You are on page 1of 17

Arab J Sci Eng (2014) 39:4405–4420

DOI 10.1007/s13369-014-1070-2

RESEARCH ARTICLE - CIVIL ENGINEERING

Selection of a Bridge Construction Site Using Fuzzy Analytical


Hierarchy Process in Geographic Information System
Abdollah Ardeshir · Nasir Mohseni ·
Kourosh Behzadian · Mark Errington

Received: 14 January 2013 / Accepted: 15 May 2013 / Published online: 4 April 2014
© King Fahd University of Petroleum and Minerals 2014

Abstract Selection of bridge construction sites over rivers Keywords Site location · Bridge construction · Analytical
is one of the most important tasks in construction feasibility hierarchy process · Fuzzy · GIS
studies. In this paper, potential sites for the construction of a
river bridge are ranked for suitability using analytical hierar-
chy process for a river in the northern part of Iran. Fuzzy logic
is used to incorporate the uncertainty associated with deci-
sion making into the model. Geographic information system
is used to facilitate the decision-making process by determin-
ing alternative sites and evaluating the selection criteria. A
floodplain is determined by the river reach using a hydraulic
simulation. This is done to identify the length and height of
potential bridges for each site. Two scenarios are analyzed
in this paper: the first ranks bridge construction sites regard-
less of the location of an existing bridge and the second ranks
bridge construction sites considering the existing bridge. The
result of this analysis shows that the existing bridge is one of
the best locations but not the best one.

A. Ardeshir · N. Mohseni
Department of Civil and Environmental Engineering,
Amirkabir University of Technology, P.O. Box: 15875-4413, 424,
Hafez Ave., Tehran, Iran 1 Introduction
e-mail: ardeshir@aut.ac.ir
River bridges play a fundamental role in road construction
K. Behzadian (B)
Environmental Research Centre, Amirkabir University
projects. This is due to the following reasons: (1) they facili-
of Technology, P.O. Box: 15875-4413, 424, Hafez Ave., tate transportation and establish vital routes between com-
Tehran, Iran munities located on both sides of the river; (2) there are
e-mail: k.behzadian-moghadam@exeter.ac.uk high fluctuations of construction costs for a river bridge from
K. Behzadian
one point to another; (3) excessively high maintenance costs
Centre for Water Systems, College of Engineering, Mathematics can occur in the future if a river bridge site is improperly
and Physical Sciences, University of Exeter, Exeter, UK selected. Bridges may be the most expensive item on a road
construction project and selection of an appropriate location
M. Errington
for bridge construction may lead to a huge saving in project
College of Engineering, Mathematics and Physical Sciences,
University of Exeter, UK expenditure. The location of the bridge also has an impact
e-mail: M.errington@exeter.ac.uk on the costs of fixing or replacing a bridge at a later date.

123
4406 Arab J Sci Eng (2014) 39:4405–4420

In feasibility studies selecting the most appropriate location fuzzy integral. Considering the uncertainty associated with
for river bridges can be as important as the design charac- input data can significantly improve the confidence of evalu-
teristics of the bridges themselves. The process of locating a ators in giving their judgments and the accuracy of selecting
river bridge consists of a comprehensive study of preliminary an optimal site for bridge construction.
engineering, hydrology and hydraulics, roadway alignment, Geographic information system (GIS) can be used as a tool
along with environmental and geological surveys [1]. for assessing the suitability of facility locations. They enable
The first step for locating a river bridge site is to iden- decision-makers to manage large volumes of spatial data
tify the main factors that affect the site location. Hydraulic, from a variety of sources and also facilitate the process of sep-
hydrological and morphological criteria are identified as pri- aration into suitable and unsuitable areas, and consequently,
mary factors as well as the construction cost of a project, identify potential site locations. Due to the increasing in
transportation of regional residents around the bridge, envi- the accuracy of spatial data gathering, GIS has been widely
ronmental and aesthetic factors. Due to the fact that effec- used for facility site locations in recent decades. Cheng and
tive factors have different levels of importance, a bridge O’Connor [21] investigated how to develop an automated
site location needs a systematic method to deal with this site layout system for temporary construction facilities using
multi-variable problem. A multiple criteria decision mak- GIS. The potential advantage of a GIS-based approach for
ing (MCDM) approach is best used to solve this variety site location is not only to reduce the time and cost of site
of decision-making, ranking facility site selection problems selection but also to provide a digital data bank for long-
influenced by some different criteria. This approach often term monitoring of the site [22]. GIS may, in turn, enable the
requires the decision-makers to provide qualitative and/or decision-maker to comprehensively evaluate a study area to
quantitative assessments for determining the performance identify other optimal alternatives.
of each alternative with respect to each criterion [2]. The Integration of GIS and multi-criteria decision analysis
MCDM technique is, therefore, a useful tool to rank the best (MCDA) can incorporate and transform geographic data
locations of bridge construction sites with respect to differ- (map criteria) and value judgments (decision-maker’s pref-
ent criteria. Some researchers detail their methods for optimal erences and uncertainties) to obtain an overall assessment of
site selection in the literature. Soroudi and Ehsan [3] define the decision alternatives [23,24]. At the time of writing, the
a new model for distributed generation planning and solve majority of GIS applications for construction site location
it using a modified NSGA method (Non-dominated Sorting used the integration between GIS and MCDA [25]. Mal-
Genetic Algorithm). They go on to state that this method czewski [25] showed in a comprehensive review that dur-
can be used in wide range of solutions of optimal location ing past two decades, there has been a significant interest
problems. in developing GIS-based MCDA in different fields. These
Analytic hierarchy process (AHP), initially proposed by include ecological sciences, urban-regional planning, waste
Saaty [4], has been widely used as a MCDM tool for locating management, hydrology and water resources, agriculture,
optimum construction sites in different applications such as forestry, natural hazards, recreation/tourism, housing/real
commercial centers [5], factories [6,7], railway stations [8], estate, geological sciences, manufacturing and cartography.
lime stone mines [9],and also for bridge risk assessments Recent applications of GIS-based MCDA for landfill con-
[10]. The traditional AHP method requires decision-makers struction site selection were presented by Guiqin et al. [26],
to explicitly state the exact value of their opinion in a pair- Sharifi et al. [27], Moeinaddini et al. [22] and Geneletti
wise comparison of alternatives [11]. This requirement is [28]. GIS–MCDA systems can potentially enhance group
hard to achieve in practice due the inherent uncertainty and decision-making processes by providing a flexible problem-
imprecision of the pair-wise comparison process [12]. To solving framework where participants can explore, under-
overcome this, fuzzy AHP was presented to solve such hierar- stand and redefine a decision making problem [29].
chical problems [13,14]. Examples of construction site selec- Despite the large number of studies in facility construction
tion problems where the fuzzy AHP method has been used site locations using GIS or MCDM, only few studies exist
are the location of fire stations within an international air- that relate to bridge construction. Hammad et al. [30] pro-
port [15], factories [16], shopping centers [17], convenience posed a methodology for bridge site and type selection using
stores [18] and mill location of a Textile Company [19]. GIS and an expert system in the early stage of road alignment
Fuzzy multiple criteria decision-making methods need to planning. They used GIS to extract the geographical data of
integrate various linguistic assessments and weights to evalu- possible sites and analyze them by the expert system to select
ate the suitability of a given location to determine an optimal possible span arrangements suitable for the site and the pos-
selection [20]. Ashrafi et al. [21] propose a model for per- sible bridge types. Finally, the overall assessment among the
sonnel selection using fuzzy linguistic variables with multi- sites is carried out based on the scaled total cost, driving com-
criteria decision making. They also developed their model by fort and landscape. Although this assessment is very useful
defining a new linguistic extension of the fuzzy measure and for road planners for rational decisions, this approach only

123
Arab J Sci Eng (2014) 39:4405–4420 4407

gives a rough idea of suitable bridge sites and types. The northern part of Iran. The methodology including the eval-
model does not include the hydraulic and hydrologic crite- uation criteria, alternatives and the proposed fuzzy AHP is
ria of the river which can have a major impact on bridge described in the following section. Following this, the result
construction costs and the suitability of different designs. of the proposed methodology application is presented and
GangaRao et al. [31] used value engineering principles to discussed. Finally, the relevant conclusions are drawn.
develop parameters for evaluation of different design scenar-
ios in low-volume bridge selection. In their study they fail
to address the issue of where to actually locate the bridges. 2 Methodology
Ostenfeld and Andersen [32] investigated the main crite-
ria used to select bridge sites and types in major bridge Figure 1 shows the flowchart of the proposed method for
projects. The bridge selection was more focused on the bridge ranking bridge construction sites using MCDM. As shown
type and site location. In their case studies this was done in the figure, the required stages for bridge site selection
by applying specified criteria locally for some pre-specified are divided into three main categories. The first stage is
locations. the development of AHP structure. This consists of defin-
A number of studies were observed in bridge construc- ing criteria and specifying alternatives. In the second stage, a
tion methods and bridge management using MCDA tech- river hydraulic model is developed by HEC-RAS1 hydraulic
niques. For instance, Sasmal and Ramanjaneyulu [33] used simulation model. HEC-RAS is public domain software by
a fuzzy AHP to evaluate the condition of existing rein- USACE and is used for hydraulic analysis of river flow [37].
forced concrete bridges. They claim that this methodology In the present analysis, HEC-RAS was used to model the sur-
helps the engineers and policy makers concerned with bridge face profile of river. The required layers in GIS based on the
management to overcome the problems related to prioriti- defined criteria are defined and provided. Fuzzy analytical
zation and to make decisions on funding related to reha- hierarchy process is performed in the final stage.
bilitation of bridges. In addition, Pan [34] analyzed fuzzy
AHP approach for choosing the best bridge construction
2.1 Main Criteria
method among three available methods. In addition, Zhao
and Chen [35] presented a fuzzy rule-based inference sys-
The main criteria are those which play a fundamental role
tem for bridge damage diagnosis and prediction. This aimed
in the selection of a bridge construction site. Groenier and
to provide bridge designers with valuable information about
Gubernick [1] conducted a comprehensive study about sig-
the impacts of design factors on bridge deterioration, which
nificant factors for choosing the best bridge construction
is an essential part of the efficient bridge management sys-
site. These factors were preliminary studies, river hydraulics,
tem. In addition, Wang and Elhag [36] proposed a fuzzy
hydrology, geomorphology of river, environmental issues
group decision-making approach for bridge risk assessment
and the effect of road alignment around a bridge construction
to evaluate bridge risk factors using linguistic terms rather
site. In our case, considering the regional conditions of the
than precise numerical values. These are Certain, Very High,
case study (which will be introduced later), three main cri-
High, Slightly High, Medium, Slightly Low, Low, Very Low
teria are taken into account: (1) transportation criterion; (2)
or None.
economic criterion; (3) morphology criterion. These criteria
At the time writing, the authors could not find any other
are described in more details in the following section. It is
systematic approach in the literature review for the bridge
important to note that the importance of the other criteria with
site location. Fuzzy-AHP is a proven technique for facility
regard to cost have been assumed to be negligible compared
selection problems which appears to be promising for use in
to the main criteria listed above (Road Safety Manual 2005).
this application. This MCDA technique coupled with GIS is
They have, therefore, been discarded for the purposes of this
able to create an organized framework to achieve a substan-
study.
tially precise method in the site location of the river bridge.
In other words, GIS environment reinforces the MCDA tech-
nique through extracting some morphological and geotechni- 2.1.1 Transportation Criterion
cal specifications related to hydrologic and hydrologic analy-
sis of the potential sites, thus leading to a more careful and It is assumed that a bridge construction site is appropriate if it
detailed analysis of the model. Consequently, we believe minimizes the total distance traveled by the regional residents
that the innovative integration of fuzzy AHP and GIS is crossing the river. For this reason for each bridge construction
able to enhance the procedure of site selection for river alternative, the total distance traveled from each residential
bridges. area of one river side to all public centers of another river side
In this paper, a fuzzy AHP technique is applied to rank
bridge construction site alternatives in a river located in the 1 Hydrologic Engineering Centers-River Analysis System.

123
4408 Arab J Sci Eng (2014) 39:4405–4420

Start

11- Performing 2- Developing hydraulic simulation model 1- Developing AHP


fuzzy-AHP in GIS structure

Transportation
3- Gathering data criterion
12- Weighting alternatives
with respect to each criteria
Economic 1-1- Defining
4- Providing required layers criterion criteria
according to the criteria
13- Performing fuzzy pair-wise
Morphology
comparison of the criteria by
criterion
linguistic terms
8- Preparing imput file for 1-2- Specifying
hydraulic model simulation alternatives
(PRE-RAS)

14- Determining overall weights of 5- Dividing the


alternatives (ranking best sites for 9- Runing hydraulic model river reaches
bridge construction) simulation (HEC-RAS)
6- Removing restricted divisions
(infeasible alternatives)

End 10- Determining the length


and the height of bridge for
each divisions (alternatives) 7- Identifing final divisions
(select alternatives)

Fig. 1 The flowchart of the proposed method

are calculated, and their sum is considered as the transporta- side of the river; N1 = the number of public centers located
tion criterion of that alternative. The transportation criterion at the northern side of the river; N2 = the number of public
for each bridge site alternative can, therefore, be written as centers located at the southern side of the river. Also, the
coefficient βi j is calculated as

N1 
M1 
N2 
M2
Dk = di jk × βi j + di jk × βi j , (1)
j=1 i=1 j=1 i=1 Pi Wj
βi j =  M × N , (2)
1 1
k=1 Pk l=1 Wl
where Dk = weight of bridge construction site k for trans-
portation criterion; di jk = the shortest road distance between
residential area i at the southern side of the river and pub- where Pi = population of residential area i at the southern side
 M1
lic center j at the northern side of the river through bridge of the river; k=1 Pk = population of all residential areas at
construction site k; di jk = the shortest road distance between the southern side of the river; W j = importance weight of
 N1
residential area i at the northern side of the river and public public center j at the northern side of the river; l=1 Wl =
center j at the southern side of the river through bridge con- sum of importance weights of public centers at the northern
struction site k; βi j = coefficient indicating the population side of the river. Note that these weights are specified by a
of residential area i at the southern side of the river and the transportation expert with respect to the type of public center.
importance of public center j at the northern side of the river; Thus, coefficient βi j is calculated the same as in (2) taking
βi j = coefficient indicating the population of residential area into account the population of residential areas at the northern
i at the northern side of the river and the importance of public side of the river and public centers at the southern side of the
center j at the southern side of the river; M1 = the number river. Obviously, the alternatives with lower weights (Dk )
of residential areas located at the southern side of the river; provide shorter daily transportation trips for residents of one
M2 = the number of residential areas located at the northern river side to have access to public centers of another river side,

123
Arab J Sci Eng (2014) 39:4405–4420 4409

thereby giving higher priority from transportation criterion perspective. This is due to that fact that the river course is
perspective. liable to change over time in these sections. When the coef-
ficient is equal to 1, this means that the relevant section has
no meandering and bending. The natural river route is in
2.1.2 Economic Criterion
a straight line which is an ideal site location. In situations
where the value of the coefficient is a lot greater than one
The most significant issues available in economic criterion
the meandering of river is larger and the site is less suitable
include the cost of purchasing the land beside the river at the
for bridge construction. The sinus coefficient of each alter-
bridge site, the construction cost of the bridge and the con-
native is calculated for the arch of the river where the bridge
struction cost of the required access roads from the nearest
is more likely to be built. Finally, the coefficients are normal-
main roads to the bridge. As most of these costs are simi-
ized with respect to each other to generate weightings for this
lar for all potential bridge sites, only the construction cost
criterion.
of a bridge and access road is considered as the economic
criterion.
In order to calculate the total cost of bridge and access 2.2 Alternatives
road construction, the characteristics of the road and the
bridge need to be estimated. In order to approximately esti- To specify potential bridge construction sites, the river needs
mate the construction cost of the bridge, the height and the to be divided into smaller divisions called alternatives. Two
length of the bridge need to be specified. Feasible and practi- main factors directly effective in making divisions are (1)
cally acceptable steps for obtaining these data are as follows equality and/or similarity in the characteristics of geomor-
[1]: (1) carrying out hydrological analyses including the fre- phology, topography and geology aspects; (2) land use
quency analysis of flood and consequently, determining a around the river. Each alternative is therefore made up of
design flood with a specified return period; (2) developing a part of the river, in which an individual bridge site is avail-
a hydraulic simulation model to calculate the floodplain and able and where every other point through that part has the
therefore, estimating the height and length of the bridge. The same criteria values.
HEC-RAS model was used as the hydraulic simulation model In addition to this, alternatives failing to meet minimum
in conjunction with GIS environment to better estimate the technical requirements for access road and bridge construc-
river floodplain [37]. The result of hydraulic analysis and tion are removed from the pool of available alternatives. The
floodplain determination is the estimation of the height and main decision factors at this stage are the minimum techni-
length of the bridge for each alternative. The construction cal conditions which consist of the limitations of the con-
cost of the access road on both sides of the river can be struction project. These are as follows: (1) large heights of
estimated based on the preliminary design of the road with the bridge column above the river leading to high construc-
respect to the local road network code [38]. tion cost; (2) cross sections which require a large amount
In order to take into account the soil effect of the bridge of cut and fill tunnels to reach the proposed bridge site; (3)
foundation, two types of foundations are assumed in the cliffs with steep slopes on one or both sides of the river;
analysis: (1) a shallow foundation for the sites with strong (4) long bridge span with respect to standard slope in the
soils; (2) a deep foundation for the site with loose and muddy roads reaching the bridge or the existence of multiple val-
soils. The cost of bridge building will, therefore, vary with leys in the waterway cross section of the proposed site.
respect to the soil type beneath the bridge foundations. When removing constrained and unfeasible alternatives for
bridge construction, final alternatives (select divisions) are
extracted as potential options and are taken forward for final
2.1.3 Morphology Criterion
screening.
Where a section of a river meanders, bridge building is more
difficult. Most rivers usually meander within their route; how- 2.3 Required Layers in GIS and Hydraulic Analysis
ever, bridge sites should be located in the sections with less
meandering [1]. In order to consider meandering conditions After defining the criteria and specifying the alternatives, the
in our model, a sinus coefficient between two points of the required data for assigning weights to the alternatives needs
river is defined as the ratio of meandering distance (i.e. nat- to be collected and analyzed. As the data and their analysis are
ural river route) to non-meandering distance. It is assumed mostly spatial, GIS is used to manage and process the data.
that the two points are the start and the end of the site loca- The data required are topographical maps, the population and
tion. The coefficient, therefore, always takes a value of 1 or their distribution in residential areas, the distribution of res-
greater. The larger the coefficient is for a possible site loca- idential areas and public centers, main and secondary roads
tion, the less favorable the site location is from morphology maps, river maps and geology and land use map. These data

123
4410 Arab J Sci Eng (2014) 39:4405–4420

Table 1 Linguistic terms for fuzzy pair-wise comparisons of the criteria


Verbal judgment of preference Description Fuzzy number

Very unimportant (VU) Criterion A is very unimportant compared with criterion B (0, 0, 1, 2)
Less important (LI) Criterion A is less important than criterion B (1, 2.5, 4)
Equal important (EI) Criterion A is as important as criterion B (3, 5, 7)
More important (MI) Criterion A is more important than criterion B (6, 7.5, 9)
Very important (VI) Criterion A is very important compared with criterion B (8, 9, 10, 10)

need to be entered as different layers in the GIS environment. functions varying between 0 and 10 [16]. Note that except
In addition, hydraulic analysis is performed through the river for the statement of “Equal Importance”, the other statements
reach using a HEC-RAS hydraulic simulation model [37]. are converse in pairs. Two steps are required when dealing
This is also carried out within the GIS environment and has with fuzzy attributes: first converting linguistic terms to fuzzy
the aim of identifying the height and length of the proposed numbers; and second, assigning crisp scores to fuzzy num-
bridge for each alternative site. bers [39]. Thus, after comparing the criteria with linguistic
terms by evaluators, the first step is taken by replacing fuzzy
numbers for subsequent calculations.
2.4 Fuzzy AHP To investigate different levels of imprecision associated
with fuzzy numbers of evaluators’ opinions, three α-cut
The base method of fuzzy AHP used for this work has been values equal to α=0, 0.5 and 1 are used to represent pes-
extracted from Buckley’s study [14]. Certain improvements simistic, moderate, and optimistic opinions of the evaluators,
have been made to some parts of Buckley’s approach. These respectively. Note that α=0 and 1 indicate that maximum
will be described in more detail later in this paper. In order and minimum fuzzy (no fuzzy) membership functions are
to incorporate the uncertainty associated with the experts’ used, respectively. Five aforementioned membership func-
judgments, triangular and trapezoidal fuzzy numbers are tions considering the α-cut concept can be written by math-
used. ematical Eqs. (3)–(7):
When comparing alternatives with respect to the criteria ⎧
discussed in the previous section, the crisp value is calcu- ⎨ X α,L = 0

0.5+0.5×(X α,R −1)[1+0.33×(X α,R −1)]
lated as the weight of alternatives for each criterion. There X (α)V U = X α,M = 0.5X α,R +0.5


is, therefore, no need to use a ‘fuzzy’ value. When perform- X α,R = 2 − α
ing pair-wise comparisons of the criteria with respect to the (3)
main objective, however, the following fuzzy values need to ⎧
be defined. ⎨ X α,L = 1 + 1.5α
Five linguistic terms were used to represent fuzzy pair- X (α) L I = X α,M = 2.5 (4)

wise comparisons of the criteria. These terms are given in X α,R = 4 − 1.5α
Table 1 along with the relevant description and the associated ⎧
fuzzy numbers. Figure 2 represents the defined membership ⎨ X α,L = 3 + 2α
X (α) E I = X α,M = 5 (5)

X α,R = 7 − 2α

1 ⎨ X α,L = 6 + 1.5α
0.9
X (α) M I = X α,M = 7.5 (6)
0.8 ⎩
0.7
VU X α,R = 9 − 1.5α
LI
0.6
EI
µ(x)

0.5 MI ⎧
0.4 VI

⎨ X α,L = 8 + α
1.5+0.67×(4.5−0.5X α,L )(9−X α,L )
0.3
X (α)V I = X α,M = 8 +
0.2 ⎪
⎩X
5.5−0.5X α,L
0.1 α,R = 10
0 (7)
0 2 4 6 8 10
x In fuzzy AHP, each element of the pair-wise comparison
matrix is presented as fuzzy numbers. Equation (8) shows a
Fig. 2 Fuzzy numbers for different linguistic terms typical matrix of fuzzy pair-wise comparison. For instance,

123
Arab J Sci Eng (2014) 39:4405–4420 4411

in this Eq. x 12,U , x 12,M and x 12,L indicate up, middle and where wi j = normalised weight of alternative j with respect
low limits of pair-wise comparison between element 1 and to criterion i; n= the number of criteria and OW j = overall
element 2, respectively. weight of alternative j.

⎡ ⎤
1 (x12,L , x12,M L , x12,MU , x12,U ) ······ (x1n,L , x1n,M L , x1n,MU , x1n,U )
⎢ (x21,L , x21,M L , x21,MU , x21,U ) ······ (x2n,L , x2n,M L , x2n,MU , x2n,U ) ⎥
Ãn×n = ⎢



······ ······ ······ ······
(xn1,L , xn1,M L , xn1,MU , x21,U ) ······ ······ 1
(8)

To normalize fuzzy numbers in the fuzzy pair-wise com- 3 Results and Discussion
parison matrix of the criteria, the geometric mean of the data
is calculated as follows: given a fuzzy number (a, b, c, d) for The case study is a reach of Dough River, which belongs
each element of the matrix, the normalized fuzzy number of to the river basin of Banial in Golestan Province located in
each criterion is calculated through Eqs. (9)–(11): the north of Iran (Fig. 3). The river is the longest tributary
⎛ ⎞1/n of Gorganrud River located in Golestan and North Khorasan
n Provinces. The river reach is approximately 15 km long and
ai = ⎝ ai j ⎠ for all i = 1, 2, . . . , n (9) located at the longitude between 55◦ 26 and 55◦ 32 east and
j=1 latitude between 37◦ 21 and 37◦22 north. Figure 3 shows

n the area map with the scale of 1:50,000.
a= ai (10) To specify the alternatives, the river reach was first divided
i=1 into 30 parts with respect to the factors for divisions discussed
bi , b, ci , c, di and d, are similarly defined for other earlier in the paper. These include similar physiographic and
fuzzy numbers. geomorphologic data (see Fig. 4). It would seem appropriate
  to take into consideration all river divisions over the entire
ai bi ci di river reach. This was not done as based on the road and bridge
μi (z) = , , , for all i = 1, 2, . . . , n (11)
d c b a code [38]; the construction of bridges is infeasible for some of
the divisions. This is due to the main requirements specified
where ai = geometric mean of the first elements of fuzzy in the code such as maximum road slope approaching the
numbers related to the comparison of criterion i to other bridge and the maximum meandering of river at bridge site.
criteria j = 1, . . . , n; ai j = first element of the fuzzy number As a result, as a preliminary refining stage, the alternatives
related to the comparison of criterion i to criterion j; and not meeting the minimum requirements of the code were
μi (z)= normalized fuzzy value of criterion i. Other elements removed from the possible alternatives before moving to the
are defined using a similar method. next step. This had the effect of eliminating 16 divisions from
After normalizing fuzzy numbers, the second step (i.e. 30 primary divisions. The remaining 14 divisions are shown
assigning crisp scores to fuzzy numbers) must be performed. in Fig. 5. If this had not been done, retaining the eliminated
For fuzzy group decision-makings, the judgment of all divisions may have led to high ranks for unfeasible bridge
experts need to be aggregated and converted into one crisp construction sites.
score (defuzzification) [40]. The center of gravity (COG) As an existing bridge, known as Kouseh Bridge, is present
technique was used to assign a crisp score to each crite- in the river reach and is located in division 20, two different
rion considering the group decision making process. This scenarios are analyzed here: (scenario 1) ranking bridge con-
is shown as follows [36]: struction sites regardless of the existing bridge and (scenario
K  2) ranking bridge construction sites considering the exist-
k=1 ( μi (z).zdz)k
z i∗ = K  for all i = 1, 2, . . . , n (12) ing bridge. The difference between the two scenarios is in
k=1 ( μi (z)dz)k the calculation of the transportation criterion. In scenario 2,
when finding the shortest route for each bridge construction
where K = the number of experts and z i∗ = weighted average alternative, the route passing from each bridge construction
(crisp score) of criterion i. Finally, overall weight of alterna- alternative must always be compared with the route passing
tive j can be written as from the existing bridge.
Figure 6 depicts residential areas and public centers in the

n
OW j = z i∗ × wi j , (13) study area as well as main access road around the river. As
i=1 it can be seen, the number of residential areas located at the

123
4412 Arab J Sci Eng (2014) 39:4405–4420

Fig. 3 Layout of case study of


Dough River

Fig. 4 Primary divisions of the


river reach

northern and southern sides of the river is M2 = 8 and M1 = 9, (Pi ) and the importance weight of the public centers (W j ) in
respectively. The public centers include one sport complex Eq. (2) are given in Tables 2 and 3, respectively. Note that
located at the northern side of the river (N1 = 1) and two the importance weights are based on the ranking on a scale
primary schools, one secondary school, one shopping center of 1–10 specified with the aid of a sociology expert. Given
(store), two clinics and one pharmacy at the southern side of the data in Tables 2 and 3, coefficients βi j and βi j can be
the river (N2 = 7). The population of these residential areas calculated for each residential area i and public center j.

123
Arab J Sci Eng (2014) 39:4405–4420 4413

Fig. 5 Select divisions


(alternatives) after removing
unfeasible divisions

Fig. 6 Location of select


cross-sections for bridge
construction, residential areas
and public centers


Thus, Eqs. (1)–(2) are used to calculate the weights of center on another river side for coefficients di jk and di jk ) is
alternatives for transportation criterion. 14 divisions are com- calculated by measuring the direct route of those points pass-
pared in scenario 1 while 13 divisions are compared in sce- ing the alternative bridge site. In scenario 2, however, after
nario 2. This is because the existing bridge division is dis- measuring the distance of the direct route between two points
carded in scenario 2. For each alternative (i.e. bridge con- passing the alternative, the distance is compared with the
struction site), the connecting distance between two points route passing Kouseh Bridge (existing bridge). The shorter
(i.e. from one residential area on one river side to each public route is then taken into account as the final route of con-

123
4414 Arab J Sci Eng (2014) 39:4405–4420

Table 2 Residential area population


Location Population Location Population

NRA 1 70,600 SRA 1 10,600


NRA 2 23,000 SRA 2 17,800
NRA 3 11,000 SRA 3 33,600
NRA 4 14,000 SRA 4 13,600
NRA 5 60,000 SRA 5 7,900
NRA 6 59,000 SRA 6 55,600
NRA 7 131,600 SRA 7 84,200
NRA 8 93,400 SRA 8 68,300
SRA 9 28,600
NRA northern residential area, SRA southern residential area

Table 3 Importance weight of public centers


Public center Importance weight

School 9
Clinic 5 Fig. 7 A typical bridge assigned for a cross-section of the river in flood
Pharmacy 4 simulation by HEC-RAS model
Shopping center (store) 3
Sport complex 2
Sinus coefficients were used as the morphology criterion
and were calculated for each division. Starting and finishing
points of the division were considered when calculating the
necting those two points when calculating Eq. (1) for the ratio of meandering distance to non-meandering distance.
alternative. Thus, the weight of each alternative bridge con- The sinus coefficients that were calculated for each division
struction site for transportation criterion is then calculated. are shown in the last column of Table 4.
The normalized weights of this criterion are shown in Table In order to decrease the effect of personal and biased judg-
9 for both scenarios. ments of experts, ten experts with different expertise includ-
The weight of alternatives for economic criterion is calcu- ing economics, hydraulics and transportation were invited to
lated based on total construction cost. This consists of (1) the participate in the survey. Table 5 gives a summary of char-
cost of the road construction from the bridge location to the acteristics of the ten experts participating in the survey. The
nearest main access road around the river and (2) the cost of selection of experts was carried out to cover the whole range
bridge construction with respect to the length and the height of relevant expertise with various ages, work experiences and
of the bridge and the type of the bridge foundation. academic degrees. The responses recorded during this survey
In order to calculate the bridge construction cost associ- are shown in Table 6. The normalized fuzzy values of the cri-
ated with a given site of alternative, appropriate dimensions teria for each expert’s judgment were calculated using Eqs.
of a typical bridge which can be fitted against simulated flood- (9)–(11), which are given in Table 7. These results were con-
plain is first identified (Fig. 7). Bridge dimensions are deter- verted to crisp scores by using Eq. (12) based on the COG
mined by HEC-RAS hydraulic model which is used for the technique. The resulting scores for the three criteria under
simulation of river floodplain. The length and height of the different levels of imprecision are presented in Table 8.
candidate bridges in each division are, therefore, estimated To give an example, the calculations related to the first
based on the simulation of floodplain in the cross section of expert (EX1) relative to the criteria are described here in
each division (Table 4). The bridge foundations constitute more detail. The pair-wise comparison matrix of the criteria
another large part of the total bridge costs. The foundation is as follows based on the linguistic terms of Table 6 and the
type for each division was determined based on the soil type opinions of the first expert:
extracted from GIS for each division. The soil and foundation
types for each division are given in Table 4. The associated
⎡ ⎤
eco trans mor ph
bridge cost for each division was then calculated based on eco EI VI VI
the bridge specifications collected in Table 4 and based on A3×3 = trans ⎣ VU EI EI ⎦ ,
recommendations in the road and bridge code [38]. mor ph VU EI EI

123
Arab J Sci Eng (2014) 39:4405–4420 4415

Table 4 Specification of divisions and the respective potential bridges


Division no. Alternative no. Bridge length (m) Bridge height (m) Soil type Foundation type Sinus coefficient

1 2 214.5 3.34 Marl and silt Deep 1.65


2 3 216 3.15 Marl and silt Deep 1.16
3 5 171.7 3.79 Marl and silt Deep 1
4 11 165.5 4.45 Marl and silt Deep 1
5 18 132.5 3.76 Marl and silt Deep 1
6 19 146.5 5.07 Marl and silt Deep 1.1
7 20 91.3 3.9 Marl and silt Deep 1.46
8 23 206 3.25 Fluvial deposits Shallow 1.06
9 24 117 3.05 Marl and silt, Fluvial deposits Deep 1.18
10 25 223 4.81 Fluvial deposits, quartz Shallow 1.95
11 26 133.5 3.97 Marl and silt, Fluvial deposits Deep 1.49
12 27 131.4 2.57 Fluvial deposits, quartz Shallow 1.3
13 28 215 2.63 Fluvial deposits Shallow 1.7
14 29 175.5 3.08 Fluvial deposits Shallow 1.24

Table 5 Specifications of experts involved in the survey


Expert no. Expertise Age (years) Academic degree Work experience (years)

Ex1 Economic expert 55 BSc 25


Ex2 Project manager 42 MSc 15
Ex3 Bridge engineer 40 MSc 15
Ex4 Project manager 55 BSc 28
Ex5 Transportation engineer 42 MSc 18
Ex6 Traffic expert 38 MSc 10
Ex7 Road and housing expert 35 BSc 12
Ex8 Hydrology engineer 42 PhD 13
Ex9 Hydraulic engineer 40 PhD 12
Ex10 Bridge engineer 45 MSc 20

Table 6 Linguistic terms of experts’ judgments in pair-wise compar- where eco = economic criterion; trans = transportation crite-
isons of the criteria rion; and morph = morphology criterion. Note that, an expert
Expert no. Comparison of the criteria is required to fill either upper or lower diagonal pair-wise
Transportation Transportation Economic comparisons. The other verbal judgments can be completed
to economic to morphology to morphology automatically for this matrix. For instance, each element of
the lower diagonal (e.g. VU in the above matrix) is the oppo-
Ex1 VU EI VI
site of the symmetrical element of the upper diagonal (e.g.
Ex2 LI EI VI VI in the same matrix). For this reason only three elements
Ex3 VU LI MI of these pair-wise comparisons were shown for each expert
Ex4 EI MI VI in Table 6. Full matricides of pair-wise comparisons were
Ex5 MI VI EI shown in this example for descriptive purposes for the rest
Ex6 VI MI LI of the analysis. In the first step, linguistic terms are replaced
Ex7 MI EI LI with fuzzy numbers shown in the following matrix:
Ex8 MI LI VU ⎡ ⎤
Ex9 LI VU LI
(3, 5, 7) (8, 9, 10, 10) (8, 9, 10, 10)
Ex10 EI LI LI
A3×3 = ⎣ (0, 0, 1, 2) (3, 5, 7) (3, 5, 7) ⎦
(0, 0, 1, 2) (3, 5, 7) (3, 5, 7)

123
4416 Arab J Sci Eng (2014) 39:4405–4420

Table 7 Normalized fuzzy values of the criteria for Experts’ judgments


Expert Criteria
Transportation Economic Morphology

Ex1 (0, 0, 0.399, 0.801) (0.319, 0.535, 1.072, 1.53) (0, 0, 0.399, 0.801)
Ex2 (0.111, 0.282, 0.364, 0.790) (0.277, 0.492, 0.658, 1.162) (0, 0, 0.269, 0.629)
Ex3 (0, 0, 0.204, 0.487) (0.281, 0.493, 0.626, 1.082) (0.141, 0.323, 0.396, 0.800)
Ex4 (0.197, 0.399, 0.485, 0952) (0.216, 0.423, 0.533, 0986) (0, 0, 0.198, 0.482)
Ex5 (0.277, 0.492, 0.658, 1.162) (0.111, 0.282, 0.364, 0.790) (0, 0, 0.269, 0.629)
Ex6 (0.281, 0.493, 0.625, 1.082) (0, 0, 0.204, 0.487) (0.141, 0.323, 0.396, 0.800)
Ex7 (0.190, 0.392, 0.392, 0.838) (0.073, 0.217, 0.217, 0.533) (0.190, 0.392, 0.392, 0.838)
Ex8 (0.141, 0.323, 0.396, 0.800) (0, 0, 0.204, 0.487) (0.281, 0.493, 0.625, 1.082)
Ex9 (0, 0, 0.204, 0.487) (0.141, 0.323, 0.396, 0.800) (0.281, 0.493, 0.625, 1.082)
Ex10 (0.106, 0.274, 0.274, 0.647) (0.106, 0.274, 0.274, 0.647) (0.240, 0.451, 0.451, 0.919)

Table 8 Crisp scores of the criteria (z*) under different levels of impre- fuzzy values of all experts for each criterion are aggregated
cision using Eq. (12) to calculate crisp values. These are shown in
z* Criteria Table 8. It is important to note that to calculate the integral
Transportation Economic Morphology in Eq. (12), the integral needs only to be calculated as a first-
order equation. The final value is then calculated by replacing
α=0 0.397 0.457 0.364 numerical values for the limits of integral solution.
α = 0.5 0.339 0.438 0.318 The weights of each alternative were calculated for both
α=1 0.292 0.428 0.279 scenarios with respect to each criterion as shown in Table 9.
In order to better compare the criteria, dimensionless forms
of absolute weights are used through normalizing by the best
In order to normalize the result, the matrix of up ( AU ), weight for each criterion. As an example, division 20 is the
middle upper ( AMU ), middle lower ( AML ) and low (AL ) lim- best division in scenario 1 with respect to economic crite-
its of pair-wise comparison are first created as follows: rion. Therefore, the weight of this alternative is 1 and the
⎡ ⎤ ⎡ ⎤ weights of other alternatives are less than 1. Finally, the over-
3 8 8 5 9 9 all weights of the alternatives can be calculated by Eq. (13)
AL = ⎣ 0 3 3 ⎦ AML = ⎣ 0 5 5 ⎦ in which their elements are calculated. These are shown in
0 3 3 0 5 5 Tables 4 and 5. Table 10 represents the overall weights of the
⎡ ⎤ ⎡ ⎤
5 10 10 7 10 9 alternatives under different levels of imprecision (α= 0, 0.5
AMU = ⎣ 1 5 5 ⎦ AU = ⎣ 2 7 7 ⎦ and 1) for both scenarios. In this Table, for each scenario, the
1 5 5 2 7 7 alternatives are ranked with respect to their relevant overall
weights. For example, it can be seen that construction sites
The values of ai and a in low limit matrix (AL ) are calcu-
(divisions) no 18 and 5 are the top sites with respect to sce-
lated based on Eqs. (9) and (10) as
narios 1 and 2 respectively.
a1 = (3 × 8 × 8)1/3 = 5.67 a2 = (0 × 3 × 3)1/3 = 0 The following can be noted from the results in scenario
1: (1) division 18 has higher priority for bridge construction
a3 − (0 × 3 × 3)1/3 = 0 a = 5.67 + 0 + 0 = 5.67
compared to other 14 divisions and should be considered
Thus, the values of bi and b for middle lower limit matrix as the best site location for a bridge construction site over
(AML ), ci and c for middle upper limit matrix ( AMU ) and di Dough River; (2) although Kouseh Bridge (Division 20) is in
and d for up limit matrix (AU ) are calculated similarly. The the first rank with respect to both economic and transporta-
fuzzy values μi (z) of the first expert’s opinion are, therefore, tion criteria, division 18 is finally selected as the best one
equal to (0.319, 0.535, 1.072, 1.53), (0, 0, 0.399, 0.801), for bridge construction site among 14 divisions because the
and (0, 0, 0.399, 0.801) for the criteria of economic, trans- weight of division 20 with respect to morphology criterion
portation and morphology, respectively. For instance, 0.319 is very low compared to division 18; (3) divisions 18, 20 and
is equal to ai (5.67) divided by d (17.768). These calculations 19 are the top three sites for bridge construction with a rela-
are repeated for each criterion for all experts. The resulting tively significant improvement compared to other divisions.
values are shown in Table 7. In the next step, the resulting This indicates the superiority of these locations with respect

123
Arab J Sci Eng (2014) 39:4405–4420 4417

Table 9 Normalized weights of alternative with respect to the criteria in both scenarios
Alternative no. Division no. Weights of alternatives in scenario 1 Weights of alternatives in scenario 2
Economic Transportation Morphology Economic Transportation Morphology

1 2 0.698 0.439 0.606 0.711 0.89 0.606


2 3 0.611 0.445 0.862 0.622 0.902 0.862
3 5 0.892 0.546 1 0.907 0.985 1
4 11 0.745 0.711 1 0.758 1 1
5 18 0.85 0.938 1 0.865 0.917 1
6 19 0.765 0.978 0.909 0.778 0.921 0.909
7 20 1 1 0.684 − − −
8 23 0.592 0.863 0.943 0.602 0.908 0.943
9 24 0.628 0.847 0.847 0.639 0.905 0.847
10 25 0.464 0.780 0.512 0.473 0.901 0671
11 26 0.547 0.768 0.671 0.556 0.9 0.671
12 27 0.69 0.732 0.769 0.702 0.897 0.769
13 28 0.633 0.79 0.588 0.644 0.911 0.588
14 29 0.982 0.665 0.804 1 0.952 0.804

Table 10 Overall weights of 14 alternatives in scenario 1 and 13 alternatives in scenario 2


Alternative no. Division no. Overall weights of the alternatives Overall weights of the alternatives
in scenario 1 in scenario 2
α=0 α = 0.5 α=1 α=0 α = 0.5 α=1

1 2 0.7139 0.6473 0.5961 0.8988 0.8058 0.7332


2 3 0.7697 0.6926 0.6319 0.9561 0.8523 0.7700
3 5 0.9884 0.8938 0.8202 1.1695 1.0492 0.9548
4 11 0.9867 0.8853 0.8055 1.1074 0.9890 0.8954
5 18 1.1248 1.0083 0.9167 1.1234 1.0077 0.9170
6 19 1.0687 0.9557 0.8667 1.0521 0.9420 0.8556
7 20 1.1030 0.9945 0.9108 – – –
8 23 0.9564 0.8517 0.7684 0.9788 0.8714 0.7859
9 24 0.9316 0.8315 0.7524 0.9596 0.8560 0.7740
10 25 0.7081 0.6304 0.5692 0.8181 0.7260 0.6527
11 26 0.7992 0.7133 0.6456 0.8556 0.7620 0.6880
12 27 0.8859 0.7949 0.7236 0.9568 0.8561 0.7769
13 28 0.8169 0.7320 0.6657 0.8700 0.7779 0.7057
14 29 1.0054 0.91122 0.8388 1.1276 1.0164 0.9303

to the criteria; (4) division of Kouseh Bridge (division 20) importance of this division with respect to transportation
is recognized as the second most optimal location in fuzzy and morphology criteria caused it to be selected as the best
AHP. This implies that this site, which has been selected division.
in the past, has not been the best location for bridge con- The following can be noted from the results in scenario
struction, although its current rank proves that a relatively 2: (1) division 5, which is 7.5 km far from Kouseh Bridge
good selection was made in the past; (5) among top five divi- (division 20), is in the highest ranking location among 13
sions, the region containing three divisions 18, 19 and 20 divisions and is recognized as the best division for bridge
are in close proximity to each other with a distance of 500 construction with respect to the existing bridge; (2) although
m. This signifies relative superiority of this region compared division 11 is in the highest ranking location with respect to
to other areas in the case study; (6) while the rank of divi- two out of three criteria. Division 5 holds the highest rank for
sion 18 is three with respect to economic criterion, the great bridge site location because its weights for all three criteria

123
4418 Arab J Sci Eng (2014) 39:4405–4420

Table 11 Relative frequency, given as a percentage, of ranking of five top sites in 10,000 times simulation of Fuzzy-AHP considering two different
states of uncertainties for model parameters
Top sites (section number) Frequency (in percentage) of ranking of five top sites in different uncertainty environments

Uncertainty: ±20 % for population of residential Uncertainty: ±20 % for population of residential
areas, ±40 % for relative importance of public centers areas, ±30 % for relative importance of public centers
and ±40 % for weight of defuzzified values of criteria and ±30 % for weight of defuzzified values of criteria

1st rank 2nd rank 3rd rank 4th rank 5th rank 1st rank 2nd rank 3rd rank 4th rank 5th rank

18 88 12 0 0 0 95 5 0 0 0
20 12 60 23 2 2 5 79 16 0 0
19 0 27 69 4 0 0 16 84 0 0
11 0 0 3 66 24 0 0 0 77 23
5 0 1 4 27 65 0 0 0 22 77

Fig. 8 Relative frequency


(percentage) of ranking of five
top sites obtained from 10,000
simulations for uncertainty
conditions: ±20 % for
population of residential areas
(Pi), ±40 % for importance
weight of public centers (Wi)
and ±40 % for weight of
defuzzified values of criteria (z*)

are the highest ones; (3) top four divisions are 5, 29, 18 and state crisp value when comparing different alternatives and
11, respectively. The relative weights of these divisions are they prefer to use linguistic terms when comparing items; (3)
in close proximity and are significantly superior compared to fuzzy values were used to compare the criteria to determine
other divisions. (4) With the exception of divisions 18 and 19, their weights which were converted to crisp values. In addi-
the top five divisions are almost as far apart from each other. tion, as each division has unique characteristics which are
This implicitly indicates that no prevailing region is observed very similar for all points inside it, if the whole length (15
in the river reach and that a spread of select divisions are Km) of the river reach was broken into 100 divisions instead
obtained in the river reach; (5) similar to scenario 1, four of 30, there would be no difference in the result.
divisions 29, 18, 19 and 5 are the best divisions indicating A sensitivity analysis is carried out for the model parame-
the importance of these sites with respect to the criteria. (6) ters by using a higher level of uncertainty to assess the ranking
Except for divisions 24 and 27, the orders of ranking are results of site selection. The fuzzy-AHP model is, therefore,
identical for all three different levels of imprecision. This assessed under different uncertainty domains of the model
implies that the order of the ranking of the bridge site location parameters as follows: The population of residential areas
is almost independent of the level of imprecision. (Pi ) with uncertainty domain of 10 %; importance weight of
Note that although the result obtained from different level public places (Wi ) with two uncertainty domains of 30 and
of imprecision levels (α = 0, 0.5 and 1) makes little differ- 40 %; and the weight of defuzzified values of criteria (z∗)
ence however, it is a useful stage in this analysis because with two uncertainty domains of 30 and 40 %. It is assumed
of the following reasons: (1) the utilization of fuzzy logic is that all these uncertain parameters follow a Gaussian PDF
mainly to assist the experts with expressing their judgment with coefficient of variation (CV) equal to their uncertainty
as linguistic terms; (2) it is often difficult for an expert to domains and mean equal to their deterministic values.

123
Arab J Sci Eng (2014) 39:4405–4420 4419

The uncertainty analysis of the model is carried out using fuzzy application in AHP can be eliminated, it is able to
Monte Carlo simulation (MCS) with the number of sample assist the experts with expressing their judgment as linguistic
sets equal to 10,000. Hence, 10,000 sets of uncertain parame- terms, which may make them respond to the survey in a more
ter values are randomly generated from the associated PDFs decisive way. Since fuzzy AHP was successfully applied for
using Monte Carlo sampling technique [41,42]. For each set bridge construction site selection in GIS environment, it may
of randomly generated parameters the fuzzy-AHP presents be further demonstrated and validated for other applications
a set of site rankings as a result. The relative frequency of of construction site selection problems.
ranking of top five sites in 10,000 simulations is shown in
Table 11 and Fig. 8. For the results are given for two different
states of uncertainties of model parameters. As it can be seen, References
the results demonstrate that the ranking of top sites is quite
1. Groenier, J.S.; Gubernick, R.: Choosing the best site for a bridge.
stable for both input data uncertainty situations. For instance, J. Transp. Res. pp. 347–354 (2007)
the bridge site for section number 18 obtained the top rank 2. Kuo, M.S.; Liang, G.S.; Huang W.C.: Extensions of the multicrite-
in 88 per cent of states for the worst uncertain situation (i.e. ria analysis with pairwise comparison under a fuzzy environment.
Int. J. Approx. Reason. 43, 268–285 (2006)
±40 % for relative importance of public centers and ±40 %
3. Soroudi, A.; Ehsan, M.: Application of a modified NSGA method
for weight of defuzzified values of criteria) while this site for multi-objective static distributed generation planning. Arab. J.
obtained the second rank in only 12 per cent of the Fuzzy- Sci. Eng. (2011). doi:10.1007/s13369-011-0077-1
AHP simulations in uncertain environment. The same was 4. Saaty, T.L.: Multicriteria Decision Making—The Analytic Hierar-
chy Process. RWS Publications, Pittsburgh (1988)
observed for the other top five sites for less uncertain envi-
5. Ko, J.: Solving a distribution facility location problem using an
ronment (i.e. ±30 % for relative importance of public centers analytic hierarchy process approach-Gwangju-Korea. In: ISAHP
and ±30 % for weight of defuzzified values of criteria). As Conference, Honolulu, Hawaii (2005)
it can be seen in the Table, the ranking position of each site 6. Yanpirat, P.; Panjarongkha, V.: Decision Support Model for site
selection of Wafer Fabrication Plant in Thailand—The MCDM
in this state is more robust compared to those in uncertain
Approach. Bangkok (2005)
environment (e.g. site no 18 is 95 per cent in the first rank 7. Mousavi, S.M.; Tavakkoli-Moghaddam, R.; Heydar, M.; Ebrahim-
in the less uncertainty state compared to 88 per cent in the nejad, S.: Multi-criteria decision making for plant location selec-
previous uncertainty environment). tion: an integrated Delphi–AHP–PROMETHEE methodology.
Arab. J. Sci. Eng. (2012). doi:10.1007/s13369-012-0361-8
8. Mohajeri, N.; Amin, G.R.: Railway station site selection using ana-
lytical hierarchy process and data envelopment analysis. J. Comput.
4 Conclusions Ind. Eng. 59, 107–114 (2010)
9. Dey, P.K.; Ramcharan, E.K.: Analytic hierarchy process helps
select site for limestone quarry expansion in Barbados. J. Envi-
The main purpose of this study was to rank potential river ron. Manage. 88, 1384–1395 (2008)
bridge construction sites using fuzzy AHP. Potential bridge 10. Wang, Y.M.; Liu, J.; Elhag, T.M.S.: An integrated AHP–DEA
construction sites were identified in GIS through dividing the methodology for bridge risk assessment. J. Comput. Ind. Eng. 54,
river reach into a number of sections. After removing unfea- 513–525 (2008)
11. Wang, T.C.; Chen, Y.H.: Applying consistent fuzzy preference rela-
sible sites, those remaining were introduced as alternatives tions to partnership selection. Omega Int. J. Manage. Sci. 35, 384–
for fuzzy AHP. Two scenarios, assuming whether or not the 388 (2007)
existing bridge is present in the analysis, were considered. 12. Deng, H.: Multicriteria analysis with fuzzy pair-wise comparison.
An analytical hierarchy process was created for the alterna- Int. J. Approx. Reason. 21, 215–231 (1999)
13. Laarhoven, Van.: Pedrcyz, W.: A fuzzy extension of Saaty’s priority
tives and the criteria. Fuzzy logic was also used in AHP to theory. Fuzzy. Set. Syst. 11, 229–241 (1983)
incorporate the imprecision of the experts’ judgments. The 14. Buckley, J.J.: Fuzzy hierarchy analysis. Fuzzy. Set. Syst. 17, 233–
application of the proposed method was demonstrated in a 247 (1985)
case study of selection of a bridge construction site. This 15. Tzeng, G.H.: Chen, Y.W.: Optimal location of airport fire station: a
fuzzy multi-objective programming and revised generic algorithm
demonstrated the power of the approach to identify potential approach. J. Transp. Plan. Technol. 23, 37–55 (1999)
options given a combination of quantitative and qualitative 16. Chu, T.C.: Selecting plant location via a fuzzy TOPSIS approach.
information and eventually finding best bridge site location. Int. J. Adv. Manuf. Technol. 28, 839–844 (2002)
The results of the sites’ ranking reveal that the superiority 17. Kuo, R.J.; Chi, S.C.; Kao, S.S.: A decision support system for
locating convenience store location through integration of fuzzy
of a division with respect to one criterion or even two cannot AHP and artificial neural network. J. Comput. Ind. 47(2), 199–214
guarantee that the division is selected as the best site. In other (2002)
words, all of the relevant criteria must be considered using an 18. Kuo, R.J.; Chi, S.C.; Kao, S.S.: A decision support system for
appropriate analytical method such as AHP to select the best locating convenience store through fuzzy AHP. J. Comput. Ind.
Eng. 37, 323–326 (1999)
alternative. While applying different levels of imprecision 19. Ertugrul, I.; Karakasoglu, N.: Comparison of fuzzy AHP and TOP-
using the α-cut concept in fuzzy AHP, no difference was SIS methods for facility location selection. Int. J. Adv. Manuf.
observed in the overall weights of alternatives. Although the Technol. (2007). doi:10.1007/s00170-007-1249-8

123
4420 Arab J Sci Eng (2014) 39:4405–4420

20. Chen, S.J.; Hwang, C.L.; Hwang, F.P.: Fuzzy multiple attributes 31. Hammad, A.; Itoh, Y.; Nishido, T.: Bridge planning using GIS and
decision making methods and applications. Springer, Berlin (1992) expert system approach. J. Comput. Civil. Eng. 7, 42–52 (1993)
21. Ashrafi, A.R.; Yuseff, R.M.; Derayatifar, A.R.: Linguistic exten- 32. GangaRao, H.; Ward, R.; and Howser, V.:Value engineering
sion of fuzzy integral for group personnel selection problem. Arab. approach to low-volume road bridge selection. J. Struc. Eng.
J. Sci. Eng. (2011). doi:10.1007/s13369-012-0491-z 114(9), 1962–1977 (1988)
22. Cheng, M.Y.; O’Connor, J.T.: Site layout of construction temporary 33. Ostenfeld, K.H.; Andersen, E.Y.: Major bridge projects—a multi-
facilities using an enhanced-geographic information system (GIS). disciplinary approach. Front. Archit. Civ. Eng. China 5(4), 479–495
J. Aut. Construct. 3, 11–19 (1994) (2011). doi:10.1007/s11709-011-0137-3
23. Moeinaddini, M.; Khorasani, N.; Danehkar, A.; Darvishsefat, A.A.; 34. Sasmal, S.; Ramanjaneyulu, K.: Condition evaluation of existing
zienalyan, M.: Siting MSW landfill using weighted linear combi- reinforced concrete bridges using fuzzy based analytic hierarchy
nation and analytical hierarchy process (AHP) methodology in GIS approach. Exp. Syst. Appl. 35(3), 1430–1443 (2008)
environment (case study: Karaj). J. Waste. Manage. 30, 912–920 35. Pan, N.F.: Fuzzy AHP approach for selecting the suitable bridge
(2010) construction method. J. Aut. Construct. 17, 958–965 (2008)
24. Malczewski, J.: GIS and Multicriteria Decision Analysis. Wiley, 36. Zhao, Z.; Chen, C.: A fuzzy system for concrete bridge damage
New York (1999) diagnosis. Comput. Struct. 80, 629–641 (2002)
25. Boroushaki, S.; Malczewski, J.: Implementing an extension of 37. Wang, Y.M.; Elhag, T.M.S.: A fuzzy group decision making
the analytical hierarchy process using ordered weighted averag- approach for bridge risk assessment. J. Comput. Ind. Eng. 53, 137–
ing operators with fuzzy quantifiers in ArcGIS. J. Comput. Geosci. 148 (2007)
34, 399–410 (2007) 38. USACE (US Army Corps of Engineers), Hydrologic Engineering
26. Malczewski, J.: GIS and Multicriteria Decision Analysis. Wiley, Centre, HEC-RAS River Analysis System, User’s Manual, Version
New York (2006) 4.1, January (2010)
27. Guiqin, W.; Guoxue, L.; Lijun, C.: Landfill site selection using 39. Ministry of Roads and Transportation Road Safety Manual (Road
spatial information technologies and AHP: a case study in Beijing, Side Safety) Iran, Manual No. 267-1 (2005)
China. J. Environ. Manag. 90, 2414–2421 (2009) 40. Aldian, A.; Taylor, M.A.P.: Fuzzy multi-criteria analysis for inter-
28. Sharifi, M.; Hadidi, M.; Vessali, E.; Mosstafakhani, P.; Taheri, K.; city travel demand modelling. J. Eastern. Asia. Soc. Transp. Stud.
Shahoie, S.; Khodamoradpour, M.: Integrating multi-criteria deci- 5, 1294–1307 (2003)
sion analysis for a GIS-based hazardous waste landfill sitting in 41. Ross, T.J.: Fuzzy logic with engineering application. McGraw-Hill,
Kurdistan Province, western Iran. J. Waste. Manag. 29, 2740–2758 New York (1995)
(2009) 42. McKay, M.D.; Conover, W.J.; Beckman, R.J.: A comparison of
29. Geneletti, D.: Combining stakeholder analysis and spatial multi- three methods for selecting values of input variables in the analysis
criteria evaluation to select and rank inert landfill sites. J. Waste. of output from a computer code. Technometrics. 211, 239–245
Manag. 30, 328–337 (2010) (1979)
30. Kye, P.A.K.: On intractable conflicts participatory GIS 43. Post, J.; Hattermann F.F.; Krysanova, V.; Suckow, F.: Parameter
applications—the search for consensus amidst competing and input data uncertainty estimation for the assessment of long-
claims and institutional demand. Ann. Assoc. Am. Geogr. 94(1), term soil organic carbon dynamics. Environ. Modell. Softw. 23(2),
37–57 (2004) 125–138 (2008)

123
Copyright of Arabian Journal for Science & Engineering (Springer Science & Business Media
B.V. ) is the property of Springer Science & Business Media B.V. and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.

You might also like