WIND TURBINE SITE SELECTION IN INDONESIA

BY

GALIH PAMBUDI

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE

REQUIREMENTS FOR THE DEGREE OF MASTER OF

ENGINEERING (LOGISTICS AND SUPPLY CHAIN SYSTEMS

ENGINEERING)

SIRINDHORN INTERNATIONAL INSTITUTE OF TECHNOLOGY

THAMMASAT UNIVERSITY

ACADEMIC YEAR 2018

WIND TURBINE SITE SELECTION IN INDONESIA

BY

GALIH PAMBUDI

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE

REQUIREMENTS FOR THE DEGREE OF MASTER OF

ENGINEERING (LOGISTICS AND SUPPLY CHAIN SYSTEMS

ENGINEERING)

SIRINDHORN INTERNATIONAL INSTITUTE OF TECHNOLOGY

THAMMASAT UNIVERSITY

ACADEMIC YEAR 2018

i

WIND TURBINE SITE SELECTION IN INDONESIA

A Thesis Presented

By

GALIH PAMBUDI

Submitted to

Sirindhorn International Institute of Technology

Thammasat University

MASTER OF ENGINEERING (LOGISTICS AND SUPPLY CHAIN SYSTEMS

ENGINEERING)

Approved as to style and content by

Advisor

(Asst. Prof. Dr. Narameth Nananukul)

Committee Member and

Chairperson of Examination Committee

(Asst. Prof. Dr. Morrakot Raweewan)

Committee Member

(Assoc. Prof. Dr. Thananya Wasusri)

NOVEMBER 2018

ii

Acknowledgements

The author gratefully acknowledges the financial support provided by the Excellent

Foreign Student Scholarship (EFS) for Graduate Student in Sirindhorn International

Institute of Technology, Thammasat University.

iii

Abstract

WIND TURBINE SITE SELECTION IN INDONESIA

by

GALIH PAMBUDI

Bachelor of Engineering, Universitas Gadjah Mada, 2016

Master of Engineering, Sirindhorn International Institute of Technology, 2018

Wind farm sites are selected in spacious regions which have more output potential

within constraint resources. Due to its spacious terrain, Indonesia has great potential

for building wind power plants, providing the perfect settings to generate electricity

using wind energy. Keeping in view the reliability and sustainability of wind farm sites,

the selection of the most suitable locations for optimal result is of prime concern to

generate greater amount of energy with less utilization of resources. In this study, the

focus is on proposing a multi-criterion approach to find the most suitable location for

building wind farms. Locations from every region of Indonesia were selected based on

two levels defined by district level to province level. All districts and provinces are

considered as Decision-Making Units (DMUs) which are used to measure the

efficiency scores using Dual Data Envelopment Analysis (DDEA) method. Two levels

are defined to find the best feasible locations within Indonesia from 165 districts and

33 provinces with major focus on geographical and structural technicality of each

DMU. The results show that South Sumatra province has the highest priority potential

for the construction of wind power plants, especially in the district of Palembang. West

Papua, Papua and Maluku provinces have descending priority based on good

infrastructure accessibility and less prone to natural disaster.

Keywords: Dual data envelopment analysis, Wind Power Plant, Site Selection,

Decision Making Unit.

iv

Table of Contents

Chapter Title

Page

Signature Page

i

Acknowledgements

ii

Abstract

iii

Table of Contents

iv

List of Tables

vi

List of Figures

vii

1 Introduction

1

1.1 Background of Propose Study

1

1.2 Problem Statement

2

1.3 Objectives of Propose Study

2

1.4 The Advantages of Propose Study

3

2 Literature Review

4

2.1 Literature Review

4

2.2 Research Gap

7

3 Research Methodology

8

3.1 Possible Factors

8

3.1.1 Level 1 criteria

8

3.1.2 Level 2 criteria

15

3.2 Methodology

18

3.2.1 Dual Data Envelopment Analysis

19

3.2.2 The Hierarchical Model for two Level DDEA

20

3.2.3 Fuzzy Primary Data Envelopment Analysis

21

v

3.2.4 Principal Component Analysis

23

4 Results and Discussion

27

4.1. Data Envelopment Analysis Results

27

4.2 The Hierarchical Model for two Level DDEA Results

32

4.3 Fuzzy Primary Data Envelopment Analysis Results

34

4.4 Principle Component Analysis Results

40

4.5 Comparison of Three Methods Result

50

5 Conclusions and Recommendations

53

5.1 Conclusion

53

5.2. Recommendations

54

References

55

Appendices

57

Appendix A Data Resources

58

A.1 Districts Level Data

58

A.2 Provinces Level Data

60

Appendix B General Optimization Model in IBM ILOG CPLEX

61

B.1 Model on Districts Level

62

B.2 Model on Provinces Level

63

vi

List of Tables

Tables

Page

2.1 The summary of case study

5

2.2 Summarizes the relevant criteria in the wind farm site selection

5

2.3 Summarizes the relevant methods in the wind farm site selection

6

3.1 Comparison of the land requirement in different power plant [16]

9

3.2 Cost analyst for 1 m length of road infrastructure

13

3.3 Wind class definitions [17]

18

4.1 Efficiency and ranking of provinces (Level 2)

27

4.2 Efficiency score of districts (Level 1)

28

4.3 Detail of hierarchical score for provinces level

31

4.4 Hierarchical score for five dominant provinces

33

4.5 Importance degree and context free grammar on HFLTS

34

4.6 Pairwise evaluations of one expert in main criteria on level 1

35

4.7 Obtained envelops for HFLTS

35

4.8 Pessimistic and optimistic preference in district level

36

4.9 The linguistic interval, interval utilities, midpoint and weights

37

4.10 Pessimistic and optimistic preference in province level

37

4.11 The constraint of the priorities for district level

38

4.12 The constraint of the priorities for province level

38

4.13 Hierarchical Score for HFLTS

39

4.14 Principal Component Analysis Results

50

4.15 Comparison of three methods result.

51

vii

List of Figures

Figures

Page

3.1 Maps of provinces in Indonesia

8

3.2 Land cost of districts in Indonesia

9

3.3 Types of road infrastructure

10

3.4 Types of road infrastructure in Medan North Sumatra

11

3.5 Distance of the primary and secondary road infrastructures in Medan

11

3.6 Total cost of infrastructure data

12

3.7 Data of population in districts

14

3.8 Ratio of usage area

14

3.9 Data of electricity consumption in Indonesia’s Provinces

15

3.10 Data of natural disaster in provinces

16

3.11 Gravity loading; a. full blade; b. spar-only simplification

17

3.12 Blade loading cases; a. edgewise bending; b. flap-wise bending

17

3.13 Data of wind velocity in provinces of Indonesia

17

3.14 Data of total area in provinces in Indonesia

18

3.15 Flow Chart of the Proposed Study

19

3.16 Extraction of Factor analysis in district level

24

3.17 Scree Plot of district level

24

3.18 Extraction Box

25

3.19 Descriptive Box

25

3.20 Rotation Box

26

3.21 Options Box

26

4.1 Hierarchical Score

31

4.2 Correlation matrix on district level

40

4.3 KMO and Bartlett’s Test on district level

41

4.4 Communalities on level 1

41

4.5 Total variance on district level

42

4.6 Component matrix on district level

43

4.7 Pattern matrix on district level

43

4.8 Structure matrix on level 1

44

viii

4.9 Scree plot for level 2

45

4.10 KMO and Bartlett’s Test for Level 2

45

4.11 Correlation matrix on level 2

46

4.12 Communalities on level 2

46

4.13 Total variance on level 2

47

4.14 Component matrix on level 2

47

4.15 Pattern matrix on level 2

48

4.16 Structure matrix on level 2

48

4.17 Component correlation matrix on level 2

49

4.18 Top five Provinces in Indonesia

52

5.1 Full Score of Three Methods

54

1

Chapter 1

Introduction

The propose study in the first chapter are divided into four sections. Section 1.1

gives background of the proposed study and show the importance of the research study.

Section 1.2 contains details of problem statement of proposed study to define the issue

of the research. Section 1.3 gives the objective of proposed study presents the

framework of this study. Section. Section 1.4 The advantages of the proposed study

provide the benefit of the study to apply in the research area.

1.1 Background

Natural energy resources such as wind energy is renewable, and is freely

available which could lead to the sustainability of energy usage. Selecting the most

suitable sites which have the optimal wind energy resource is a complicated decision-

making process. It is considered as primary concern based on the sustainability and

reliability aspects. The selection of the optimal locations is very important including

several factors the topography of the area and the usage of the decision support models

could fulfill the requisites and shows the optimal outcome. It means modelling,

formulation and determining solution of the site problem that can be implemented in

establishing facilities in the selected area. Different literatures show that there are

different approaches for selecting the optimal location for wind power plant site, as

follows Haydar et al. [1] defining the optimal area in university for a station of wind

observation based on Analytical Hierarchy Process (AHP) approach. Bhatnagar et al.

[2] the establishment of gas stations and power plants using location factor as multi-

criteria. Afshartous et al. [3] to determine the location of the coast guard air station

based on Improved Optimization Model. Gamboa et al. [4] determining wind plant site

selection as a multi-criterion used social framework. Choudhary et al. [5] determined

site selection of thermal power plant based on Fuzzy DEA. As it is evident from the

previous studies, the site selection is of prime concern for establishing a facility at some

place. It needs multitude factors to be considered, making the decision hard and

required complex modeling. In this study, the method based on Dual Data Envelopment

2

Analysis (DEA) approach with multi-criteria is used as site selection mechanism for

wind power plants construction within Indonesia. In Indonesia the energy demand is

growing dramatically than population. At present, Indonesia have six main types of

power plant use gas, steam turbines, combined cycle, geothermal, diesel engine, and

hydro-power where fossil fuel is the major energy generation [6]. In this decades, for

genererating the electricity in Indonesia, the resources up to 96% using fossil fuel and

just 4% uses renewable energy. Hence, the government policy targets a portion of

renewable energy resources to be increased up to 17% in 2025 [7]. The current energy

policy in Indonesia is central in Fossil fuel. Decreasing of fossil fuel resources and

growing environmental concerns are challanging viewpoint in Indonesia’s energy

policy which leads to the propose of using renewable energy to increase energy

efficiency [8]. Indonesia as a archipelago country, having huge potential for wind

power generation because of high wind rate in most of the regions. The criteria of wind

turbine site selection should be selected carefully before making decisions.

1.2 Problem Statement

Determining the potential of using the wind power in the possible region is

important. In spite of the comprehensiveness in location considered for the optimization

of wind power plants, the criteria and the method for the site selection that will be used

to compare the potential of the region must be carefully selected. Location problem

includes simulation, formulation and model in establish the facility in every region

which is likely to have multiple factors and is difficult for the analysis. The quantitative

approach must be used to determine the suitable locations.

1.2 Objectives of Propose Study

This study considers an integrated mathematical approach for location

optimization of wind plants. Determining all criteria that significantly influences the

establishment of a wind farm in Indonesia is important. The implementation of the

proposed approach to decide the most suitable location for building of a wind power

plant in Indonesia is based on a Dual Data Envelopment Analysis (DEA) for wind farm

power plant.

3

1.4 The Advantages of Propose Study

The advantages of this proposed study hopefully can be used as the alternative

approach to decide site selection, generally in any case and especially in wind plant

power plant. This proposed study can help improve the reseach which have correlation

with location optimization in wind farm location on the other location.

4

Chapter 2

Literature Review

As a consideration of the literature, the proposed study refers several studies

which have been reviewed as a reference. Section 2.1 show the insight of the literature

review. Section 2.2 presents research gap which is used in the proposed study. Herewith

is further description of the research and the comparison of the previous studies.

2.1 Literature Review

Data envelopment analysis (DEA) is for analyzing the performance efficiency

of the comparable units called decision-making units (DMUs) as quantitative method.

Every DMU performs the same purpose by using ratio between input and output criteria

which are characterized by the modeled system [9]. Several references which have used

DEA for site selection such as Ertek et al. [10] for determining the efficiency of on-

shore wind turbines they provided data centric analysis. Saglam, U [11] The goal of

those paper was to evaluate quantitatively efficiencies of 39 states wind power

performance for electricity generation by using multi-criteria methods as DEA. Wu et

al. [12] in China to perform efficiency assessment of wind power plant used based on

two stage of DEA. These studies identified potential inefficicient factors and try to seek

out the factor which can improve the performance of wind farm. Azadeh et al. [9]

provided wind farm site selection under uncertainty using Hierarchical Fuzzy DEA.

Since traditional DDEA models cannot be used to combine the indicators especially in

qualitative data. Sueyoshi et al. [13] proposed an approach improvement as Range

Adjusted Measure (RAM) which is as integrated of DEA. Seiford et al. [14] proposed

the results from multi-stage DEA involved the input and output criteria which are

validated by Numerical Taxonomy and Principal Component Analysi. In this study,

the efficiency of DMUs in the selection of most suitable location for wind farm plant is

based on land cost, road accessibility, infrastructure cost, population density, supply

demand, natural vulnerability, wind velocity and total area. This research proposes a

multi-criterion apporach based on Data Envelopment Analysis (DEA) for analyzing the

most feasible wind farm site selection in Indonesia.

5

According of the literature that have been reviewed, the summary of the case

study is shown in Table 2.1. Table 2.2. lists the criteria which are significant influence

in the site selection of wind farm. Therefore, the methods based on the quantitative

approach that have been used are shown in Table 2.3. Further information shows in the

describe as below:

Table 2.1 The summary of case study

No Author

Year Case Study

1

Saglam, Ümit

[11]

2017 efficiency assessments of 39 state’s wind power location

using A two-stage data envelopment analysis in the United

States

2

Yunna Wu, et

al [12]

2016 Efficiency assessment of wind farms location using two-stage

data envelopment analysis in China

3

Azadeh, Ali et

al [15]

2013 Location optimization of wind power generation systems

under uncertainty using hierarchical fuzzy DEA in Iran

4

Azadeh, Ali et

al [9]

2010 Location optimization of wind plants by an integrated

hierarchical Data Envelopment Analysis in Iran

5

Ertek, Gürdal

et al [10]

2012 Insights into the efficiencies of wind turbines using data

envelopment analysis

Table 2.2. Summarizes the relevant criteria in the wind farm site selection

No Author

Year DMU Input

Output

1

Saglam,

Ümit [11]

2017 39

Total

Project

Investment

($),

Annual Land Lease

Payments ($)

Average

wind

blow,

Wind Industry Employment,

Annual

Water

Savings

(Gallons),

CO2

Emissions

Avoided

(Tons)

2

Yunna Wu,

et al [12]

2016 42

Auxiliary electricity

consumption,

Wind power density

Electricity

generated,

Average wind blow

6

No Author

Year DMU Input

Output

3

Azadeh, Ali

et al [15]

2013 25

Level 1 Land Cost

Level 2 Intensity of

natural

disasters

occurrence,

Level 1 Population and human

labor, Distance of power

distribution networks,

Level 2 Average wind blow,

Quantity of proper geological

areas, Quantity of proper

topographical

areas,

Consumer proximity

4

Azadeh, Ali

et al [9]

2010 25

Level 1 Land Cost

Level 2 Intensity of

natural

disasters

occurrence,

Level 1 Population and human

labor, Distance of power

distribution

networks,

Level 2 Average wind blow,

Quantity of proper geological

areas, Quantity of proper

topographical areas

5

Ertek, Gürdal

et al [10]

2012 74

Diameter of Plant

Nominal

Wind

Speed

Nominal Output (kW)

Table 2.3. Summarizes the relevant methods in the wind farm site selection

No Author

Year

Primary

DEA

PCA NT Tobit

Hypothes

es Testing

1

Saglam, Ümit [11]

2017

v

v

2

Yunna Wu, et al [12]

2016

v

v

3

Azadeh, Ali et al [15]

2013

v

v

v

4

Azadeh, Ali et al [9]

2010

v

v

v

5

Ertek, Gürdal et al [10]

2012

v

v

Where: DEA (Data Envelopment Analysis), PCA (Principal Component Analysis), NT

(Numerical Taxonomy).

7

2.2 Research Gap

The research gap of this proposed study is wind farm site selection in province

of Indonesia using multi-criteria approach based on hierarchical dual Data

Envelopment Analysis (DEA). The integrated data envelopment analysis will be

applied on two levels of DEA, the first level considers finding the best suitable province

in Indonesia and the second level focuses on sub-district within the province based on

the distance from remote areas. The possible factors used in the districts level as defined

by land cost, population in region, ratio of free usage area, primary road, secondary

road, tertiary road, and total cost of infrastructure. In the provinces level as defined by

wind velocity, population in province, total area, electricity consumption, less of land

slide, flood, earthquake and volcanic eruption. Determining the efficiency for districts

and province level based on Hierarchical Dual Data Envelopment Analysis. Hesitant

Fuzzy Linguistic Term Set (HFLTS) for determining the weight for importance criteria.

The validation of the significant criteria based on Principal Component Analysis

(PCA). Finally, comparing three methods for deciding the most suitable location for

wind turbine site selection in Indonesia.

8

Chapter 3

Research Methodology

In this chapter further information about the criteria and methods used in this

proposed study is described. Section 3.1 provides description of the possible factors

which have influence to the wind farm site selection. Section 3.2 presents the methods

which are applied in this proposed study.

3.1 Possible Factors

Based on the literature review, the proposed factors used in this study are

districts (Level 1) and provinces (Level 2) of Indonesia as shown in Figure 3.1. The

integrated model for wind farm site selection organizes the factors into two levels

defined as input and output. The optimization technique is based on Dual Data

Envelopment Analysis method to find the most efficient location. The integrated level

criteria are developed to select the most suitable location in term of province of

Indonesia.

Fig.3.1 Maps of provinces in Indonesia

3.1.1 Level 1 Criteria

The objective of using level 1 criteria is to determine the most suitable province

in Indonesia for establishment of wind farm plant based on the efficiency of the

location. The Level 1 criteria are:

Land cost by districts in Indonesia: the land cost has become an important

criterion due to the unprecedented increase in Indonesian population, which must be

included for site selection. For selecting a wind farm site, it requires more spacious area

9

as compared with other energy sources. Table 3.1 shows the comparison of the amount

of the land required for the construction of each kind of facility [16]. Area required for

wind farm is up to 9900 km

2

/GW/year which is 283 times more than coal plant. Its

means that the land cost is the main important criterion for wind farm site selection.

Figure 3.2 shows the data of land costs in some districts in Indonesia.

Fig.3.2 Land cost of districts in Indonesia

Table 3.1 Comparison of the land requirement in different power plant [16]

Technology

Land use in km2/GW per year

Biomass

25,600

Wind power plant

9,900

Hydroelectric

7,900

Solar PV

630

Coal

35

Oil

20

Natural gas

20

Nuclear

10

The type of road infrastructure: Good road accessibility to the constructed

facility is one of the most importance considered factor for reliable, timely

transportation and distribution of goods to and from the facility. Different road facilities

(i.e., primary, secondary and tertiary roads) have different distribution lead time which

10

can affect the accessibility to the facility. Data of the road infrastructure are shown in

Figure 3.3. Primary and secondary roads are main roads and can be used for the

transportation of heavy goods. On contrary to this, tertiary roads are mostly used as

connectors to the primary and secondary roads such as, small bike roads and village

roads. Hence, for timely distribution of goods to and from the wind farm need the most

effective routes. In this study, primary and secondary road infrastructures are used as a

preference indicator for wind power plant construction.

Fig.3.3 Types of road infrastructure

Relevant data, especially types of road infrastructure within each district in

every province of Indonesia, will be used. Firstly, finding the ArcGIS Maps for every

district from the Ministry of Public Works and Public Housing of Indonesia’s data

representing the infrastructures maps in every region such as types of road as well as

natural resources. By using the ArcGIS, national roads, province roads are defined as

primary road, in the maps are shown as dark red line. The district and regional roads

are considered as secondary road type represented as light red line on the maps. The

tertiary road is one of important criterion which should be careful determined and set,

due to the limited resources in ArcGIS. They can be determined by looking for village

roads or small roads which are less than 3.6 m in width. In here, the types of road

infrastructure by ArcGIS maps and google maps. Fig. 3.4. represents the difference

types of primary roads and secondary roads.

11

Fig.3.4 Types of road infrastructure in Medan North Sumatra

The border point between the primary and secondary roads is used to determine

the distance of the roads. Fig.3.5 shows the distance of primary and secondary roads in

Medan Region. It shows that primary road as the significance distance in Medan City

is 18.6 km and for secondary road is 9.8 km. The distance is based on a spotted location

in remote region which is suitable to establish a wind farm. Due to the good

infrastructure in the main region of Medan, there is no need for tertiary road so this

region just have 0 km of it and do not need to build additional infrastructures.

Fig.3.5 Distance of the primary and secondary road infrastructures in Medan

12

Total cost of infrastructure (IDR): The construction of the infrastructure

requires a lot of capital cost incurred. So, selection of the site with less incurring cost

for building new roads to access the facility is also very much important to avoid extra

expenses thus increasing the overheads of the construction projects. The data in this

study is shown in Figure 3.6 which are the sites selected by the minimum capital cost

for the construction of tertiary roads with the shortest distance. In other words, if in any

case the construction of the infrastructure is inevitable and unavoidable, tertiary roads

are given preference over primary and secondary roads. In our approach we prefer the

construction of tertiary road as compared with secondary road if the width of the

dispatching vehicle is up to 5m. As the average width of the tertiary roads in Indonesia

is up to 3.6m and we included 1.4m to accommodate the convenience of shipment. So,

the minimum cost of infrastructure is another input criterion in DDEA model.

Fig.3.6 Total cost of infrastructure data

The total cost of infrastructure for each region is collected based on Widarno,

B et al (2015). The included components are labors, materials, tools and equipment as

shown on Table 3.2. as cost analyst for each 1 m length of road. As a result, the total

area of 1 m

3

is approximately 150,000 (IDR) (1 USD as 13,994.25 IDR). As an example,

from Fig.3.6. consider one of regions in South Sumatra where the selected region to

build infrastructure is Pagar Alan. In Pagar Alan, there are up to 159.1 km as tertiary

13

road and needed to expand the road for shipping the wind power materials to the

location. The total cost for building tertiary roads in Pagar Alan is 159.1 x 1000 m x

150,000 IDR which is 23,865,000,000 IDR.

Table 3.2 Cost analyst for 1 m

length of road infrastructure

Population by district in Indonesia: for maintenance and operational

technicality in case of emergency the wind power plant should be established in regions

with easily available human resources. It can likely decrease the resources management

for the transportation cost of labors, accommodation of labors and expert availability

when needed. The choice of a centered place with easy accessibility of human resources

is an important output indicator in the DDEA modal calculation. Figure 3.7 shows the

population in districts of Indonesia.

No Component

Dimension

Coefficient

Cost (IDR)

Total cost (IDR)

A

1

Labor

Hour

0.221

7,500

1,651

2

Foreman

Hour

0.0314

12,500

393

B

1

Aggregate B class

1m x 3.6m

1.2

140,000

68,000

C

1

Wheel loader

Hour

0.0314

375,000

11,775

2

Dump truck

Hour

0.1655

150,000

24,825

3

Motor grader

Hour

0.0092

355,000

3,266

4

Vibratory loader

Hour

0.008

316,365

25,310

5

Pneumatic tire loader

Hour

0.0115

345,725

2,976

6

Water tanker

Hour

0.0383

153,240

5,870

7

Assisted tools

Hour

1

D

145,066

Total cost of labor per m

3

Labors

Materials

Equipment and tools

14

Fig.3.7 Data of population in districts

Ratio of free usage area: Areas with greater value of free area usage ratio near

to one are considered more suitable for establishment of the wind power plants. The

free usage area means the ratio between total area divided by population in each region.

The more available land is preferred and used as output in DDEA modal calculation.

Figure 3.8 shows the data of the free usage area ratio.

Fig.3.8 Ratio of free usage area

15

3.1.2 Level 2 Criteria

In the second level, the criterion of DDEA is to find the most appropriate

province in Indonesia for constructing wind power plants based on the geographical

and technical structures as input and output indicators for DDEA modal calculation.

The indicators in this level are mentioned as below:

Electricity consumption: This criterion based on the consumption of the

electricity which have been recorded in every province by Giga Watt per Hour (GW/h)

including general activity of electricity usage which are shown on Figure 3.9.

Fig.3.9 Data of electricity consumption in Indonesia’s Provinces

Natural disaster: The probability occurrence of natural disasters in the region

have significant impact on wind farm site selection. The damage caused by natural

disasters such as flooding, volcanic eruption, earth quakes, and land sliding have

menaces effect on site selection. Figure 3.10. shows the data of natural disasters in

provinces of Indonesia. These disasters may accumulate extra cost of maintenance, thus

increasing the maintenance and operational overheads of wind power plants. Selection

of the safe sites is very core fundamental in decision making for selecting locations for

new facility. These four main parameters (i.e., flooding, volcanic eruption and earth

quakes, land sliding) are included in the list of natural disasters as input indicator.

16

Fig.3.10 Data of natural disaster in provinces

Wind velocity: The wind velocity is the most important and primary criteria

which must be included in the model. Every province has different wind rate based on

the geographical features. Areas with greater wind velocity are the most suitable

locations for economic growth of energy generation. Based on the data on Figure 3.11

shows that several provinces have varieties of wind velocities. Low wind speed (LWS)

and high wind speed (HWS) are based on different configurations such as wind

resources, aerodynamics, and structural design/ analysis [17]. The aerodynamics loads

are smaller per unit length for the LWS blades but the increased span means that total

forces are closer or larger than the equivalent HWS blade. Due to that for construction

a wind farm in Indonesia should use the technology namely low speed wind turbines.

The design on LWS and HWS blades differ in the blade’s lengths and the magnitude of

aerodynamic loads [17]. Average wind speed in provinces on Indonesia are including

in low to medium of wind speed according to Table 3.3. Low and medium wind speed

sites are mostly classified on Class II-IV. The design of low speed wind farm mostly

based on the blades structural design where blades typically lengthened versions up to

39 m. The materials for modern wind blades are primarily glass fiber reinforced

polymer structures.

17

Fig.3.11 Gravity loading; a. full blade; b. spar-only simplification

A wind turbine blade in low wind speed is a cantilever which is shown on Figure

3.11. Gravity loading causes edgewise bending, as illustrated in Figure 3.12, the

direction shows reverse twice per full rotation and on the maximum loading condition

as flap-wise bending when the wind direction is perpendicular on the blade [17].

Fig.3.12 Blade loading cases; a. edgewise bending; b. flap-wise bending

Fig.3.13 Data of wind velocity in provinces of Indonesia

18

Table 3.3 Wind class definitions [17]

Total area: Every province has different land use activities such as industrial

zones, housing schemes, available landscapes with respect to the total area of the

province. The province which has more spacious area is preferred. Because of the

importance of the available land it is considered as a great influential indicator in site

selection as output parameter.

Fig.3.14 Data of total area in provinces in Indonesia

Population by province in Indonesia: Generally, higher population in a

province is preferred hence it implies a higher supply for electricity. Areas with more

human being are given priority in order to minimize the cost of energy distribution to

the far or bounder places.

3.2 Methodology

In this study have a multi-criteria approach to find the most suitable location for

establishment of wind farms. Based on this study, the concept of location as efficiency

in sub-region is defined for wind plants location. Figure 3.15. presents the flow chart

of this proposed study. The proposed study is starting by defining of input and output

Parameter (m/s)

Class I

Class II

Class III Class IV

Average wind speed

10

8.5

7.5

6

19

factors that already mention before, finding the data and analyst it, then measuring the

amount of the decision-making unit to verify the amount of input and output. Location

analysis by Dual Data Envelopment Analysis (DEA) in two level and combine it to get

the rank of the DEA results, after that would like to validate the result of DEA using

principal component analysis model to verify the significant influence of the criteria to

the DMU rank. The location optimization of the wind plant by DEA model are shown

by the most suitable location in sub-region of province in Indonesia.

Fig.3.15 Flow Chart of the Proposed Study

The propose methodologies which are used in this research topics are explaining

briefly as below.

3.2.1 Dual Data Envelopment Analysis

This research considering multi-criteria approach to find the most suitable

location for founding a wind farm. Hence, this propose study consider location based

on the provinces and districts in Indonesia and find the appropriate location. Every

20

province and district in here become the decision-making units (DMUs) which are used

to measure efficiency score. Data envelopment analysis is a non-parametric and

multivariate method to measure DMUs implementaion. In this research, the method

based on the hierarchical dual form of DEA (DDEA) is used. The DMUs are calculated

using a mathematical method as Linear programming using empirical data of inputs

and output, then measure the performance scores using the ratio between input and

output to compare the performance scopes generated. The measure of efficiency for

DMU is given by following linear programming [18].

Minimize



1

. .

,

1, 2,...,

n

j

ij

io

j

s t

X

X

i

m





=



=



1

,

1, 2,...,

n

j rj

ro

j

Y

Y

r

s





=



=



0,

j

j



 

(1)

where:

𝜃

: the efficiency of DMU,

𝜆

𝑗

: weight given to DMU,

𝑖

: input of DMU i,

𝑟

: output of DMU r.

3.2.2 The Hierarchical Model for two Level DDEA

In this section, the hierarchical of total efficiency scores from two levels of

DDEA is illustrated. In Level 1, all districts are considered, where the districts in

province k is represented by a set J

k

using index j

k

. Level 2 as provinces level, where

there are

K provinces, and each province is given by a subscript k. Combining results

from both levels consists of three steps [9], which are explained as follows:

Step 1: Normalize in Level 1 by scaling each efficiency value by the average efficiency

of group

k .

/

,

k

k

k

k

k

kj

kj

k

k

kj

k

j

J

f

e

e e

e

J







=

= 













(2)

k

J represents the number of members in set

k

J .

Step 2: Calculate the combine efficiency by multiplying the scaled value of

k

kj

f

with the

efficiency of Level 2 (

k

e ).

k

k

kj

kj

k

g

f

e

=



(3)

21

Step 3: Scaling the value of

k

kj

g

.

k

kj

h

is the total score between two levels.

 

,

,

min

1

k

k

k

k

kj

kj

k j

kj

h

g

R R

g

=



=

(4)

3.2.3 Fuzzy Primary Data Envelopment Analysis

In the scope of the study, wind turbine site selection which is one of the most

important problems related to sustainable energy have many alternatives and multi

criteria decision-making. Due to the uncertainty of decision makers in criteria choices,

the hesitant decision-making approach based on hesitant fuzzy linguistic term sets

(HFLTS) is chosen for solution of the complex problem. The upper bound and lower

bound weight ratio between criteria are used for calculating on hierarchical primary

data envelopment analysis. The algorithm hesitant fuzzy linguistic term sets are

proposed by Yavuz et al [19] which provides the capability to deal with hesitancy of

decision makers in assessment. The main of HFLTS is aim to advance flexibility and

completeness of linguistic importance based on the fuzzy linguistic approach.

Linguistic term is relating to language name which is used mostly in fuzzy to define the

uncertainty relation. Context free grammar such as at most, between and so on is the

figure for dealing with uncertainty relation. This algorithm combines the linguistics

term sets with context free grammar to handle the complexity of multi-criteria problems

with hierarchical structure using fuzzy approach.

The steps of the algorithm are shown as below,

Step 1. Defining the linguistic term sets S .

S = {no importance (n), very low importance (vl), low importance (l), medium

importance (m), high importance (h), very importance (vh), absolute importance (a)}.

Step 2. Defining the context-free grammar G

H.

G

H

= {lower than, greater than, at least, at most, between, and}.

Step 3. Collecting the preference relations provided by experts (

k

p

).

Step 4. Transforming the preference relations into HFLTS.

Step 5. Obtaining the envelope between pesimistic and optimistic preference relation

.

22

1

1

n

i

i

n





=





= 









,

i



= round assigns in integer number.

Step 6. Computing the pessimistic and optimistic collective preference by linguistic

aggregation.

Step 7. Build the intervals utilities for the collective preference

Step 8. Normalize the collective interval vector to get the weight scores.

Ten experts from academician, NGO on renewable energy, Integrated energy

and environmental planning and policy of Indonesia, Engineers in wind turbine project

in Indonesia, and Technical officer at ASEAN Center for Energy have been asked to

evaluate the wind turbine site selection criteria in Indonesia using their expertise by

filling the fuzzy questionnaire.

Step 9. For every input and output (q, r), the weight ratio v

q

/ u

r

must be bounded by L

qr

(lower bound) and U

qr

(upper bound) as L

qr

≤ v

q

/ u

r

≤ U

qr.

The example of the weight

ratio is the relation on lower bound as pessimistic in district level between land cost and

population in region is 3.00 and upper bound is 4.20 (see Table 4.8). The lower bound

weight ratio is (land cost (LC)/population in region (PinR)) ≥ 3.00. The upper bound

weight ratio is similar as (LC/PinR) ≤ 4.20. The same procedure is carried to all criteria

to calculate the priorities.

The fuzzy set can be in combined into the primary Data Envelopment Analysis

is determined by Amy H.I Lee at al [20]. In the early stage the fuzzy analytic hierarchy

process is applied to extract expert’s questionnaire to set the pairwise comparison

values which have been introduced from step 1 to step 9. The bounded weight ratio is

designed to measure the data envelopment analysis (DEA) efficiency of a specific

DMU. DMU is a unit under evaluation in here as provinces and districts level. The

primary data envelopment analysis can be expressed by [20]:

1

1

Max

R

r rk

r

Q

q

qk

q

u Y

v X

=

=





(5)

23

Subject to

1

1

1

R

r

rk

r

Q

q

qk

q

u Y

v X

=

=







1

1

,

1,... ..., R

, q

1,...q..., Q

r

Q

q

qk

q

q

Q

q

qk

q

u

r

r

v X

v

v X





=

=



=



=





,

1,... ,

1...

q

qr

qr

r

v

L

U

r

R q

Q

u





=

=

Where

q

v

is the weight given to the q-th input and

r

u is the weight output to the

r-th output.

qk

X

is the amount of the q-th input of the k’-th DMU,

rk

Y is the amount of

the r-th output. Q is the number of inputs and R is the number of outputs and K is the

DMUs.

3.2.4 Principal Component Analysis

Reducing the number of variables under study and consequently ranking and

analysis of decision-making units (DMUs). The objective of PCA [12] is to reduce

ineffective indicators and also as a ranking methodology for determination the

efficiency of different units from the results of DEA. Discussing about principal

component analysis in here using IBM SPSS for knowing the importance of component.

The illustration how to find the importance criteria is applied in district level. The first

step, knowing how many components to extract in the analysis and looking on the Scree

plot by going to Analysis menu then dimension reduction and choose factor analysis is

illustrated in Fig. 3.16.

24

Fig.3.16 Extraction of Factor analysis in district level

Scree plot help to look for how many components should be extracted is shown

in Fig 4.17.

Fig.3.17 Scree Plot of district level

Looking at the break seems to be at about after the first three components so the

first three components definitely look like meaningful legitimate components and then

there’s a specific estrade component and it looks like the third component might be

something worth extracting. There is a more sophisticated approach to evaluate how

many components should extract an analysis in parallel analysis.

Based on the scree plot, will be extracted total for three eigen values consisting

of two eigen values which have values greater than 1 and one eigen value close to 1

from the analysis that’s why have to do it in three steps to analyze again in dimension

reduction. Choosing analyze with correlation matrix due to the variable are measured

in different units, this implies normalizing all variables using division by their standard

deviation is given in Fig. 3.18.

25

Fig.3.18 Extraction Box

The next step is chosen descriptive box and checklist on Coefficient, KMO and

Bartlett’s test of sphericity, and Univariate Descriptive is shown in Fig. 3.19. Going to

get the descriptive box to look at correlation matrix on Coefficient and KMO and

Bartlett’s test of sphericity as ferocity to tell whether should actually be doing of

component analysis to begin with and would typically want to look at univariate

descriptive x in any case.

Fig.3.19 Descriptive Box

Rotation Box is chosen for the next step is illustrated in Fig. 3.20. Choosing

Direct Oblimin as the rotation method. Direct Oblimin is an approach to produce an

oblique factor rotation that means the factors solution can be actually correlated with

each other and mostly used as familiar. If the factor solution is the most appropriate an

orthogonal uncorrelated effective solution then yield can be shown as a more or less

26

oblique orthogonal factor solution. Correlations between the three components that

have been extracted.

Fig.3.20 Rotation Box

Another options that’s good is wanting to sort the components factor loadings

more accurately. In this case to be sorted by size which makes it much easier to interpret

a component pattern matrix is given in Fig. 3.21 on Options Box.

Fig.3.21 Options Box

After interpreting the results, the significance criteria are obtained by principal

component analysis are used to measure the efficiency of the location both on district

level and province level. The multi-criteria approach based on the hierarchical Dual

Data Envelopment Analysis in Sub Section 3.2.1 and 3.2.2 are used to measure

efficiency score.

27

Chapter 4

Results and Discussion

4.1. Data Envelopment Analysis Results

In the proposed hierarchical Dual Data Envelopment Analysis model, 33

provinces at Level 2 and 165 districts at Level 1 in Indonesia are used to define DMUs

for wind farm sites. The data are collected from the Statistical Department of Indonesia,

Internal Ministry of Indonesia, Indonesian Agency for Meteorology, Climatology, and

Geophysics, and The National Land Agency of Indonesia. Overall data are mentioned

in the Appendix A. Measuring the data assessment based on DDEA and hierarchical

methods from section 3.2.1 and 3.2.2. Level 1 calculates for measuring the performance

score for districts level. Level 1 becomes the basic level for combining with the score

from Level 2 where is provinces level.

The score of efficiency at the provincial level are shown in Table 4.1. The

province efficiency represents the priority of each province based on the location

resources.

Table 4.1 Efficiency and ranking of provinces (Level 2)

No

Province

Efficiency No Province

Efficiency

1

West Papua

1.000

18

East Kalimantan

0.500

2

Papua

1.000

19

Riau

0.408

3

Maluku

1.000

20

East Java

0.449

4

East Nusa Tenggara

1.000

21

Southeast Sulawesi

0.423

5

Gorontalo

0.991

22

Jambi

0.457

6

South Sumatra

0.949

23

DI Yogyakarta

0.329

7

West Kalimantan

0.808

24

Central Sulawesi

0.387

8

West Sulawesi

0.791

25

West Sumatra

0.319

9

Lampung

0.816

26

Bengkulu

0.397

10

North Maluku

0.842

27

North Sulawesi

0.344

11

Central Kalimantan

0.744

28

DKI Jakarta

0.362

12

South Sulawesi

0.761

29

Bali

0.331

13

South Kalimantan

0.695

30

Banten

0.258

14

Riau Islands

0.656

31

Central Java

0.207

15

Aceh

0.525

32

West Nusa Tenggara

0.207

16

North Sumatra

0.525

33

West Java

0.063

17

Bangka Belitung Islands

0.536

28

The 165 districts efficiency and rankings as Level 1 from 33 provinces of

Indonesia are given in Table 4.2. It shows that the most suitable district for establishing

a wind power plant is in Palembang, one of district in province of South Sumatra. The

location of this districts is on the remote of the province, one of the public facilities as

good transportation infrastructure to ship wind power plant materials by both river and

road transportation is mainly advantages. The geographical location also giving benefit

to the regions due to Palembang is less occur able to natural disasters. The wind rate as

natural resources with a decent average wind speed that can be used for economical

electricity generation.

Table 4.2 Efficiency score of districts (Level 1)

Province

District

Eff

Rnk Province

District

Eff

Rnk

Aceh

Lhokseumawe

0.348 73

West Nusa

Tenggara

Mataram

0.179 120

Banda Aceh

0.338 75

Bima

0.478 48

Langsa

0.330 76

Dompu

0.234 102

Subulussalam

0.099 152

East

Lombok

0.702 25

Sabang

0.854 15

Sumbawa

0.537 42

North

Sumatra

Medan

0.403 60

East Nusa

Tenggara

Kupang

0.201 109

Tebing Tinggi

0.123 142

Alor

0.251 94

Tanjung Balai

0.126 140

Belu

0.467 50

Pematangsiantar 0.190 115

Ngada

0.248 95

Padang

Sidempuan

0.183 118

Southwest

Sumba

0.370 65

West

Sumatra

Padang

0.564 38

West

Kalimantan

Pontianak

0.950 12

Bukit Tinggi

0.069 158

Singkawang 0.158 129

Payakumbuh

0.175 121

Bengkayang 0.267 88

Pariaman

0.120 143

Landak

0.358 67

Solok

0.098 154

Kubu Raya

0.495 46

Riau

Pekanbaru

0.162 127

Central

Kalimantan

Palangka

Raya

0.054 161

Dumai

0.100 150

Seruyan

0.389 61

Kampar

0.114 146

Gunung

Mas

0.406 59

Rokan Hilir

0.509 44

South

Barito

0.211 107

Siak

0.593 36

Pulang

Pisau

0.497 45

29

Province

District

Eff

Rnk Province

District

Eff

Rnk

Jambi

Jambi

0.156 132

South

Kalimantan

Banjarmasin 0.380 63

Sungaipenuh

0.166 123

Banjarbaru

0.203 108

Merangin

0.200 110

Balangan

0.127 139

Sarolangun

0.199 111

Barito Kuala 0.621 32

Tebo

0.268 86

Tabalong

0.412 58

South

Sumatra

Palembang

1.000 1

East

Kalimantan

Balikpapan

0.096 155

Pagar Alam

0.136 137

Samarinda

0.116 144

Lubuk Linggau

0.338 74

Bontang

0.219 104

Prabumulih

0.256 93

Paser

0.297 81

Lahat

0.420 57

Tarakan

0.244 96

Bengkulu

Bengkulu

0.665 27

North

Sulawesi

Manado

0.561 39

Kaur

0.275 84

Bitung

0.240 101

Rejang Lebong

0.661 28

Tomohon

0.155 133

Seluma

1.000 1

Minahasa

0.358 69

Kepahiang

0.704 24

Kotamobagu 0.164 124

Lampung

Bandar

Lampung

0.431 55

Central

Sulawesi

Palu

0.130 138

Metro

0.257 92

Parigi

Moutong

0.356 70

Pesawaran

0.912 13

Donggala

0.212 106

Tanggamus

0.535 43

Banggai

0.424 56

Mesuji

0.474 49

Poso

0.641 30

Bangka

Belitung

Islands

Pangkal Pinang 0.164 125

South

Sulawesi

Makasar

0.990 11

Bangka

0.281 83

Palopo

0.618 33

Belitung

0.163 126

Sidrap

0.743 17

West Bangka

0.192 114

Parepare

0.453 53

East Belitung

0.144 135

Maros

0.719 22

Riau

Islands

Batam

0.266 89

Southeast

Sulawesi

Kendari

0.046 164

Bintan

0.061 159

Baubau

0.656 29

Tanjungpinang

0.262 90

Muna

0.366 66

Lingga

0.159 128

Kolaka

0.724 19

Karimun

0.213 105

Wakatobi

0.303 79

DKI

Jakarta

South Jakarta

0.111 147

Gorontalo

Gorontalo

City

0.354 71

Central Jakarta

0.138 136

North

Gorontalo

0.537 41

East Jakarta

0.242 99

Bone

Bolango

0.385 62

West Jakarta

0.145 134

Pohuwato

0.810 16

North Jakarta

0.185 117

Boalemo

0.609 34

30

Province

District

Eff

Rnk Province

District

Eff

Rnk

West Java

Bandung

0.275 85

West

Sulawesi

Mamuju

0.460 52

Bogor

0.108 149

Majene

0.314 78

Sukabumi

1.000 7

Polewali

Mandar

1.000 1

Tasikmalaya

0.633 31

Mamasa

0.728 18

Cimahi

0.157 130

North

Mamuju

0.267 87

Central

Java

Semarang

0.243 98

Maluku

Ambon

0.687 26

Jepara

0.867 14

Seram

0.348 72

Pekalongan

0.440 54

Tual

0.719 21

Surakarta

0.092 156

Aru

1.000 1

Magelang

0.482 47

Buru

0.547 40

DI

Yogyakarta

Yogyakarta

0.024 165

North

Maluku

Ternate

0.100 151

Sleman

0.190 116

Sula

0.156 131

Bantul

0.181 119

Morotai

0.225 103

Kulon Progo

0.076 157

Tidore

0.196 112

Gunung Kidul

0.577 37

Halmahera

0.172 122

East Java

Surabaya

0.462 51

West Papua

Manokwari

0.301 80

Pasuruan

0.373 64

Sorong

0.110 148

Malang

0.193 113

Fakfak

1.000 7

Kediri

0.282 82

Sorong City

0.705 23

Probolinggo

0.721 20

Bintuni

1.000 7

Banten

Tangerang

0.124 141

Papua

Jayapura

0.358 68

Serang

1.000 1

Merauke

0.243 97

Lebak

1.000 7

Biak

Numfor

0.260 91

Cilegon

0.319 77

Nabire

0.240 100

Pandeglang

1.000 1

Mimika

0.606 35

Bali

Denpasar

0.116 145

Gianyar

0.052 162

Buleleng

0.098 153

Bangli

0.049 163

Klungkung

0.054 160

Both efficiency between Level 1 and Level 2 are combined as hierarchical score

using hierarchical model for two level DMU in sub section of 3.2.2. Hierarchical Score

means the final score of hierarchical for two level of district and province as the

combination of efficiency score by both levels using DDEA. The ranking of province

is based on full efficiency. The result shows that the best location is in The South

31

Sumatra province, especially in Palembang district. West Papua, Papua, and Maluku

provinces also have high efficiency scores which are shown on Figure 4.1.

Fig.4.1 Hierarchical Score

Table 4.3 Detail of hierarchical score for provinces level

Province

The most

influence

district

Eff

District

Rank

of

Dist.

Eff

Hierarchical

Score

Ranking

South

Sumatera

Palembang

1.000

1

0.949

1.3404

1

West Papua

Fakfak

1.000

7

1.000

1.3391

2

Papua

Mimika

0.606

35

1.000

1.1661

3

Maluku

Aru

1.000

1

1.000

1.1055

4

East Nusa

Tenggara

Belu

0.467

50

1.000

1.1005

5

Gorontalo

Pohuwato

0.810

16

0.991

1.0745

6

West

Kalimantan

Pontianak

0.950

12

0.808

0.9024

7

West Sulawesi Polewali

Mandar

1.000

1

0.791

0.7802

8

Lampung

Pesawaran

0.912

13

0.816

0.7796

9

North Maluku Morotai

0.225

103

0.842

0.7532

10

Central

Kalimantan

Pulang

Pisau

0.497

45

0.744

0.6971

11

32

Province

The most

influence

district

Eff

District

Rank

of

Dist.

Eff

Hierarchical

Score

Ranking

South

Sulawesi

Makasar

0.990

11

0.761

0.6149

12

South

Kalimantan

Barito

Kuala

0.621

32

0.695

0.6015

13

Riau Island

Batam

0.266

89

0.656

0.4984

14

Aceh

Sabang

0.854

15

0.525

0.3848

15

North

Sumatera

Medan

0.403

60

0.525

0.3447

16

Bangka

Belitung

Island

Bangka

0.281

83

0.536

0.3061

17

East

Kalimantan

Paser

0.297

81

0.500

0.2888

18

Riau

Siak

0.593

36

0.408

0.2517

19

East Java

Surabaya

0.462

51

0.449

0.2421

20

South East

Sulawesi

Kolaka

0.724

19

0.423

0.2414

21

Jambi

Tebo

0.268

86

0.457

0.2038

22

DI Yogyakarta Gunung

Kidul

0.577

37

0.329

0.201

23

Central

Sulawesi

Poso

0.641

30

0.387

0.188

24

West

Sumatera

Padang

0.564

38

0.319

0.1821

25

Bengkulu

Seluma

1.000

1

0.397

0.1766

26

North

Sulawesi

Manado

0.561

39

0.344

0.1493

27

DKI Jakarta

East

Jakarta

0.242

99

0.363

0.1415

28

Bali

Denpasar

0.116

145

0.331

0.125

29

Banten

Serang

1.000

1

0.258

0.0874

30

Central Java

Jepara

0.867

14

0.207

0.0591

31

West Nusa

Tenggara

East

Lombok

0.702

25

0.207

0.0518

32

West Java

Sukabumi

1.000

7

0.063

0.0064

33

4.2 The Hierarchical Model for two Level DDEA Results

The methodology for combining the results of two levels between province and

district levels using three steps procedure from section 3.2.2. The hierarchical scores

show the Dual DEA Results based on province level. Firstly, collecting two levels

efficiency. Then using first step to normalize every district with average districts in one

province. Repeating for every province, in here showing for 5 dominant provinces

33

which will discuss further with another methodology in the following sub section. After

normalizing the district efficiency, then combining with province level by multiplying

it with province efficiency which will give the hierarchical score for every district. The

last step is scaling the hierarchical district score with the highest one in each province

to get the maximize one. Finally, the hierarchical score for each province is determined

by averaging the results from the third step. The detail results are shown in Table 4.4.

Table 4.4 Hierarchical score for five dominant provinces

Province

Prov

Eff

District

Dist

Eff

Step 1

Step 2

Step 3

Hierarchical

Score

South

Sumatra

0.949

Palembang

1.000

2.325

2.208

4.874

1.340

Pagar Alam 0.136

0.316

0.300

0.090

Lubuk

Linggau

0.338

0.786

0.747

0.557

Prabumulih 0.256

0.595

0.565

0.319

Lahat

0.420

0.978

0.928

0.862

West

Papua

1.000

Manokwari 0.301

0.483

0.483

0.233

1.339

Sorong

0.110

0.176

0.176

0.031

Fakfak

1.000

1.605

1.605

2.576

Sorong

City

0.705

1.131

1.131

1.280

Bintuni

1.000

1.605

1.605

2.576

Papua

1.000

Jayapura

0.358

1.049

1.049

1.100

1.166

Merauke

0.243

0.713

0.713

0.508

Biak

Numfor

0.260

0.760

0.760

0.578

Nabire

0.240

0.703

0.703

0.495

Mimika

0.606

1.775

1.775

3.150

Maluku

1.000

Ambon

0.687

1.041

1.041

1.083

1.105

Seram

0.348

0.527

0.527

0.278

Tual

0.719

1.089

1.089

1.187

Aru

1.000

1.514

1.514

2.293

Buru

0.547

0.828

0.828

0.686

East Nusa

Tenggara

1.000

Kupang

0.201

0.653

0.653

0.427

1.101

Alor

0.251

0.818

0.818

0.669

Belu

0.467

1.519

1.519

2.308

Ngada

0.248

0.805

0.805

0.649

Southwest

Sumba

0.370

1.204

1.204

1.450

34

4.3 Fuzzy Primary Data Envelopment Analysis Results

Due to a lot remaining for the expert to decide with their subjective judgement

and expertise. Ten experts have been informed with the objective information and asked

to fill the significance of decision-making criteria using their expertise. After the

importance degree and the context free grammar are built in the first and the second

steps which are shown in the Table 4.5, then collecting preference relations were

collected from experts. The fuzzy questionnaire based on importance degree and

context free grammar to apply with the criteria in Level 1 and Level 2 are designed. In

here we do not show all relations matrices here, we show one example for all steps. The

illustration here is one of seven main criteria in district level. For province level, we

show the result as well in the following steps to combine the results by HFLTS to get

the hierarchical score using data envelopment analysis method.

Table 4.5 Importance degree and context free grammar on HFLTS

Number

Importance Degree

Context free grammar

0

No importance (n)

lower than

1

Very low importance (vl)

greater than

2

Low importance (l)

at least

3

Medium importance (m)

at most

4

High importance (h)

between

5

Very high importance (vh)

and

6

Absolute importance (a)

The expert evaluation data shows in Table 4.6. is the one of expert evaluation

of the main criteria in district level with respect to the goal. Firstly, shows as discreate

sets and then converted to intervals. For example, the first expert preference the land

cost (LC) in relation to population in region (PinR) is “at least low importance” in

relation of linguistic terms and can be expressed in the discreate set as {low importance

(l), absolute importance (a)} as the interval set term [l,a], similarity for all relation term

set between every criteria in one expert linguistic evaluations. These evaluations are

proposed for ten experts for every level. After converting the relations term to interval,

the data were collected to determine envelops based on expert evaluations which are

shown in Table 4.7.

35

Table 4.6 Pairwise evaluations of one expert in main criteria on level 1

LC

PinR

RF

PR

SR

TR

TCI

Expert1s Linguistic Evaluations

LC

-

at least

l

between

l and m

is m

between

l and m

between

l and m

is l

PinR at most h

-

is vh

between

h and vh

between

h and vh

at most

vh

between h

and vh

RF

between m

and h

is vl

-

between

h and vh

between

h and vh

at most

h

between

vl and l

PR

is m

between

vl and l

between

vl and l

-

is h

is vh

between h

and vh

SR

between m

and h

between

vl and l

between

vl and l

is l

-

is h

is l

TR

between m

and h

at least

vl

at least

l

is vl

is l

-

is vl

TCI

is h

between

vl and l

between

h and

vh

between

vl and l

is h

is vh

-

Table 4.7 Obtained envelops for HFLTS

E1

LC

PinR

RF

PR

SR

TR

TCI

LC

-

[l,a]

[l,m]

[m,m]

[l,m]

[l,m]

[l,l]

PinR

[n,h]

-

[vh,vh]

[h,vh]

[h,vh]

[n,vh]

[h,vh]

RF

[m,h]

[vl,vl]

-

[h,vh]

[h,vh]

[n,h]

[vl,l]

PR

[m,m]

[vl,l]

[vl,l]

-

[h,h]

[vh,vh]

[h,vh]

SR

[m,h]

[vl,l]

[vl,l]

[l,l]

-

[h,h]

[l,l]

TR

[m,h]

[vl,a]

[l,a]

[vl,vl]

[l,l]

-

[vl,vl]

TCI

[h,h]

[vl,l]

[h,vh]

[vl,l]

[h,h]

[vh,vh]

-

In the interval set for every evaluation represent the pessimistic in left hand site

and optimistic in right hand side as [P,O]. In here we show the calculation for

pessimistic and optimistic preference using two operations. The scale of the importance

degree is shown in Table 4.5. to the linguistic terms. Table 4.8. shows the pessimistic

and optimistic values. For instance, we show one of the examples for pessimistic and

optimistic preference by land cost (LC) with respect to Population in Region (PinR)

criteria is calculated as follows:

36

Pessimistic preference.

Optimistic preference.

Table 4.8 Pessimistic and optimistic preference in district level

Level 1

LC

PinR

RF

PR

SR

TR

TCI

P

O

P

O

P

O

P

O

P

O

P

O

P

O

LC

-

-

3.0 4.2 3.2 4.0 2.7 4.0 1.9 4.2 2.6 4.2 1.4 2.4

PinR

1.8 3.0 -

-

3.4 4.4 3.0 3.6 3.7 4.6 3.0 4.1 2.4 2.7

RF

2.0 2.8 1.6 2.6 -

-

2.1 3.2 2.0 4.3 1.6 3.5 1.0 1.4

PR

2.1 2.8 2.4 3.0 2.8 3.9 -

-

3.5 4.0 4.7 5.1 1.2 2.2

SR

1.8 4.1 1.5 2.3 1.7 4.0 2.0 2.5 -

-

3.0 4.0 1.3 2.0

TR

1.9 3.4 1.9 3.0 2.5 4.4 0.9 1.3 2.0 3.0 -

-

1.3 1.6

TCI

3.6 4.6 3.3 3.5 4.6 5.0 3.8 4.8 4.0 4.5 4.4 4.7 -

-

The next step is looking for the linguistic intervals. The linguistic intervals are

calculated by using the average of pessimistic and optimistic values. For example, using

in one criterion as land cost (LC) as follows:

(

)

12

12

12

12

1

1

1

1

1

1

1

1

1

1

1

( , 2)

( ,1)

( , 3)

( ,1)

( , 4)

( , 0)

(

, 5)

( , 4)

( , 6)

( , 4)

10

1

(2 1 3 1 4 0 5 4 6 4)

10

(3.00)

( ,.00)

l

vl

m

vl

h

n

vh

h

a

h

L

L

L

m

L

P

P

P

P

−

−

−

−

−

−

−

−

−

−

−

−

−

−





= 

+

+

+

+

+

+

+

+

+













= 

+ + + + + + + + +









= 

=





















(

)

12

12

12

12

1

1

1

1

1

1

1

1

1

1

1

( , 6)

( , 2)

( , 4)

( , 3)

( , 4)

( , 2)

( , 6)

(vh, 5)

( , 6)

( , 4)

10

1

(6 2 4 3 4 2 6 5 6 4)

10

(4.20)

(h,.20)

a

l

h

m

h

l

a

a

h

L

L

L

L

P

P

P

P

+

−

−

−

−

−

−

−

−

−

−

+

+

+





= 

+

+

+

+

+

+

+

+

+













= 

+ + + + + + + + +









= 

=





















(

)

(

) (

)

(

)

12

12

6

6

1

1

1

,

6

1

1

(3.00 3.20 2.70 1.90 2.60 1.40) ,

(4.20 4.00 4.00

4.20 4.20 2.40)

6

6

(2.467), (3.833

,.467 ,

,

167

)

.

L

L

l

h

P

P

−

+































 





+

+

+

+

+



+

+

+

+

+



 









 







−









37

The linguistic intervals are converted to interval utilities as known as the value

to get the midpoint by the average between pessimistic and optimistic values. The

weight value is obtained by normalizes the midpoint.

The linguistic interval, interval utilities, midpoint and weights of all seven

criteria in district level are given in Table 4.9.

Table 4.9 The linguistic interval, interval utilities, midpoint and weights

Criteria

Linguistic

intervals

interval utilities

Midpoints Weights

P

O

P

O

LC

l,.467

h,-.167

2.467

3.833

3.150

0.150

PinR

m,-.117 h,-.267

2.883

3.733

3.308

0.158

RF

l,-.283

m,-.033

1.717

2.967

2.342

0.112

PR

m,-.217 h,-.500

2.783

3.500

3.142

0.150

SR

l,-.117

m,.150

1.883

3.150

2.517

0.120

TR

l,-.250

m,-.217

1.750

2.783

2.267

0.108

TCI

h,-.050

vh,-.483

3.950

4.517

4.233

0.202

After getting the weight ratio for every criterion in district level. We need to

look for the efficiency by using the ratio as constraint of criteria. Table 4.8. is ratio

relation of criteria in for province level is given in Table 4.10.

Table 4.10 Pessimistic and optimistic preference in province level

Level

2

WV

PinP

TA

EC

LL

LF

LE

LVE

P

O

P

O

P

O

P

O

P

O

P

O

P

O

P

O

WV

-

-

5.1 5.5 4.3 5.4 3.2

5

4.5

5

4.2 4.7 3.5 4.7 3.1 5.3

PinP

0.7

0.9

-

-

3

3.9 2.5 3.5

3

4.2 3.4 4.5 3.5 4.8 3.1

4

TA

0.6

1.7

2.1

3

-

-

2.5 3.1 2.7

4

1.4 3.4 2.2

4

2.1 3.8

EC

1

2.8

2.5 3.5 2.9 3.5

-

-

2.8 4.3 2.7 4.1 3.1 4.6 3.5 4.5

LL

1

1.5

1.8

3

2

3.4 1.8 3.2

-

-

3.8 3.9 2.4 3.3 3.1 3.5

LF

1.3

1.8

1.5 2.6 2.5 4.6 1.9 3.2 1.9 2.2

-

-

2.7 2.9 2.7 3.4

LE

1.3

2.5

1.2 2.5 1.8 3.5 1.4 2.9 2.7 3.9 3.1 3.3

-

-

3.2

4

LVE

0.7

2.9

2

2.9 1.9 3.7 1.5 2.5 2.5 2.9 2.6 3.3

2

2.8

-

-

3.150

3.150 3.308 2.342 3.142 2.517 2.267 4.233

0.150

Weights

Weights

=

+

+

+

+

+

+

=

38

Pairwise comparison matrix is performed based on the fuzzy aggregation in

Table 4.8 for district level and Table 4.10 for province level. The constraints show the

lower bound and upper bound values as pessimistic and optimistic priorities in fuzzy

matrix, for showing the example of the priority range in district level as Step 9 in Sub-

Section 3.2.3. The constraints of the priorities for each criterion are given in Table 4.11

for district level and Table 4.12 for province level. Due to the space constraints in here

just for showing one example of the constraints for district level as Palembang district.

South Sumatra as representing for province level. The same procedure is applied for

each region in district and province level to get location efficiency.

Table 4.11 The constraint of the priorities for district level

Table 4.12 The constraint of the priorities for province level

WV

PinP

TA

EC

LL

LF

LE

P

O

P

O

P

O

P

O

P

O

P

O

P

O

WV

-

-

0.0

0

0.0

0

0.0

0

0.0

0

0.5

6

0.8

8

0.7

9

0.8

8

0.9

6

1.0

7

0.8

6

1.1

5

Pin

P

0.1

2

0.1

6

-

-

0.0

0

0.0

0

0.4

4

0.6

1

0.5

3

0.7

4

0.7

8

1.0

3

0.8

6

1.1

8

TA

0.1

1

0.3

0

0.0

0

0.0

0

-

-

0.4

4

0.5

4

0.4

7

0.7

0

0.3

2

0.7

8

0.5

4

0.9

8

EC

0.1

8

0.4

9

0.0

0

0.0

0

0.0

0

0.0

0

-

-

0.4

9

0.7

5

0.6

2

0.9

3

0.7

6

1.1

3

LL

0.1

8

0.2

6

0.0

0

0.0

0

0.0

0

0.0

0

0.3

2

0.5

6

-

-

0.8

7

0.8

9

0.5

9

0.8

1

LC

PinR

RF

PR

SR

TR

TCI

P

O

P

O

P

O

P

O

P

O

P

O

P

O

LC

-

-

0.1

9

0.2

6

0.5

2

0.6

5

0.4

1

0.6

0

0.1

8

0.4

0

0.3

5

0.5

7

0.2

3

0.3

9

Pin

R

0.4

1

0.6

9

-

-

0.5

6

0.7

2

0.4

5

0.5

4

0.3

5

0.4

3

0.4

1

0.5

6

0.3

9

0.4

4

RF

0.4

6

0.6

4

0.1

0

0.1

6

-

-

0.3

2

0.4

8

0.1

9

0.4

1

0.2

2

0.4

8

0.1

6

0.2

3

PR

0.4

8

0.6

4

0.1

5

0.1

9

0.4

6

0.6

4

-

-

0.3

3

0.3

8

0.6

4

0.6

9

0.2

0

0.3

6

SR

0.4

1

0.9

4

0.0

9

0.1

4

0.2

8

0.6

5

0.3

0

0.3

8

-

-

0.4

1

0.5

4

0.2

1

0.3

3

TR

0.4

3

0.7

8

0.1

2

0.1

9

0.4

1

0.7

2

0.1

4

0.2

0

0.1

9

0.2

8

-

-

0.2

1

0.2

6

TCI

0.8

2

1.0

5

0.2

1

0.2

2

0.7

5

0.8

2

0.5

7

0.7

2

0.3

8

0.4

2

0.6

0

0.6

4

-

-

39

WV

PinP

TA

EC

LL

LF

LE

P

O

P

O

P

O

P

O

P

O

P

O

P

O

L

F

0.2

3

0.3

2

0.0

0

0.0

0

0.0

0

0.0

0

0.3

3

0.5

6

0.3

3

0.3

9

-

-

0.6

6

0.7

1

L

E

0.2

3

0.4

4

0.0

0

0.0

0

0.0

0

0.0

0

0.2

5

0.5

1

0.4

7

0.6

8

0.7

1

0.7

5

-

-

Hierarchical DEA is run to evaluate the total score between district and province

level for wind turbine site selection after getting ratio of weight by HFLTS as seen in

Table 4.13, the result shows that, considering expert judgement on the importance of

significant criteria, South Sumatra as the most appropriate location for establishing

wind turbine power plant, following by west Papua, Papua, Maluku, and East of Nusa

Tenggara, respectively.

Table 4.13 Hierarchical Score for HFLTS

No Province

Eff

District

Eff

Dist

Step 1

Step 2

Step 3

Hierarchical

Score

1

South

Sumatra

0.725

Palembang

0.6076

4.9965

3.6238 13.1317

2.626

Pagar

Alam

0.0000

0.0001

0.0001

0.0000

Lubuk

Linggau

0.0001

0.0010

0.0007

0.0000

Prabumulih 0.0001

0.0010

0.0008

0.0000

Lahat

0.0002

0.0013

0.0010

0.0000

2

West

Papua

0.729

Manokwari 0.1218

0.7429

0.5414

0.2931

1.979

Sorong

0.0001

0.0004

0.0003

0.0000

Fakfak

0.0005

0.0032

0.0024

0.0000

Sorong

City

0.6972

4.2520

3.0989

9.6031

Bintuni

0.0002

0.0015

0.0011

0.0000

3

Papua

0.885

Jayapura

0.0017

2.9959

2.6500

7.0227

1.695

Merauke

0.0002

0.3528

0.3121

0.0974

Biak

Numfor

0.0000

0.0487

0.0431

0.0019

Nabire

0.0002

0.3285

0.2906

0.0844

Mimika

0.0007

1.2740

1.1269

1.2700

4

Maluku

0.908

Ambon

0.6807

1.5134

1.3738

1.8874

1.388

Seram

0.0002

0.0004

0.0003

0.0000

Tual

0.7079

1.5738

1.4287

2.0413

Aru

0.8598

1.9117

1.7355

3.0118

Buru

0.0003

0.0008

0.0007

0.0000

40

No Province

Eff

District

Eff

Dist

Step 1

Step 2

Step 3

Hierarchical

Score

5

East

Nusa

Tenggara

0.639

Kupang

0.0030

3.2750

2.0918

4.3758

0.938

Alor

0.0003

0.3644

0.2327

0.0542

Belu

0.0003

0.3631

0.2320

0.0538

Ngada

0.0005

0.5607

0.3581

0.1283

Southwest

Sumba

0.0004

0.4368

0.2790

0.0779

4.4 Principle Component Analysis Results

Based on the scree plot in Fig. 3.17., will be extracted for total three eigen values

consisting of two eigen values which have values greater than 1 and one eigen value

close to 1 from the analysis that’s why have to do it in two steps to analyze again in

dimension reduction. Choosing analyze with correlation matrix due to the variable are

measured in different units, this implies normalizing all variables using division by their

standard deviation.

Fig.4.2 Correlation matrix on district level

Looking at the correlation on Fig 4.2 between Primary and Secondary road have

positive correlation 0.795 they seem to hang together when the primary road is needed

in wind turbine site criteria, Secondary can also necessary. But there is also some

negative correlation such as Land cost and population which are different in usual but

cannot expect too much such a thing as slightly not significant. We can see a lot of

positive correlation mostly on Primary, Secondary, Tertiary and Total Cost of

Infrastructure that very consistent, Overall have a lot positive correlation but there are

41

also have some negative correlation that’s not going to be real straight forward one

component extraction effects on the scree-plot. That true as three component extraction.

Fig.4.3 KMO and Bartlett’s Test on district level

The Bartlett’s test of sphrericity will be non-significant because see on Fig. 4.3.

in this case is statistically significant basically telling that at least one statistically

significant correlation matrix. On the Kaiser-Meyer-Olkin measure of Sampling

Adequacy is also more effect size measure is determining whether use principal

component analysis or not. 0.695 or up to 0.70 or higher is great the lower point is on

less than 0.40 this is the rule time that generally used. Overall on the correlation matrix,

KMO that are over than 0.40 and The Bartlett’s test is statistically significant this will

make confidence to perform the component analysis on district level.

Fig.4.4 Communalities on level 1

The Communalities is output from SPSS that shows the extraction based on

three components being extracted as shown on Fig. 4.4. Communalities represent

variance that have been counted from Component analysis. We can see that ratio of free

usage area have the largest amount of variance that being explained by component

42

analysis solutions as 99.6% of the significant criteria on wind turbine site selection,

following with total cost infrastructure and tertiary road, respectively. Overall the

communalities are good due to more than 50 % for each criterion.

Fig.4.5 Total variance on district level

The real important thing that should be interpret on column extraction sums of

squared loadings that have been extracted from the three components factor solutions

as given in total variance on Fig. 4.5. These are the eigenvalues 3.375 for the first

component, 1.835 for the second component, and following by the third component is

1.158. Overall the extraction sums of squared loadings have more than 1. The

cumulative percentage of variance and these are the rotated component solution

eigenvalues. SPSS technically calls rotations sums of squared loading as the

eigenvalues in the rotated component solutions which is oblique on scree plot that we

used for the first step of the analysis.

43

Fig.4.6 Component matrix of district level

The component matrix on Fig. 4.6 of component loadings is basically the

extraction method based on the unrotated solution just for showing the initial value and

we don’t need to interpret it.

Fig.4.7 Pattern matrix on district level

When we have oblique rotated component solution, we really want to use the

pattern matrix are given in Fig. 4.7. The pattern matrix is to help to identify the nature

of the components and what have here in the first component is total cost of

infrastructure is 92.2%, tertiary road is 92.2%, secondary road is 90.3%, and primary

road is 87.8% all loading nicely on this first component so wind turbine site selection

44

criteria seem to hang together to trade together as significant criteria. But that’s not

always exactly true because these factor loadings are not on all 0.95 but they are high

enough to suggest a pretty strong pattern. These the rest of the listed criteria as land

cost and ratio of free space don’t seem to load very strong only the exception would be

ratio of free space what do you use as a statistically significant component loading.

The second component has two major loadings that are land cost is 91.2% and

population 89.9% then it has negative component loading and if look at total cost of

infrastructure and tertiary road. its component loading to the first component and the

second component its one’s positive and one’s negative either one is very high positive

in first component. Mostly we choose positive value for both components and have a

significant decision or both have difference value on less of negativity.

The third component has one major loading that is ratio of free space area as

highly positive 99.8%. It is totally hanged to trade as significant criteria. So, we can

conclude that majority the percentage of significant criteria have more that 87% as in

the first component is total cost of infrastructure, tertiary, secondary and primary road,

for the second component is land cost and population, and the last component is ratio

of free usage area.

Fig.4.8 Structure matrix on level 1

The last table is the structured matrix which is actually the correlation between

each variable in the analysis and that sequence of respective component from the most

significant criteria to less significant criteria as given in Fig. 4.8.

45

On the District level we can conclude that have seven significant criteria which

influence on wind turbine site selection in Indonesia such as total cost infrastructure,

tertiary road, secondary road, primary road, population and land cost and we reduce

ratio of free space based on analysis of principal component analysis.

Due to same steps to do analysis using principal component. In province level

we can interpret the results directly.

Fig.4.9 Scree plot for level 2

Based on the scree plot which have been shown on Fig. 4.9, we can see on the

oblique shows on the first three component which have eigenvalue more than 1. Due to

that 3 components look like have meaning components and we can use it on extraction

analysis.

Fig.4.10 KMO and Bartlett’s Test for Level 2

46

The KMO and Bartlett’s results test is given in Fig. 4.10 show that for KMO

measure of sampling adequacy is 0.660 is good enough to determine using principal

component analysis. The Bartlett’s test of sphrericity will be non-significant because in

this case is statistically significant basically telling that at least one statistically

significant correlation matrix. Overall on the correlation matrix, KMO that are over

than 0.40 and The Bartlett’s test is statistically significant this will make confidence to

perform the component analysis on province level.

Fig.4.11 Correlation matrix on level 2

Fig. 4.11 shows the correlation matrix on province level. Looking at wind

velocity column we can see that majority have positive correlation such as wind

velocity with population, electricity consumption, earthquake, volcanic eruption and

landslide, but have some negative correlation with total area and flood. Correlation

between wind velocity and population have positive 0.419 as dominant. Also looking

at other column, we can say as generally, overall have a lot positive correlation but

there are also have some negative correlation. It is meaning that more than one

component has extraction effects on the scree plot. We use three components for

extraction analysis.

Fig.4.12 Communalities on level 2

47

The results of variance being showed in the communalities table is given on Fig.

4.12 We can see that majority have good value of extraction. Population have the

highest amount of variance as 84.2% and wind velocity even have less amount still

more than 10 % for saying as a statistical criterion on 42.3%.

Fig.4.13 Total variance on level 2

The rotation sums of squared loadings are the most important thing to interpret

as the extraction results from three components factor solutions which are given in Fig.

4.13. These are the result 2.666 for the first component, 1.837 for second component

and the last as 1.118 for third component. These results show the position of component

after rotated component as scree plot on the early analysis.

Fig.4.14 Component matrix on level 2

48

The component matrix shows the initial component as unrotated solution as

given in Fig. 4.14. In here just show the early step before being rotated.

Fig.4.15 Pattern matrix on level 2

The pattern matrix helps to identify oblique rotated component solution is given

on Fig. 4.15. In first component we can see the positive loadings is population,

electricity consumption, earthquake, wind velocity, land slide and volcanic eruption.

The second component have 5 major loadings as volcanic eruption, landslide, flood,

earthquake and population. Wind velocity and electricity consumption have decreased

on few negativities and have not impact on changes. On third components just have two

loadings positive as Total area and earthquake but majority on negativity neither one is

very high. Overall, we can conclude that each criterion has positive loadings even in

just one component and have an impact on solution as the group of components. Due

to that, we can conclude that all of criteria have statistically significant and we do not

need to reduce the criterion.

Fig.4.16 Structure matrix on level 2

49

As we have discussed the function of structure matrix on Fig. 4.16 to show the

sequence of influence criteria. Looking at last criterion on total area, even on first and

two components give negative loadings but have a great trend on last component as

87% its look like increasing trend and have been impacted by rotated component

solution.

Fig.4.17 Component correlation matrix on level 2

Component correlation matrix is correlation between each component based on

rotated component solutions is given in Fig. 4.17. Looking at component 1 shows that

have correlation with component 2. Due to negativity on component 3 so component 1

have not correlation with it. Differently at component 2 have positive correlation for

component 1 and component 3. It is showing the good relation from rotated component

solution.

We can conclude that population, electric consumption, earthquake, wind

velocity, flood, volcanic eruption, landslide, and total area as the significant criteria

which are influencing on province level of wind turbine site selection in Indonesia.

Following results of significance criterion, multivariable ranking method namely

Principal component Analysis (PCA) is used for verifying a hierarchical DEA result.

The PCA ranking result is given in Table 4.14.

50

Table 4.14 Principal Component Analysis Results

Province

Prov

Eff

District

Dist

Eff

Step 1

Step 2

Step 3

Hierarchical

Score

South

Sumatra

0.949

Palembang

1.000

2.325

2.208

4.874

1.340

Pagar Alam 0.136

0.316

0.300

0.090

Lubuk

Linggau

0.338

0.786

0.747

0.557

Prabumulih 0.256

0.595

0.565

0.319

Lahat

0.420

0.978

0.928

0.862

West

Papua

1.000

Manokwari 0.301

0.483

0.483

0.233

1.339

Sorong

0.110

0.176

0.176

0.031

Fakfak

1.000

1.605

1.605

2.576

Sorong

City

0.705

1.131

1.131

1.280

Bintuni

1.000

1.605

1.605

2.576

Papua

1.000

Jayapura

0.358

1.049

1.049

1.100

1.166

Merauke

0.243

0.713

0.713

0.508

Biak

Numfor

0.260

0.760

0.760

0.578

Nabire

0.240

0.703

0.703

0.495

Mimika

0.606

1.775

1.775

3.150

Maluku

1.000

Ambon

0.687

1.041

1.041

1.083

1.105

Seram

0.348

0.527

0.527

0.278

Tual

0.719

1.089

1.089

1.187

Aru

1.000

1.514

1.514

2.293

Buru

0.547

0.828

0.828

0.686

East Nusa

Tenggara

1.000

Kupang

0.201

0.653

0.653

0.427

1.101

Alor

0.251

0.818

0.818

0.669

Belu

0.467

1.519

1.519

2.308

Ngada

0.248

0.805

0.805

0.649

Southwest

Sumba

0.370

1.204

1.204

1.450

4.5 Comparison of Three Methods Result

The top five are chosen based on the three methods which are discussed.

Hierarchical data envelopment analysis (DEA) is one of multi-variable approach which

can measure efficiency score, in here the result represents the total score of the province

combine with district level. Hesitant fuzzy linguistic term set is used for measuring

uncertainty criteria which can influence in site selection, in this study the judgement,

expertise and advisement by the expert needed to evaluate the importance of the

51

criterion. After getting the hesitant criteria which have been proven, validation is

needed to validate the significance criteria. The top five suitable locations for

establishing wind turbine power plant in Indonesia are South Sumatra, Papua, West

Papua, Maluku and East Nusa Tenggara provinces, respectively as shown in Fig. 4.18

as geographically location.

Table 4.15 Comparison of three methods result.

Province

DEA

Rank

Fuzzy

DEA

Rank

PCA

Rank

South Sumatra

1.340

1

2.626

1

1.340

1

West Papua

1.339

2

1.979

2

1.339

2

Papua

1.166

3

1.695

3

1.166

3

Maluku

1.105

4

1.388

4

1.105

4

East Nusa Tenggara

1.101

5

0.938

5

1.101

5

The Table 4.15 shows that although using three methods differently still giving

the same priority ranking. Expert judgement can help with multi complex decision and

uncertainty condition. The fuzzy DEA results based on the expert advice giving the

same priority as South Sumatra have the highest priority to build a wind farm. It shows

that the importance criteria relation with specific bound weight are optimal used. The

significance criteria which are obtained based on principal component analysis shows

that the ratio of free usage area and total cost of infrastructure are highly influence to

the results in district level. The ratio of free usage area in South Sumatra is high. It

shows that more space area in one region is advantages. The availability of the

infrastructure of primary road and secondary road in each region have improved in

South Sumatra as primary concern. In this region do not need to build some additional

infrastructure in tertiary road. It can be decreasing the total cost of infrastructure. The

minimize total cost of infrastructure is preferable to influence the efficiency score. In

the province level some criteria such as population, electricity consumption and less of

natural disaster are the most influence to the total efficiency score. The availability

human resources in South Sumatra shows that higher spreading population is in

Palembang district. It can decrease the resources management for the transportation and

accommodation cost of labors. Electricity consumption as the demand in South

52

Sumatra is high are needed to show the amount of electricity distribution to the center

area. The establishment of wind farm in South Sumatra can help as the alternative

energy resources to fulfill the electricity demand. The third influence criteria in South

Sumatra is less of natural disaster. The South Sumatra as geographically located in the

Sumatra Island based on the occurrence of the disaster shows that in this region have

less of landslide, earthquake and volcanic eruption. Due to the reasons that South

Sumatra can be decided as the most suitable location to build a wind farm power plant.

Fig.4.18. Top five Provinces in Indonesia

53

Chapter 5

Conclusions and Recommendations

5.1 Conclusion

Wind energy as natural energy resources is a renewable, freely available and

environmentally compared with other sources of fossil fuel, such as coal, and oil. In

this study, three methods are proposed to decide the most suitable location based on the

multi criteria approach. The hierarchical DDEA to determine the integrated efficiency

scores of DMUs between the district level and the province level. The Fuzzy Data

Envelopment Analysis is used for measuring the bound weight ratio for specific DMUs

based on the expert judgement, advice and expertise. The validation based on principal

component analysis to know the significance of criteria which are influences to the

wind turbine site selection in Indonesia. The possible factors used in the districts level

as defined by land cost, population in region, ratio of free usage area, primary road,

secondary road, tertiary road, and total cost of infrastructure. In the provinces level as

defined by wind velocity, population in province, total area, electricity consumption,

less of land slide, flood, earthquake and volcanic eruption. This method was applied to

33 provinces and 165 districts of Indonesia. The final result shows that the South

Sumatra province has the highest efficiency score which is the most economical

location for constructing a wind farm as given in Fig. 5.1. The most significant criteria

which influence on wind turbine site selection based on principal component analysis

in district level is ratio of free usage area, following by total cost of infrastructure, and

tertiary road. Population, electricity consumption and total area influence in province

level. This study is a milestone for policy makers, government and private stakeholders

in decision making for selecting the most suitable sites for wind power plant

construction in Indonesia. The hierarch DEA results can be used to assist decision

maker on selecting the most suitable wind farm site. The proposed approach can be

considered as an alternative solution, and an early study for policy makers.

54

Fig.5.1. Full Score of Three Methods

5.2. Recommendations

Further improvement could be on criterion specification, which includes social,

environmental, economic, and technical aspects. The final site selection will be more

practical, if opinions from experts, policy makers, government, and private stakeholders

are also considered in the analysis. Collecting more specific data is better approach to

improve the advance analysis in wind turbine site selection.

55

References

[1] A. Haydar, E. Senol and K. Eylem, “Multi-criteria selection for a wind

observation station location using analytic hierarchy process,” Renewable

Energy, vol. 29, pp. 1383-92, 2004.

[2] R. Bhatnagar and A. Sohal , “Supply chain competitiveness: measuring the impact

of location factors, uncertainty and manufacturing practices,” Technovation, vol.

25, pp. 443-56, 2005.

[3] D. Afshartous, Y. Guan and A. Mehrotra, “US Coast Guard air station location

with respect to distress calls: a spatial statistics and optimization based

methodology,” European Journal of Operational Research, vol. 196, p. 1086–96,

2009.

[4] G. Gamboa and G. Munda, “The problem of wind farm location: a social

multicriteria evaluation frame work,” Energy Policy, vol. 35, p. 1564–83, 2007.

[5] D. Choudhary and R. Shankar, “An STEEP-fuzzy AHP-TOPSIS framework for

evaluation and selection of thermal power plant location: A case study from

India,” Energy, vol. 42, pp. 510-21, 2012.

[6] R. Dutu, “Challenges and policies in Indonesias energy sector,” Energy Policy,

vol. 98, pp. 513-519, 2016.

[7] M. H. Hasan, W. K. Wuzammil, T. M. I. Mahlia, A. Jannifar and I. Hasanuddin,

“A review on the pattern of electricity generation and emission in Indonesia from

1987 to 2009,” Renewable and Sustainable Energy Reviews, vol. 16, no. 5, pp.

3206-3219, 2012.

[8] S. Mujiyanto and G. Tiess, “Secure energy supply in 2025: Indonesias need for

an energy policy strategy,” Energy Policy, vol. 61, pp. 31-41, 2013.

[9] A. Azadeh, S. F. Ghaderi and M. R. Nasrollahi, “Location optimization of wind

plants in Iran by an integrated hierarchical Data Envelopment Analysis,”

Renewable Energy, vol. 36, no. 5, pp. 1621-1631, 2011.

[10] G. Ertek, M. M. Tunç, E. Kurtaraner and K. D, “Insights into the efficiencies of

on-shore wind turbines: A data-centric analysis,” in International Symposium on

Innovations in Intelligent Systems and Applications, Trabzon, 2012.

56

[11] Ü. Sağlam, “A two-stage data envelopment analysis model for efficiency

assessments of 39 state: wind power in the United States,” Energy Conversion

and Management, vol. 146, pp. 52-67, 2016.

[12] Y. Wu, Y. Hu, X. Xiao and C. Mao, “Efficiency assessment of wind farms in

China using two-stage data envelopment analysis,” Energy Conversion and

Management, vol. 123, pp. 46-55, 2016.

[13] T. Sueyoshi and M. Goto , “Measurement of Returns to Scale and Damages to

Scale for DEA-based operational and environmental assessment: how to manage

desirable (good) and undesirable (bad) outputs,” European Journal of

Operational Research, vol. 211, no. 1, pp. 76-89, 2010.

[14] L. M. Seiford and J. Zhu, “Modeling undesirable factors in efficiency evaluation,”

European Journal of Operational Research, vol. 142, no. 1, pp. 16-20, 2002.

[15] A. Azadeh, A. R. Golkhandan and M. Moghaddam, "Location optimization of

wind power generation–transmission systems under uncertainty using

hierarchical fuzzy DEA: A case study," Renewable and Sustainable Energy

Reviews, vol. 30, pp. 877-885, 2014.

[16] R. Ramanathan, “Comparative risk assessment of energy supply technologies: a

data envelopment analysis approach.,” Energy, vol. 26, pp. 197-203, 2001.

[17] R. H. Barnes, E. V. Morozov and K. Shankar, "Improved methodology for design

of low wind speed specific wind turbine blades," Composite Structures, vol. 119,

pp. 677-684, 2015.

[18]

[19]

[20]

M. Toloo, M. Barat and A. Masoumzadeh, “Selective measures in data

envelopment analysis,” Annals of Operations Research, vol. 226, no. 1, p. 623–

642, 2015.

M. Yavuz, B. Oztaysi, S.C. Onar and C. Kahraman, "Multi-criteria evaluation of

alternative-fuel vehicles via a hierarchical hesitant fuzzy linguistic model,"

Expert Systems with Applications, vol. 42, p. 2835-2848, 2015.

A.H.I. Lee, H.Y. Kang, C.Y. Lin and K.C. Shen, "An Integrated Decision-Making

Model for the location of a PV Solar Plant," Sustainability, 7, 13522-41, 2015

57

Appendices

58

Appendix A

Data Resources

A.1 Districts Level Data

Output

Province

District

Port

Primary

Secondary Tertiary

Total Cost of Insfrastructure (Rp)

Land Cost (m^2) (IDR) Population Ratio of Free Usage Area

1 Lhokseumawe

International Samudera Pasee Port

15.3

2.6

0

0

700000

188221

0.000961954

2 Banda Aceh

Ulee Lheue Port

6

1.2

0

0

900000

235305

0.000260768

3 Langsa

Kuala Langsa Port

7.9

0.5

0

0

700000

178334

0.001471452

4 Subulussalam

Tapak Tuan Port

102

25.2

41.8

6270000000

600000

78801

0.01765206

5 Sabang

Sabang Port

1.3

0.2

0

0

700000

38077

0.004018174

6 Medan

Belawan International Port

18.8

8.8

0

0

7900000

2465469

0.000107485

7 Tebing Tinggi

Kuala Tanjung Port

19.3

20.5

3

450000000

1000000

169786

0.000182583

8 Tanjung Balai

Tanjung Tiram Port

13.2

34.3

12.2

1830000000

900000

165763

0.000650507

9 Pematangsiantar

Tanjung Tiram Port

33

33.5

6.3

945000000

900000

278055

0.000200176

10 Padang Sidempuan

Angin Sibolga Port

42.8

34.4

10.9

1635000000

700000

225544

0.000508371

11 Padang

Indonesian Port II

6.7

2.1

0

0

2000000

872271

0.000795235

12 Bukit Tinggi

Indonesian Port II

28.8

63.4

8.2

1230000000

1000000

113326

0.00022272

13 Payakumbuh

Indonesian Port II

80.1

33.5

14.7

2205000000

350000

125608

0.00067846

14 Pariaman

Indonesian Port II

28.7

20.1

11.3

1695000000

400000

85485

0.000773586

15 Solok

Indonesian Port II

34.8

16.6

8.2

1230000000

350000

63672

0.001119644

16 Pekanbaru

Duku Port

4.5

2.4

0

0

12500000

855221

0.000739306

17 Dumai

Dumai Port

3.5

3.3

0

0

7000000

264084

0.006147211

18 Kampar

Roro sei Paknik Port

112

125

43.1

6465000000

4000000

722328

0.015205654

19 Rokan Hilir

Bandar Seribu Kubah Port

52.1

23.7

14.1

2115000000

700000

625642

0.014195962

20 Siak

Tanjung Buton Port

33.7

14.6

17.4

2610000000

400000

407093

0.020327493

21 Jambi

Pelita Jambi Port

6

7.4

0

0

5000000

602187

0.00017194

22 Sungaipenuh

Indonesian Port II

98.4

113

35.5

5325000000

300000

101325

0.003863805

23 Merangin

Indonesian Port II

157

120

83.6

12540000000

900000

329077

0.023334964

24 Sarolangun

Pelita Jambi Port

88.3

72.4

32.4

4860000000

900000

309621

0.019972805

25 Tebo

Pelita Jambi Port

66.1

75.1

67.6

10140000000

700000

323554

0.019968846

26 Palembang

Indonesian Port II

1.3

0

0

0

2000000

1548064

0.000238504

27 Pagar Alam

Indonesian Port II

142

78.6

59.1

8865000000

500000

136244

0.00465092

28 Lubuk Linggau

Indonesian Port II

72.9

91.1

28.3

4245000000

300000

208225

0.001928203

29 Prabumulih

Indonesian Port II

33.6

35.7

23.7

3555000000

400000

188082

0.001339522

30 Lahat

Indonesian Port II

87.7

91.8

41.1

6165000000

500000

418845

0.012681875

31 Bengkulu

Bengkulu Pelindo Port

11.1

6.1

0

0

700000

360495

0.00042081

32 Kaur

Linau Port

44.5

18.9

6.1

915000000

300000

123236

0.019223685

33 Rejang Lebong

Bengkulu Pelindo Port

64.3

33.2

11.4

1710000000

200000

268569

0.006106364

34 Seluma

Bengkulu Pelindo Port

31.8

38.1

19.4

2910000000

100000

204790

0.011721471

35 Kepahiang

Bengkulu Pelindo Port

41.6

44.3

11.5

1725000000

100000

144418

0.004604689

36 Bandar Lampung

Panjang Port

7.9

5.5

0

0

3500000

1166761

0.000253694

37 Metro

Panjang Port

25.3

17.2

4.7

705000000

350000

161799

0.000381894

38 Pesawaran

Panjang Port

22.1

25.6

10.1

1515000000

337000

542984

0.004131816

39 Tanggamus

Piers Attorney Port

33.2

17.5

5.8

870000000

700000

634643

0.004759526

40 Mesuji

Mesuji Port

37.8

31.6

8.4

1260000000

337000

302524

0.007219262

41 Pangkal Pinang

Balam Base port

3.4

1.2

0

0

1600000

202959

0.000440483

42 Bangka

Balam Base port

26.8

13.5

6

900000000

700000

304944

0.009676137

43 Belitung

Tanjung Pandan Port

13.9

7.6

3.2

480000000

800000

152250

0.015064762

44 West Bangka

Muntok Port

38.9

16.2

5

750000000

600000

179711

0.015695255

45 East Belitung

Belitung Port

11.1

25.5

13.8

2070000000

650000

109564

0.022880782

46 Batam

Batam Port

8.3

7.3

0

0

5000000

1029808

0.000932455

47 Bintan

Bintan Lagoon International Port

15.1

12.4

4.4

660000000

2500000

140169

0.009404433

48 Tanjungpinang

Sri Bintan Putra Port

3

1.2

0

0

1000000

203008

0.00071209

49 Lingga

Jagoh Port, Dabo Singkep

11.4

5.9

3.9

585000000

700000

87463

0.025916902

50 Karimun

Tanjung Batu Port

13.2

5.5

2.4

360000000

1000000

237002

0.003851233

51 South Jakarta

Tanjung Priok Port

16.1

15.5

0

0

24600000

2113411

7.30194E-05

52 Central Jakarta

Tanjung Priok Port

6.8

10.5

0

0

21000000

1114581

4.69952E-05

53 East Jakarta

Tanjung Priok Port

9.9

10.6

0

0

20000000

2852887

6.40404E-05

54 West Jakarta

Tanjung Priok Port

12.9

8.4

0

0

22000000

2234397

5.56929E-05

55 North Jakarta

Tanjung Priok Port

7.5

7.1

0

0

20000000

1647853

8.4953E-05

56 Bandung

Patimban port

59.4

44.6

0

0

11000000

2339463

7.16703E-05

57 Bogor

Tanjung Priok Port

36.2

30.4

3

450000000

11000000

982469

0.000120614

58 Sukabumi

Ratu Port

39.8

12.1

9.9

1485000000

1700000

2436729

0.001701338

59 Tasikmalaya

Cirebon port

31.7

63.1

18.3

2745000000

1700000

1640647

0.00155499

60 Cimahi

Tanjung Priok Port

84.3

37.7

24.6

3690000000

2000000

513176

7.65235E-05

61 Semarang

Tanjung Mas Port

5.6

3.9

0

0

14800000

1621384

0.000230531

62 Jepara

Kartini Port

2.7

2.6

0

0

1700000

1141236

0.00092816

63 Pekalongan

Nusantara Port

3.1

2.6

0

0

2676000

911277

0.000918491

64 Surakarta

Tanjung Mas Port

60.4

38.3

10.5

1575000000

5000000

552118

8.33336E-05

65 Magelang

Tanjung Mas Port

39.2

29.5

10.2

1530000000

1700000

1261661

0.000874189

66 Yogyakarta

Tanjung Mas Port

65.6

59.9

4.6

690000000

21000000

407617

7.97317E-05

67 Sleman

Tanjung Mas Port

57.4

53.1

10.5

1575000000

4500000

1062801

0.000540854

68 Bantul

Tanjung Mas Port

60.2

80.4

7.7

1155000000

4500000

912937

0.000556588

69 Kulon Progo

Tanjung Mas Port

84.5

25.2

28.6

4290000000

4000000

409568

0.001431459

70 Gunung Kidul

Tanjung Mas Port

60.8

60.5

39.2

5880000000

700000

749155

0.001910713

12

West Java

13

Central Java

14

DI Yogyakarta

9 Bangka Belitung Islands

10

Riau Islands

11

DKI Jakarta

6

South Sumatra

7

Bengkulu

8

Lampung

3

West Sumatra

4

Riau

5

Jambi

Input

1

Aceh

2

North Sumatra

59

Output

Province

District

Port

Primary

Secondary Tertiary

Total Cost of Insfrastructure (Rp)

Land Cost (m^2) (IDR) Population Ratio of Free Usage Area

71 Surabaya

Tanjung Perak Port

5.1

1.8

0

0

21000000

2805906

0.000124929

72 Pasuruan

Pasuruan Port

8.1

10.2

3.5

525000000

5000000

1553563

0.0009488

73 Malang

Pasuruan Port

27.4

22.2

5.4

810000000

4000000

808945

0.000179592

74 Kediri

Tanjung Perak Port

50.6

50.1

16.4

2460000000

3500000

1440425

0.000962251

75 Probolinggo

Probolinggo Port

24.7

11.4

8

1200000000

1000000

1072101

0.001582136

76 Tangerang

Tanjung Priok Port

17.1

13.1

1.6

240000000

15900000

1566190

9.82831E-05

77 Serang

Karangantu Port

10.7

11.1

3.1

465000000

1000000

1401036

0.001237855

78 Lebak

Binuangeun Port

16.9

40.1

24.3

3645000000

700000

1133671

0.003022535

79 Cilegon

Merak Port

8.9

3.2

0

0

1572000

387543

0.000452853

80 Pandeglang

Labuan Port

17.9

16.3

5.1

765000000

700000

1139061

0.002411539

81 Denpasar

Indonesian, Benoa Port

4.6

4.9

0

0

15000000

632016

0.000202178

82 Gianyar

Benoa Port

26.5

7.2

0

0

12000000

485377

0.000758174

83 Buleleng

Buleleng Port

20.5

7.2

3.5

525000000

10000000

805883

0.001693459

84 Bangli

Benoa Port

31.8

26.7

7.4

1110000000

5000000

261240

0.001878388

85 Klungkung

Nusa Penida Port

4.5

5.6

2.8

420000000

5000000

211862

0.001486817

86 Mataram

Lembar Port

11

10.4

2.5

375000000

2300000

408900

0.000149914

87 Bima

Bima Port

15.4

16.6

17.1

2565000000

700000

519078

0.006560921

88 Dompu

Bima Port

16.1

40.4

5.8

870000000

600000

211051

0.011331574

89 East Lombok

Lembar Port

11.6

30.1

3.5

525000000

1500000

1279949

0.00096157

90 Sumbawa

Badas Port

15.4

24

8.3

1245000000

600000

503978

0.013183075

91 Kupang

Tenau Port

7.7

3.1

2.3

345000000

2500000

433970

6.03268E-05

92 Alor

Feri Port

6.3

15

10

1500000000

700000

207283

0.013819754

93 Belu

Atapupu port

7.7

11.2

10.6

1590000000

300000

218880

0.00587066

94 Ngada

Aimere port

34.5

6.3

5.1

765000000

450000

162721

0.010114736

95 Southwest Sumba

Waikelo Port

5.1

10

12.1

1815000000

700000

300671

0.004923854

96 Pontianak

Indonesian Port II

0.95

0

0

0

2500000

651139

0.000165556

97 Singkawang

Sedau Port

6.2

2.1

1.4

210000000

1700000

230216

0.002189248

98 Bengkayang

Teluk Suak Port

23.6

68.6

52.3

7845000000

700000

280168

0.018115845

99 Landak

Dwikora Port

79.2

24.2

33.6

5040000000

600000

391767

0.022756128

100 Kubu Raya

Dwikora Port

19.3

13.8

22

3300000000

800000

596421

0.011666625

101 Palangka Raya

Sampit Port

93.3

84

41.5

6225000000

3000000

249429

0.009619972

102 Seruyan

Sigintung Port

11.4

8.5

7.5

1125000000

700000

141334

0.11606549

103 Gunung Mas

Sampit Port

96.6

95.8

72.2

10830000000

400000

135872

0.079523375

104 South Barito

Sampit Port

138

108

35.1

5265000000

700000

121557

0.072640819

105 Pulang Pisau

Sampit Port

96.1

42.4

58.8

8820000000

300000

122143

0.073659563

106 Banjarmasin

Trisakti Port

4

0

0.6

90000000

2500000

635688

0.000113263

107 Banjarbaru

Trisakti Port

26.7

5.4

3.3

495000000

850000

216600

0.001712835

108 Balangan

Trisakti Port

92.4

46.7

85.4

12810000000

600000

121429

0.015468298

109 Barito Kuala

Trisakti Port

14.2

16.7

8.6

1290000000

300000

303193

0.009883012

110 Tabalong

Semayang Port

120

78.1

29.2

4380000000

300000

230847

0.016318037

111 Balikpapan

Semayang Port

5.4

1.5

1.9

285000000

15000000

597625

0.000881824

112 Samarinda

TPK Palaran Port

5.6

13.6

8.8

1320000000

10300000

752845

0.001040055

113 Bontang

Tanjung Laut Port

1

0.21

0.39

58500000

1000000

161356

0.002520514

114 Paser

Semayang Port

5

48.3

102

15300000000

900000

240043

0.03220623

115 Tarakan

Malundung Port

3.3

1.9

0.7

105000000

800000

179079

0.000402057

116 Manado

ASDP Manado Port

0.7

1.8

0.3

45000000

16000000

461636

0.00034068

117 Bitung

Bitung Port

1.4

5

3.5

525000000

1133000

218520

0.001386097

118 Tomohon

ASDP Manado Port

4.1

22.1

0.4

60000000

700000

96411

0.001184512

119 Minahasa

Amurang Port

11.9

17.4

15.3

2295000000

600000

331647

0.000337316

120 Kotamobagu

Amurang Port

4.8

62.7

23.3

3495000000

600000

120597

0.000564359

121 Palu

Pantoloan Port

3.2

18.1

1.8

270000000

3500000

359350

0.001099374

122 Parigi Moutong

Tinombo Port

56.2

16.6

10.4

1560000000

700000

439799

0.011573264

123 Donggala

Donggala Port

51.6

45.2

6.5

975000000

850000

288686

0.014808754

124 Banggai

Luwuk Port

16.9

10.7

7.8

1170000000

600000

355415

0.027215227

125 Poso

Laut Poso Port

1.3

1.2

1

150000000

600000

238400

0.029833263

126 Makasar

Soekarno Hatta Makasar Port

1.4

9.2

6.3

945000000

14070000

1651146

0.00012068

127 Palopo

Tanjung Ringgit Port

1.2

1.9

0.6

90000000

300000

180130

0.001404486

128 Sidrap

Awerange Port

29

51.1

0.4

60000000

450000

317691

0.005927867

129 Parepare

Indonesian Port 4

0.8

0.8

0

0

500000

175040

0.00056747

130 Maros

Soekarno Hatta Makasar Port

25.4

11.2

6.7

1005000000

300000

395081

0.004098198

131 Kendari

Indonesian Port 4

6.8

4.1

0

0

9300000

331013

0.000908998

132 Baubau

Murhum Port

2.2

0.6

0

0

300000

152143

0.001452581

133 Muna

Nusantara Raha Port

9.8

15.2

12.6

1890000000

400000

223982

0.008581761

134 Kolaka

Kolaka Port

7.3

1.3

2

300000000

300000

204044

0.016092558

135 Wakatobi

Mola Port

6.4

1.5

1.5

225000000

350000

107898

0.005185824

136 Gorontalo City

Indonesian Port 4

2.7

1.9

0

0

700000

191897

0.000414754

137 North Gorontalo

Anggrek Port

2.7

5.5

0

0

300000

122124

0.013724984

138 Bone Bolango

Indonesian Port 4

20.7

15.2

10.2

1530000000

250000

157215

0.012621633

139 Pohuwato

Marisa Port

4.1

19.1

0.4

60000000

150000

136448

0.031105696

140 Boalemo

Tilamuta Port

8.4

15.7

15.8

2370000000

150000

141796

0.010754041

141 Mamuju

Mamuju Port

2.8

15.9

2.7

405000000

700000

290672

0.017200453

142 Majene

Palipi Port

49.3

5.9

3.2

480000000

300000

164107

0.005775744

143 Polewali Mandar

Tanjung Silopo Port

34.9

6.2

3.3

495000000

250000

513180

0.003460092

144 Mamasa

Tanjung Silopo Port

50.7

27.9

12.9

1935000000

150000

200977

0.014956338

145 North Mamuju

Pasangkayu Port

55.7

8.8

12.6

1890000000

500000

205774

0.014791713

146 Ambon

Ambon Port

5.9

0.8

0

0

700000

372249

0.000802178

147 Seram

Kobi Sadar Port

45.8

31.2

12.2

1830000000

350000

125684

0.051159097

148 Tual

Tual Port

1.7

0.4

0

0

150000

82955

0.003066602

149 Aru

Dobo Port

1.8

3.5

0

0

150000

100766

0.080904472

150 Buru

Namlea Port

66.2

6.8

6

900000000

200000

128720

0.03831821

151 Ternate

A Yani Port

1.9

1.4

0

0

5836000

213274

0.000522286

152 Sula

Sanana Port

16.5

11

10.2

1530000000

700000

122726

0.026924368

153 Morotai

Morotai Ferry Port

14.1

24.8

5.5

825000000

350000

63033

0.039281012

154 Tidore

Trikora Tidore Port

3.3

5

0

0

700000

103171

0.015951479

155 Halmahera

Tongute Port

8.8

8.9

5.9

885000000

600000

130137

0.01309543

156 Manokwari

Manokwari Port

2.2

2.4

0

0

2000000

190337

0.016740203

157 Sorong

Arar Port

15.5

17.2

9.4

1410000000

1700000

120956

0.05410422

158 Fakfak

Fak Fak Port

2.6

2.4

2.5

375000000

700000

83072

0.132854391

159 Sorong City

Sorong Port

2.5

0.6

0

0

500000

272349

0.002411024

160 Bintuni

Bintuni Port

7

6.6

5

750000000

450000

75410

0.276366927

161 Jayapura

Indonesian Port 4

4.3

6.3

1.5

225000000

2000000

162199

0.068786799

162 Merauke

Merauke Port

57.6

40.5

17.5

2625000000

1500000

219438

0.200835771

163 Biak Numfor

Laut Biak Port

4.4

7.2

8

1200000000

500000

138401

0.018800442

164 Nabire

Nabire Port

21.6

27.5

14

2100000000

700000

163390

0.068012791

165 Mimika

LPMAK Port

3.8

26.6

6.7

1005000000

600000

303376

0.071307552

Input

33

Papua

30

Maluku

31

North Maluku

32

West Papua

27

Southeast Sulawesi

28

Gorontalo

29

West Sulawesi

24

North Sulawesi

25

Central Sulawesi

26

South Sulawesi

21

Central Kalimantan

22

South Kalimantan

23

East Kalimantan

18

West Nusa Tenggara

19

East Nusa Tenggara

20

West Kalimantan

15

East Java

16

Banten

17

Bali

60

A.2 Provinces Level Data

Output

Input

Province

Wind Velocity

(m/s)

Population

Area

(sq.km)

Electricity

Consumption

(Gwh)

Landslide

(times)

Flood

(times)

Eartquake

(times)

Volcanic eruption

(times)

Aceh

5.3

4494410

57,956.00

2119

273

1 649

1 228

0

North Sumatra

2.7

12982204

72,981.23

8704

569

807

191

194

West Sumatra

2.4

4846909

42,012.89

3063

225

306

78

6

Riau

2.9

5538367

87,023.66

3586

24

512

0

0

Jambi

5.9

3092265

8,201.72

1083.79

58

518

40

0

South Sumatra

5.5

7450394

50,058.16

4783

145

745

2

0

Bengkulu

3.9

1715518

91,592.43

785.43

151

213

56

0

Lampung

4.0

7608405

16,424.06

3571.00

82

508

5

0

Bangka Belitung Islands

4.2

1223296

19,919.33

861.52

4

58

0

0

Riau Islands

5.7

1679163

34,623.80

2695

13

51

0

0

DKI Jakarta

6.9

9607787

664.01

41329

0

151

0

0

West Java

4.7

43053732

35,377.76

44071

1 578

1 193

412

5

Central Java

3.3

32382657

9,662.92

20408

1 222

1 273

129

1

DI Yogyakarta

10.2

3457491

32,800.69

2484

77

76

27

2

East Java

4.3

37476757

3,133.15

30825

665

1 218

207

43

Banten

13.3

10632166

47,799.75

8575

150

531

19

0

Bali

2.4

3890757

5,780.06

4594

150

58

0

0

West Nusa Tenggara

6.4

4500212

18,572.32

1402.30

46

286

68

0

East Nusa Tenggara

7.0

4683827

48,718.10

749.76

581

445

97

17

West Kalimantan

8.8

4395983

147,307.00

1989.63

65

616

0

0

Central Kalimantan

5.0

2212089

153,564.50

1048.64

23

534

0

0

South Kalimantan

3.0

3626616

38,744.23

2187.64

40

623

0

0

East Kalimantan

5.3

3553143

129,066.64

3007.30

55

409

4

0

North Sulawesi

4.1

2270596

13,851.64

1302.58

308

353

102

102

Central Sulawesi

5.3

2635009

11,257.07

948.78

205

731

158

0

South Sulawesi

3.9

8034776

61,841.29

4479.46

280

728

22

0

Southeast Sulawesi

4.0

2232586

46,717.48

703.59

123

702

175

0

Gorontalo

5.9

1040164

16,787.18

398.82

73

323

99

0

West Sulawesi

2.2

1158651

38,067.70

258.70

157

159

8

0

Maluku

4.5

1533506

46,914.03

509.51

122

233

43

0

North Maluku

5.0

1038087

31,982.50

329.44

52

285

143

63

West Papua

5.0

760422

319,036.05

455.58

54

88

160

0

Papua

8.5

2833381

99,671.63

763.32

251

308

341

0