Tunnelling and Underground Space Technology 24 (2009) 173–184

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology

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

Underground mining method selection by decision making tools

Serafettin Alpay a, Mahmut Yavuz b,*
a
Department of Industrial Engineering, Eskisehir Osmangazi University, 26030 Bademlik, Eskisehir, Turkey
b
Department of Mining Engineering, Eskisehir Osmangazi University, 26480 Meselik, Eskisehir, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: Underground mining method selection is one of the most important decisions that mining engineers have
Received 28 January 2008 to make. Choosing a suitable underground mining method to extract a mineral deposit is very important
Received in revised form 3 July 2008 in terms of economics, safety and productivity of mining operations. In real life, underground mining
Accepted 29 July 2008
method selection is one of the multiple attribute decision making (MADM) problems and decision makers
Available online 17 September 2008
have always some difficulties in making the right decision in the multiple criteria environment. Analytic
hierarchy process (AHP) and Yager’s methods are the MADM tools and can be used for selection of the
Keywords:
best underground mining method by considering the problem criteria. In this study, a computer program
Analytic hierarchy process
Multiple attribute decision making
(UMMS) based on the AHP and the Yager’s method was developed to analyze the underground mining
Yager’s method method selection problems and produce the best underground mining method swiftly for different
Underground mining method selection deposit shapes and ore bodies.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction important to re-analyze the decision made before carrying it out.

The method that the decision makers generally use for this aim
Mineral exploitation in which all extractions are carried out be- is the sensitivity analysis on the final decision.
neath the earth’s surface is termed as underground mining. Under-
ground methods are employed when the depth of the deposit, the
2. Literature review
stripping ratio of overburden to ore (or coal or stone), or both be-
come excessive for surface exploitation (Hartman and Mutmansky,
2.1. Studies on mining method selection
2002).
Underground mining method should be primarily selected to
The problem of underground mining method selection has been
make use of underground resources optimally. Besides, the ground
studied in the literature. Boshkov and Wright (1973) proposed a
control on the mining areas, planning the ventilation system,
classification system which was one of the first qualitative classifi-
decreasing the maintenance costs of gallery, developing new min-
cation schemes. Morrison (1976) suggested a selection chart for
ing panels and preparing the underground production schedule are
mining method selection. Laubscher (1981) proposed a selection
also directly related to underground mining method selection,
methodology of an appropriate mass underground mining method
such like geology of deposit. So, underground mining method
based on rock mass classification system. Nicholas (1981) pre-
selection process is extremely important in mine designs.
sented a classification system for selection of the optimum mining
To make the right decision on underground mining method
method via numerical ranking with quantitative analysis. Hartman
selection, all known criteria related to the problem should be taken
(1987) developed a selection chart based on the geometry of the
into consideration. Increasing the number of criteria in decision
deposit and the ground conditions of the ore zone for select-
making process makes the problem more complex, but the right-
ing mining method. Miller-Tait et al. (1995) modified the
ness of the decision also increases. Because of arising complexity
Nicholas’ system and developed the UBC mining method selection
in the decision process, many conventional methods consider only
process.
limited number of criteria. So, there is a need for alternative meth-
ods, which can consider all known criteria related to underground
2.2. Studies on decision making techniques in mining
mining method selection in the decision making process.
Once selected a mining method, as it is nearly impossible to
A review of the literature reveals that decision making tech-
change it owing to the rising costs and mining losses, it is very
niques have been used for a variety of specific mining applications.
Bascetin and Kesimal (1999) used Yager’s method for selection of
* Corresponding author. Tel.: +90 222 239 37 50 3421; fax: +90 222 239 36 13. an optimum coal transportation system from pit to the power
E-mail address: myavuz@ogu.edu.tr (M. Yavuz). plant. Karadogan et al. (2001) solved an underground mining

0886-7798/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tust.2008.07.003
174 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184
method selection problem by using the Yager’s method and they
used Satty’s analytic hierarchy process (AHP) method for pair-wise
comparison of the criteria. Elevli et al. (2002) selected a new verti-
cal shaft or ramp system by comparing the weighted alternative
criteria for a small-scale underground mine on the basis of total
investment cost, ore transport unit cost and net present value of
overall project for various depths. Kesimal and Bascetin (2002)
used the Yager’s method for solving equipment selection problem
in open pit mine. Samanta et al. (2002) used the AHP method for
selection of open cast mining equipment. Bitarafan and Ataei
(2004) solved the similar problem by using the Yager’s method
with Satty’s AHP method and they also used fuzzy dominance
method in their analysis. Elevli and Demirci (2004) selected most
suitable underground ore transport system for a chromate mine
by using the one of the multiple attribute decision making (MADM)
method namely Preference Ranking Organization MeTHod for
Enrichment Evaluation (PROMETHEE). Kazakidis et al. (2004) used
the AHP method and analyzed five different mining scenarios such
as drilling technology investment analysis, ground support design,
tunneling systems design, shaft location selection and mine plan-
ning risk assessment. Ataei (2005) used the AHP method for the
problem of selection of a new alumina cement plant location in
East-Azerbaijan province of Iran. Acaroglu et al. (2006a, 2006b)
used the Yager’s and the AHP method for selection of roadheaders
in tunneling applications. Bascetin et al. (2006) developed a com-
puter program using the Yager’s method for equipment and mining
method selection in mining. Yavuz et al. (2008) used the AHP
method for selection of optimum support type in the main haulage
road in Lignite colliery.
In this paper, a computer program (UMMS) based on the AHP
and the Yager’s method for underground mining method selection
was developed. The UMMS provides two different AHP models so
that the decision makers can analyze the selection problem based
on either only main criteria or main criteria with their sub-criteria.
The UMMS also provides a Yager’s method which is one of the fuz-
zy multiple attribute decision making (FMADM) methods as an
alternative way to analyze the problem. Another important func-
tion provided by the UMMS is the sensitivity analysis. The decision
makers can carry out the sensitivity analysis whenever he/she
needs to observe the sensitivity of final solution against the varia-
tions in the main criteria considered.
3. Theory review
3.1. AHP model
The AHP method developed by Saaty (1980) gives an opportu-
nity to represent the interaction of multiple factors in complex
unstructured situations. The method is based on the pair-wise
comparison of components with respect to attributes and alterna-
tives. A pair-wise comparison matrix n  n is constructed, where n
is the number of elements to be compared. The method is applied
for the hierarchy problem structuring (Saaty, 2000).
After the hierarchy structuring the pair-wise comparison matrix
is constructed for each level, where a nominal discrete scale from 1
to 9 (Table 1) is used for the evaluation (Saaty, 1980).
The next step is to find the relative priorities of criteria or alter-
natives implied by this comparison. The relative priorities are
worked out using the theory of eigenvector. For example, if the pair
comparison matrix is A, then,
ðA  kmax  IÞ  w ¼ 0 ð1Þ
To calculate the eigenvalue ‘‘kmax” and eigenvector
w = (w1, w2, . . . , wn), weights can be estimated as relative priorities
of criteria or alternatives (Saaty, 2000).
Table 1
Scale for pair-wise comparisons (Saaty, 1980)
Relative Definition Explanation
intensity
1 Of equal value Two requirements are of equal value
3 Slightly more value Experience slightly favors one
requirement over another
5 Essential or strong value Experience strongly favors one
requirement over another
7 Very strong value A requirement is strongly favored and its
dominance is demonstrated in practice
9 Extreme value The evidence favoring one over another is
of the highest possible order of
affirmation
2, 4, 6, 8 Intermediate values When compromise is needed
between two adjacent
judgments
Since the comparison is based on the subjective evaluation, a
consistency ratio is required to ensure the selection accuracy.
The consistency index (CI) of the comparison matrix is computed
as follows:
CI ¼ ðkmax  nÞ=ðn  1Þ ð2Þ
where kmax is maximal or principal eigenvalue and n is the matrix
size. The consistency ratio (CR) is calculated as
CR ¼ CI=RI ð3Þ
where ‘‘RI” denotes random consistency index. Random consistency
indices are given in Table 2 (Saaty, 2000).
As a general rule, a consistency ratio of 0.10 or less is considered
acceptable. In practice, however, consistency ratios exceeding 0.10
occur frequently.
3.2. FMADM model
FMADM methods have been developed due to the lack of preci-
sion in assessing the relative importance of attributes and the per-
formance ratings of alternatives with respect to an attribute. The
problem of FMADM is to select/prioritize/rank a finite number of
alternatives by evaluating a group of predetermined criteria. Thus,
to solve this problem, an evaluation method to rate and rank, in or-
der of preference, the set of alternatives must be constructed (Chen
and Klein, 1997).
Although a large number of FMADM methods have been ad-
dressed in the literature, the focus of this paper is on Yager’s
(1978) method. This method is general enough to deal with both
multiple objectives and multiple attribute problems and follows
the max–min method of Bellman and Zadeh (1970), with the
improvement of Saaty’s method, which considers the use of a reci-
procal matrix to express the pair-wise comparison of the criteria
and the resulting eigenvector as subjective weights. The weighting
procedure uses exponentials based on the definition of linguistic
hedges, proposed by Zadeh (1973).
On describing multiple attribute decision making problems,
only a single objective is considered, namely the selection of the
best alternative from a set of alternatives. The decision method as-
sumes the max–min principle approach. Formally, let
A = {A1, A2, . . . , Am} be the set of alternatives, C = {C1, C2, . . . , Cn} be
the set of criteria, which can be given as fuzzy sets in the space
of alternatives. Hence, the fuzzy set decision is the intersection of
all criteria: lD ðAÞ ¼ min½lc1 ðAÞ; lC 2 ðAÞ; . . . ; lC n ðAÞ. For all (Ai) e A,
and the optimal decision is yielded by, lD ðA Þ ¼ maxA lD ðAÞ, where
A* is the optimal decision. Main difference in this approach is that
the importance of criteria is represented as exponential scalars.
The rationale behind using weights (or importance levels) as expo-
nents is that the higher the importance of criteria, the larger should
 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184
Table 2
Consistency indices of randomly generated reciprocal matrices (Saaty, 2000)
Order of matrix 1, 2 3 4 5 6
RI value 0.00 0.58 0.90 1.12 1.24
be the exponent, giving the minimum rule (Riberio, 1996).
Formally:
n o
lD ðAÞ ¼ min ½lC1 ðAÞa1 ; ½lC2 ðAÞa2 ; . . . ; ½lCn ðAÞan
Yager suggests the use of Saaty’s method for pair-wise comparison
of the attributes and the use of the resulting eigenvector expresses a
decision maker’s empirical estimate of the level of importance of
alternatives for a given criterion.
4. Methodology
The approach proposed in this study comprises the following
steps:
(a) Determination of the method: The method may be AHP or
FMADM.
(b) Determination of the AHP model type (AHP method only): The
decision maker should select one of two AHP models. The
hierarchy structures of the AHP models are shown in Figs.
1 and 2., respectively.
Total of 36 criteria which have been classified in six main
groups by Hartman and Mutmansky (2002) are proposed by the
first model to analyze an UMMS problem in details. These criteria
are given in Table 3. On the other hand, when the first model is se-
lected, the number of pair-wise comparison matrices increases
Fig. 1. First AHP model including all criteria and their sub-criteria.
Fig. 2. Second AHP model including only main criteria.
175
7 8 9 10 11 12 13 14 15
1.32 1.41 1.45 1.49 1.51 1.48 1.56 1.57 1.59
quite and this also makes the time needed to reach a solution much
longer. If the time is restricted and/or the six main group criteria
without their sub-criteria are enough to make a decision, the sec-
ð4Þ ond AHP model can be used. The first level of the both models is
the general object. In the first model, six main group criteria are
placed in the second level and their sub-criteria are placed in the
third level. In the second model, only main group criteria are con-
sidered in the decision making process and placed in the second le-
vel of the model. The last level of either of the models is the
alternatives.
(c) Determination of the set of alternatives: The alternatives of the
AHP models and FMADM are determined based on the UBC
approach by Miller-Tait et al. (1995). The UBC approach is
simply a modified version of the Nicholas approach and
numerically ranks deposit characteristics of ore geometry
and rock mechanics characteristics of the ore zone, footwall
and hanging wall. The rankings are then summed together
with the higher rankings being the more favorable or likely
methods. Each ranking consists of a number ‘‘0 to 6” or
‘‘10, 49”. ‘‘49” completely eliminates a mining method
from being feasible. A value of ‘‘0” strongly suggests that this
characteristic makes that particular mining technique less
attractive. The value ‘‘6” indicates a very favorable charac-
teristic for that particular mining method. To determine
the set of valid alternatives, the UMMS uses UBC final
176 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184

Table 3 (e) Constructing the membership levels of each criterion (FMADM

Main criteria and their sub-criteria (Hartman and Mutmansky, 2002) method only): The decision maker should define the mem-
C1.Spatial characteristics of the C4.Economic considerations bership levels of each criterion after conferring with the
deposit experts on the subject.
S11.Size S41.Reserves (f) Producing the best alternative: The last step of underground
S12.Attitude S42.Production rate
S13.Depth S43.Mine life
mining method selection is to produce the best alternative
S14.Regularity of the ore S44.Productivity along with the final rank of the valid alternatives by per-
boundaries forming the selected method.
S15.Existence of previous mining S45.Comparative mining costs (g) Performing sensitivity analysis: The decision makers can per-
S46.Comparative capital costs
form the sensitivity analysis on the final rank of the alterna-
C2.Geologic and hydrologic C5.Technological factors tives considered in the decision process.
conditions
S21.Mineralogy and petrography S51.Recovery
S22.Chemical composition S52.Dilution
S23.Deposit structure S53.Flexibility of the method to changing 5. Computer implementation
conditions
S24.Planes of weakness S54.Selectivity of the method
S25.Uniformity of grade S55.Concentration or dispersion of workings
The UMMS is the main program, which calls several subroutines
S26.Alteration and weathered S56.Ability to mechanize and automate as illustrated in Fig. 3. The UMMS also provides a database (UMMS-
zones DAT) in which two tables are placed. The tables store the descrip-
S27.Existence of strata gases S57.Capital and labor intensities tive information about the alternatives and criteria. That
C3.Geotechnical properties C6.Environmental concerns information can be examined by the decision maker in any time
S31.Elastic properties S61.Ground control to maintain integrity of of the decision process.
openings
The UMMS controls the input of user data, the input of param-
S32.Plastic or viscoelastic behavior S62.Subsidence or caving effects at the surface
S33.State of stress S63.Atmospheric control eters, the output of results, and the inference via subroutines.
S34.Rock mass rating S64.Availability of suitable waste disposal
areas 5.1. Input parameters
S35.Other physical properties S65.Workforce
affecting competence
S66.Comparative safety conditions of the
suitable mining methods (a) The method parameter: This parameter may be 1 (AHP) or 2
(FMADM).
(b) The AHP model parameter: This parameter may be 1 (first
ranking of the appropriate alternatives generated by follow- model) or 2 (second model). The model parameter is impor-
ing the UBC rating system given in Tables 4–6 (Miller-Tait tant because of determining the number of pair-wise com-
et al., 1995). parison matrices as well.
(d) Constructing pair-wise comparison matrices: For the AHP (c) The set of alternatives: This set simply contains the names of
method, the pair-wise comparison of the main criteria is the valid alternatives generated by the decision maker.
performed first and then the pair-wise comparison of the Using the user interface related to the UBC ranking system
alternatives based on each criterion follows. The number of provided by the UMMS, the decision maker generates the
comparison matrices varies according to the selected AHP set of alternatives to be valid for the mining method selec-
model. On the other hand, one comparison matrix of the cri- tion problem considered.
teria (i.e. main criteria) is enough for the FMADM method. (d) The name of the data file: Data file is used for saving and
However, there are some general limitations on human per- reloading the pair-wise comparison values entered by the
formance and the total number of criteria to be handled decision maker in any comparison level of the AHP method
simultaneously should be seven plus two (=9) maximum and in the first level of the FMADM method. For the AHP
(Yavuz and Alpay, 2007a, 2007b; Yavuz 2007). Thus, only method, the number of data files may vary according to
main criteria are used for pair-wise comparison for the the selected AHP model. A typical data file contains the com-
FMADM method. parison values calibrated on the numerical scale given in

Table 4
Ranking of geometry/grade distribution for different mining methods (Miller-Tait et al.,1995)

Mining method General shape Ore thickness Ore plunge Grade distribution Depth
M T/P I VN N I T VT F I S U G E SH I D
Open pit mining 4 2 3 1 2 3 4 4 3 3 1 3 3 2 4 0 49
Block caving 4 2 0 49 49 0 3 4 3 2 4 3 2 2 2 3 3
Sublevel stoping 3 4 1 10 1 3 4 3 2 1 4 4 4 3 3 4 2
Sublevel caving 3 4 1 49 49 0 4 4 1 1 4 3 2 2 3 2 2
Longwall mining 49 4 49 4 3 0 49 49 4 0 49 4 1 0 2 2 3
Room and pillar 0 4 2 4 3 1 49 49 4 0 49 4 2 0 3 3 2
Shrinkage stoping 0 4 2 4 4 0 49 49 49 0 4 3 2 2 3 3 2
Cut and fill stoping 1 4 4 3 4 4 1 0 1 3 4 2 3 4 2 3 4
Top slicing 1 2 0 1 1 0 2 1 4 2 0 2 1 1 2 1 1
Square set stoping 0 1 4 4 3 2 0 0 2 3 2 0 1 3 1 1 2
M = Massive VN = Very narrow F = Flat U = Uniform S = Shallow
T/P = Tabular or platy N = Narrow I = Intermediate G = Gradational I = Intermediate
I = Irregular I = Intermediate S = Steep E = Erratic D = Deep
T = Thick
VT = Very thick
 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184 177

Table 5
Rock mechanics characteristics-rock mass rating (Miller-Tait et al.,1995)

Mining method Ore zone Hanging wall Footwall

VW W M S VS VW W M S VS VW W M S VS
Open pit mining 3 3 3 3 3 2 3 4 4 4 2 3 4 4 4
Block caving 4 3 2 0 49 3 3 3 2 2 3 3 3 2 2
Sublevel stoping 1 3 4 4 4 49 0 3 4 4 0 0 2 3 3
Sublevel caving 3 4 3 1 0 4 4 3 2 2 1 2 3 3 3
Longwall mining 6 6 4 2 2 6 5 4 3 3 – – – – –
Room and pillar 49 0 3 5 6 49 0 3 5 6 – – – – –
Shrinkage stoping 0 1 3 3 3 0 0 2 4 4 0 0 2 3 3
Cut and fill stoping 0 1 2 3 3 3 5 4 3 3 3 3 2 2 2
Top slicing 3 2 1 1 0 0 0 2 3 3 0 0 1 2 2
Square set stoping 4 4 1 0 0 4 4 1 0 0 3 1 0 0 0

RMR ratings: very weak (VW) = 0–20, weak (W) = 21–40, moderate (M) = 41–60, strong (S) = 61–80, very strong (VS) = 81–100.

Table 6 5.2. Output of results

Rock mechanics characteristics-rock substance strength ratings (Miller-Tait et al.,
1995) The overall result/final matrix: After the decision making pro-
Mining method Ore zone Hanging wall Footwall cess through the AHP or FMADM is completed, all calculated over-
all ratings of the valid alternatives are presented in a matrix form
VW W M S VW W M S VW W M S
and the alternative with the highest rating is proposed as the most
Open pit mining 4 3 3 3 3 3 4 4 3 3 4 4
preferred mining method.
Block caving 4 2 1 0 4 3 2 0 4 3 2 1
Sublevel stoping 0 2 4 4 0 1 4 5 0 1 3 3
Sublevel caving 2 3 3 2 4 3 2 1 1 2 2 2 5.3. Subroutines
Longwall mining 6 5 2 1 6 5 2 2 – – – –
Room and pillar 0 0 3 6 0 0 2 6 – – – – 5.3.1. Metodsel and modsel
Shrinkage stoping 0 1 3 4 0 1 3 4 0 2 3 3
These routines use the same user interface. The decision maker
Cut and fill stoping 0 1 3 3 3 5 4 2 1 3 2 2
Top slicing 3 2 1 0 3 2 2 2 2 2 1 1 can select the method by this user interface and input the method
Square set stoping 4 3 1 0 4 2 1 0 3 2 0 0 parameter. On the other hand, if the decision maker selects AHP as
the method, he/she also selects the model parameter. After selec-
RSS ratings: very weak (VW) = <5, weak (W) = 5–10, moderate (M) = 10–15, strong
(S) = >15. tion process is completed, method and/or model parameter va-
lue(s) is (are) returned to the UMMS by the routines.

Table 1. The number of comparisons is n  (n  1)  2,
where n is the number of elements with the considerations
that diagonal elements are 1. The other comparison values
are simply the reciprocals of the earlier comparisons. An
example of data file with five elements is given in Table 7.
(e) Change value in percent: This value denotes a change
(increase or decrease) in the weights of the main criteria
and used for the sensitivity analysis.
5.3.2. Altersel
This routine provides a user interface with which the decision
maker can perform the UBC approach to generate the valid alterna-
tives and return the appropriate set of them to the UMMS. This
user interface can also be used for getting the detailed information
about the alternatives provided by GETINFO subroutine.
5.3.3. Getinfo
This routine accesses UMMSDAT and provides the decision
makers the information stored in the tables by a user interface.

UMMS
METODSEL MODSEL

LOADMTRX PROCFUZZY ALTERSEL

PROCAHP
UMMS

SAVEMTRX
DAT

GETINFO

COMPARA PERFORSENSITIV
MULTMTRX

FINSOL

Fig. 3. The UMMS and 12 external subroutines.

178 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184

Table 7 tives varies at a specific percentage change (increase and decrease)

Example of data files from pair-wise comparison matrix of the spatial characteristics in the weights of the main criteria.
main criterion with sub-criteria

Spatial characteristics main Size Attitude Depth Regularity Existence 5.3.9. Procfuzzy
criterion This routine asks decision makers for the pair-wise comparisons
Size 1 1/5 1/7 1/3 1/4 of the main criteria and membership levels of each criterion and
Attitude 5 1 1/5 3 2 performs Yager’s FMADM method by calling FINSOL routine to pro-
Depth 7 5 1 7 4
Regularity 3 1/3 1/7 1 1/2
duce the final ratings of the alternatives. The ratings are then re-
Existence 4 1/2 1/4 2 1 turned to the UMMS.

The routine can also manipulate the information by adding, delet-
ing and updating operations by the decision makers.
5.3.4. Loadmtrx
This routine accesses any data file which was already saved to
disk by SAVEMTRX routine and loads the data in the data file into
the corresponding comparison matrices in the AHP method or into
the corresponding comparison matrix at the first level of the
FMADM method.
5.3.5. Savemtrx
This routine can save any comparison matrix data as a data file
into the disk.
5.3.6. Multmtrx
Multiplying of matrices is the most used operation in the AHP
method. On the other hand, this operation is also used for the
FMADM method. This routine takes two matrices as parameters
and returns the multiplied matrix.
5.3.10. Procahp
This routine asks decision makers for constructing the pair-wise
comparison matrices of the alternatives and criteria related to the
selected AHP model. By those matrices, the routine performs the
AHP method and calls the FINSOL routine to produce the final rat-
ings of the alternatives. The ratings are then returned to the UMMS.
5.3.11. Finsol
This routine computes the final rating of the alternatives and
ranks them as well. For the FMADM method, the ratings are com-
puted by using the max–min Bellman and Zadeh (1970) principle.
Computed ratings in a vector form are returned to corresponding
caller routine: PROCAHP or PROCFUZZY.
The UMMS takes the final rating and proposes the decision ma-
ker the alternative with the highest rating as the most preferred
underground mining method. After that time, the decision maker
can either accept the method proposed by the UMMS or make a
sensitivity analysis on the final solution.
6. Case study: selection of underground mining method for
Eskisehir-Karaburun chromite ore

The example is about selection of underground mining method

5.3.7. Compara
for Eskisehir-Karaburun chromite ore. The total ore reserve of the
This routine computes eigenvector ‘‘w”, eigenvalue ‘‘kmax”, CI
Karaburun chromite mine is calculated as 230 000 tons and subsi-
and CR values for any pair-wise comparison matrix A and controls
dence is not prevalent. Table 8 shows the physical and mechanical
if CR is 0.10 or less. If CR is more than 0.10, the routine raise an
characteristics of the deposit.
exception to the decision makers so that they can fix their pair-
wise comparison values. The computation procedure is performed
6.1. The AHP method
through the following steps:

The AHP method was selected in the initial window of the

(a) using the theory of eigenvector, compute ‘‘w”,
UMMS as shown in Fig. 4 and method parameter was set to 1.
(b) compute kmax,
For a quick solution, the second AHP model was selected for anal-
8n 9 ysis so the model parameter was set to 2. By choosing the AHP
> P >
>
> a  w >
j>
1 n <
X ij =
j¼1
kmax ¼ ð5Þ
n >
> wi >
>
i¼1 >
: >
; Table 8
Technical parameters of Karaburun chromite mine (Kahriman et al., 1996)
where aij is the element at row i and column j of A, Parameter Quality
(c) compute CI using Eq. (2),
Inclination +1010 between +1060 : 50–75°,
(d) compute CR using Eq. (3), average 65° NE
(e) evaluate CR. If CR is 0.10 or less, then return A is consistent +960 between +1010 : 40–50°, average
else return inconsistent to the UMMS by raising an excep- 65° NE
tion for inconsistency. Thickness 1.5–3.5 m, average 2.7 m
Dip N 80° W
Type Irregular
Grade distribution Uniform
5.3.8. Perforsensitiv Depth Shallow
This routine takes the final rating of the alternatives and per- Specific gravity for hanging wall, 2.630, 2.586 and 3.244
footwall and ore
forms the sensitivity analysis on this rating by increasing/decreas-
Uniaxial strength of ore 40 MPa
ing the weights of the main criteria by a percent value. This percent Uniaxial strength of hanging wall 49 MPa
value is asked from the decision maker. After input of a percent va- Uniaxial strength of footwall 53 MPa
lue, the routine re-performs the AHP or FMADM process for every RQD for walls and ore 72% and 40 %
increasing/decreasing in the main criteria and instantaneously RMR for hanging wall, footwall and ore 47, 53 and 34
RSS for hanging wall, footwall and orea 18.63, 20.49 and 12.33
shows the variations on the final rating of the alternatives. So,
a
the decision maker can observe how the final rating of the alterna- Rock substance strength (RSS) values were calculated by the authors.
 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184
Fig. 4. The initial windows of the UMMS.
model, we also determine the criteria to be used for the decision
making process.
To determine the valid alternatives for the problem, appropriate
selections according to the problem parameters given in Table 8
are made in the UBC rating window of the UMMS as shown in
Fig. 5. At this point, it should be noted that the rock mass rating
(RMR) values were updated according to Bieniawski (1989). It
should also be noted that the rock substance strength (RSS) values
were calculated by dividing each uniaxial strength of ore, hanging
wall and footwall to principal stress and principal stress values
were calculated by multiplying each gravity for ore, hanging wall
and footwall with 100 m average depth.
From Fig. 5, available alternatives are open pit mining, shrink-
age stoping, cut and fill stoping, sublevel stoping, square set stop-
Fig. 5. The window related to the UBC ratings.
179
ing, and top slicing. Although the open pit mining method is
included in the available alternatives, the decision maker will be
asked for whether he/she let this alternative still remain in the va-
lid alternatives set or exclude it from the set. If the decision maker
decides to exclude it, the open pit mining method will no longer be
available further in the decision process and remaining methods
will stay in the alternatives set. For the problem, the open pit min-
ing method was excluded and the AHP method was performed.
The current alternatives of the model are A1: shrinkage stoping,
A2: cut and fill stoping, A3: sublevel stoping, A4: square set stoping
and A5: top slicing. The pair-wise comparisons of the main criteria
were made and given in Fig. 6. It is apparent that spatial character-
istics of the deposit is the most important criterion (priority of
0.429), then economic considerations follows (priority of 0.265).
On the other hand, the pair-wise comparison of the alternatives
based on each criterion should also be performed. Thus, six matri-
ces were formed. As there are five alternatives, the matrix order
was 5  5. The pair-wise comparison of the alternatives by each
criterion is presented in Tables 9–14. The values in the tables were
captured from the corresponding pair-wise comparison matrices of
the UMMS. It is readily observed from the tables that the most suit-
able alternatives are A4, A4, A4, A5, A2 and A5 when judged by the
criteria of C1, C2, C3, C4, C5 and C6, respectively.
The overall rating of each alternative is calculated by summing
the product of the relative priority of each criterion and the relative
priority of the alternative considering the corresponding criteria. For
example, the overall rating of alternative A1 = (0.429 
0.095) + (0.046  0.097) + (0.082  0.085) + (0.265  0.125) + (0.149
 0.115) + (0.028  0.157) = 0.107. The overall ratings calculated by
the UMMS are presented in Fig. 7. From Fig. 7, it is obvious that A4
(square-set stoping) with a rating of 0.295 is the most preferred min-
ing method. In Table 15, all calculated priorities and final ratings are
summarized.
As the comparisons are based on the subjective evaluation, the
CR values should be calculated using Eq. (3) to ensure the selection
accuracy. Table 16 presents kmax, CI, RI, and CR of the correspond-
ing matrices. For all matrices, the CI and CR are less than 0.10. So,
the logically substantiated decision of constructing the pair-wise
comparison of the criteria or alternatives was made.
180 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184

Fig. 6. Pair-wise comparison matrix of the main criteria.

Table 9 Table 12
Pair-wise comparison matrix of alternatives for spatial characteristics of deposit (C1) Pair-wise comparison matrix of economic considerations (C4)

A1 A2 A3 A4 A5 Priorities A1 A2 A3 A4 A5 Priorities
A1 1 1/3 3 1/5 1/2 0.095 A1 1 1/3 3 2 1/4 0.125
A2 3 1 4 1/4 2 0.215 A2 3 1 6 5 1/3 0.284
A3 1/3 1/4 1 1/6 1/3 0.051 A3 1/3 1/6 1 1/2 1/5 0.051
A4 5 4 6 1 3 0.488 A4 1/2 1/5 2 1 1/6 0.073
A5 2 1/2 3 1/3 1 0.150 A5 4 3 5 6 1 0.467

kmax = 5.208, CI = 0.052 and CR = 0.046 6 0.1. kmax = 5.217, CI = 0.054 and CR = 0.048 6 0.1.

Table 10 Table 13
Pair-wise comparison matrix of alternatives for geologic and hydrologic conditions Pair-wise comparison matrix of technological factors (C5)
(C2)
A1 A2 A3 A4 A5 Priorities
A1 A2 A3 A4 A5 Priorities A1 1 1/5 3 2 1/4 0.115
A1 1 1/2 2 1/4 1/3 0.097 A2 5 1 6 4 3 0.476
A2 2 1 3 1/3 1/2 0.160 A3 1/3 1/6 1 1/2 1/5 0.052
A3 1/2 1/3 1 1/5 1/4 0.062 A4 1/2 1/4 2 1 1/3 0.089
A4 4 3 5 1 2 0.417 A5 4 1/3 5 3 1 0.267
A5 3 2 4 1/2 1 0.263
kmax = 5.239, CI = 0.060 and CR = 0.053 6 0.1.
kmax = 5.068, CI = 0.017 and CR = 0.015 6 0.1.

Table 11 Table 14
Pair-wise comparison matrix of geotechnical properties (C3) Pair-wise comparison matrix of environmental concerns (C6)

A1 A2 A3 A4 A5 Priorities A1 A2 A3 A4 A5 Priorities
A1 1 1/5 2 1/3 1/2 0.085 A1 1 1/3 3 4 1/4 0.157
A2 5 1 7 1/2 6 0.372 A2 3 1 5 2 1/2 0.269
A3 1/2 1/7 1 1/5 1/3 0.050 A3 1/3 1/5 1 1/2 1/6 0.055
A4 3 2 5 1 4 0.382 A4 1/4 1/2 2 1 1/3 0.095
A5 2 1/6 3 1/4 1 0.111 A5 4 2 6 3 1 0.423

kmax = 5.301, CI = 0.075 and CR = 0.067 6 0.1. kmax = 5.348, CI = 0.087 and CR = 0.078 6 0.1.

6.2. Sensitivity analysis priority of one of the criteria increases, the priorities of the remain-
ing criteria must decrease proportionately, and the global priorities
A sensitivity analysis can be performed to see how sensitive the of the alternatives must be recalculated. Sensitivity analysis can
alternatives will change with the importance of the criteria. As the also be used to determine the most important or critical criterion
 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184 181

Fig. 7. Final solution screen of the UMMS.

Table 15 spatial characteristics of the deposit was increased by 8%. This case
Overall result/final matrix was captured from the corresponding sensitivity window of the
A1 A2 A3 A4 A5 Priority UMMS and illustrated in Fig. 8.
It should also be noted that cut and fill stoping method was also
C1 0.095 0.215 0.051 0.488 0.150 0.429
C2 0.097 0.160 0.062 0.417 0.263 0.046 the best alternative when the decreasing of 19% in the priority of
C3 0.085 0.372 0.050 0.382 0.111 0.082 technological factors and the decreasing of 21% in the priority of
C4 0.125 0.284 0.051 0.073 0.467 0.265 economic considerations occur. It can be concluded from the sen-
C5 0.115 0.476 0.052 0.089 0.267 0.149 sitivity analysis that the final solution of the proposed AHP model
C6 0.157 0.269 0.055 0.095 0.423 0.028
is mainly sensible to increasing of spatial characteristics of the de-
Overall 0.107 0.284 0.052 0.295 0.261
posit and decreasing of technological factors and economic
considerations.

6.3. The FMADM method

Table 16
kmax, CI, RI, and CR of different matrices FMADM was selected and method parameter was set to 2. The
Goal C1 C2 C3 C4 C5 C6 alternatives considered in the AHP method were still valid for this
method. The pair-wise comparison matrix of the main criteria was
Weight 1 0.429 0.046 0.082 0.265 0.149 0.028
kmax 6.132 5.208 5.068 5.301 5.217 5.239 5.348
also same with that in the AHP method and it is given in Fig. 6. The
CI 0.026 0.052 0.017 0.075 0.054 0.060 0.087 decision maker was asked to define the membership levels of each
RI 1.24 1.12 1.12 1.12 1.12 1.12 1.12 criterion after conferring with experts on this subject according to
CR 0.021 0.046 0.015 0.067 0.048 0.053 0.078 Fig. 9 (Chen and Klein, 1997). In the figure, VH, H, MLH, M, MLL, L
and VL denote very high, high, more or less high, medium, more or
less low, low and very low respectively. Table 17 shows the mem-
by computing the absolute or percentage amount by which the bership levels of each criterion judged by the experts.
weight of any criterion must be changed in order to cause a switch The respective weights of the criteria were obtained from the
in the ranking of either the top alternative or in any pair of alterna- eigenvector of the matrix given in Fig. 6, i.e. eigenvec-
tives (Triantaphyllou and Sánchez, 1997). tor = {0.429, 0.046, 0.082, 0.265, 0.149, 0.028}. The eigenvector cor-
By increasing or decreasing the values of the Eigenvector of responds to the weights to be associated with the memberships
main criteria in the pair-wise comparison matrix, the variations of each criterion. The exponential weighting was consequently de-
on the final rank of the alternatives were observed. There was no fined from each criterion as: a1 = 0.429, a2 = 0.046, a3 = 0.082,
change in the judgement evaluations in the final rank of the alter- a4 = 0.265, a5 = 0.149, a6 = 0.028. Table 18 shows the membership
natives (i.e. 1. Square set stoping, 2. Cut and fill stoping, 3. Top slic- decision function of each criterion according to Yager’s (1978)
ing, 4. Shrinkage stoping and 5. Sublevel stoping) when eigenvector method.
value of each criterion increased/decreased up to 7%. This means According to the max–min Bellman and Zadeh (1970) principle,
that square set stoping method will still be the most preferred the final rating of the alternatives was determined as below:
method for the problem when the percentage increasing or
decreasing is up to 7%. On the other hand, the best alternative lD ðAÞ ¼ minfðA1 =0:637Þ; ðA2 =0:743Þ; ðA3 =0:501Þ; ðA4 =0:757Þ;
was changed to cut and fill stoping method when the priority of A5 =0:743Þg
182 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184

Fig. 8. Sensitivity analysis window of the UMMS.

More or More or
Very Low Low Medium Less High High Very High
Less Low
(VL) (L) (M) (MLH) (H) (VH)
(MLL)
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.05

0.1
0.15

0.2

0.25

0.3
0.35

0.4

0.45

0.5
0.55

0.6
0.65
0.7

0.75
0.8

0.85
0.9

0.95
1

Fig. 9. Linguistic model for fuzzy numbers (Chen and Klein, 1997).

Table 17 Table 18
Membership level of each criterion Membership decision function of each criterion by Yager’s method

A1 A2 A3 A4 A5 A1 A2 A3 A4 A5
C1 0.35 0.5 0.2 0.95 0.5 C1 0.637 0.743 0.501 0.978 0.743
C2 0.2 0.35 0.05 0.95 0.65 C2 0.929 0.953 0.871 0.998 0.980
C3 0.2 0.65 0.05 0.8 0.35 C3 0.876 0.965 0.782 0.982 0.918
C4 0.2 0.5 0.2 0.35 0.95 C4 0.653 0.832 0.653 0.757 0.986
C5 0.35 0.65 0.05 0.5 0.5 C5 0.855 0.938 0.640 0.902 0.902
C6 0.2 0.5 0.2 0.35 0.8 C6 0.956 0.981 0.956 0.971 0.994

yielding the result: lD(A*) = max{lD(A4)} = 0.757 which selects that terms of fuzzy dominance method explained by Bitarafan and Ataei
square set stoping method as most preferable. The program’s out- (2004).
put by the UMMS in Yager’s method section is presented in After reaching a solution, the decision maker can perform the
Fig. 10. At this section, the UMMS finds an alternative solution in sensitivity analysis in a similar way as previously presented in Sec-
 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184
Fig. 10. The program’s output by the UMMS in Yager’s method section.
tion 6.2 for the AHP method so as to analyze the sensitivity of the
solution.
7. Conclusions
In this paper, a computer program for underground mining
method selection (UMMS) was developed to eliminate the difficul-
ties in taking into consideration many decision criteria simulta-
neously in the underground mining method selection process
and to guide the decision makers to select the optimal under-
ground mining method.
In the study, two methods, AHP and FMADM, and two AHP
models are proposed by the UMMS. Using the first AHP model, it
is possible to analyze an underground mining method selection
problem with respect to the total of 36 criteria clustered in six
main groups. This also makes possible to increase the consistency
of decisions made for selection of the optimal mining method. On
the other hand, if the time is restricted and/or the six main group
criteria without their sub-criteria are enough to make a decision,
the second AHP model and/or the FMADM method can be used
as the alternative ways.
An important support of the UMMS to the decision making pro-
cess is that the decision makers can perform the sensitivity analy-
sis after reaching a solution. In this way, the decision makers can
observe how the proposed solution varies at a specific percentage
change (increase and decrease) in the weights of the main criteria.
References
Acaroglu, O., Ergin, H., Eskikaya, S., 2006a. Analytical hierarchy process for selection
of roadheaders. J. S. Afr. Inst. Min. Metall. 106, 569–575.
Acaroglu, O., Feridunoglu, C., Tumac, D., 2006b. Selection of roadheaders by fuzzy
multiple attribute decision making method. Trans. Inst. Min. Metall. A 115,
A91–A98.
Ataei, M., 2005. Multicriteria selection for alumina-cement plant location in East-
Azerbaijan province of Iran. J. S. Afr. Inst. Min. Metall. 105, 507–514.
Bascetin, A., Kesimal, A., 1999. The study of a fuzzy set theory for the selection of an
optimum coal transportation system from pit to the power plant. Int. J. Surf.
Min. Reclam. Environ. 13, 97–101.
Bascetin, A., Oztas, O., Kanli, A.I., 2006. EQS: a computer software using fuzzy logic
for equipment selection in mining engineering. J. S. Afr. Inst. Min. Metall. 106,
63–70.
183
Bellman, R.E., Zadeh, L.A., 1970. Decision making in a fuzzy environment. Manage.
Sci. 17, 141–164.
Bieniawski, Z.T., 1989. Engineering Rock Mass Classification. John Wiley & Sons,
New York. 237 p.
Bitarafan, M.R., Ataei, M., 2004. Mining method selection by multiple criteria
decision making tools. J. S. Afr. Inst. Min. Metall. 104, 493–498.
Boshkov, S.H., Wright, F.D., 1973 (Basic and parametric criteria in the selection,
design and development of underground mining systems). SME Mining
Engineering Handbook. SME-AIME, New York.
Chen, C., Klein, C.M., 1997. An efficient approach to solving fuzzy MADM problems.
Fuzzy Set Syst. 88, 51–67.
Elevli, B., Demirci, A., 2004. Multicriteria choice of ore transport system for an
underground mine: application of PROMETHEE methods. J. S. Afr. Inst. Min.
Metall. 104, 251–256.
Elevli, B., Demirci, A., Dayi, O., 2002. Underground haulage selection: shaft or ramp
for a small-scale underground mine. J. S. Afr. Inst. Min. Metall. 102, 255–
260.
Hartman, H.L., 1987. Introductory Mining Engineering. John Wiley, New Jersey.
Hartman, H.L., Mutmansky, J.M., 2002. Introductory Mining Engineering. John
Wiley, New Jersey.
Kahriman, A., Ceylanoğlu, A., Demirci, A., Arpaz, E., Görgülü, K., 1996. Selection of
optimum underground mining method for Eskisehir-Karaburun chromite ore
(in Turkish). In: Proceeding of the 3rd National Rock Mechanics Symposium, pp.
47–60.
Karadogan, A., Bascetin, A., Kahriman, A., Gorgun, S., 2001. A new approach in
selection of underground mining method. In: Proceeding of the International
Conference-Modern Management of Mine Producing, Geology and Environment
Protection, pp. 171–183.
Kazakidis, V.N., Mayer, Z., Scoble, M.J., 2004. Decision making using the analytic
hierarchy process in mining engineering. Trans. Inst. Min. Metall. A 113, A30–
A42.
Kesimal, A., Bascetin, A., 2002. Application of fuzzy multiple attribute decision
making in mining operations. Miner. Resour. Eng. 11, 59–72.
Laubscher, D.H., 1981. Selection of Mass Underground Mining Methods. Design and
Operation of Caving and Sublevel Stoping Mines. SME-AIME, New York.
Miller-Tait, L., Panalkis, R., Poulin, R., 1995. UBC mining method selection. In:
Proceeding of the Mine Planning and Equipment Selection Symposium, pp.
163–168.
Morrison, R.G.K., 1976. AW Philosophy of Ground Control. McGill University,
Montreal, Canada.
Nicholas, D.E., 1981. Method Selection – A Numerical Approach. Design and
Operation of Caving and Sublevel Stoping Mines. SME-AIME, New York.
Riberio, R.A., 1996. Fuzzy multiple attribute decision making: a review and new
preference elicitation techniques. Fuzzy Set Syst. 78, 155–181.
Saaty, T.L., 1980. The Analytic Hierarchy Process. McGraw-Hill, Pittsburgh.
Saaty, T.L., 2000. Fundamentals of Decision Making and Priority Theory with the
Analytic Hierarchy Process. RWS, Pittsburgh.
Samanta, B., Sarkar, B., Murherjee, S.K., 2002. Selection of opencast mining
equipment by a multi-criteria decision-making process. Trans. Inst. Min.
Metall. A. 111, A136–A142.
184 S. Alpay, M. Yavuz / Tunnelling and Underground Space Technology 24 (2009) 173–184
Triantaphyllou, E., Sánchez, A., 1997. A sensitivity analysis approach for some
deterministic multi-criteria decision making methods. Decision Sci. 28, 151–
194.
Yager, R.R., 1978. Fuzzy decision making including unequal objectives. Fuzzy Set
Syst. 1, 87–95.
Yavuz, M., 2007. Analytical hierarchy process for selection of roadheaders, by O.
Acaroglu, H. Ergin, and S. Eskikaya in the Journal of SAIMM, vol. 106, No. 8, pp.
569–575. J. S. Afr. Inst. Min. Metall. 107, 138.
Yavuz, M., Alpay, S., 2007a. Mining method selection by multiple criteria decision
making tool by M.R. Bitarafan and M. Ataei in the Journal of SAIMM, vol. 104,
No. 9, pp. 493–498. J. S. Afr. Inst. Min. Metall. 107, 137.
Yavuz, M., Alpay, S., 2007b. EQS: a computer software using fuzzy logic for
equipment selection in mining engineering by A. Basçetin, O. Öztasß, and A.I.
Kanli in the Journal of SAIMM, vol. 106, No. 1, pp. 63–70. J. S. Afr. Inst. Min.
Metall. 107, 138.
Yavuz, M., Iphar, M., Once, G., 2008. The optimum support design selection by using
AHP method for the main haulage road in WLC Tuncbilek colliery. Tunn.
Undergr. Space Technol. 23, 111–119.
Zadeh, L.A., 1973. Outline of a new approach to the analysis of complex systems and
decision process. IEEE Trans. 3, 28–44.