Desalination 313 (2013) 44–50

Contents lists available at SciVerse ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

A hybrid fuzzy multi-criteria decision making approach for desalination

process selection
Seyed Ali Ghassemi, Shahnaz Danesh ⁎
Department of Civil Engineering, Faculty of Engineering, Ferdowsi University of Mashhad, Mashhad, Iran

H I G H L I G H T S

► A hybrid model was developed based on the fuzzy-AHP and TOPSIS methods.
► Fuzzy-AHP was used to determine the weights of the criteria and sub-criteria.
► TOPSIS method was used to calculate the final ranking of the desalination technologies.
► A real world application of the model demonstrated its feasibility and reliability.

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

Article history: In this paper an integrated two-step model was developed based on the fuzzy-AHP and TOPSIS methods. The
Received 14 March 2012 performance and reliability of the model were then evaluated in a real world case study concerning the selection
Received in revised form 27 November 2012 of the most suitable desalination technology for the treatment of brackish groundwater typical of an area located
Accepted 9 December 2012
in north-east of Iran. The desalination technologies included in this study were reverse osmosis, electrodialysis,
Available online 5 January 2013
ion exchange, multistage flash distillation, multi-effect distillation, and vapor compression. The comparison of
Keywords:
the technologies was based on various environmental, technical and economical criteria and sub-criteria. The
Fuzzy-AHP fuzzy-AHP was used to analyze the structure of the selection process and to determine the weights of the criteria
TOPSIS and sub-criteria, and the TOPSIS method was used to calculate the final ranking of the technologies. The outcome
Desalination results of the two-step model revealed that electrodialysis, with a closeness coefficient value of 0.7547, was the
Brackish water most applicable desalination technology for the study area. Moreover, sensitivity analysis demonstrated that any
Electrodialysis variation in the criteria weights does not affect the outcome of the model.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction (MED), and vapor compression (VC) [1]. In contrast, membrane-based

desalination techniques use different types of membrane to separate
In recent years, the scarcity of fresh water resources in countries dissolved solids from water. The most popular membrane-based desali-
located in the arid and semi arid regions of the world has become nation technologies are reverse osmosis (RO) and electrodialysis (ED).
more critical due to resource limitations, global warming and the in- In ion exchange process (IE) the ions in the solution are exchanged by
crease in water consumption. As such, the use of brackish groundwater the ions of a resin [2].
and the application of different technologies to improve their quality for Because of the advantages and disadvantages associated with each
various consumptions have gained special attention in water resource desalination technology, the selection of the optimum technique for
management programs. any specific area is a complicated task due to the diversity of objectives
At the present, there are many different types of desalination tech- and constraints that should be considered and satisfied simultaneously.
nologies available in the market, each with its own technical specifica- With these types of problems, decision makers cannot go through the
tions and applicability. These technologies can be categorized in three standard single criteria mathematical programming techniques to find
general groups of distillation, membrane-based and ion exchange. In the best option. Moreover, like most other real-life problems, there al-
distillation processes, water is transformed into vapor and then is con- ways exists a lack of sufficient data which will add an extra dimension
densed into a liquid state. Commercially available technologies of this of complexity. In such situations where the decision maker confronts
type include multistage flash distillation (MSF), multi-effect distillation many criteria and constraints, multi criteria decision making (MCDM)
methods can offer a proper solution, as they provide techniques for com-
paring and ranking many criteria and choices. Another major advantage
⁎ Corresponding author at: Ferdowsi University of Mashhad, Azadi Sq., Mashhad, Khorasan
Razavi, Iran. Tel.: +98 9153100003; fax: +98 511 8788805.
of most MCDM techniques is their ability to analyze both quantitative
E-mail addresses: [email protected] (S.A. Ghassemi), and qualitative criteria together. Many techniques and methodologies
[email protected] (S. Danesh). are reported in the literature for MCDM [3]. Among most popular ones

0011-9164/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.desal.2012.12.008
 S.A. Ghassemi, S. Danesh / Desalination 313 (2013) 44–50 45

are the analytic hierarchy process (AHP) [4], the technique for order Table 1
preference by similarity to an ideal solution (TOPSIS) [5], elimination Triangular fuzzy scale of preferences [26].

and choice corresponding to reality (ELECTRE) [6], preference rank- Linguistic terms Triangular fuzzy Triangular fuzzy
ing organization method for enrichment of evaluations (PROMETHEE) scale reciprocal scale
[7], decision-making trial and evaluation laboratory (DEMATEL) [8], Just equal (1, 1, 1) (1, 1, 1)
analytic network process (ANP) [9], and Vlsekriterijumska Optimizacija Equally important (1/2, 1, 3/2) (2/3, 1, 2)
I Kompromisno Resenje (VIKOR) [10]. Weakly important (1, 3/2, 2) (1/2, 2/3, 1)
Strongly more important (3/2, 2, 5/2) (2/5, 1/2, 2/3)
In regard to desalination technologies, Hajeeh and Al-Othman
Very strong more important (2, 5/2, 3) (1/3, 2/5, 1/2)
[11] used a two-stage AHP process to select the most appropriate al- Absolutely more important (5/2, 3, 7/2) (2/7, 1/3, 2/5)
ternative. Seven criteria were selected and used in order to identify
the most suitable desalination technology from four desalination
plants. Mohsen and Al-Jayyousi [12] also applied a five-step AHP according to the extent of uncertainty and ambiguity present in the
model to evaluate various desalination technologies. The criteria decision making problem. In a literature a range of linguistic scales in-
adopted for evaluation were based on technical, economic, and envi- cluding 5-point, 6-point and 7-point has been reported [25]. In this
ronmental aspects. Hajeeh [13] presented a hierarchy model based research, a 6-point triangular fuzzy scale of preferences was used
on the fuzzy set theory to deal with the desalination technology (Table 1). This scale was proposed by Kahraman et al. [26] and used
selection problem. The linguistic values were used to assess the rat- for solving fuzzy decision making problems [27–29].
ings and weights for the technology evaluating factors. The selection There are many alternatives as solution methods to perform on the
process was limited to six factors and three commercially available fuzzy-AHP based structured model on MCDM problems [23,30–34].
desalination technologies including MSF, MED, and RO. Bick and Among the most reliable and simple ones is the Chang [30] extent anal-
Oron [14] developed an AHP-based decision making approach to ysis method which is used in this study. The method is used to determine
select the best post-treatment technology for a specific seawater re- the extent of an object to be satisfied for the goal. The Chang's [30] meth-
verse osmosis plant. Post-treatment systems were evaluated based od includes several steps which are summarized as follows [21,26].
on seven criteria. A sensitivity analysis was performed to examine Assume X = {x1, x2, …, xn} be an object set, and G = {g1, g2, …, gm}
the response of alternatives when the relative importance rating of be a goal set. According to the method of extent analysis, each object
each criterion was changed. is taken and extent analysis is performed for each goal, gi respectively.
In this study, an integrated model consisting of fuzzy-AHP and Therefore, the m extent analysis values for each object can be ob-
TOPSIS was established to provide a stepwise methodology for the selec- tained, with the following signs:
tion of the optimum desalination technology among different available
1 2 m
technologies. The model was then applied in a case study to demon- M̃ gi ; M̃gi ; …; M̃ gi ; i ¼ 1; 2; …; n
strate its applicability in a real world pilot study and prove its reliability.
j
The fuzzy-AHP has a strong ability to handle the uncertainty and am- where all the M̃ gi ; j ¼ 1; 2; …; m are TFNs. A TFN is represented by
biguity present in deciding the priorities of different alternatives in an three parameters: the least possible value, the most possible value,
MCDM situation. Moreover, it allows for approximate values and infer- and the highest possible value, here are represented by l, m and u
ences as well as incomplete or ambiguous data (fuzzy data) as opposed respectively.
to only relying on crisp data (binary yes/no choices) [15]. TOPSIS is an
Step 1 The value of the fuzzy synthetic extent with respect to the ith
efficient model in handling the sensible attributes and there is no limit
object is defined as follows:
in terms of number of criteria, sub-criteria or alternatives. As a result,
the integration of AHP-fuzzy and TOPSIS can provide a strong base for 2 3−1
X
m Xn X
m
the analysis of complex decision problems [16–20]. Furthermore, the S~i ¼ M̃ ⊗4j ~
M 5 :
j
ð1Þ
gi gi
AHP-fuzzy and TOPSIS methods can be easily programmed by using a j¼1 i¼1 j¼1
spreadsheet to automate the decision making process.
The following sections cover respectively: a brief description of m ~j
To obtain ∑j¼1 M gi
, the fuzzy addition operation of the m extent anal-
fuzzy-AHP and TOPSIS, the proposed integrated methodology, the ysis values for a particular matrix is performed as shown in
application of the model in a real world situation concerned with the Eq. (2):
selection of the optimal desalination technology, and the conclusions.
2 3−1
X
m Xm X
m Xm
2. Fuzzy-AHP method ~
M ¼4
j
lj ; mj ; uj 5 : ð2Þ
gi
j¼1 j¼1 j¼1 j¼1
AHP is an MCDM technique developed by Saaty [4] for evaluating
different alternatives against a set of selected criteria in order to de-
n m ~ j ~j
termine the best alternative. AHP assumes that criteria can be And to obtain ∑i¼1 ∑j¼1 M gi the fuzzy addition operation M gi (j = 1,
expressed in a hierarchical structure. In this model, the criteria are 2, …, m) values are performed as in Eq. (3):
compared pairwise and the final decision is made based on the results
!
of these comparisons [21]. In conventional AHP for pairwise compar- X
n X
m X
n X
n X
n
~ j¼
M li ; mi ; ui : ð3Þ
isons of the criteria an arbitrary value (acquired by mainly decision gi
i¼1 j¼1 i¼1 i¼1 i¼1
makers) is allocated to each criterion. Therefore, due to the high
degree of uncertainty involved in the allocated values, the results can-
not be completely reliable [22]. To reduce the degree of uncertainty
The inverse of the vector above is then computed as presented
and vagueness associated with the conventional AHP, different ver-
in Eq. (4):
sions of the fuzzy-AHP methods were developed [23]. In general, in
the fuzzy-AHP models a linguistic approach is applied, in which the
optimism/pessimism conceptual rating attitude of decision-makers 2 3−1 !
is taken into account. Because of the linguistic approach, triangular Xn X
m
1 1 1
4 ~ 5 ¼
M j
; ; : ð4Þ
fuzzy numbers (TFNs) are used to quantify conceptual preferring n n n
i¼1 j¼1
gi ∑i¼1 ui ∑i¼1 mi ∑i¼1 li
ratings of criteria [24]. The linguistic scale of TFNs can be chosen
46 S.A. Ghassemi, S. Danesh / Desalination 313 (2013) 44–50

~ 1 and M
Fig. 1. The intersection between M ~ 2.

Step 2 As M ~ and M ~ are two TFNs, the degree of possibility of M

~ ¼ where i = 1, 2, …, k.
1 2 2
ðl2 ; m2 ; u2 Þ≥M~ ¼ ðl ; m ; u Þ is defined as: Assume that:
1 1 1 1

  h i  
d ðAi Þ ¼ minV S~i ≥S~k :
′
~
~ ≥M ð8Þ
V M 2 1 ¼ supy≥x minμ M
~ ðxÞ; minμ M
~ ðyÞ
1 2
ð5Þ

and can be equivalently expressed as follows: For k = 1, 2, …; k ≠ i.

8 Then the weight vector is given by:
>
> 1;  T
  < if m2 ≥m1 ′ ′ ′ ′
~
~ ≥M 0; W ¼ d ðA1 Þ; d ðA2 Þ; …; d ðAn Þ ð9Þ
V M ¼ μ ~ ðdÞ ¼ l −u if l1 ≥u2 ð6Þ
2 1 M 2 >
> 1 2
; otherwise
:
ðm2 −u2 Þ−ðm1 −l1 Þ
where Ai (i = 1,2, …, n) are n elements.
Step 4 Via normalization, the normalized weight vectors are defined
where d is the ordinate of the highest intersection point D
~ as:
between μ M~ 1 and μ M~ 2 , as shown in Fig. 1. To compare
 M 1
and M~ , both values of V M ~ ≥M~ and V M~ ≥M~ should T
2 2 1 1 2
W ¼ ðdðA1 Þ; dðA2 Þ; …; dðAn ÞÞ ð10Þ
be calculated.
Step 3 The degree of possibility for a convex fuzzy number to be where W is a non-fuzzy number.
greater than k convex fuzzy numbers Mi (i = 1, 2, …, k) can The issue of consistency of fuzzy pairwise comparison matrices
be defined by: in fuzzy-AHP is another subject that needs to be taken into con-
sideration. The consistency of a comparison matrix in conven-
    tional AHP is measured by the consistency ratio [4]. But the
~ M
~ ;M~ ; …; M
~ ~ ~
V M≥ 1 2 k ¼ minV M≥M i ð7Þ
results of fuzzy synthetic decision are fuzzy numbers, thus, it

Define criteria and sub criteria for evaluation of desalination technologies

Literature review
Determining alternative desalination technologies

Construct fuzzy pairwise comparison matrices using triangular fuzzy numbers Expert opinions
Fuzzy AHP

Calculate fuzzy relative importance weights of matrices

Check the consistency of the matrices

Obtain the evaluation criteria, sub criteria and alternatives weights

Evaluate the desalination technologies using TOPSIS method

TOPSIS

Calculate PIS and NIS, and separation measures

Rank the preference order for desalination technologies and selecting the optimal technology

Fig. 2. Schematic diagram of the proposed model for desalination technology selection.
 S.A. Ghassemi, S. Danesh / Desalination 313 (2013) 44–50 47

Table 2
Desalination technology evaluation criteria, sub-criteria and definition.

Criteria Sub-criteria Definition

Environmental (C1) Brine management (SC1) Handling and disposal of brine

Air pollution (SC2) Water vapor as a byproduct of thermal desalination processes
Technical (C2) Operational complexity (SC3) Skill required to operate technology
Reliability (SC4) The ability of technology to perform steadily under stated conditions
Expandability (SC5) Ability of technology to accommodate additions to its capacity or capabilities
Adaptability (SC6) Compatibility of technology with quality of influent water
Water recovery (SC7) Product water relative to the input water flow
Treated water quality (SC8) Salinity of product water
Economical (C3) Capital costs (SC9) Purchase of mechanical equipments, installations and other incidental construction work
Operating costs (SC10) Wages and the funds spent for the energy, the products, services and maintenance

should be converted to real numbers which could be done Step 2 The normalized decision matrix is then weighted by multiply-
by employing any of the defuzzification techniques [35]. The ing the normalized matrix with the weights of the criteria:
most common techniques are right value, center of area, and

α-cut method [36]. In this research the center of area approach vij ¼ wi r ij ; j ¼ 1; 2; …; J; i ¼ 1; 2; …; n: ð12Þ
was used because: a) it is simple and does not need an analyst's
personal judgment [37], b) the center of area method is the Step 3 PIS (maximum values) and NIS (minimum values) are deter-
most prevalent and physically most appealing of all the mined respectively as:
defuzzification methods [38], and C) it gives the same ranking   
 
with most of the other methods [39]. A ¼ v1 ; v2 ; …; vn ; ð13Þ

− − − −
3. TOPSIS method A ¼ fv1 ; v2 ; …; vn g: ð14Þ

TOPSIS is one of the major techniques in dealing with MCDM Step 4 The distance of each alternative from PIS and NIS is calculated
problems. It is based upon the concept that the best alternative is as follows:
the one that has the shortest distance from the positive ideal solution
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uX
(PIS) and the longest distance from the negative ideal solution (NIS).
 u n  
 2
The PIS is the solution that maximizes the benefit criteria and min- di ¼t vij −vj ;j ¼ 1; 2; … , J ð15Þ
imizes the cost criteria; whereas the NIS has an opposite logic, j¼1

i.e., maximizes the cost criteria and minimizes the benefit criteria vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uX
[40]. The TOPSIS method considers the distances to both PIS and the − u n  − 2

NIS, simultaneously. Due to its rationality, logic, and computational di ¼t vij −vj ;j ¼ 1; 2; …; J: ð16Þ
j¼1
simplicity, TOPSIS has been widely applied in many research areas as-
sociated with a selection of various alternatives and their risk analysis
[16,41–44]. Step 5 The closeness coefficient of each alternative (CCi) relative to
The application of the TOPSIS model includes the following steps its distance from PIS and NIS is then calculated by using the
[45]: following equation:

Step 1 Decision matrix is normalized by using Eq. (11): d−

i
CC i ¼ : ð17Þ
di þ d−
i
wij
r ij ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; j ¼ 1; 2; …; J; i ¼ 1; 2; …; n: ð11Þ Step 6 Finally, the ranking of alternatives is determined by comparing
J 2
∑j¼1 wij
the CCi values.

Desalination Technology Selection

C1 C2 C3

SC1 SC2 SC3 SC4 SC5 SC6 SC7 SC8 SC9 SC10

IE VC MED MSF ED RO

Fig. 3. The decision hierarchy of desalination technology selection.

48 S.A. Ghassemi, S. Danesh / Desalination 313 (2013) 44–50

Table 3 so many different desalination technologies being available in the

The pairwise comparison matrix for criteria. market, as well as, the environmental, technical and economical con-
C1 C2 C3 straints to be considered, the final decision making for the most suit-
able technology, was not an easy task to achieve [47]. Therefore, the
C1 (1, 1, 1) (0.44, 0.57, 0.8) (0.67, 1, 2)
C2 (1.25, 1.75, 2.25) (1, 1, 1) (1, 1.5, 2) proposed methodology was applied to analyze the situation and to
C3 (0.5, 1, 1.5) (0.5, 0.67, 1) (1, 1, 1) simplify the decision making process, effectively. For this purpose,
in the first step, an expert team of five members including three experts
from the province's Water and Wastewater Company and the authors,
was formed. The experience and viewpoints of these members were
Table 4 used throughout the entire course of this study, following the procedure
Priority weights of sub-criteria.
described below.
Sub-criteria Priority weight Sub-criteria Priority weight

SC1 1.000 SC6 0.271 5.1. Identification of criteria

SC2 0.000 SC7 0.184
SC3 0.087 SC8 0.192 The criteria and sub-criteria of importance for comparison of
SC4 0.192 SC9 0.500 desalination technologies were determined by the expert team mem-
SC5 0.073 SC10 0.500
bers based on their backgrounds and experiences. The final outcome
of their decision included three criteria and ten sub-criteria which
4. Proposed methodology are summarized in Table 2.
In the next step, available desalination technologies were researched
The proposed model for the water desalination technology selection by the team members and six desalination processes including IE, VC,
integrates the fuzzy-AHP and TOPSIS methods and includes three basic MED, MSF, ED and RO were chosen to be included in this study. The de-
stages: (1) identification of the criteria, sub-criteria and alternatives to cision hierarchy was then structured using desalination technologies,
be used in developing the decision hierarchy structure, (2) fuzzy-AHP criteria and sub-criteria, as shown in Fig. 3.
computations, and (3) evaluation of the alternatives and determination The decision hierarchy included four levels: The overall goal “the se-
of the final ranking by TOPSIS. lection of the optimum desalination process” was considered to be the
In the first stage, alternative desalination technologies, criteria first level of the hierarchy, the criteria and sub-criteria were located
and sub-criteria of importance, are selected and the decision hierar- in the second and third level, respectively, and the alternative desalina-
chy is structured. The objective is placed in the first level and the tion technologies were included in the fourth level.
criteria, sub-criteria and alternative desalination technologies are
placed in the second, third and fourth level, respectively. The deci- 5.2. Weights of criteria
sion hierarchy should be approved by the decision-making team.
In the next stage, the criteria used in desalination technology selec- After forming the decision hierarchy structure, the weights of the
tion are assigned weights using fuzzy-AHP. As described in Section 2, criteria were calculated by the fuzzy extent analysis method, described
linguistic values are used to allocate the criteria weights. in Section 2. For this purpose, the expert team members were given the
In the third stage, by using the TOPSIS procedure, the desalination task of forming pairwise comparison matrices by using the linguistic
technologies are ranked. Ranking is done according to CCi values, in a scale given in Table 1. Since it was not possible to make mathematical
descending order. The desalination technology which obtains the calculations using linguistic terms, each term was given a TFN. In order
maximum CCi value is selected as the optimum desalination technol- to obtain the fuzzy pairwise comparison matrix, the arithmetic means
ogy. The schematic diagram of the proposed model for the desalina- of the fuzzy scores given by the team members were calculated with
tion technology selection is provided in Fig. 2. the results summarized in Table 3.
In the next step, the weights of all the criteria were determined
based on the fuzzy-AHP procedure. For this purpose, by using the values
5. Application of the proposed model reported in Table 3, the synthesis values with respect to the main goal
were calculated by using Eq. (1) as shown below:
The proposed methodology was applied to an area located in a semi
arid region, north-east of Iran, with an annual average precipitation  
1 1 1
of about 150 mm/yr [46]. The water consumption in this area is sup- C1 ¼ ð2:11; 2:57; 3:8Þ⊗ ; ; ¼ ð0:168; 0:271; 0:516Þ
12:55 9:49 7:36
plied mainly by groundwater resources. The average total dissolved
solids content of the water samples taken from a few wells in the  
1 1 1
study area indicated a value of about 2623 mg/lit, categorizing the aqui- C2 ¼ ð3:25; 4:25; 5:25Þ⊗ ; ; ¼ ð0:259; 0:448; 0:713Þ
12:55 9:49 7:36
fer as a brackish source. Due to the critical limitation of water resources
in this area, the desalination of brackish water has gained special atten-
 
tion in water resource management to extend water supply for domes- 1 1 1
C3 ¼ ð2; 2:67; 3:5Þ⊗ ; ; ¼ ð0:159; 0:281; 0:476Þ:
tic consumption. However, as mentioned in the previous sections, with 12:55 9:49 7:36

Table 5
Evaluation of alternative desalination technologies with respect to brine management (SC1) sub-criterion.

Alternatives ED RO IE MSF MED VC w

ED (1, 1, 1) (5/4, 7/4, 9/4) (3/2, 2, 5/2) (3/2, 2, 5/2) (3/2, 2, 5/2) (5/4, 7/4, 9/4) 0.146
RO (4/9, 4/7, 4/5) (1, 1, 1) (1/2, 1, 3/2) (1, 3/2, 2) (1, 3/2, 2) (1, 3/2, 2) 0.055
IE (2/5, 1/2, 2/3) (2/3, 1, 2) (1, 1, 1) (1/2, 1, 3/2) (1/2, 1, 3/2) (1, 1, 1) 0.101
MSF (2/5, 1/2, 2/3) (1/2, 2/3, 1) (2/3, 1, 2) (1, 1, 1) (1, 1, 1) (1, 1, 1) 0.101
MED (2/5, 1/2, 2/3) (1/2, 2/3, 1) (2/3, 1, 2) (1, 1, 1) (1, 1, 1) (1, 1, 1) 0.371
VC (4/9, 4/7, 4/5) (1/2, 2/3, 1) (1, 1, 1) (1, 1, 1) (1, 1, 1) (1, 1, 1) 0.226
 S.A. Ghassemi, S. Danesh / Desalination 313 (2013) 44–50 49

Table 6 using TFNs, based on the evaluation of alternative desalination tech-

Final pairwise comparison of desalination technologies with respect to sub-criteria. nologies, the fuzzy evaluation matrix was established.
Criteria Sub-criteria IE VC MED MSF ED RO In the next step, using the same procedure as described in Section 2,
the desalination technologies were compared pairwised, with respect
C1 SC1 0.146 0.055 0.101 0.101 0.371 0.226
SC2 0.197 0.126 0.159 0.096 0.197 0.224 to each sub-criterion, and their weights were calculated. Due to the
C2 SC3 0.259 0.034 0.065 0.058 0.242 0.341 extensiveness of the results, the outcome of the pairwise evaluation
SC4 0.131 0.163 0.163 0.126 0.131 0.286 of the alternative technologies with respect only to the brine manage-
SC5 0.166 0.157 0.145 0.155 0.145 0.231
ment sub-criterion (SC1), is shown in Table 5. The final weights of the
SC6 0.165 0.000 0.000 0.000 0.500 0.335
SC7 0.236 0.108 0.068 0.107 0.284 0.196 desalination technologies with respect to all sub-criteria are presented
SC8 0.000 0.293 0.284 0.293 0.000 0.129 in Table 6.
C3 SC9 0.181 0.132 0.067 0.029 0.259 0.331 In the next step, using the results presented in Tables 4 and 6, and
SC10 0.189 0.078 0.136 0.082 0.225 0.290 the priority weights of the criteria, the weights of desalination technol-
ogies were calculated which are shown in Table 7. Finally, by following
the TOPSIS steps and calculations, the ranking of the desalination tech-
nologies was determined. The final ranking results are summarized in
The above fuzzy values were compared by using Eq. (6), and the
Table 8.
obtained degree of possibility values were as follows:
Based on the CCi (closeness coefficient) values, the ranking of the
desalination technologies, in descending order (from the most prefer-
V ðC1≥C2Þ ¼ 0:59 V ðC1≥C3Þ ¼ 0:97
able technology to the least one), was ED, RO, IE, VC, MED and MSF. As
a result, Electrodialysis (ED) with a CCi value of 0.7547, was the best
V ðC2≥C1Þ ¼ 1 V ðC2≥C3Þ ¼ 1 desalination technology for the study area included in this research.

V ðC3≥C2Þ ¼ 0:57 V ðC3≥C1Þ ¼ 1 : 5.4. Sensitivity analysis

Then, priority weights were calculated by using Eq. (8):
Decision makers may need sensitivity analysis to evaluate the
′
effect of changing the priority weights of the attributes on the ranking
d ðC1Þ ¼ minð0:57; 1Þ ¼ 0:57 of alternatives. To investigate the impact of changing criteria weights
′
d ðC2Þ ¼ minð1; 1Þ ¼ 1 on the selection of the best desalination technology, a sensitivity
′
d ðC3Þ ¼ minð0:59; 0:97Þ ¼ 0:59: analysis was performed in this research. For this purpose, the weights
of two decision attributes were changed in three different scenarios,
Priority weights with respect to the main goal form w′ = (0.57, 1, while the weights of the other attributes were kept constant. In
0.59) vector which should be normalized. After normalization, the other words, the weight of the first criterion C1 was changed with
priority weights were (0.264, 0.463, 0.273). As the results indicate, C2, and C3 sequentially. Moreover, the weights of C2 and C3 were ex-
the technical criterion (C2) was determined to be the most important changed while the weight of C1 remained the same. Then the close-
criterion in the desalination technology selection. Consistency ratio of ness coefficients (CCi) were calculated by using the TOPSIS method.
the pairwise comparison matrix was calculated to be 0.02 (b0.1). There- The results of the sensitivity analysis are shown in Table 9. It should
fore, the weights were shown to be consistent and as such, they were be mentioned that the first row of the table (Main) indicates the re-
used in continuation of the selection process. sults gained for the case study. The graphical representation of the
The similar calculations were performed for each group of sub- sensitivity analysis is illustrated in Fig. 4.
criteria, separately. The comparison tables are not all presented here. As the CCi values in Table 9 show, any exchange in the criteria
But, the calculated priority weights of the sub-criteria are shown in weights does not affect the results of the model. However, in this re-
Table 4. search, ED obtained the highest CCi value under all the conditions
It can be seen that the weight of the air pollution sub-criterion tested. Therefore, ED was selected as the most appropriate technolo-
(SC2) is zero. It means that this sub-criterion is not important in the gy for the desalination of brackish water in the study area.
analysis and can be omitted in further steps of evaluation.
6. Conclusions
5.3. Evaluation of alternatives and determination of the final rank
In this paper, by using fuzzy-AHP and TOPSIS, an integrated two-
At this stage of the decision procedure, the team members were step model was established to evaluate and select the optimum desa-
asked to establish the decision matrix, by pairwise comparison of lination technology among the processes e.g., IE, VC, MSF, MED, ED
the alternatives with respect to each sub-criterion, separately. Then and RO. The evaluation was based on various environmental, techni-
cal and economical criteria and sub-criteria. The model was then ap-
plied to a real world case study (desalination of a brackish water
Table 7 source with TDS of 2623 mg/lit), and its applicability and reliability
Weighted evaluation matrix of the desalination technologies.

Criteria Sub-criteria IE VC MED MSF ED RO

C1 SC1 0.0385 0.0145 0.0267 0.0267 0.0979 0.0597 Table 8

SC2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Final evaluation and ranking of the alternative desalination technologies.
C2 SC3 0.0104 0.0014 0.0026 0.0023 0.0097 0.0137 Alternatives di− Ranking d⁎i Ranking CCi Ranking
SC4 0.0116 0.0145 0.0145 0.0112 0.0116 0.0254
SC5 0.0056 0.0053 0.0049 0.0052 0.0049 0.0078 IE 0.0426 3 0.0720 3 0.3716 3
SC6 0.0207 0.0000 0.0000 0.0000 0.0627 0.0420 VC 0.0298 4 0.1013 6 0.2272 4
SC7 0.0201 0.0092 0.0058 0.0091 0.0242 0.0167 MED 0.0285 5 0.0971 4 0.2267 5
SC8 0.0000 0.0260 0.0252 0.0260 0.0000 0.0115 MSF 0.0278 6 0.1006 5 0.2159 6
C3 SC9 0.0247 0.0180 0.0091 0.0040 0.0353 0.0451 ED 0.1000 1 0.0325 1 0.7547 1
SC10 0.0258 0.0106 0.0186 0.0112 0.0307 0.0396 RO 0.0774 2 0.0406 2 0.6558 2
50 S.A. Ghassemi, S. Danesh / Desalination 313 (2013) 44–50

Table 9
The sensitivity analysis results.

Conditions Weights CCi values

W1 W2 W3 IE VC MED MSF ED RO

1 (Main) 0.264 0.462 0.273 0.3716 0.2272 0.2267 0.2159 0.7547 0.6558
2 0.273 0.462 0.264 0.3743 0.2300 0.2278 0.2157 0.7519 0.6617
3 0.264 0.273 0.462 0.4307 0.2177 0.1917 0.1371 0.7780 0.7377
4 0.462 0.264 0.273 0.3598 0.1427 0.1701 0.1567 0.8504 0.6009

[16] S. Balli, S. Korukoglu, Operating system selection using fuzzy AHP and TOPSIS
methods, Math. Comput. Appl. 14 (2009) 119–130.
[17] A.T. Gumus, Evaluation of hazardous waste transportation firms by using a two step
fuzzy-AHP and TOPSIS methodology, Expert Syst. Appl. 36 (2009) 4067–4074.
[18] S. Önütt, S. Soner, Transshipment site selection using the AHP and TOPSIS
approaches under fuzzy environment, Waste Manag. 28 (2008) 1552–1559.
[19] T. Yang, M.-C. Chen, C.-C. Hung, Multiple attribute decision-making methods for the
dynamic operator allocation problem, Math. Comput. Simul. 73 (2007) 285–299.
[20] İ. Ertuğrul, N. Karakaşoğlu, Performance evaluation of Turkish cement firms with
fuzzy analytic hierarchy process and TOPSIS methods, Expert Syst. Appl. 36 (2009)
702–715.
[21] G. Buyukozkan, O. Feyzioglu, E. Nebol, Selection of the strategic alliance partner in
logistics value chain, Int. J. Prod. Econ. 113 (2008) 148–158.
[22] Y. Chian-Son, A GP-AHP method for solving group decision-making fuzzy AHP
problems, Comput. Oper. Res. 29 (2002) 1969–2001.
[23] J.J. Buckley, Fuzzy hierarchical analysis, Fuzzy Set Syst. 17 (1985) 233–247.
[24] G.S. Liang, M.-J.J. Wang, Personnel selection using fuzzy MCDM algorithm, Eur. J.
Oper. Res. 78 (1994) 22–33.
[25] T.Y. Chen, T.C. Ku, C.W. Tsui, Determining attribute importance based on triangu-
Fig. 4. Model results based on the sensitivity analysis.
lar and trapezoidal fuzzy numbers in (z) fuzzy measures, in: The 19th Interna-
tional Conference on Multiple Criteria Decision Making, 2008, pp. 75–76.
were well demonstrated. Based on the model evaluation, ED with the [26] C. Kahraman, T. Ertay, G. Buyukozkan, A fuzzy optimization model for QFD
planning process using analytic network approach, Eur. J. Oper. Res. 171 (2006)
CCi value of 0.7547 was the most suitable technology for the case 390–411.
study in this research. The obtained CCi values for other technologies [27] M.T. Isaai, A. Kanani, M. Tootoonchi, H.R. Afzali, Intelligent timetable evaluation
(RO, IE, VC, MED, and MSF) were 0.6558, 0.3716, 0.2272, 0.2267, using fuzzy AHP, Expert Syst. Appl. 38 (2011) 3718–3723.
[28] Y. Ju, A. Wang, X. Liu, Evaluating emergency response capacity by fuzzy AHP and
0.2159, respectively. Results also showed that membrane-based tech-
2-tuple fuzzy linguistic approach, Expert Syst. Appl. 39 (2012) 6972–6981.
nologies (ED and RO) have priority over the distillation technologies. [29] M. Haghighi, A. Divandari, M. Keimasi, The impact of 3D e-readiness on e-banking
The sensitivity analysis indicated that changes made in the criteria development in Iran: a fuzzy AHP analysis, Expert Syst. Appl. 37 (2010) 4084–4093.
weights do not affect the final output of the model. In general, it can [30] D. Chang, Applications of the extent analysis method on fuzzy AHP, Eur. J. Oper.
Res. 95 (1996) 649–655.
be concluded that the developed two-step model can be applied reliably [31] L. Mikhailov, Deriving priorities from fuzzy pairwise comparison judgements,
to the decision making situations where there are many options avail- Fuzzy Set Syst. 134 (2003) 365–385.
able and the criteria for the final decision are uncertain and ambiguous. [32] P.J.M. van Laarhoven, W. Pedrycz, A fuzzy extension of Saaty's priority theory,
Fuzzy Set Syst. 11 (1983) 199–227.
[33] Y.-M. Wang, T.M.S. Elhag, Z. Hua, A modified fuzzy logarithmic least squares method
References for fuzzy analytic hierarchy process, Fuzzy Set Syst. 157 (2006) 3055–3071.
[34] M. Celik, I. Deha Er, A.F. Ozok, Application of fuzzy extended AHP methodology on
[1] A. Assem, Prioritizing desalination strategies using multi-criteria decision analy- shipping registry selection: the case of Turkish maritime industry, Expert Syst.
sis, Desalination 250 (2010) 928–935. Appl. 36 (2009) 190–198.
[2] D.E. Weiss, The role of ion-exchange desalination in municipal water supplies, [35] H. Deng, C.H. Yeh, Simulation-based evaluation of defuzzification-based ap-
Desalination 1 (1966) 107–128. proaches to fuzzy multiattribute decision making, in: IEEE Systems, Man and Cy-
[3] P. Vincke, Multicriteria Decision-aid, Wiley, 1992. bernetics, Part A: Systems and Humans, 2006, pp. 968–977.
[4] T.L. Saaty, The Analytic Hierarchy Process: Planning, Priority Setting, Resource [36] R. Zhao, R. Govind, Algebraic characteristics of extended fuzzy numbers, Inform.
Allocation, McGraw-Hill International Book Co., 1980. Sci. 54 (1991) 103–130.
[5] C.L. Hwang, K. Yoon, Multiple Attribute Decision Making: Methods and Applications: [37] S.-H. Tsaur, T.-Y. Chang, C.-H. Yen, The evaluation of airline service quality by
A State-of-the-art Survey, Springer-Verlag, 1981. fuzzy MCDM, Tour. Manag. 23 (2002) 107–115.
[6] B. Roy, The outranking approach and the foundations of ELECTRE methods, [38] U.R. Tuzkaya, S. Önüt, A fuzzy analytic network process based approach to
Theory Decis. 31 (1991) 49–73. transportation-mode selection between Turkey and Germany: a case study, Inform.
[7] J.P. Brans, P. Vincke, A preference ranking organisation method, Manag. Sci. 31 Sci. 178 (2008) 3133–3146.
(1985) 647–656. [39] Z. Ulukan, C. Ucuncuoglu, Economic analysis for evaluation of IS projects, Inf. Syst.
[8] E. Fontela, A. Gabus, The DEMATEL Observer, Battelle Institute, Geneva Research Technol. Manag. 7 (2010) 233–260.
Center, 1976. [40] J.M. Benitez, J.C. Martin, C. Roman, Using fuzzy number for measuring quality
[9] T.L. Saaty, Decision Making with Dependence and Feedback: The Analytic Net- of service in the hotel industry, Tour. Manag. 28 (2007) 544–555.
work Process: The Organization and Prioritization of Complexity, Rws Publica- [41] I. Chamodrakas, N. Alexopoulou, D. Martakos, Customer evaluation for order accep-
tions, 2001. tance using a novel class of fuzzy methods based on TOPSIS, Expert Syst. Appl. 36
[10] S. Opricovic, in: Multicriteria Optimization of Civil Engineering Systems, 2, Faculty (2009) 7409–7415.
of Civil Engineering, Belgrade, 1998, pp. 5–21. [42] C. Chen, C. Lin, S. Huang, A fuzzy approach for supplier evaluation and selection in
[11] M. Hajeeh, A. Al-Othman, Application of the analytical hierarchy process in the supply chain management, Int. J. Prod. Econ. 102 (2006) 289–301.
selection of desalination plants, Desalination 174 (2005) 97–108. [43] T.C. Chu, Y.C. Lin, A fuzzy TOPSIS method for robot selection, Int. J. Adv. Manuf.
[12] M.S. Mohsen, O.R. Al-Jayyousi, Brackish water desalination: an alternative for Technol. 21 (2003) 284–290.
water supply enhancement in Jordan, Desalination 124 (1999) 163–174. [44] F. Ye, Y.-N. Li, Group multi-attribute decision model to partner selection in the
[13] M.A. Hajeeh, Fuzzy approach for water desalination plants selection, in: 4th formation of virtual enterprise under incomplete information, Expert Syst. Appl.
IASME/WSEAS International Conference on Geology and Seismology, University 36 (2009) 9350–9357.
of Cambridge, UK, 2010, pp. 53–61. [45] H.-J. Shyur, H.-S. Shih, A hybrid MCDM model for strategic vendor selection,
[14] A. Bick, G. Oron, Post-treatment design of seawater reverse osmosis plants: boron Math. Comput. Model. 44 (2006) 749–761.
removal technology selection for potable water production and environmental [46] IRIMO, Meteorological Year Book of Iran, Ministry of Roads and Transportation,
control, Desalination 178 (2005) 233–246. Tehran, 2009.
[15] B. Ağırgün, Ranking B2C web sites with AHP and TOPSIS under fuzzy environment, [47] A.-S. Khalid Z, Precise way to select a desalination technology, Desalination 206
in: Fen Bilimleri Enstitüsü Dergisi, 1, Nevşehir Üniversitesi, 2012, pp. 65–78. (2007) 29–35.