See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/264538259

Land suitability assessment for wind power plant site selection using ANP-
DEMATEL in a GIS environment: case study of Ardabil province, Iran

Article  in  Environmental Monitoring and Assessment · August 2014

DOI: 10.1007/s10661-014-3883-6 · Source: PubMed

CITATIONS READS

38 1,118

5 authors, including:

Ali Azizi Bahram Malekmohammadi

University of Tehran University of Tehran
9 PUBLICATIONS   61 CITATIONS    35 PUBLICATIONS   308 CITATIONS   

SEE PROFILE SEE PROFILE

Hamid Jafari Hossein Nasiri

University of Tehran University of Tehran
33 PUBLICATIONS   611 CITATIONS    14 PUBLICATIONS   129 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Environmental Impact Assessment of Groundwater depletion in Ardabil Plain View project

Comprehensive studies of groundwater resources in the Khanmirza plain View project

All content following this page was uploaded by Hossein Nasiri on 10 August 2014.

The user has requested enhancement of the downloaded file.

Environ Monit Assess
DOI 10.1007/s10661-014-3883-6
Land suitability assessment for wind power plant site selection
using ANP-DEMATEL in a GIS environment: case study
of Ardabil province, Iran
Ali Azizi & Bahram Malekmohammadi &
Hamid Reza Jafari & Hossein Nasiri &
Vahid Amini Parsa
Received: 25 January 2014 / Accepted: 11 June 2014
# Springer International Publishing Switzerland 2014
Abstract Wind energy is a renewable energy resource
that has increased in usage in most countries. Site selection
for the establishment of large wind turbines, called wind
farms, like any other engineering project, requires basic
information and careful planning. This study assessed the
possibility of establishing wind farms in Ardabil province
in northwestern Iran by using a combination of analytic
network process (ANP) and decision making trial and
evaluation laboratory (DEMATEL) methods in a geo-
graphical information system (GIS) environment.
DEMATEL was used to determine the criteria relation-
ships. The weights of the criteria were determined using
ANP and the overlaying process was done on GIS. Using
13 information layers in three main criteria including en-
vironmental, technical and economical, the land suitability
map was produced and reclassified into 5 equally scored
divisions from least suitable to most suitable areas. The
results showed that about 6.68 % of the area of Ardabil
province is most suitable for establishment of wind tur-
bines. Sensitivity analysis shows that significant portions
of these most suitable zones coincide with suitable divi-
sions of the input layers. The efficiency and accuracy of the
A. Azizi : B. Malekmohammadi (*) : H. R. Jafari :
V. Amini Parsa
Graduate Faculty of Environment, University of Tehran,
Tehran, Iran
e-mail: Malekb@ut.ac.ir
H. Nasiri
Faculty of Geography, University of Tehran,
Tehran, Iran
hybrid model (ANP-DEMATEL) was evaluated and the
results were compared to the ANP model. The sensitivity
analysis, map classification, and factor weights for the two
methods showed satisfactory results for the ANP-
DEMATEL model in wind power plant site selection.
Keywords Wind power plants (WPPs) . Geographical
information system (GIS) . Site selection . ANP-
DEMATEL . Iran–Ardabil province
Introduction
Energy is a critical factor in the sustainable development
of human societies. Growing demand for energy, in-
creasing standards of living, the danger of global
warming, and the shortage of fossil fuel energy re-
sources require that global attention be focused on the
use of renewable energy. One main advantage of new
energy sources is the reduction in CO2 emissions (Aydin
et al. 2010). One kilowatt hour of electricity produced
by wind power plants (WPPs) decreases emission of
about 1 kg of CO2 over that of fossil power plants
(Sana 2010). The use of wind as a renewable energy
supply has increased globally in comparison with other
renewable resources of energy (Alamolhoda 2010).
The first fundamental step in the development of new
energy is the careful study of potentials (Dutra and
Szklo 2008). Next is selection of an appropriate site
for the establishment of equipment and facilities to make
use of the potentials. Site selection for the establishment

of large wind turbines, called wind farms, like any other
engineering project, requires basic information and care-
ful planning. The selection of multiple factors results in
the multiplicity of information layers and efforts to find
an appropriate solution for the analysis of the informa-
tion layers. Finding a suitable analysis system can en-
courages decision makers to adopt this system, which is
very fast, easily applicable, and highly precise.
Recently, geographic information systems (GIS) in-
tegrated with Multi Criteria Decision Making (MCDM)
models have been widely applied to environmental
planning and engineering issues. Analytic network pro-
cess (ANP) is one MCDM method that is a generaliza-
tion of the analytic hierarchy process (AHP). These two
models were proposed by Saaty (Saaty 1996, 2001).
The decision making trial and evaluation laboratory
(DEMATEL) was developed by Gabus and Fontela
(1972) to solve complicated global problems. In recent
years, there has been an increase in the use of ANP for a
variety of issues, such as municipal solid waste
(Aragonés-Beltrán et al. 2010), investment (Lee et al.
2011), and technology (Lee et al. 2009). ANP that uses
DEMATEL techniques has been used in several studies
for municipal solid waste management (Tseng 2009),
the healthcare sector (Taghizadeh Herat et al. 2012),
management (Tsai and Chou 2009; Vujanović et al.
2012), education (Chen and Chen 2010) and brand
marketing (Wang and Tzeng 2012).
Yang et al. (2008) used ANP and DEMATEL to
solve dependence and feedback problems to suit the
real world. They used ANP to overcome
interdependence and feedback between criteria and
alternatives in MCDM methods; also they adopted
DEMATEL to determine the degree of influence of
criteria. Their study provides a numerical example to
illustrate the proposed method.
GIS has been successfully applied to wind farms.
Tegou et al. (2010) developed an integrated frame-
work to evaluate land suitability for wind farm site
selection that combined multicriteria analysis
(MCA) with GIS on the island of Lesvos in
Greece. A set of environmental, economic, social,
and technical constraints based on recent Greek
legislation identified potential sites for wind power
installation. The area under consideration was eval-
uated using a variety of criteria, such as wind
power potential, land cover type, electricity de-
mand, visual impact, land value, and distance from
the electrical grid. AHP was applied to estimate the
Environ Monit Assess
criteria weights to establish their relative impor-
tance for site evaluation.
Al-Yahyai et al. (2012) applied MCDM using AHP
with an Ordered Weigh Averaging (OWA) aggregation
function to derive a wind farm land suitability index and
classification under a GIS environment in Oman. The
selection criteria considered were economic (distance to
road, terrain, slope), social (urban area), environmental
(historical sites, wildlife, natural reserves) and technical
(wind power density, energy demand, percentage of
sustainable wind, turbulence intensity, sand dunes).
GIS with/without MCDM has been successfully ap-
plied to wind farm site selection. Janke (2010) and van
Haaren and Fthenakis (2011) studied wind farm site
selection using GIS. Kaya and Kahraman (2010) and
Tegou et al. (2010) studied wind farm site selection
using GIS and AHP. Bennui et al. (2007) applied GIS
integrated with MCDM for effective site selection for
large wind turbines in Thailand. Wind speed informa-
tion, elevation, slope, highways and railways, built-up
areas, forest zones, and scenic areas were used as criteria
for site selection.
Moiloa (2009) used GIS for strategic wind energy
site selection. Three case studies for GIS-based models
for wind energy facility site selection were used in the
USA, England, and South Africa. These countries were
selected to illustrate how GIS has been applied by
planners and decision makers. Sliz-Szkliniarz and Vogt
(2011) evaluated wind energy potential using a GIS-
based approach in Poland. They developed an approach
to support the decision-making process for site selection
for wind energy projects using GIS.
GIS with MCDM was used by Aydin et al. (2010) to
create a decision support system for site selection of
wind turbines in western Turkey. Wind energy potential
and environmental fitness/acceptability were used as
decision criteria for site selection. OWA was used as a
multi-objective decision-making (MODM) tool to eval-
uate alternatives for fuzzy environmental objectives.
Taghilo (2011) used GIS with MCDM for wind farm
site selection. He used GIS and MCDM methods to
consider three groups of stockholders to optimize dif-
ferent alternatives for wind farms in Zanjan province in
Iran. He used five multicriteria methods (index overlay,
AHP, fuzzy AHP, TOPSIS, fuzzy TOPSIS) to find op-
timal alternatives for wind turbines and found that
TOPSIS results were most acceptable.
A review of the literature indicates that the combined
use of ANP and DEMATEL has not been applied in the
Environ Monit Assess
GIS environment. Thus far, a comprehensive research
has not been done using a combination of these models
for WPP site selection. Although ANP has been used to
overcome interdependence and feedback problems be-
tween criteria or alternatives, it could not determine the
strength of relationships between criteria. DEMATEL
was used to address this inadequacy.
The present study applied ANP and DEMATEL in
GIS to determine suitable sites for WPPs. ANP was used
to consider the relationship between criteria and
DEMATEL to determine the degree of influence of
criteria for WPP site selection. ANP-DEMATEL in
GIS was used for WPP site selection in Ardabil province
in northwestern Iran.
Study area
A 10-year study of wind in Iran shows that many re-
gions are windswept and suitable for the production of
electricity, especially in summer. These areas include the
Fig. 1 Location and specification of Ardabil Province in Iran
coastal regions of the Oman Sea and Khuzestan prov-
ince, eastern regions, Persian Gulf islands, Manjil,
Rafsanjan, Ardabil, and Bijar (Ghandomkar 2009).
The total capacity of established wind power plants at
present is less than 2 % of the total estimated wind
potential in the country. The nominal capacity of all
power plants in Iran in 2009 was 34,000 MW
(Alamolhoda 2010). It is clear that the country needs
to diligently develop its wind energy potential for dif-
ferent applications.
Ardabil province is located in northwestern Iran,
bordering the Republic of Azerbaijan and the prov-
inces of East Azarbaijan, Zanjan, and Gilan.
Ardabil province is located at 37°12′ to 39°47′
north latitude and 47°13′ to 48°55′ east longitude
(Fig. 1). The total area of the province is
17,953 km2 (1.1 % of the area of Iran). The prov-
ince has 10 counties, 24 cities, 27 districts, 69 sub-
districts, and 2,210 villages. The city of Ardabil is
the provincial capital and is located about 70 km
from the Caspian Sea.
 Environ Monit Assess

Ardabil province encompasses two thirds of the Materials and methods

Azerbaijan mountain range, which strongly influences
local wind strength. Its average elevation is about
1,400 m. The lowest elevation is 100 m in Parsabad
and Bilesavar and the highest elevation is 4,811 m at
Sabalan. The maximum temperature is 35 °C in summer
and the minimum is -25 °C in winter. The variation in
elevation and air pressure in different seasons, especially
during the summer months, provides excellent condi-
tions for wind.
Methodology
The process of WPP site selection is shown in Fig. 2.
First (in step 1), all WPP criteria for the province was
gathered; if some information for site selection was not
available, an attempt was made to consider the most
influential technical, economic, and environmental pa-
rameters. In step 2, using local conditions, a literature

Start

1-Identifying of criteria

Accessibility Local conditions Opinions of experts Literature review

2- Review and criteria selection

Environmental Technical Economical

Slope D.F. City

Wind average

Elevation D.F. Rural

Wind continuity

D.F. Fault D.F. Airport

Wind density

D.F. River Land cover/use

D.F. Protected area D.F. Road

GIS Database

3- Identifying exclusionary areas

4- Determining strength of relationships between criteria's obtained from DEMATEL model

5- Criteria and subcriteria weights obtained from ANP using the Supper decisions software

6- Overlaying the layers in GIS

7- Reclassification in scored classes

8-Removing exclusionary areas

moved

9- Extracting final Land suitability map

10- Sensitivity analyses

End

Fig. 2 The overall process of the WPPs site selection in ANP-DEMATEL method (D.F. distance from)
Table 1 List and description of criteria for WPPs site selection
Environ Monit Assess

Criteria Description

Wind potential Calculating wind potential in a geographical region is the first and the most fundamental issue to make use of wind energy, its assessment and its
other specifications. Annual wind speed average, wind power density and percentage of windy days are very important features for choosing a
site for wind power plants (Zahedi et al. 2005). It should be noted that a wind potential map plays an important role in the selection of the
candidate sites for the wind farm and this is the most important document supporting the financial evaluations of the installations (Lee et al. 2010)
Landuse/cover Two kinds of land use should be considered in site selection for the establishment of wind turbines: the first use relates to the lands that cannot be
used as wind farms like forests, ponds, protected areas, animal habitats and military sites. The second kind of use relates to the lands that can be
used as wind farms, although they should be selected in a way that they do not damage the environment and require the lowest costs for the
establishment of wind turbines (Taghilo 2011)
Elevation Depending on the topography of the region, wind speed and wind direction changes very much (Talinli et al. 2011). The general principle that an
increase in height makes difficult the possibility of making facilities, is correct again (Farajisabokbar et al. 2010). Therefore in selecting the
appropriate site, we should avoid the high lands, because they increase the investment costs (Bennui et al. 2007)
Slope One of the natural factors, which have great impacts on the selection of suitable sites for making facilities and infrastructures is the land slope.
Because considering the direction and percentage of the slope, the land use can affect the constructional expenses (surface water drainage, land
leveling, etc.) (Farajisabokbar et al. 2010)
Distance from residential area (cities Perhaps the most important factor in limitation of installing wind turbines in residential areas is their loud noises (Moiloa 2009). Regarding the
and villages) distance from the residential areas, it should be mentioned that the physical development of the city is a dynamic and continuous process.
Therefore, the surrounding lands should be considered for the future developments in residential areas
Distance from airport Wind turbines can interfere with signals of aviation radars. So, they would require a significant buffer around areas such as airports (Moiloa 2009).
Wind turbines also interfere with electromagnetic waves in a way that they reflect, scatter or break electromagnetic waves and interfere with the
telecommunication networks
Distance from water extent The turbine towers, roads and related constructions destroy the vegetations and the natural habitats. Wind turbines also present a collision risk to
birds and may have negative effects on the bird populations, especially migratory birds (Moiloa 2009). Finally, since the river routes are dynamic
ever-changing and there is also risk of flood, the wind farms’ being away from the river-beds will increase the safety of the facilities
Distance from main road network Transportation network has special importance from two viewpoints: 1- Closeness of the turbines to the roads on the one hand affects the road
transportation, because of the turbines’ loud noises and on the other hand the roads will be shadowed, because of the turbine blades. Moreover,
there will be some changes in the visual landscape. 2- Because of the immense weight of the equipment, the wind farms’ closeness to the roads is
of great importance because the transportation costs for the establishment of wind turbines are too high (Moiloa 2009; Taghilo 2011)
Distance from protected area Protected areas play a critical role in the protection and conservation of the world's natural and cultural resources
Distance from Fault Wind farm sites like other constructional projects should be away from the fault lines in order to avoid dangerous hazards caused by these lines
 Environ Monit Assess

review and the opinions of experts, different criteria
were reviewed and 13 criteria were selected in three
main groups. These were the most important criteria
for selecting suitable sites and establishing WPPs
and are shown in Fig. 2 and explained in Table 1.
Thirteen information layers were prepared and con-
verted to a raster format. Preliminary spatial analyses
were performed on them to implement ANP in the
GIS and to establish a database.
In step 3, exclusionary areas were identified and
an exclusionary map was prepared. In step 4,
DEMATEL was used to determine the relation-
ships between criteria and in step 5; ANP was
used to weight the criteria. Overlaying was done
using GIS in step 6 and the land suitability map
was reclassified into five equally scored zones
from least suitable to most suitable in step 7. In
step 8, the exclusionary areas were removed using
the exclusionary map and final land suitability
map was extracted in step 9. Finally, in step 10,
a sensitivity analysis was carried out.
The GIS datasets of the 13 criteria were collected
from sources shown in Table 2. Since wind speed was
measured at a height of 10 m and the height of the wind
turbines is 40 m, the speed of the wind at a height of
40 m was calculated by Eq. (1):
 a
V H
¼ ð1Þ
Vr Hr
Where V is wind speed at height H (desired) and Vr is
wind speed at height Hr (baseline) and a is ground
roughness coefficient.
An overview of ANP and DEMATEL
In ANP procedures, the first step is to construct a con-
ceptual model and to determine relationships between/
among clusters and nodes. Determining relationships
between/among criteria is very important in ANP be-
cause pairwise comparisons depend on this phase.
Because there are different degrees of influence and
relationship between the criteria in the real world, as-
sumptions of equal influence and relationship seem to
be irrational. Thus, in order to solve this problem, in this
study DEMATEL method is used. In the following,
descriptions of these two methods are presented.
ANP
ANP is an extension of the AHP; indeed, it is the general
form of the AHP. These two models are proposed by
Saaty (Saaty 1996, 2001). AHP is a well-known tech-
nique that breaks down a decision-making problem into
several levels in such a way that they form a hierarchy
with unidirectional hierarchical relationships between
levels (Aragonés-Beltrán et al. 2010) but ANP is a
nonlinear structure with bilateral relationships (Fig. 3).
So the main innovation of the ANP is its network

Table 2 Data sources for input layers in land suitability assessment for Ardabil Province

Row Layers Source

1 Wind speed average Synoptic station data from Ardabil Province

2 Percentage of windy days
3 Wind power density
4 Distance from urban areas 1:25,000 topographic maps prepared by the (National Cartographic Center of Iran (NCC))
5 Distance from rural areas
6 Distance from rivers
7 Distance from protected areas
8 Distance from main roads
9 Distance from the airport
10 Slope
11 Elevation
12 Land cover/use Images taken from the TM sensor in 2010
13 Distance from fault lines 1:100,000 geological map prepared by the (National Cartographic Center of Iran (NCC))
Environ Monit Assess
Fig. 3 Structural difference
between a hierarchy (a) and a
network (b)
structure, which enables interactions between elements
situated in different clusters and dependencies between
the elements in the same cluster to be taken into account
(Nekhay et al. 2009; Tseng 2009). In this paper, ANP is
used to obtain the weight of the criteria. The application
steps of ANP can be described in the following steps.
Step 1: As mentioned above, the first step is to con-
struct a conceptual model and to determine
relationships between/among clusters and
nodes. The conceptual model (Fig. 4) is con-
structed and the relationships (shown by ar-
rows in the conceptual model) between/among
clusters, and nodes are determined.
Fig. 4 An example for designing conceptual model in ANP
Step 2: Criteria are compared, using Super Decisions
software (Fig. 5), in the whole network in
order to form an unweighted supermatrix by
pairwise comparisons (the same as the AHP).
In this phase, decision makers compare two
elements. Pairwise comparisons are made with
the grades ranging from 1–9 (Vasiljević et al.
2012). A reciprocal value of each number is
used to express the inverse comparison. The
values of pairwise comparisons are allocated
in the comparison matrix and local priority
ve cto r i s de r i v ed fr om e i ge nv ec t o r.
Consistency of pairwise matrix like the AHP
must be less than 0.1 (Sener et al. 2011).
 Environ Monit Assess

Fig. 5 Pairwise comparisons in supper decisions software

Step 3: The weights obtained from the previous steps Wij is a local priority vector obtained from the
are introduced into the supermatrix that in- corresponding pairwise comparison, representing
cludes the entire network components and rep- the importance of the elements in the ith cluster
resents their inter relationships. In this step, on an element in the jth cluster (Lee et al. 2009).
supermatrix is called the initial supermatrix. Step 4: The cluster weights should be calculated in this
Equation (2) shows the general form of the step in order to weigh the initial supermatrix.
supermatrix. When the cluster weight matrix has been ob-
tained, the initial supermatrix can be weighted
C1 C2 Cn
e1 1 e1 m1 e2 1 e 2 m 2 e n 1 e nm n by multiplying the cluster weights matrix by an
e1 1
C1
e1 2 W11 W12 W 1n initial supermatrix (Nekhay et al. 2009). The
e1 m1
new obtained matrix is known as the weighted
e2 1 supermatrix.
C 2 e2 2 W12 W 22 W2n
W = Step 5: The final step consists of multiplying the
e2 m 2
weighted supermatrix n times by itself until
e n1 the limit supermatrix is reached. Some super
C n en 2
matrices may have a cyclicity effect. As a
en m n W n1 Wn2 W nn
result, two or more final limit super matrixes
may be obtained. The main property of the
limit supermatrix is that its columns are equal
Ck is the kth cluster (k =1, 2,…, N) which has
and represent the global priority vectors
nk elements denoted as ek1, ek2, …, eknk. A matrix
(Nekhay et al. 2009).
segment, Wij, represents a relationship between
the ith cluster and the jth cluster. Each column of
DEMATEL

Table 3 Exclusionary zones in WPPs site selection
Row Zones
1 Buffer zones within 0.5 km from main road
2 Buffer zones within 0.4 km from river
3 Reservation areas
4 Buffer zones within 2 km from urban areas
5 Buffer zones within 0.5 km of rural areas
6 Buffer zones within 3 km from airport
7 Others
In order to solve complicated global problems,
Pixel value
DEMATEL method had been applied at the end of
1971 by Gabus and Fontela (Gabus and Fontela 1972).
0 The DEMATEL technique constructs the interrelations
0 between factors/criteria to build a network relations map
0 (NRM) (Tzeng et al. 2007; Yang et al. 2008; Liou et al.
0 2007). In this paper, DEMATEL is prepared in the six
0 following steps:
0
1 Step 1: In order to pairwise comparisons, the five-
point scale regarding the level of impacts
Environ Monit Assess

Fig. 6 Exclusionary map for wind power plants site selection in Ardabil Province

of particular dimensions is used. The mea-
surement criteria of 0, 1, 2, 3, and 4 are
used to illustrate no influence, very low
influence, low influence, high influence
and very high influence, respectively
(Tseng and Lin 2009).
Step 2: The direct-influence matrix is constructed
based upon the degrees of relative impacts
derived from the pair comparisons in
Eq. (3). An n×n direct-influence matrix A
with the directly observed relations is obtain-
ed where aij denotes the degree of impacts of
 Environ Monit Assess

the i factor on the j factor (Taghizadeh Herat Step 5: Summing each column and row to obtain D
et al. 2012). and R Eqs. (7) and (8):
2 3 !
a11 ⋯ a1 j ⋯ a1n X n
6⋮ ⋮ ⋮7 D¼ T ij ¼ ½d i Šn x 1 ð7Þ
6 7
A¼6 a
6 i1 ⋯ a ij ⋯ a 7
in 7 ð3Þ j¼1
4⋮ ⋮ ⋮5
an1 ⋯ anj ⋯ ann
!
Step 3: The normalized matrix obtains from Eqs. (4) Xn
 
R¼ T ij ¼ rj 1 x n ð8Þ
and (5): i¼1

X ¼ s⋅A ð4Þ
2 3 Here, di is the sum of each row in T and the
6 7 rows show the degrees of direct and indirect
6 1 1 7
6
s ¼ min6 ; 7 ð5Þ impacts over the other criteria, and rj is the sum
Xn
  Xn
  7
4 max  
aij max j  aij 5
 of each column in T where columns indicates
i
j¼1 i¼1 the degrees of influences from other criteria.
Numeric algorithm variable di, therefore, repre-
sents the factors that influence others, rj repre-
Where the all matrix diagonals are equal to sents factors that are influenced by others, di +rj
zero and the sum of each row and column does represents the strength of relationships between
not exceed 1. factors, di −rj represents the strength of influ-
Step 4: Deriving the full relationship matrix T from the ences among factors (Lee et al. 2011).
Eq. (6): Step 6: Determining a threshold value in order to
ignore the minor effects is necessary to isolate
T ¼ X þ X 2 þ … þ X k ¼ T ¼ X ð1−X Þ−1
the relationship structure of the factors and
ð6Þ finally obtain NRM (Yang and Tzeng 2011).

Table 4 The final weights for criteria land suitability assessment for Ardabil Province

ANP ANP-DEMATEL

Criteria Weight Rank Criteria Weight Rank

Wind speed average 0.228 1 Wind speed average 0.375 1

Wind power density 0.153 2 Land cover/use 0.155 2
Percentage of windy days 0.144 3 Wind power density 0.135 3
Distance from the airport 0.088 4 Slope 0.088 4
Distance from urban areas 0.080 5 Percentage of windy days 0.059 5
Distance from fault lines 0.059 6 Distance from urban areas 0.050 6
Slope 0.052 7 Elevation 0.037 7
Distance from rivers 0.039 8 Distance from the airport 0.031 8
Distance from rural areas 0.038 9 Distance from rural areas 0.027 9
Land cover/use 0.036 10 Distance from main roads 0.025 10
Distance from main roads 0.032 11 Distance from protected areas 0.011 11
Distance from protected areas 0.030 12 Distance from rivers 0.006 12
Elevation 0.021 13 Distance from fault lines 0.000 13
Environ Monit Assess

Fig. 7 Input data layers in wind power plants site selection for Ardabil Province
 Environ Monit Assess

Fig. 8 Land suitability maps in ANP and ANP-DEMATEL for Ardabil Province

Results and discussion D E M AT E L w a s d e v e l o p e d i n M AT L A B

(MathWorks 2005) and relationships between the criteria
ANP-DEMATEL and ANP methods were used for and strongness of them were identified. The results of
WPP site selection. The ANP model was developed DEMATEL were applied to refine the ANP conceptual
and the results of the two methods were compared to model. Super Decisions software (Adams and Saaty
evaluate the efficiency and accuracy of the hybrid mod- 2003) was used for ANP to obtain the final weights;
el. Recognized experts were selected from different the final weights of the criteria for ANP-DEMATEL and
organizations in the study region. Overall, 35 expert ANP are shown in Table 4. In ANP-DEMATEL, wind
opinions were applied in this study. Opinions were speed was weighted highest and the fault was weighted
collected from 10 environmentalists, 10 urban and rural
planners, 5 alternative energy experts, 5 water resources
Table 5 The areas of classes of final land suitability map
experts, and 5 GIS experts by questionnaire and by
forums of experts. Classes ANP-DEMATEL ANP
After establishing a database, areas that were unsuit-
Area Area (%) Area Area (%)
able for site selection were identified as exclusionary
(km2) (km2)
areas based upon a literature review and knowledge of
local conditions. Table 3 shows the distance from roads, Exclusion zone 5,089 28.35 5,089 28.35
rivers, reserves (forests, lakes, ponds), protected areas, Less suitable 1,625 9.05 451 2.51
urban and rural areas, and airports. Seven layers were Suitable 4,643 25.83 4,447 24.77
developed to extract the exclusionary layer. Figure 6 Moderate suitable 3,901 21.73 4,856 27.05
shows the exclusionary map for the WPP site selection High suitable 1,496 8.33 2,099 11.69
in Ardabil province. About 28.4 % of the region was Extremely suitable 1,199 6.68 1,011 5.63
excluded and rest is considered to be areas with Total 17,953 100 17,953 100
potential.
Environ Monit Assess
Fig. 9 Sensitivity analysis results for ANP and ANP-DEMATEL
lowest (no weight), having maximum and minimum
relationships, respectively, with the other parameters.
ANP established relationships regardless of strength,
thus criteria with weak relations were paired with others
in the comparison process. This means that the overall
weight is not distributed properly among the criteria, as
observed in Table 4. The weights and priorities of the
criteria differ significantly between ANP and ANP-
DEMATEL. A comparison of priorities of the criteria
from the questionnaires, used to identify the most influ-
ential criteria, and models' results in Table 4 emphasized
the better performance of the hybrid model.
Figure 7 is based on information from Tables 1 and 2
to prepare input layers and convert them to raster format.
By applying the weights in Table 4, overlaying was
done in GIS and a land suitability map was obtained.
The final zoning maps of the ANP and hybrid models
were reclassified and are presented in Fig. 8. Table 5
shows the five zones derived from the hybrid and ANP
models. A comparison of the models in Table 5 indicates
that the zones were different between models. The im-
proper distribution of weights in the ANP model
(Table 4) are oriented toward the center divisions (suit-
able, moderately suitable, highly suitable), despite the
existence of high and low potential areas. The total area
of the three middle zones in the ANP model is 7.62 %
more than in the ANP-DEMATEL model. The amount
of extremely suitable divisions in the ANP-DEMATEL
model (6.68 %) was greater than for ANP (5.63 %).
ANP-DEMATEL determined that about 1,199 km2
of the total area of the province is very suitable for wind
turbines. The most suitable areas are east of the city of
Ardabil near the political boundaries of the town of
Namin. Most of these areas are located in northeast
and southeast of Ardabil city (Fig. 8). This area has
significant wind potential and is located in an area with
relatively suitable slope. These areas are also primarily
dry land and pastures, which decreases the cost of land,
but still excludes and preserves land suitable for farm-
ing. In ANP, part of extremely suitable areas which are
located in northeast and southeast of Ardabil city are
classified into high suitable class. According to local
conditions and input layers, these areas have the poten-
tial to be placed in the extremely suitable class, but due
to improper weights distribution in ANP model, less
area are located in extremely suitable class in compare
to ANP-DEMATEL model.
Figure 9 shows the results of the sensitivity anal-
ysis for both models for the suitable zones (high and
extremely suitable) for each criterion (Fig. 7) and
the final map (Fig. 8). This figure shows the degree
to which each criterion coincides with the final map.
As seen, all input layers in both models coincide
well with the final map (+62 %) except for the
elevation layer. Because the wind parameter (speed
and continuity) increases in suitability with height,
the suitability of the elevation layer decreased. A
comparison of the sensitivity analyses shows that
the overlaps between suitable zones (high and ex-
tremely) of the final map and the input layers are
more compatible with the ANP-DEMATEL model.
These results show that the ANP-DEMATEL model
is more accurate and efficient than the ANP model
for finding suitable areas for WPPs.

Conclusions
The decline in fossil fuel reserves coincides with
global warming, much of which is caused by burn-
ing fossil fuels. In recent decades, the use of renew-
able energy has been emphasized. Wind energy is of
great interest among renewable energy sources be-
cause it is abundant, economical, and easy to use.
Finding appropriate sites to utilize this clean energy
requires consideration of several factors that make
analysis difficult. Selecting suitable sites for WPPs
involves combining several complicated parameters,
which necessitates the use of GIS in combination
with MCDM methods.
The use of ANP was tested because possible
interactions between elements are neglected in the
AHP method. Although ANP considers relation-
ships between criteria, it is unable to determine
the strengths of the relationships between criteria.
The results of ANP in this study show that the
lack of consideration of this issue causes a devia-
tion of the model results from actual conditions.
To overcome this shortcoming, DEMATEL was
used in combination with ANP. The results show
greater accuracy and applicability of ANP-
DEMATEL for determining direct and indirect re-
lationships and the strength of these relationships
for land suitability assessment.
Using DEMATEL to draw relationships between
the criteria in ANP increased the accuracy of de-
termining the relative importance of the criteria for
land suitability assessment. Where there was only
a slight relationship between a criterion and other
criteria, ANP-DEMATEL increased the accuracy of
the process. A weak plausible relationship between
fault criteria ranks it in sixth place in the ANP
model. This criterion is important, but the results
of the hybrid model show a weak relationship
between this criterion and the others, thus, it was
given no weight in the hybrid model.
The application of ANP-DEMATEL for Ardabil
province shows that the most suitable areas are located
east of the capital city. This area represents about 6.68 %
of the total area of the province. There is good potential
in this location for exploitation of wind energy.
Acknowledgments The authors would like to thank the two
reviewers for their constructive comments on correction and im-
provement of the manuscript.
Environ Monit Assess
References
Adams, W., & Saaty, R. (2003). Super Decisions Software Guide.
Super Decisions, 9.
Alamolhoda, S. (2010). Wind power. Safir omid, Iranian Oil- Gas
and Petrochemical Publications, 4, 61–67 (In persian).
Al-Yahyai, S., Charabi, Y., Gastli, A., & Al-Badi, A. (2012). Wind
farm land suitability indexing using multi-criteria analysis.
Renewable Energy: An International Journal, 44, 80–7.
Aragonés-Beltrán, P., Pastor-Ferrando, J. P., García-García, F., &
Pascual-Agulló, A. (2010). An analytic network process
approach for siting a municipal solid waste plant in the
metropolitan area of Valencia (Spain). Journal of
Environmental Management, 91(5), 1071–1086.
Aydin, N. Y., Kentel, E., & Duzgun, S. (2010). GIS-based envi-
ronmental assessment of wind energy systems for spatial
planning: a case study from Western Turkey. Renewable
and Sustainable Energy Reviews, 14(1), 364–373. doi:10.
1016/j.rser.2009.07.023.
Bennui, A., Rattanamanee, P., Puetpaiboon, U., Phukpattaranont,
P., & Chetpattananondh, K. (2007) Site selection for large
wind turbine using GIS. In PSU-UNS International
Conference on Engineering and Environment (pp. 561-566).
Chen, J.-K., & Chen, I. (2010). Using a novel conjunctive MCDM
approach based on DEMATEL, fuzzy ANP, and TOPSIS as
an innovation support system for Taiwanese higher educa-
tion. Expert Systems with Applications, 37(3), 1981–1990.
Dutra, R., & Szklo, A. (2008). Assessing long-term incentive
programs for implementing wind power in Brazil using GIS
rule-based methods. Renewable Energy, 33(12), 2507–2515.
Farajisabokbar, H., Nasiri, H., Hamzeh, M., Talebi, S., & Rafee, Y.
(2010). Analysis of land suitability to locate rural service
centres using PROMETHEE II and AHP methods in GIS
environment (Case study in Soltanieh-Abhar city). Journal of
Rural Development, 2, 95–118.
Gabus, A., & Fontela, E. (1972). World problems, an invitation to
further thought within the framework of DEMATEL. Geneva,
Switzerland: Battelle Geneva Research Center.
Ghandomkar, A. (2009). Evaluation of wind energy potential in
Iran. Journal of Geography and Environmental Planning, 36,
85–100.
Janke, J. R. (2010). Multicriteria GIS modeling of wind and solar
farms in Colorado. Renewable Energy, 35(10), 2228–2234.
doi:10.1016/j.renene.2010.03.014.
Kaya, T., & Kahraman, C. (2010). Multicriteria renewable energy
planning using an integrated fuzzy VIKOR & AHP method-
ology: the case of Istanbul. Energy, 35(6), 2517–2527.
Lee, H., Kim, C., Cho, H., & Park, Y. (2009). An ANP-based
technology network for identification of core technologies: a
case of telecommunication technologies. Expert Systems with
Applications, 36(1), 894–908. doi:10.1016/j.eswa.2007.10.
026.
Lee, K.-H., Jun, S.-O., Pak, K.-H., Lee, D.-H., Lee, K.-W., & Park,
J.-P. (2010). Numerical optimization of site selection for
offshore wind turbine installation using genetic algorithm.
Current Applied Physics, 10(2), S302–S306.
Lee, W.-S., Huang, A. Y., Chang, Y.-Y., & Cheng, C.-M. (2011).
Analysis of decision making factors for equity investment by
DEMATEL and Analytic Network Process. Expert Systems
with Applications, 38(7), 8375–8383.
Environ Monit Assess
Liou, J. J., Tzeng, G.-H., & Chang, H.-C. (2007). Airline safety
measurement using a hybrid model. Journal of Air Transport
Management, 13(4), 243–249.
MathWorks, I. (2005). MATLAB: the language of technical com-
puting. Desktop tools and development environment, version
7 (Vol. 9): MathWorks.
Moiloa, B. (2009). Geographical Information Systems for
Strategic Wind Energy Site Selection. Vrije Universiteit
Amsterdam Netherlands.
Nekhay, O., Arriaza, M., & Boerboom, L. (2009). Evaluation of
soil erosion risk using Analytic Network Process and GIS: a
case study from Spanish mountain olive plantations. Journal
of Environmental Management, 90(10), 3091–3104. doi:10.
1016/j.jenvman.2009.04.022.
Saaty, T. L. (1996). Decision making with dependence and feed-
back: the analytic network process.
Saaty, T. L. (2001). The analytic network process: decision making
with dependence and feedback: RWS Publ.
Sana (2010). Data Collection Guide for Non Government renew-
able power plant construction (Vol. 1). Tehran: Renewable
energy organization of Iran.
Sener, S., Sener, E., & Karaguzel, R. (2011). Solid waste disposal
site selection with GIS and AHP methodology: a case study
in Senirken t-Ulu borlu (Isparta) Basin, Tu rk ey.
Environmental Monitoring and Assessment, 173(1–4), 533–
554. doi:10.1007/s10661-010-1403-x.
Sliz-Szkliniarz, B., & Vogt, J. (2011). GIS-based approach for the
evaluation of wind energy potential: a case study for the
Kujawsko–Pomorskie Voivodeship. Renewable and
Sustainable Energy Reviews, 15(3), 1696–1707. doi:10.
1016/j.rser.2010.11.045.
Taghilo, A. (2011). Calculation of wind potential for wind power
construction plant (case study in Zanjan province). Tehran,
Iran: Tarbiat Modarres University.
Taghizadeh Herat, A., Noorossana, R., Parsa, S., & Shariatmadari
Serkini, E. (2012). Using DEMATEL—analytic network
process (ANP) hybrid algorithm approach for selecting im-
provement projects of Iranian excellence model in healthcare
sector. African Journal of Business Management, 6(2), 627–
645. doi:10.5897/ajbm11.2148.
Talinli, I., Topuz, E., Aydin, E., & Kabakcı, S. B. (2011). A holistic
approach for wind farm site selection by FAHP. Wind farm:
technical regulations, potential estimation and siting assess-
ment. InTech, Croatia, 213-234.
Tegou, L.-I., Polatidis, H., & Haralambopoulos, D. A. (2010).
Environmental management framework for wind farm siting:
methodology and case study. Journal of Environmental
View publication stats
Management, 91(11), 2134–2147. doi:doi:10.1016/j.
jenvman.2010.05.010.
Tsai, W.-H., & Chou, W.-C. (2009). Selecting management sys-
tems for sustainable development in SMEs: a novel hybrid
model based on DEMATEL, ANP, and ZOGP. Expert
Systems with Applications, 36(2), 1444–1458.
Tseng, M.-L. (2009). Application of ANP and DEMATEL to
evaluate the decision-making of municipal solid waste man-
agement in Metro Manila. Environmental Monitoring and
Assessment, 156(1–4), 181–197.
Tseng, M.-L., & Lin, Y. H. (2009). Application of fuzzy
DEMATEL to develop a cause and effect model of municipal
solid waste management in Metro Manila. Environmental
Monitoring and Assessment, 158(1–4), 519–533.
Tzeng, G.-H., Chiang, C.-H., & Li, C.-W. (2007). Evaluating
intertwined effects in e-learning programs: a novel hy-
brid MCDM model based on factor analysis and
DEMATEL. Expert Systems with Applications, 32(4),
1028-–1044.
van Haaren, R., & Fthenakis, V. (2011). GIS-based wind farm site
selection using spatial multi-criteria analysis (SMCA): eval-
uating the case for New York State. Renewable and
Sustainable Energy Reviews, 15(7), 3332–3340. doi:10.
1016/j.rser.2011.04.010.
Vasiljević, T. Z., Srdjević, Z., Bajčetić, R., & Miloradov, M. V.
(2012). GIS and the analytic hierarchy process for regional
landfill site selection in transitional countries: a case study
from Serbia. Environmental Management, 49(2), 445–458.
Vujanović, D., Momčilović, V., Bojović, N., & Papić, V. (2012).
Evaluation of vehicle fleet maintenance management indica-
tors by application of DEMATEL and ANP. Expert Systems
with Applications, 39(12), 10552–10563.
Wang, Y.-L., & Tzeng, G.-H. (2012). Brand marketing for creating
brand value based on a MCDM model combining
DEMATEL with ANP and VIKOR methods. Expert
Systems with Applications, 39(5), 5600–5615.
Yang, J. L., & Tzeng, G.-H. (2011). An integrated MCDM tech-
nique combined with DEMATEL for a novel cluster-
weighted with ANP method. Expert Systems with
Applications, 38(3), 1417–1424.
Yang, Y.-P. O., Shieh, H.-M., Leu, J.-D., & Tzeng, G.-H. (2008). A
novel hybrid MCDM model combined with DEMATEL and
ANP with applications. International Journal of Operations
Research, 5(3), 160–168.
Zahedi, M., Salahi, B., & Jamil, M. (2005). Calculation of wind
density and potentials for use of wind energy in Ardabil.
Journal of Research in Geography, 53, 41–55.