underground mining. In H.
0.0344
0.0213
0.0159
0.0539
0.0136
0.0102
0.0102

G. Okuno, & M. Ali (Vol. Eds.), New trends in applied artificial intelligence decision
α=1

support systems, proceedings of the 20th international conference on industrial, engineering

and other applications of applied intelligent systems, IEA/AIE 2007, Kyoto, Japan, June
26–29: Vol. 2007, (pp. 334–343). . https://doi.org/10.1007/978-3-540-73325-6_33.
Ataei, M., Shahsavany, H., & Mikaeil, R. (2013). Monte Carlo Analytic Hierarchy Process
α = 0.8

0.0371
0.0229
0.0162
0.0583
0.0137
0.0105
0.0107

(MAHP) approach to selection of optimum mining method. International Journal of

Mining Science, 23(4), 573–578. https://doi.org/10.1016/j.ijmst.2013.07.017.
Azadeh, A., Osanloo, M., & Ataei, M. (2010). A new approach to mining method selection
based on modifying the Nicholas technique. Applied Soft Computing, 10(4),
α = 0.6

0.0376
0.0235
0.0162
0.0592
0.0136
0.0106
0.0111

1040–1061. https://doi.org/10.1016/j.asoc.2009.09.002.
Chang, C. W., Wu, C. R., Lin, C. T., & Chen, H. C. (2008). Evaluating and controlling
silicon wafer slicing quality using fuzzy analytical hierarchy and sensitivity analysis.
International Journal of Advanced Manufacturing Technology, 36(3–4), 322–333.
α = 0.4

https://doi.org/10.1007/s00170-006-0831-9.
0.0381
0.0239
0.0162
0.0599
0.0136
0.0107
0.0113

Dehghani, H., Siami, A., & Haghi, P. (2017). A new model for mining method selection
based on grey and TODIM methods. Journal of Mining and Environment, 8(1), 49–60.
https://doi.org/10.22044/jme.2016.626.
Gorai, A. K., Kanchan, Upadhyay, A., Tuluri, F., Goyal, P., & Tchounwou, P. B. (2015). An
α = 0.2
λ = 0 (pessimistic)

0.0384
0.0243
0.0164
0.0602
0.0136
0.0109
0.0115

innovative approach for determination of air quality health index. The Science of the
Total Environment, 533, 495–505. https://doi.org/10.1016/j.scitotenv.2015.06.133.
Gupta, S., & Kumar, U. (2012). An analytical hierarchy process (AHP)-guided decision
model for underground mining method selection. International Journal of Mining,
0.0385
0.0247
0.0165
0.0602
0.0137
0.0111
0.0117

Reclamation and Environment, 26(4), 324–336. https://doi.org/10.1080/17480930.

α=0
Moderate-RMR of footwall).

2011.622480.
Hsu, P. F., & Chen, B. Y. (2007). Developing and implementing a selection model for
bedding chain retail store franchisee using Delphi and Fuzzy AHP. Quality and
Cut and Fill stoping

Quantity, 41(2), 275–290. https://doi.org/10.1007/s11135-006-9004-z.

Shrinkage Stoping

Squareset Stoping

Laarhoven, V., & Pedrycz, W. (1983). A fuzzy extention of saaty's priority theory. Fuzzy
Sublevel Stoping

Room and Pillar

Sublevel Caving

Sets and Systems, 11(1–3), 229–241. https://doi.org/10.1016/S0165-0114(83)

Block Caving

80082-7.
Lee, A. R. (1995). Application of modified fuzzy AHP method to analyze bolting sequence of
Table 9

structural jointsDoctoral Dissertation. PA, USA: Lehigh University Bethlehem.

Mikaeil, R., Naghadehi, M. Z., Ataei, M., & Khalokakaie, R. (2009). A decision support

16
B.C. Balusa, A.K. Gorai Journal of Sustainable Mining 18 (2019) 8–17

system using fuzzy analytical hierarchy process (FAHP) and TOPSIS approaches for method for location selection: A hybrid analysis. Applied Mathematics and
selection of the optimum underground mining method. Archives of Mining Sciences, Computation, 206(2), 598–606. https://doi.org/10.1016/j.amc.2008.05.111.
54(2), 349–368. Tatiya, R. R. (2013). Surface and underground excavations: Methods, techniques and equip-
Miller-Tait, L., Pakalnis, R., & Poulin, R. (1995). UBC mining method selection. Mine ment. CRC Press.
planning and equipment selection (pp. 163–168). Rotterdam: Balkema. Tsai, H. Y., Chang, C. W., & Lin, H. L. (2010). Fuzzy hierarchy sensitive with Delphi
Naghadehi, M. Z., Mikaeil, R., & Ataei, M. (2009). The application of fuzzy analytic method to evaluate hospital organization performance. Expert Systems with
hierarchy process (FAHP) approach to selection of optimum underground mining Applications, 37(8), 5533–5541. https://doi.org/10.1016/j.eswa.2010.02.099.
method for Jajarm Bauxite Mine, Iran. Expert Systems with Applications, 36(4), Tseng, M. L., & Lin, Y. H. (2008). Selection of competitive advantages in TQM im-
8218–8226. https://doi.org/10.1016/j.eswa.2008.10.006. plementation using fuzzy AHP and sensitivity analysis. Asia Pacific Management
Namin, E. S., Shahriar, K., Ataee-Pour, M., & Dehghani, H. (2008). A new model for Review, 13(3), 583–599.
mining method selection of mineral deposit based on fuzzy decision making. Journal Yavuz, M. (2015). The application of the analytic hierarchy process (AHP) and Yager's
of the South African Institute of Mining and Metallurgy, 108(7), 385–395. method in underground mining method selection problem. International Journal of
Saaty, T. L. (1980). The analytic hierarchy process: Planning, priority setting, resource allo- Mining, Reclamation and Environment, 29(6), 453–475. https://doi.org/10.1080/
cation. New York: McGraw-Hill. 17480930.2014.895218.
Tabari, M., Kaboli, A., Aryanezhad, M. B., Shahanaghi, K., & Siadat, A. (2008). A new

17