Long-Term Infrastructure Planning of Airport System in Developing Country

Transcription

1 Technische Universiteit Delft Long-Term Infrastructure Planning of Airport System in Developing Country

2

3 Long-Term Infrastructure Planning of Airport System in Developing Countries By In partial fulfillment of the requirements for the degree of Master of Science in Transport, Infrastructure, and Logistics at the Delft University of Technology, To be defended publicly on Wednesday, 27 September 2017 at AM Graduation Committee: Chairman: Prof. Dr. G.P. van Wee Full Professor, Department of Transport and Logistics Faculty of Technology, Policy, and Management First Supervisor: Dr. Milan Janic Department of Transport and Planning Faculty of Civil Engineering and Geosciences Second Supervisor: Dr. Jan Anne Annema Department of Transport and Logistics Faculty of Technology, Policy, and Management An electronic version of this thesis is available at

4

5 Summary Air transportation is one of the vital systems in the world since it contributes enormous economic and social benefits. The development of airport as the key infrastructure which enables passengers to fly, needs a huge investment. Without a proper investment planning, the development will cause an enormous sunk cost and other consequential problems such as the failure with airport development of Mattala Rajapaksa International Airport in Sri Lanka. Thus, planning an airport a new airport or expansion of existing airport is a real challenge. In developing an Airport Master Plan (AMP), a guideline from IATA, FAA or ICAO can be used by airport operator or designer. These standard guidelines merely provide the lists of elements which need to be provided. Nonetheless, they do not equip the users with a calculation method to determine the ideal numbers (e.g., the number of runways). The process of determining the amount or size of infrastructure required in the master plan is specifically called airport infrastructure planning. By having the plan, an airport operator can prepare the budget to fund the expansion project. The previous study which conducted similar problem is the one in India. This study proposed the long-term plans of future investment in airport system. However, the passenger demand model, which only utilizes the relationship with GDP, is deemed to be insufficient. Moreover, the result is presented in the amount of investment required, not in specific value or unit of the infrastructure. The problem of estimating the future infrastructure is also faced by Angkasa Pura II (AP2) which manages 13 airports in Indonesia. Currently, most of the airports handled traffic more than its stated capacity. Per 2015, many airports in AP2 suffer from the over-capacity situation. The occupancy level of 7 out of 13 airports was in the over-capacity situation. In response to this condition, AP2 need to find the right amount of infrastructure to be built by However, the problems are the lack of performance data, i.e., delay data and lack of capability to assess the infrastructure utilization. Hence, the convenient method to calculate the future airport infrastructure in strategic level with limited input is needed. By having such method, the airport operators or airport authorities, especially in developing countries, can plan their strategic infrastructure expansion in an accountable way and independently. Accordingly, this research aims at helping airport operator to make a convenient method for estimating the required infrastructure to balance the future demand. The main research question of this study is: What is the convenient method for an airport system owner in developing countries to assess the infrastructure requirement of the airports for balancing the future demand considering the uncertainties in the future? First, the literature review is conducted to answer the research question. The econometric model is found popular for analyzing the airport demand since this model has underlying factors that have a relationship with airport demand. Many literatures attempt to explain the airport demand evolution with various explanatory variables. The variables can be categorized as economic activity, demographic structure, and tourism-related indicators. The common model structures found in the literature are the linear model and log-log model. The infrastructure design parameter that are estimated in the model are the number of runways, aircraft parking stands in apron-gate complex and passenger terminal size. The analytical models to calculate are found. A comprehensive uncertainty analysis on an airport is found. The uncertainties can be classified into three categories: (1) uncertainties affecting airport demand, (2) uncertainties affecting airport capacity, and (3) uncertainties affecting both, demand and capacity. The results from literature review is then combined into a method to estimate the required infrastructure in the future. It consists of two main models, the demand model, and the infrastructure model. The goal of demand model is to find the best model to explain the growth of airport traffic. Therefore, the model can be used to calculate the future demand. The infrastructure model, on the other hand, use the result of demand model to estimate the appropriate infrastructure to balance the demand. In the last step, the i

6 uncertainty is explored and the scenario analysis is conducted to test how robust the output of the model is. The final recommendation is made based on the result of this scenario analysis. Application to the case of AP2 The proposed method then applied to AP2 case. The general demand is developed for all airports. From the demand model, it is found that GDRP, Population, and Domestic Tourist can explain the number of domestic passengers. The number of international passengers is explained by GDRP, Population, Export, Import and International Tourist. The dummy variables which explain hub position for the two airports, CGK and KNO, have a significant contribution to the passenger number. Then, demand models for each airport are developed as the general model has poor forecasting accuracy. After estimating the demand model for each airport using four possible configuration, the linear model formulation which includes all proposed explanatory variables is the best model to forecast the future demand of AP2 based on the average MAPE compared to the log-log model. Then, the future passenger number is forecasted using the models. In the base case, the total passenger handled by AP2 in 2030 is 174,757,647 passengers from 83,151,110 passengers in It means the traffic grows 210% for 15 years or equivalent to 5.08% per year. In the base case, the required number of runways should be added by 2, the number of parking stands should be 17 stands and passenger terminal should be expanded by 411,980 m 2. The plausible uncertainty that may affect the plan is also analyzed. There are various uncertainties which might play a role in system of airports. The uncertainty considered in Schiphol Airport is taken as point of departure in analyzing the uncertainty in AP2. It is found that the uncertainties that affect airport operator may differ from the others. The scale of operation of the airport operator and the location of the airports are the sources of this difference. From these uncertainties, several scenarios are made to test the robustness of the recommendation. The result shows that the required infrastructure is different from the base case. Knowing that the uncertainty can alter the infrastructure estimation, we consider giving a range of plausible outcome rather than a single number. Hence, the final recommendation for AP2 is summarized in the following table. Airport Code Additional runway 1 Without Uncertainty With Uncertainty Additional parking stands 2 Without Uncertainty With Uncertainty Additional passenger terminal size 3 Without Uncertainty With Uncertainty CGK , ,939 KNO ,858 HLP - - 5,830 2,865 14,209 PLM ,109 10,016 39,944 PNK ,876 8,846 27,661 PDG ,048 16,468 55,591 PKU ,187 26,830 53,306 BDO , , ,020 BTJ - - 4,178 2,948 8,118 TNJ - - 3,290 1,975 5,441 PGK ,949 18,212 39,405 DJB ,227 4,173 31,854 DTB - - 8,546 7,406 10,273 Total , , ,619 Notes: 1) The additional runway is in parallel and have enough distance to achieve maximum capacity 2) The parking stands can be as gate or remote 3) The passenger terminal size is in m 2 ii

7 This range of recommendation shows that the best value is hard to determine when uncertainty analysis is added in estimating airport infrastructure. Therefore, the decision makers can use this information to make a detailed plan, but they should also include flexibility in the design within the master plan. For example, the development can take place in huge area but conducted in several modular infrastructures (gradual construction). Thus, in the execution, before the next module is constructed, the decision makers can reevaluate current demand situation and budget availability to decide whether the construction project can proceed at that year or not. To conclude, this study shows that for the airport operator in the developing countries with a lack of detailed data can use a relatively simple method to estimate the required airport infrastructures. The proposed method uses a combination of simple models: the demand model using a multiple linear regression, infrastructure sizing model which calculates the number of runway, the number of aircraft parking stands, and the size of passenger terminal. It should be kept in mind that the method also takes uncertainties into account by using scenario analysis which result in in a range of plausible required infrastructure in each airport. The method developed in this thesis is in principle the same as the one in a developed country. However, its simplicity allows airport operators in developing countries with limited data availability to have an estimation of their future infrastructure requirement. iii

8 iv

9 Acknowledgments Completing the two years of the master program in Transport, Infrastructure, and Logistics is certainly not a piece of cake, especially finishing this thesis. Nevertheless, I enjoyed the process of learning many topics in the classes. Many exams and group projects have been passed. Finally, this is the time when this memorable journey should be ended. First and foremost, I would like to express a gratitude to my graduation committees, Prof Bert van Wee, Dr. Milan Janic, and Dr. Jan Anne Annema. All of you has a great contribution to my thesis project and myself as a person. For Prof Bert van Wee, thank you for becoming my graduation chair. You are a really kind and warm person who always trust and give me a chance to move and step forward. Now, I understand how hard is to deal with policy-making. For Dr. Milan, thank you for always open for my questions. You are such a warm heart person I have ever met. Although we had never met before this thesis, you already showed a positive enthusiasm on my first time delivering the idea of the thesis project. Although it changed a lot during the process, you are always motivating me in every occasion we met. Dr. Jan Anne Annema, you are my favorite lecturer since the first day of my study life at TIL until the day I graduate. You always can give me positive comments and helpful suggestions on everything. Maybe my gratitude cannot repay your favor to me. Again, thank you for all of you. Second, I would like to thank Angkasa Pura II, especially for Mr. Gautsil, Mrs. Diah, and all people in Directorate Corporate Strategic Planning and Performance Management of AP2. Warm welcome and openness from all of you were something that I appreciated. I also learned many things, from the most obvious one, how AP2 run the business, until the random topic such as what should I do after my graduation. Next, I take this opportunity to express my gratitude to my family. Even though I am living thousands of miles away, you always support me and send me your best prayer. Thank you. Special thanks to Indonesian Endowment Fund for Education (LPDP) for providing me the opportunity to study in TU Delft for the past two years. Last but not least, let me also send final thanks to all my friends, especially my fellow Indonesian friends in Delft, my TIL International friends, and all my friends in Indonesia who still keep in contact with me while I m here, in Delft. Thank you being available for me when I need you. Delft, September v

10 vi

11 Table of Contents SUMMARY... I ACKNOWLEDGMENTS... V TABLE OF CONTENTS... VII LIST OF FIGURES... IX LIST OF TABLES... X GLOSSARY... XI 1 INTRODUCTION... 1 BACKGROUND... 1 RECENT PRACTICES IN AIRPORT PLANNING... 1 AIRPORT INFRASTRUCTURE PLANNING... 2 AIRPORT PLANNING PROBLEM IN ANGKASA PURA II (INDONESIA)... 2 RESEARCH OBJECTIVE AND RESEARCH QUESTION... 5 RESEARCH METHODOLOGY... 6 THESIS LAYOUT LITERATURE REVIEW... 9 METHODS FOR ANALYZING AND FORECASTING AIRPORT DEMAND... 9 ANALYSIS OF AIRPORT INFRASTRUCTURE AIRSIDE INFRASTRUCTURE LANDSIDE INFRASTRUCTURE UNCERTAINTY IN AIRPORT SYSTEM PROPOSED METHOD ASSUMPTIONS METHOD FOR ANALYZING AND FORECASTING AIRPORT DEMAND MODEL STRUCTURE MODEL ESTIMATION AND ANALYSIS DEMAND FORECASTING METHOD FOR SIZING THE INFRASTRUCTURE NUMBER OF RUNWAYS NUMBER OF APRON-GATE STANDS SIZE OF PASSENGER TERMINAL METHOD FOR ANALYZING UNCERTAINTY UNCERTAINTY IDENTIFICATION SCENARIO DEVELOPMENT APPLICATION OF THE PROPOSED METHOD AIRPORT DEMAND MODEL AVAILABLE INPUT DATA MODEL STRUCTURE MODEL ESTIMATION AND ANALYSIS DEMAND FORECASTING FUTURE AIRPORT DEMAND SIZING THE INFRASTRUCTURE INPUT DATA RUNWAY SYSTEM APRON-GATE COMPLEX PASSENGER TERMINAL SUMMARY OF INFRASTRUCTURE NEEDS UNCERTAINTY ANALYSIS UNCERTAINTY IDENTIFICATION SCENARIO DEVELOPMENT RESULT AND DISCUSSION vii

12 5 CONCLUSIONS AND RECOMMENDATIONS CONCLUSIONS DISCUSSION POLICY IMPLICATION PRACTICAL RECOMMENDATION CONTRIBUTION SCIENTIFIC CONTRIBUTION PRACTICAL CONTRIBUTION LIMITATION AND FUTURE RESEARCH REFERENCES APPENDIX viii

13 List of Figures Figure 1.1. The map of Indonesia and 13 Airports of Angkasa Pura II. Source: AP2 s website Figure 2.1. Alternative Forecasting Techniques. Source: adapted from (ICAO, 2006)... 9 Figure 2.2. The causal relation between economy and air transport system. Source: (Ishutkina & Hansman, 2009) Figure 2.3. A conceptual model of the uncertainty in the airport system. Source: own work Figure 3.1 Proposed Method Source: own work Figure 4.1. Relationship between Annual Passengers and Peak Hour Coefficient Source: own work Figure C.0.1. Econometric Model Development Flowchart. Source: (ICAO, 2006) ix

14 List of Tables Table 1.1 List of Airport in AP2. Source: AP2 s document... 3 Table 1.2 Method to answer Research Question. Source: own work... 6 Table 2.1 Type of Functional relationship. Source: adapted from (ICAO, 2006) Table 2.2. Hourly capacity and annual service volume of various runway configuration Source: Compiled from (Horonjeff, McKelvey, Sproule, & Young, 2010) and (Whitford, 2003) Table 2.3. Typical Peak Hour Passenger (FAA) and Peak Hour Coefficient (PM178/2015) Source: (FAA, 1988) and (Indonesia Ministry of Transportation, 2015) Table 2.4. Standard space requirement for passenger terminal per TPHP. Source: (Horonjeff, McKelvey, Sproule, & Young, 2010) Table 2.5. Uncertainties in Schiphol Airport. Source: (Kwakkel, Walker, & Wijnen, 2008) Table 3.1. Peak Hour Coefficient (PM178/2015) Source: (Indonesia Ministry of Transportation, 2015) Table 4.1. Explanatory Variables. Source: own work Table 4.2. Parameter estimation of General Domestic Passenger Demand model in AP2 Source: own work Table 4.3. Parameter estimation of International Passenger Demand in AP2 Source: own work Table 4.4. MAPE of the models. Source: own work Table 4.5. Domestic Passenger Model of AP2 Airports Source: own work Table 4.6. International Passenger Model of AP2 Airports Source: own work Table 4.7. Total Passenger in 2015 and 2030 Source: own work Table 4.8. Total ATM in 2015 and 2030 Source: own work Table 4.9. Input for Facility model Source: own work Table Required Number of Runways Source: own work Table Required parking stands in Apron-Gate Complex Source: own work Table Required Passenger Terminal Size in 2015 in m 2 Source: own work Table Occupancy level comparison Source: own work Table Peak Hour Coefficient of Empirical Data and PM178/2015 Source: own work Table Passenger terminal size requirement in 2030 (base scenario) Source: own work Table Base infrastructure recommendation for AP2 Source: own work Table Uncertainty modification to fit the condition of AP2. Source: own work Table List of Uncertainties in AP2 Source: own work Table 4.19 Total infrastructure required in 2030 Source: own work Table Final infrastructure recommendation for AP2 by 2030 Source: own work Table B.0.1. Airport Demand Model in Literature. Source: own work Table D.0.2. Fleet Mix in AP2. Source: processed from AP2 s documents Table D.0.3. Characteristics of Arrival Operation Table D.0.4. Separation rules for arrival (in NM) Table D.0.5. Characteristics of departure operation Table D.0.6. Separation rules for departure (in sec) Table D.0.7. Average Turn Around Time and Fleet Mix. Source: Processed from AP2 s documents x

15 Glossary ATM Air Traffic Movement AP2 Angkasa Pura 2 DGCA FAA GDP GDRP IATA ICAO MoT MSOE PHC TPHP Directorate General of Civil Aviation Federal Aviation Administration (US) Gross Domestic Product Gross Domestic Regional Product International Air Transport Association International Civil Aviation Organization Ministry of Transportation Ministry of State-Owned Enterprise Peak Hour Coefficient Typical Peak Hour Passenger xi

16 xii

17 1 Introduction Background Recent Practices in Airport Planning Air transportation is one of the vital systems in the world since it contributes enormous economic and social benefits (ATAG, 2005; ATAG, 2016). Some of the benefits are the generation of jobs, tourism stimulation, and access to remote areas. Besides its importance, aviation is also a promising industry. The world s air traffic for the past few years has shown an annual average of six percent growth rate (Airbus, 2017; Boeing, 2017). One of the elements of the system is the airport. The airport is the key infrastructure which enables passengers to fly, that links the passengers with the aircraft. However, the construction of airport infrastructure needs a huge investment. Without a proper investment planning, the development of airport will cause an enormous sunk cost and other consequential problems. One recent example is the failure with airport development in Sri Lanka. The government finished building the Mattala Rajapaksa International Airport in 2013, which turned out to be the emptiest international airport in the world (Shepard, 2017). Per 2016, it is reported that it only served two flights per day in the airport which cost $190 million. Thus, planning an airport a new airport or expansion of existing airport is a real challenge. As stated in the example, careless analysis on the airport demand can result in a serious problem. On the other hand, insufficient supply of capacity can lead to congestion problem which results in lower consumer welfare. In 2014, it was estimated that total additional fare premium 2.1 billion paid by passengers in Europe because of the capacity scarcity (Burghouwt, et al., 2017). In addition, the airport is a complex system with interrelated sub-systems (ICAO, 1987). Scholars have shown similar interest by developing methods to manage the airport infrastructure efficiently, such as terminal planning (Solak, Clarke, & Johnson, 2009), apron capacity estimation (Mirkovic & Tosic, 2014) and runway capacity (Bubalo & Daduna, 2011). These methods, however, require advanced calculation of the capacity of the airport sub-systems such as simulation and optimization. IATA has published a guideline for the airport operator or designer to develop a complete master plan of an airport (IATA, 2004). With this guideline, a checklist of all infrastructures and facilities needs to be prepared by airport operator to obtain a complete plan and the so-called Airport Master Plan (AMP). A similar guideline has also been set up by FAA of the United States (FAA, 2005). These standard guidelines merely provide the lists of elements which need to be provided. Nonetheless, they do not equip the users with a calculation method to determine the ideal numbers (e.g., the number of runways). The most fundamental part of an AMP is to forecast airport demand. There are numerous forecasting methods, from conventional to advanced ones. The examples for the advanced methods are Banging Holt Winter (Dantas, Oliveira, & Repolho, 2017) and ARIMAX (Andreoni & Postorino, 2006). Yet, these advanced methods are lacking in the explanation capability with regard to the previous trend. Thus, the conventional econometric model is still favorable since it provides the user with a simple causal relationship that is easy to understand. However, Sismanidou & Tarradellas (2017), reviewed the master plan of Madrid-Barajas Airport which only used a linear relationship between demand and GDP in the model, and they concluded that such method is insufficient. Hence, there is still no consensus on the best forecasting model. Other than the aforementioned concerns, forecasting also requires significant attention to be given towards uncertainty. In this case, uncertainty does affect not only the demand level but also other aspects which, eventually, influence the decision-making process. A trend in the aviation industry, such as market liberalization is one of the examples of this uncertainty. In a market liberalization, the increased level of competition forces the airlines to optimize their benefits in many ways, including using the hub-and-spoke network (Burghouwt & Huys, Deregulation and the Consequences for Airport Planning in Europe, 2003). It means that, when uncertainty unfolds, there are consequences to be borne by the airport operators.

18 Introduction 2 The crucial involvement of uncertainties in airport planning can also be seen in the development of recent studies. Kwakkel, Walker, and Wijnen (2008) have identified the uncertainties in Schiphol Airport as a first attempt to analyze the uncertainty in Airport Planning. As the follow-up, Kwakkel, Walker, & Marchau (2010) have shown a new way to develop an Adaptive Airport Strategic Planning (AASP). Other than Schiphol Airport, the same approach has not been found to be applied in another airport in the literature. Airport Infrastructure Planning The goal of having a Master Plan is to have a detailed design to make decision on each airport facility and infrastructure to fulfill future demand. There are many aspects covered in Airport Master Plan. Starting from the selection of project location for a new airport, infrastructure design decision, layout design and finally a financial analysis of the plan. The process of determining the amount or size of infrastructure required in the master plan is specifically called airport infrastructure planning. By having the plan, an airport operator can prepare the budget to fund the expansion project. One example of the long-term plans of future investment in airport system was conducted in India (Singh, Dalei, & Raju, 2016). It shows how the required future investment in expanding the capacity of the airports is projected. However, the passenger demand model by Singh, Dalei & Raju (2016) still uses the relationship between GDP and demand to forecast future demand. As mentioned before, this method was found to be problematic, as argued by Sismanidou & Tarradellas (2017). Their study only provides the reader with information about the total investment in the future. The explanation of how the study derived investment value is simply based on the past investment in the last five years. It does not clearly explain how the detailed requirements for the expansion plan should be provided. Infrastructure planning shall be used as a foundation for future investment. This is due to the reason that expanding an infrastructure is a discrete decision that cannot linearly be withdrawn from the projected demand. For example, a demand of 100,000 aircraft movements (ATM) a year can be served by a single runway. If the future demand is doubled, it does not necessarily mean adding a new runway since a runway can accommodate up to ATM a year. Nevertheless, Singh, Dalei & Raju (2016) derived the future infrastructure investment from the projected demand. Some infrastructures, e.g., runway, have a lumpy cost that cannot be estimated without knowing how much adjustment is required. Moreover, the aggregate level of analysis (national levels) gives little information about the requirement in each airport across the country. Whereas, the demand evolution in each airport may differ. Airport Planning Problem in Angkasa Pura II (Indonesia) Indonesia has a unique transportation challenge: the sovereignty of the country is connected by the sea. Air and maritime transport are critical systems which complement each other. However, a city at thousands of kilometers distance needs some days or even weeks to be reached by ships. Yet, it can be visited within hours using an airplane. Thus, the role of air transportation becomes more crucial in such archipelagic country. The importance of air transportation in Indonesia calls for much attention to be given to the airport operators. The airport operators are responsible for making sure the benefit or air transport is supported from the ground. The management of Airports in Indonesia is given to three entities: Angkasa Pura I (AP1), Angkasa Pura II (AP2), and Technical Operation Unit of the Indonesian Directorate General of Civil Aviation (DGCA). AP2 operates 13 airports out of 300 airports in Indonesia (see Figure 1.1). AP2 is a one of the State- Owned Enterprises (SOEs) in Indonesia which is also constrained by the regulations issued by the Ministry of Transportation (MoT). The Ministry of SOE (MSOE) controls the business aspects of the company, and the MoT regulates the operational process and safety standard of the operations. The basic regulation for the operational guideline is the Indonesian Aviation Act No.1/2009. Additionally, the international airports in AP2 also need to comply with other international regulations, e.g., IATA and ICAO.

19 Introduction 3 Figure 1.1. The map of Indonesia and 13 Airports of Angkasa Pura II. Source: AP2 s website. Balancing capacity and demand of the airports is a serious issue in AP2 (Table 1.1). Per 2015, many airports in AP2 suffer from the over-capacity situation. The occupancy level of 7 out of 13 airports was in the over-capacity situation (see the last column in Table 1.1Error! Reference source not found.). Even t he condition in one of the airports, CGK, was at the level of 247%. Thus, planning appropriate infrastructure expansions in the future of those airports is important. The difference in the demand level is a result of different market served by each airport which has various economic activities in which, to some extent, affect the dynamics in the airport. These differences call for more attention to be given to each of the airports. The brief introduction to each airport and the historical traffic development from 2001 to 2016 can be seen in Appendix A. Table 1.1 List of Airport in AP2. Source: AP2 s document No Name Location Province 1 Soekarno-Hatta International Airport Cengkareng - Jakarta IATA Code Passenger Capacity 1 (2015) Passenger Movement 2 (2015) Jakarta CGK 22,000,000 54,291, % Occupancy Level 3 2 Kualanamu International Airport 3 Halim Perdanakusuma Airport 4 Sultan Mahmud Badaruddin II Airport 5 Supadio International Airport 6 Minangkabau International Airport 7 Sultan Syarif Kasim II International Airport 8 Husein Sastranegara Airport 9 Sultan Iskandar Muda International Airport 10 Raja Haji Fisabilillah Airport Kualanamu - Medan North Sumatra KNO 9,000,000 8,004,791 89% Jakarta Jakarta HLP 1,900,000 3,059, % Palembang South Sumatra PLM 3,000,000 3,384, % Pontianak West PNK 2,400,000 2,713, % Kalimantan Padang West PDG 3,000,000 3,169, % Sumatra Pekanbaru Riau PKU 3,500,000 2,670,046 76% Bandung West Java BDO 2,400,000 3,146, % Banda Aceh Aceh BTJ 1,000, ,721 75% Tanjung Pinang Riau Islands TNJ 1,200, ,936 22%

20 Introduction 4 11 Depati Amir Airport Pangkal Pinang Bangka- Belitung Islands PGK 500,000 1,658, % 12 Sultan Thaha Airport Jambi Jambi DJB 1,500,000 1,168,219 78% 13 Silangit Airport Siborong- Borong North Sumatra DTB 100,000 17,784 18% Notes: 1) The design capacity of the passenger terminal 2) The actual passenger traffic in the airport 3) Occupancy level is based on the actual passenger traffic and the design capacity of the passenger terminal The MSOE and MoT have different interests on AP2. The MSOE has a focus on making AP2 into a profitable company that can supply an income to the country. On the other hand, the focus of MoT is to ensure the airports fulfill the standard of safety, security, and service of the aviation industry. Consequently, every investment initiative, especially the capital-intensive project (e.g., infrastructure improvement), should be studied carefully. It is in line with the blueprint of Indonesia s Air Transportation in In this document, the direction of air transport development policy in infrastructure is: The effort to improve the ability of airport services through the construction and development will be based purely on the investment feasibility and follow market mechanism (Demand Analysis) which are conducted by the airport operator without the aid of government subsidies, with the government's authority in certifying its operation. (Indonesia s Air Transportation Blueprint , p.vi-29) The above statement implies that AP2 needs a convenient method or tool to verify its long-term development policy to comply with such government directive. The long-term investment in this context is the expansion of airport infrastructure of the existing airports (i.e., construction of terminals and runways in the airport landside and airside area). Unfortunately, in AP2, the studies related to the expansion of the airports are always conducted with the help of external party (i.e., airport consultant). In this case, AP2 only provides data and information to be further developed by the consultants. Unfortunately, based on the author s personal communication with the Directorate of Corporate Planning of AP2, the recommendations given by the consultants are not critically reviewed at times. AP2 merely provide feedbacks/approvals to the proposed design. This is mainly due to the limited capability of AP2 in evaluating this kind of decision. As a result, there is no proper transfer of knowledge during the process. Until now, AP2 is still lacking in conducting an airport planning by themselves. The AP2 is indicated to possess this knowledge gap that shares disadvantageous in several ways. The involvement of external parties in this matter incurs a high cost. In this case, AP2 faces an informational disparity in which it cannot ascertain the validity and the convenience of the results presented by the external parties. Therefore, it is argued that AP2 need to acquire the capability to analyze and to assess their long-term plan. The manager of corporate planning of AP2 said that in defining the future growth of demand, they are always looking at the macro outlook of the global aviation industry, social and economic development of Indonesia, and other relevant factors. But, the link between the factors is not clear. The expected growth is then decided according to the deliberation between internal experts in AP2. It means that many of the decisions made by the company are based on the intuitive judgment which may not be reliable all the time. In a way, an intuitive approach can also be interpreted as a qualitative method (ICAO, 2006). Nevertheless, the problem with such approach arises when there is a significant deviation between the forecast and the realization levels. AP2 always finds it difficult to explain why such deviation occurs. The baseless decision made in the beginning is unaccounted for. Besides, the assumptions in this method are often not well-documented. Eventually, it becomes even more difficult for the company to learn from the

21 Introduction 5 past challenges. Moreover, despite the importance of taking uncertainty into account, this particular aspect of airport planning has not been addressed in the current master plan. Noting the importance of airport infrastructure planning, determining a proper infrastructure estimation method for AP2 becomes salient. The study by Singh, Dalei, & Raju (2016) attempted to address a similar problem in India. However, this approach is limited in considering some of the most important factors, e.g., the difference in the economic growth of each region, into the demand forecast model. Additionally, the study does not provide a detailed recommendation for the infrastructure development. Moreover, the aspect of uncertainty is not taken into account. These issues are considered as the gap in solving AP2 infrastructure planning problem, in which this thesis aims to cover. To conclude, there is a gap to have a convenient method of calculating the future airport infrastructure at the strategic level. Hence, the airport operators or airport authority, especially in developing country, can plan their strategic infrastructure expansion in an accountable way and independently. This research tries to propose a method for estimating a long-term infrastructure for airport system owner in order to manage the expansion of their airport. Research Objective and Research Question The main objective of this research is to develop a practical method for an airport operator to predict the future demand of a set of airports and assess their available infrastructure to balance the future demand. The focus of this study is to analyze the infrastructure planning in macroscopic level. A more detailed study is still needed to be conducted comprehensively especially for a specific airport that is indicated having an inadequate infrastructure. Based on the research objective, the following research question is formulated: What is the convenient method for an airport system owner in developing countries to assess the infrastructure requirement of the airports for balancing the future demand considering the uncertainties in the future? The main research question can be solved by answering three sub-questions. First, a review of the current methods that can be used to explain the evolution of airport demand and predict future airport demand. SRQ1. What is the best method to explain the development of demand and forecast the future demand of airports? 1.1. What are the determinants of airport demand in an airport system? 1.2. How to choose the best model to forecast airport demand in an airport system? Second, after having the traffic forecast, the sizing of infrastructure should be conducted. There are many infrastructures should be developed in an airport system such as runway, taxiway, apron, and terminals. The appropriate method should be selected based on the available methods to estimate the capacity. SRQ2. How to estimate the required infrastructure to balance the future demand? 2.1. How to estimate the required infrastructure to serve the demand? 2.2. What are the infrastructures to be expanded in the network? Third, after having the estimated infrastructure as the base case, the airport operator should assess it, since the uncertainty may affect the required infrastructure. Thus, the uncertainty should be analyzed and incorporated before going to the final recommendation. SRQ3. What are the uncertainties that need to be considered in developing the airport system and how these uncertainties can be included in the decision-making process?

22 Introduction 6 Answering the abovementioned questions will result in a practical method to help an airport operator in developing countries to estimate the required infrastructure in balancing the future demand. Then, the developed method is applied in the available case study. Research Methodology In resolving the research questions, both qualitative and quantitative methods are used. The qualitative method of the literature review is conducted to answer the early sub-research questions. Quantitative methods, such as regression analysis, is used when the proposed infrastructure estimation method is applied to the case study. The summary of SRQs along with chosen method and relevant section throughout this thesis is depicted in Table 1.2 Table 1.2 Method to answer Research Question. Source: own work SRQ Method Relevant Section 1. Method for analyzing and forecasting airport demand 1.1 Determinants of airport demand 1.2 Model for demand forecast 2. Method for estimating airport infrastructure 2.1. Model to estimate the required infrastructure 2.2. Infrastructure that can be improved 3. Incorporating Uncertainty Literature Review Case Study Literature Review Case Study Regression analysis Literature Review Case Study Literature Review Case Study Literature Review Case Study Chapter 2.1, Chapter 3.2.1, and Chapter 4.1 Chapter 2.1, Chapter 3.2, and Chapter 4.1 Chapter 2.2 Chapter 3.4, and Chapter 4.2 Chapter 2.2 Chapter 2.3, Chapter 3.4, and Chapter 4.3 SRQ1 is answered by exploring the literature that discussed used and tested variables in developing airport demand model. At first, we searched in Scopus with these keywords: TITLE-ABS-KEY (airport AND demand AND determinant) TITLE-ABS-KEY (forecasting AND airport AND demand) and in google search engine with this keyword: Analysis and forecast of air transport demand After finding the relevant article, the backward and forward snowball methods are conducted. All relevant cited papers in the article are checked. The papers that are not relevant or discussed another topic are filtered out. This process results in 15 papers as can be seen in Chapter 2.1 and Appendix B. Then, several variables are selected for proposed method (Chapter 3.2.1) and examined using the data of the case study in Chapter 4.1. SRQ2 is answered by using literature review and mathematical model formulation. The first step is finding the current analytical method that has been employed in Scopus with these keywords: TITLE-ABS-KEY (runway AND capacity) OR (runway AND estimation) TITLE-ABS-KEY (apron AND capacity) OR (apron AND estimation) TITLE-ABS-KEY (airport AND terminal AND capacity) OR (airport AND terminal AND estimation) In addition, the documents from ICAO, FAA, a handbook of airport planning, and the documents regarding regulation from Indonesia (as the case study) are also used as input materials. All discovered methods in the literature are presented in Chapter 2.2. The relevant calculation models are adapted to the proposed

23 Introduction 7 method (Chapter 3.4). Further adjustment is then executed after the models are applied to the case study (Chapter 4.2). SRQ3 is answered by using literature study. The first step is exploring the method to analyze the uncertainty in airport system and how to incorporate the uncertainty to the planning process (Chapter 2.3). After knowing how to analyze the uncertainty, the uncertainty in the case study is explored and analyzed (Chapter 4.3). Next, these uncertainties are included in the model to increase the robustness of the recommendation. Thesis Layout Chapter 1: Introduction This chapter explains the background story of this research, the objective, and the formulated research question. Finally, this chapter is closed by presenting the methodology to answer the research questions and the layout of the thesis report. Chapter 2: Literature Review This chapter presents the literature reviews with regard to methods for analyzing and forecasting demand in the airport, models for analyzing airport infrastructure, and uncertainty in airport system. Chapter 3: Proposed Method This chapter explains the step-by-step process in developing the recommendation of long-term infrastructure at the airports. Chapter 4: Application of the Proposed Method This chapter shows the implementation of the method to the available case study. The results and analysis of all tests cases are also presented. Chapter 5: Conclusions and Recommendations This final chapter recaps the outcome of the thesis and the recommendations for future research.

24 Introduction 8

25 2 Literature Review The relevant literature to answer the research question are discussed in this section. This chapter is presented to answer the research questions and as the input for the development of the method in Chapter 3. The first topic is the methods to develop airport demand model. Second, the explanation of the airport infrastructures and the methods to estimate them to balance the demand in airport planning process. Third, the explanation about the uncertainty in airport system and how the uncertainty is currently addressed. Methods for Analyzing and Forecasting Airport Demand Having reliable forecasts in aviation sector play a critical role in the planning process (ICAO, 1987). Consistent and dependable prospective demand model enables airport planner to design adequate infrastructures. There are a lot of modeling techniques used in aviation planning. ICAO published a manual for air transport planner to forecast the future traffic (ICAO, 2006). The methods classification can be categorized into three main categories: Quantitative, Qualitative, and Decision Analysis (see Figure 2.1). Forecasting Techniques Quantitative Qualitative Decision Analysis Time-series Causal Methods Delphi Technique Market research and industry surveys Trend Projection Regression Analysis Technological forecasting Probabilistic analysis Decomposition Methods Simultaneous Equation Model Bayesian analysis Spatial Equilibrium System dynamics Figure 2.1. Alternative Forecasting Techniques. Source: adapted from (ICAO, 2006) The quantitative methods use historical data to be analyzed and develop forecast based on the mathematical formulation. The quantitative methods can be divided into two subcategories: time series analysis and causal methods. Some of the common techniques used for this category are Trend Projection and Regression Analysis. Time series method is the most common quantitative methods used in practice. However, this approach is limited by the inability to explain the causes of past growth and projected growth in the future. Development of

26 Literature Review 10 The qualitative methods are usually used in the situation in which there is lack information about past data. Thus, judgment of the experts is used to estimate the future. Because these methods rely on the experience of the expert, it is also realized as intuitive methods. Regression analysis as the main method in causal methods is commonly utilized for modeling air travel demand in the literature. One of the advantages of this method is this approach can also help the analyst explain the driver of the airport demand and further predict the future demand based on the expected future situation although there should always be an intuitive story behind the determinants used in the model (Janic, 2013). Since, scholars and practitioners curious about the story behind the growth, many studies on air travel demand found in the literature. All indicators from the literature (see Appendix B) can be classified into (1) economic, (2) demographic, (3) tourism, and (4) others. Economic The first dominant factor stands out in literature is the economic activity of the region. (Ishutkina & Hansman, 2009) show the conceptual relation between economy and air transport system (Figure 2.2). They conclude that the correlation between air travel in a country and GDP does exist. However, the mechanism and magnitude of the relation is differed between one country and another. This result is aligned with many studies that argue economic activity, most of the case is represented by GDP, influence air travel demand such as in Saudi Arabia (BaFail, 2004), German (Grosche, Rothlauf, & Heinzl, 2007), Bangladesh (Wadud, 2014) (Wadud, 2014), China (Yao & Yang, 2012), and Sweden (Kopsch, 2012). Besides GDP, the similar indicator that represent economic activity are total expenditure (BaFail, Abed, & Jasimuddin, 2000; Abed, Ba-Fail, & Jasimuddin, 2001; BaFail, 2004), GDP per capita (Valdes, 2015), or oil GNP (BaFail, Abed, & Jasimuddin, 2000; Abed, Ba-Fail, & Jasimuddin, 2001). The other economic indicator used in literature is the trade. It is assumed that the export and import activity also attract the passenger flow, especially for business trip. (BaFail, 2004) include import value in the final demand model and (Yao & Yang, 2012) use the ratio of trade over GDP to the model. Demography Figure 2.2. The causal relation between economy and air transport system. Source: (Ishutkina & Hansman, 2009) The next dominant factor is population or demographic condition. It is a natural thing to assume that more people in the region will lead to more passenger using the airport. Most of the studies have proven that population has a statistically significant relation with air travel demand such as in (BaFail, Abed, &

27 Literature Review 11 Jasimuddin, 2000), (Abed, Ba-Fail, & Jasimuddin, 2001), (BaFail, 2004), (Abbas, 2004), (Grosche, Rothlauf, & Heinzl, 2007), (Wadud, 2011), (Wadud, 2014), (Yao & Yang, 2012), (Kopsch, 2012), (Baikgaki & Daw, 2013), and (Sivrikaya & Tunç, 2013). Tourism The tourism has also been mentioned as the determinant of air travel demand. (Devoto, Farci, & Lilliu, 2002) make a demand model for airports in Sardinia that focus on tourism indicators. This study uses population number, number of tourist arrivals, number of tourist bed, and bed per capita. (Sivrikaya & Tunç, 2013) used bedding capacity to express the tourism development in their model and (Abbas, 2004) includes the number of foreign tourist in their model. Other Besides economic and demographic, there are many other variables that are being introduced to the regression model. Most of the cases to test whether the introduced variable significantly influences the dynamics of air travel demand. Yao and Yang (2012) conclude that economic growth, industrial structure, population density, and openness has a positive impact on air transport. In the study that used origin-destination (O&D) data, the additional use of geographical characteristics and service related data is found. The distance from two airports is used to represent geographical impedance to the demand (Grosche, Rothlauf, & Heinzl, 2007) (Sivrikaya & Tunç, 2013). The servicerelated data that used are the airfare (Kopsch, 2012) (Sivrikaya & Tunç, 2013). The recap of literature that studies the relationship between air travel demand and its driving factors can be seen in Appendix B. Besides the variable issue, the several functional relationships are common in the econometric analysis (see Table 2.1). The linear structure uses a standard linear relationship of all dependent variables to the independent variables, and the log-log structure uses additional natural logarithm transformation to both, the dependent and independent variables. In the literature, the log-log model is utilizing the natural logarithm (Wadud, 2014) as well as the base logarithm (Sivrikaya & Tunç, 2013). Although this difference, parameter interpretation of both is the same. Table 2.1 Type of Functional relationship. Source: adapted from (ICAO, 2006) No Type of structure 1 Linear Basic Formula N Y = β 0 + β i x i + ε i=1 Interpretation β denotes the marginal effect of x on y. The increase (or decrease) of x will linearly increase (or decrease) y. 2 Log-log ln(y) = β 0 + β i ln x i + ε Notes: Y is the dependent, x i is the independent variable and β is model parameter N i=1 β represents the elasticity of x on y. Thus, this specification can be used if it is expected that the percentage increases in x lead to constant percentage changes in y. (e.g., constant demand elasticity). In deciding the type of functional relationship to be used for an econometric traffic forecast, the analyst can choose the appropriate type through judgment, experimentation, experience and or prior knowledge of the problem. The relationship validity can be examined empirically through tests against actual historical data. Both structure, linear and log-log are almost equally employed by researchers. (BaFail, Abed, & Jasimuddin, 2000) (Abed, Ba-Fail, & Jasimuddin, 2001) (Abbas, 2004) employed Linear model structure. On the other hand, (Grosche, Rothlauf, & Heinzl, 2007) (Wadud, 2011) (Yao & Yang, 2012) (Kopsch, 2012) (Baikgaki & Daw, 2013) (Valdes, 2015) present log-log model structure in modeling airport demand. Unfortunately, no literature compares both functional structure side-by-side. Hence, in the method development, the efficiency of both models is utilized and examined.

28 Literature Review 12 Analysis of Airport Infrastructure The element of the airport is divided into two major components: the airside system and the landside system. Both systems are interdependent each other. Different type of studies is performed in planning the airport such as facility planning and financial planning. All the studies can usually be categorized into three levels: the system planning level, the master planning, and the project planning level. An airport system plan is a general overview of aviation facilities required to serve the future demand of a specific region or country. The National Plan of Integrated Airport System (NPIAS) is one of Airport System Plan of the United States which is stating the future development of the airports. This type of study is also used as an input for a detailed study of the airport master plan. Therefore, the Airport master plan contains the concept of ultimate development of an airport. Airside Infrastructure Airport airside system mainly consists of Runway, Taxiway, and Apron. These elements form the capability of an airport to serve aircraft. A sufficient number of the runway, the taxiway, and apron size is important to meet the forecast traffic demand Runway The runway is an essential part of the airport. In most small airports, the number of the runway is not a problem because the Air Traffic Movements (ATM) is usually below 35 operation/hour. In the large airport, the calculation of runway capacity become critical to check if an additional runway should be constructed. Unfortunately, estimating runway capacity is not a simple task. Many factors affect the capacity, namely the layout or configuration, separation requirement, weather condition, the wind rose, a mix of aircraft, the sequence of operations, quality of ATM, exit location, and noise considerations. Most of the available methods see the runway as a server in the queueing system. Hence, the factors related to arrival-departure or landing-takeoff procedure are the main factor in defining runway capacity. The calculation of runway capacity formulation can be approximated using the following formula: Where 1 C m = E( T ij ) (1 + n dp nd ) (1) C m = Capacity of the runway to proceed mixed operations E( T ij )= Expected value of interarrival time n d p nd = number of departures which can be released each gap between arrivals = probability of releasing n d departures in each gap For guiding purpose, some of the approximation of runway capacity based on several runway configurations are presented in Table 2.2. Many runway configurations are used such as intersecting runways and open-v runways, but the maximum capacity can be achieved with parallel configuration (Whitford, 2003; Horonjeff, McKelvey, Sproule, & Young, 2010). Table 2.2. Hourly capacity and annual service volume of various runway configuration Source: Compiled from (Horonjeff, McKelvey, Sproule, & Young, 2010) and (Whitford, 2003) Number of Runways Runway Configuration Hourly Capacity VFR IFR Annual Service Volume 1 Single Two parallel (near) Two parallel Two parallel (far) Two parallel + one runways Two pairs of parallel

29 Literature Review 13 In this part, the method to estimate capacity of runway system is found. This will be the input to calculate the required number of runways in an airport Taxiway Network Taxiway network is the pavement area that connects various part of the airport, especially the runway and the apron area. The layout of taxiway can determine the capacity of the runway system especially in peak hour of the busy airport. Several taxiway components that can increase the capacity are Rapid Exit Taxiways (RET), parallel taxiway, and queueing taxiway. The capacity of taxiway is hard to be studied alone but together with runway system through simulation such as in (Scala, Mota, & Delahaye, 2016) and (Uehara, Hiraishi, & Kobayashi, 2015). Additionally, the capacity of a taxiway is generally far exceeding the capacity of the gates or the runways except for the cross-runway taxiways (Ashford, Mumayiz, & Wright, 2011). Since there is no analytical method can be used to estimate a parameter of the taxiway, the sizing of taxiway is omitted and assumed sufficient Apron-Gate Complex The study of Apron capacity and infrastructure requirements is not as much as runway system in the literature because, in an airside system, the runway is the most problematic one. However, some studies are found that addresses the planning issue of the apron. Basically, the number of aircraft parking stands at the apron depends on aircraft movements by aircraft type during the peak hour and their gate occupancy time (ICAO, 1987). Mirkovic & Tosic (2014) shows a comprehensive review of the current state of the art on a study of apron capacity. Two approaches are discussed: microscopic model (simulation) and macroscopic model (analytical model). The simulation model is suitable for the planning process when there is sufficient input and usually in the operational planning stage. When rough data is used or in the strategic planning stage, an analytical model is more suitable. There are two different analytical apron capacity estimation models found. The first, the basic model, assumes that all aircraft can use all stands available at the airport. The second, the restricted model, assumes that a stand can accommodate the aircraft class they are designed for and all smaller-sized aircraft. The first formula to calculate the number of aircraft stands is as follows: Where S T i N i α = required number of aircraft stands = gate occupancy time in minutes of aircraft group i = number of arriving aircraft group i during peak hour = number of extra aircraft stands as spare S = ( T i 60 N i) + α (2) The second method is pretty similar to the first model. The only difference is the second model uses the stand-use restriction to higher aircraft size. The first model is found suitable for the case to calculate the number of parking stands in Apron-Gate Complex. Landside Infrastructure The landside system of an airport is mainly focused on the study of the passenger terminal. Many processes are happening in the terminal because it is the area where interaction between passengers, airlines, quarantines, customs, and immigration officers are located Passenger Terminal In designing passenger terminal, numerous aspects are considered until in the end, the size of the terminal is found. However, for long-term planning purposes, there is not enough time to analyze the required terminal size for each specific airport. Thus, the forecasted annual volume of passengers is usually being used for planning and design of terminal area and facilities. In order to do that, one modest approach is by estimating the magnitude of operation in peak hour. The peak demand rather than annual demand must be determined to evaluate facility requirements since capacity utilization of airport facilities becomes more critical during peak hour or peak day (ICAO, 1987).

30 Literature Review 14 The annual forecast, thus, has to be converted into peak periods for both, aircraft movements (ATM) and passenger throughput. There is a belief that annual passenger has a relationship with its peak hour passenger. The use of peak periods, especially peak hour period, has been defined differently in practice. ICAO (1987) mentioned that the typical peak hour is common as 30 th or 40 th busy hour. Busy hour rate (BHR) is the term that used in the UK that is the value of passenger flow which only 5 percent of annual traffic experience this level of service. FAA in the US uses typical peak hour passengers (TPHP) in defining their airport size. The FAA has made the conversion guideline to translate annual passenger traffic into an hourly passenger traffic (Table 2.3). As the problem in the case study is in Indonesia, the regulation about airport management in Indonesia are sought. The Indonesian government published a ministerial decree No. 178 the year 2015 or PM178/2015 as a guideline for airport operator in Indonesia to comply the airport design of national regulation. However, it is found that the Peak Hour Coefficient in PM178/2015 is the same as TPHP of FAA (Table 2.3). Table 2.3. Typical Peak Hour Passenger (FAA) and Peak Hour Coefficient (PM178/2015) Source: (FAA, 1988) and (Indonesia Ministry of Transportation, 2015) Annual Passenger (in million Pax) > % 20-29, % 10-19, % 1-9, % 0,5-0, % 0,1-0, % <0, % TPHP as a percentage of Annual Traffic or Peak Hour Coefficient The term TPHP does not represent the maximum passenger demand of the airport. It is, however, well above the average demand and considers periods of high airport usage. Whitford (2003) mentioned a different formula to compute TPHP which is as follows: TPHP = ENP 0.9 for the large airport (over passengers) (3) TPHP = ENP 0.9 for the small airport (below passengers) (4) Where, ENP equals annual enplanements. From three methods found, two methods have the same formulations which are the FAA and PM178/2015 method. Thus, the two methods are considered as converting method from annual traffic into peak hour traffic. The result of peak hour traffic then used as input for further estimation steps. After having the TPHP, the next step is calculating the overall size of the passenger terminal. Overall estimation of Passenger terminal has found in some literature such as in (Whitford, 2003) and (Horonjeff, McKelvey, Sproule, & Young, 2010). Directorate General of Civil Aviation of Ministry of Transportation of Indonesia also has a standard rule in managing Indonesian airports which are stated in PM178/2015 (Indonesia Ministry of Transportation, 2015). Ashford and Wright present a common measure used for long-term planning of airport which is 150 ft 2 (14m 2 ) per TPHP for domestic passengers and 250 ft 2 (24m 2 ) per TPHP for international passengers (Whitford, 2003).

31 Literature Review 15 FAA has stated the space requirements for the terminal area which is between 0.08 and 0.12 ft 2 (0.007 to m 2 ) per annual passenger. Another estimation is also can be made by applying a size of 150 ft 2 (14 m 2 ) of gross terminal building per design peak-hour passengers. Horonjeff et.al., (2010) presents approximation value that can be a guideline for airport designer which is based on Typical Peak Hour Passengers volume (Table 2.4). The values in the table should provide a reasonable level of service and tolerable occupancy level (Horonjeff, McKelvey, Sproule, & Young, 2010). Table 2.4. Standard space requirement for passenger terminal per TPHP. Source: (Horonjeff, McKelvey, Sproule, & Young, 2010) Component Space Required (m 2 ) Component Space Required (m 2 ) Ticket Lobby 1 Total Gross Area Baggage Claim 1 Domestic 25 Departure Lounge 2 International 30 Waiting rooms 1.5 Immigration 1 Customs 3 Amenities 2 Airline Operations 5 Directorate General of Civil Aviation of Ministry of Transportation published a standard area requirement for Passenger terminal in PM178 No Terminal area per peak hour passenger is 14 m 2 for domestic and 17 m 2 for international excluding the space needed for circulation, utility, and concession rooms. In this regulation, the coefficient to translate annual passenger to peak hour passenger is also provided (Table 2.3). In addition to the basic terminal size, the size of the commercial area is up to 30% of the operational area. Additional space for circulation, utility, and concession rooms are 20% of the total terminal size. In this part, all important infrastructure in an airport has been presented. The methods to estimate these components has been also found. This will be input to develop the method to estimate the infrastructure. Uncertainty in Airport System A general definition of uncertainty is any departure from the unachievable ideal of complete determinism (Walker, et al., 2003). In developing any model, uncertainty is always inherent in it. Uncertainty is not as simple as a lack of knowledge but as an increase in knowledge might lead to things that we don t know, thus, increase the uncertainty. To help in understanding the uncertainty that is essential in model-based decision support, Walker, et.al., (2003) develop a framework which has three dimensions: location, level, and nature of uncertainty. The location of uncertainty is a dimension that describes the source of uncertainty. It can be in the input or within the model. The level of uncertainty explains how uncertain is the uncertainty. The last dimension is the nature of uncertainty. This dimension tries to categorize whether the uncertainty is something that can be studied (epistemic) or it is a natural variability of the phenomenon. This framework has been applied in the case of Schiphol Airport (The Netherlands) to examine the emerged uncertainties (Kwakkel, Walker, & Wijnen, 2008). As the follow-up, the concept of Adaptive Airport Strategic Planning is introduced (Kwakkel, Walker, & Marchau, 2010). This is the first attempt that multiple uncertainties are considered in Airport Planning. Previously, the planning approach that handles the full range of uncertainties has seen limited application (Hansman, Magee, De Neufville, Robins, & Roos, 2006). Since the uncertainty analysis in the airport system is introduced by (Kwakkel, Walker, & Wijnen, 2008), there is no other systematic analysis of the uncertainty in the airport system. Principally, the uncertainty in (Kwakkel, Walker, & Wijnen, 2008) can be categorized into three parts: (1) uncertainties affecting airport demand, (2) uncertainties affecting airport capacity, and (3) uncertainties affecting both, demand and capacity. By classifying the uncertainty, the relation of demand and infrastructure requirement (or capacity) can be conceptualized (Figure 2.3).

32 Literature Review 16 Figure 2.3. A conceptual model of the uncertainty in the airport system. Source: own work. Uncertainties in Schiphol Airport (SA) (Kwakkel, Walker, & Wijnen, 2008) analyze the uncertainties that are stated in policy documents and other relevant uncertainties that are considered of being important. The list of the uncertainties is presented in Table 2.5. The uncertainties have been categorized using the conceptual model in Figure 2.3. Table 2.5. Uncertainties in Schiphol Airport. Source: (Kwakkel, Walker, & Wijnen, 2008) Impact Location Demand Side Supply side Both The uncertainty Aviation Demand Demography Modal Choice Landside Accessibility Aviation Technology Air Traffic Management Technology Weather Patterns Implementation uncertainty Land use developments System operations Rules and Regulations Evaluation of outcomes Fleet mix The demand in SA is influenced by the aviation demand uncertainty which is resulted from the several incidents occurred in the past (e.g., epidemics and terrorist issue) and from the impact of several airlines strategic decisions (e.g., LCC operation and merger of airlines). The impact of oil price and economic development is also questioned. In addition, the demography structure, the distribution of people living in the country, age composition and ethnicity in the future might change the demand pattern. Since located in the advanced transportation network, the development of competing for modal choice and accessibility to the airport can interfere the demand of SA. In the supply side, the new aviation technology, i.e., aircraft and new ATM technology can influence the operational procedure of the airport and leads to extending the capacity of the airport. In the tactical level, unstable weather pattern can reduce the capacity of the runway. New rules and regulations in system operation of airports can adjust the capacity of the airport. For example, if in the future the process in check-in counter is replaced by automatic check-in using biometric identification, the area for a queue can be reduced significantly. Thus, the overall capacity of the passenger terminal may increase. The changes in the evaluation of outcome such as the noise impact can also alter the capacity of the airport since its affect the regulation of the airport. Last, the land use development can also alter the airport capacity. In accordance with the limit on noise, the land use development in surrounding airport area may shrinkage the airport capacity.

33 Literature Review 17 The uncertainty of the fleet mix is the result of privatization and liberalization of airlines network. This uncertainty can affect both the demand uncertainty and supply uncertainty of the airport. In the extreme case, the use of one type of aircraft will reduce the separation time and increase the capacity of a runway. On the other hand, the increased use of high capacity aircraft will reduce the air traffic movements to be handled. Hence, the composition of the fleet in the future may affect the supply and demand side of the airport infrastructure.

34 Literature Review 18

35 Proposed Method 19 3 Proposed Method Based on the available case study and the literature review, a method to estimate the future required infrastructure is developed (Figure 3.1). The model consists of two main building blocks, the demand model, and the infrastructure model. The goal of demand model is to find the best model to explain the growth of airport traffic. Therefore, the future demand can be calculated using the estimated model. The infrastructure model, on the other hand, uses the result of demand model to estimate the appropriate infrastructure to balance the demand. Figure 3.1 Proposed Method Source: own work The econometric model is chosen to model airport demand because it is necessary to find underlying factors that have a relationship with airport demand. Many studies in the literature attempted to explain the airport demand with various explanatory variables. Both, linear and log-log structure are produced and compared to find the most suitable model to forecast the demand. In developing a set of explanatory variables, a logical reasoning comes at a first place based on the literature study. Later, stepwise regression is employed to find the simplest model or the model with the less explanatory variables.

36 Proposed Method 20 The number of domestic passengers and international passenger is chosen to be forecasted. The number of air traffic movements (ATM) are derived from the passenger number since it depends on the number of passengers. The number of ATM is approximated by dividing the passenger number with the Average Passenger per Aircraft (APA) in basis year per airport. The analytical model is used to sizing the required infrastructure components at the airports. An analytic model is chosen as the most convenient approach because it is easier to understand and does not require a lot of data compared to the simulation models. In addition, this study focuses only on macro-scale analysis. Hence, in the end, a detailed study will still be needed if indicated so. The critical infrastructure that are estimated are the runway, the apron-gate complex and the passenger terminal. These three infrastructures are known as the most capital-intensive infrastructure in airport system. The choice of these three is also supported by the survey results conducted by ACI on the perceived constraints faced by airports in balancing the future demand. Apron, runway, and terminal are the top-three source of constraints for airports in Region outside Europe and North America (ACI, 2011). In the last step, the uncertainty is analyzed and explored whether it can be used to check the robustness of the output. The relevant scenarios are made and tested to the model. The final recommendation is made based on the result of this scenario analysis. In the next part, the model will be described in more detail. But first, the assumptions in developing the methods are explained. Assumptions In this part, the necessary assumptions in developing the method are presented. These assumptions are made to enable the method development process. The assumptions are as follows: 1. Assumptions for the method The availability of operational data is limited 2. Assumptions for analytical model The future of airport demand can be analyzed from the past An analytical model can be used to estimate the required facility of airports in strategic level (macroscopic). The additional runway in runway system can be placed in parallel. Thus, the optimal capacity of the runway can be achieved. 3. Assumptions based on input data The average number of passenger per aircraft (APA) remains the same for each passenger category (i.e., domestic and international) Operational procedure of landing and take-off remains same between current situation and the future Number of ATM grows proportionally with the growth of passenger numbers. Method for Analyzing and Forecasting Airport Demand Model Structure In the demand model, a number of domestic passengers and international passenger are analyzed and forecasted separately. From the literature review in 2.1, it is known that the main drivers of airport demand are economic activity, population, trade, and tourism. Two model structures are dominant in use to model the airport demand: Linear and log-log models. Both models are adapted and analyzed first. Later, the results are compared to find the more suitable structure to forecast the airport demand.

37 Proposed Method 21 In analyzing the demand, the division between domestic passenger and the international passenger is used. The domestic passenger is not necessarily the local passengers, and the international passenger does not only mean the foreigners. The indicators that proposed as explanatory variables for the domestic passenger are the GDRP, Population, and Domestic Tourist. The indicators that assumed to have a relationship with the international passengers are GDRP, population, export value, import value and international tourist. It is assumed that tourism can also be the variables that drive airport demand. Specifically for international passenger, the additional variables of export and import are to represent the magnitude activity with other countries that can attract foreigner. The mathematical formulation is expressed as: For the linear model PaxDom = β 0d + β 1d GDRP + β 2d Pop + β 3d DomTour (5) PaxInt = β 0i + β 1i GDRP + β 2i Pop + β 3i Imp + β 4i Exp + β 5i IntTour (6) For the log-log model (natural logs or ln) ln PaxDom = α 0d + α 1d ln GDRP + α 2d ln Pop + α 9d ln DomTour ln PaxInt = α 0i + α 1i ln GDRP + α 2i ln Pop + α 3i ln Exp + α 4i ln Imp + α 5i IntTour (7) (8) Where: PaxDom PaxInt GDRP Pop Exp Imp DomTour IntTour = Annual Domestic Passenger = Annual International Passenger = Gross Domestic Regional Product (in USD) = Population of the region = Export value of the region (in USD) = Import value of the region (in USD) = Number of Domestic Tourist = Number of International Tourist Model Estimation and Analysis The model parameter both in linear and log-log model structures are estimated. There are also two regression approaches: multiple linear regression and stepwise regression. In the multiple linear regression, all variables are kept in the model, but in the stepwise regression, only the significant variables in the model remained in the final model. By doing this, a model with fewer variables can be achieved with the considerably higher fit with the data. This parameter estimation process is conducted in MATLAB. Then, the estimated parameters are analyzed, and necessary interpretation is made. This result then become the input to the regulator or the airport operator itself to explain how the past demand evolves. Demand Forecasting From the combination of two model structure and two model estimation process, there are four possible models: 1. All Variables-Linear 2. All Variables-Log-log 3. Stepwise-Linear 4. Stepwise-Log-log

38 Proposed Method 22 Before doing the forecasting process, the best model should be chosen. The Mean Average Percentage Error (MAPE) is used as an indicator to select the appropriate model. Because the goal in the forecasting process is to get a model with high forecasting power, thus, a model with smaller MAPE is preferred. Besides the model, the other step is preparing the forecast input. It can be using scenarios or using time series analysis. In this study, it is assumed that the previous trend of all the variables has a relationship with the historical data. The exponential smoothing method is used to forecast the explanatory variables in the assumption that the development of the socio-economic condition in the past will sustain to the future. After the model are selected, and future input date is ready, the future demand is calculated using the selected model. This result becomes the input for further steps. Method for Sizing the Infrastructure Before jumping into the model of each infrastructure, the number of ATM in peak hour and Passenger in Peak Hour are calculated first. The annual passenger is converted to a number of Passenger in Peak Hour. The method can be used explained in 2.2. Then, the number of ATM in Peak Hour are calculated by dividing Passenger in Peak Hour with the Passenger per Aircraft Ratio. In the proposed model, the three main infrastructures are estimated: number of runways, number of stands in apron-gate complex, and size of the passenger terminal. In the following parts, the method for sizing each infrastructure is explained. Number of Runways In calculating a number of runways, an assumption is made. The additional runway will be built with enough space between the old runway and a new one. Hence, under this assumption, the ultimate capacity of the runway can be achieved. The number of runways is calculated by dividing the ATM in peak hour with the ultimate capacity of the runway. Thus, the ideal number of runway for an airport is R = N u r (9) Where, R N u r = number of runways required = number of ATM (landing and taking off) in peak hour = number of ATM (landing and taking off) can be handled in a runway or ultimate capacity (ops/hour) The ultimate capacity of a runway (u r ) is the number of landing and taking-off which can be handled by a runway. A runway is assumed served both arrival and departure flight in the same proportion. Hence, the capacity of a runway with mixed operation (landing and take-off) is Where, u r = u a + u d 2 (10) u r u a u d = number of ATM (landing and taking off) can be handled in a runway or ultimate capacity (ops/hour) = number of landing operation can be handled in a runway or ultimate arrival capacity (ops/hour) = number of taking-off operation can be handled in a runway or ultimate departure capacity of a runway (ops/hour) The ultimate arrival capacity (u a ) is:

39 Proposed Method 23 u a = T t a (11) t a = p i p j t a,ij ij (12) Where t a,ij = { d ij v j max (t ai ; d ij v j ), v i v j + γ ( 1 v j 1 v i ), v i > v j (13) u a T t a p i, p j t a,ij t ai d ij γ v i, v j = ultimate arrival capacity of a runway = the period of which the arrival capacity is estimated (1 hour) = minimum average interarrival time at the runway threshold for all combinations of arrival sequences = proportion of aircraft class i and j in the arrival traffic mix = interarrival time between leading and trailing aircraft = runway occupancy time by the arriving aircraft = minimum wake-vortex distance-based separation rule = length of final approach path (nm) = speed of aircraft along the path γ (knots) The ultimate departure capacity (u d ) is u d = T t d (14) Where u d T t a p i, p j t a,ij t ai d ij γ v i, v j t d = p i p j t d,ij ij (15) t d,ij = max (t d,ij ; t d,ij0 (t dj t di ) γ d ( 1 v jd 1 v id )) (16) = ultimate departure capacity of a runway = the period of which the arrival capacity is estimated (1 hour) = minimum average interarrival time at the runway threshold for all combinations of arrival sequences = proportion of aircraft class i and j in the arrival traffic mix = interarrival time between leading and trailing aircraft = runway occupancy time by the arriving aircraft = minimum wake-vortex distance-based separation rule = length of final approach path (nm) = speed of aircraft along the path γ (knots) Number of Apron-Gate Stands There are three aircraft types considered in which can park in the apron; small, large, and heavy. The ATM in peak hour per aircraft type is calculated by transform peak hour passenger volume using average number of passenger per aircraft (APA) and fleet mix. The gate occupancy time is the time for an aircraft to maneuver in and out of an aircraft stand, deplane and enplane passengers, baggage, cargo and various standard services. The required number of aircraft stands at the apron-gate complex is as follows:

40 Proposed Method 24 Where S T i T i N i U i = required number of aircraft stands S = = gate occupancy time in minutes of aircraft group i = parking time in minutes of aircraft group i = number of arriving aircraft group i during peak hour = utilization of the aircraft stands ( T i + P i 60 N i) U i (17) Size of Passenger Terminal In calculating passenger terminal, the guideline of Airport Authority in Indonesia (PM178/2015) is used as the basis calculation because after all, the detailed master plan will be reviewed by Indonesia s DGCA. Regulation regarding terminal size follows these rules: The terminal area per passenger during peak hours is 14 m 2 /Pax for domestic and 17 m 2 /Pax for international which does not include the area required for circulation, utilities and concession spaces Airport service providers shall provide facilities for operational services of at least 70% and facilities for commercial services of at most 30% of the total area of passenger terminals minus the circulation and utility space of 20% The passengers in peak hour are calculated based on Table 3.1. Table 3.1. Peak Hour Coefficient (PM178/2015) Source: (Indonesia Ministry of Transportation, 2015) Annual Passenger (in million Pax) Peak Hour Coefficient (% of annual pax) > % 20-29, % 10-19, % 1-9, % 0,5-0, % 0,1-0, % <0, % Based on this regulation, the formulation is: Where T P i A i ρ σ = required size of passenger terminal = number of passenger type i in peak hour T = P ia i ρ (1 σ) = terminal area needed per passenger i in peak hour = ratio of operational area and commercial area = ratio of circulation area from total terminal area (18) Calculation using methods in Ashford et.al., (2011) and Horonjeff et al. (2010) are performed as a comparison of the result.

41 Proposed Method 25 Method for Analyzing Uncertainty Developing strategic infrastructure plan is under an assumption that the future can have a similar pattern with the past. However, sometimes uncertainties play a role in shaping the future. By having noticed by the possible uncertainties in the future, the airport owner can have a better plan. Uncertainty Identification The uncertainty in estimating the required infrastructures for AP2 in the future is analyzed using Walker s framework (Walker, et al., 2003). Comprehensive analysis of all possible uncertainties needs to be discussed to warn the airport the range of uncertainties that might happen in the future and affect airport system. The listed uncertainty in (Kwakkel, Walker, & Wijnen, 2008) is used as reference point. Scenario Development The identified uncertainty is then used to improve the robustness of infrastructure recommendation. From all the uncertainties, two scenarios, high and low, are made for this input: Explanatory variables in Demand Model Peak Hour Coefficient (PHC) in Demand Model Average number of Passenger per Aircraft (APA) in Demand Model and Fleet Mix in Infrastructure Model The APA and Fleet Mix are combined since both are closely interrelated. Thus, six possible scenarios are tested and further, the final recommendation is proposed.

42 Proposed Method 26

43 Application of the Proposed Method 27 4 Application of the Proposed Method Airport Demand model Available Input Data The annual number of passengers from in AP2 airports is collected. The explanatory variables data is collected from National Statistics Bureau of Indonesia. All the explanatory variables and the units are summarized in Table 4.1. The figure of all traffic data in AP2 is also illustrated in Appendix B. Some variables are transformed to get more coherent information, such as the GDRP. At first, the GDRP data is collected in Rupiah (currency of Indonesia). This GDRP is transformed to US dollar (US$) using historical currency exchange for each respective year. It is more practical to use the international currency standard (US$) because the following up of this study is to estimate the total budget of the infrastructure and financing study which in many cases, US dollar is preferred. Table 4.1. Explanatory Variables. Source: own work No Determinant Category Explanatory Variables Variable code Unit 1 Economic Gross Domestic Regional Product Export (FOB) Import (CIF) GDRP Exp Imp Million US$ Million US$ Million US$ 2 Demographic Population size Pop Thousands 3 Tourism Number of Domestic Tourists Number of International Tourists DomTour IntTour Thousands Thousands Model structure The goal in this step is to find the general demand model for AP2. The presence of two hub airports, CGK and KNO, is expected to be the problem in the estimation process. CGK is a main hub of Indonesia and KNO is a secondary hub for western part of Indonesia, i.e., Sumatera. Two dummy variables, Hub CGK and Hub KNO, are introduced to the general demand model to handle this situation. Using explanatory variables coding in Table 4.1, the model structures become: For the linear model PaxDom = β 0d + β 1d GDRP + β 2d Pop + β 3d DomTour + β 4d Hub CGK,d + β 5d Hub KNO,d (19) PaxInt = β 0i + β 1i GDRP + β 2i Pop + β 3i Imp + β 4i Exp + β 5i IntTour + β 6i Hub CGK,i + β 7i Hub KNO,i (20) For the log-log model (natural logs or ln) ln PaxDom = α 0d + α 1d ln GDRP + α 2d ln Pop + α 9d ln DomTour + Hub CGK,d + Hub KNO,d (21) ln PaxInt = α 0i + α 1i ln GDRP + α 2i ln Pop + α 3i ln Exp + α 4i ln Imp + α 5i IntTour + α 6i Hub CGK,i + α 7i Hub KNO,i (22)

44 Application of the Proposed Method 28 Where: PaxDom PaxInt GDRP Pop Exp Imp DomTour IntTour HubCDK HubKNO = Annual Domestic Passenger at specific airport = Annual International Passenger at specific airport = Gross Domestic Regional Product = Population of the region = Export value of the region (in USD) = Import value of the region (in USD) = Number of Domestic Tourist = Number of International Tourist = Dummy variable for specific airport, CGK and KNO. It is binary value of 1 if it is the airport, otherwise 0. All parameters α and βare estimated with multiple linear regression analysis. The parameters α and β have different interpretation because of different model structures. The value ofαis the elasticity of dependent variable on the independent variable and βthe marginal effect of independent variable on the dependent variable. Model Estimation and Analysis Demand models are generated using stepwise linear regression method to find the best model with a small number of variables and high model fit. The estimation is performed in MATLAB. The general demand model of all airports and each passenger category are shown in Table 4.2 and Table 4.3. Table 4.2. Parameter estimation of General Domestic Passenger Demand model in AP2 Source: own work Linear Log-log Variable Full Stepwise Full Stepwise Intercept 413, , (41.94) * GDRP * * 0.39 Pop (0.10) * (0.10) * 3.44 DomTour (747.67) *** (747.67) *** (0.07) HubCGK 24,268, * 24,268, * 3.70 * 3.71 * HubKNO 3,681, * 3,681, * 1.86 * 1.81 * R Notes: *significant at 99%, **significant at 95%, ***significant at 90%. From the result of estimated parameters of Domestic Passenger Model in Table 4.2, some insights can be derived. First, all variables are significant to explain the demand because all of them still appear in the stepwise model. Second, the log-log model shows a lack of capability to explain the domestic passenger demand using the panel data. Third, the role of CGK and KNO as a hub airport has a huge impact on their demand as expected. The CGK and KNO have additional 24,2 million and 3,6 million domestic passengers respectively compared to the others. Fourth, surprisingly, the coefficient of the population and domestic tourist that are expected to have a positive relation, are found to be negative.

45 Application of the Proposed Method 29 Variable Table 4.3. Parameter estimation of International Passenger Demand in AP2 Source: own work Linear Log-log Full Stepwise Full Stepwise Intercept 51, , * * GDRP * * 1.69 * 1.56 * Pop (0.04) * (0.04) * (1.83) * (1.83) * Export (80.89) * (79.94) * (0.77) * (0.77) * Import * * (0.05) IntTour (0.07) HubCGK 7,896, * 7,899, * 5.70 * 5.62 * HubKNO 1,315, * 1,314, * 4.57 * 4.56 * R Notes: *significant at 99%, **significant at 95%, ***significant at 90%. The insights can also be derived from the international passenger model as shown in Table 4.3. First, the hub also has a significant contribution to the international passenger number of CGK and KNO. Second, the coefficient tells that CGK and KNO have additional 7,9 million and 1,3 million domestic passengers respectively compared to the others. Third, Import and International Tourist are found have an expected positive sign. Fourth, however, the export value is found to be negative. Based on this panel data result, the models confirm the relationship of airport demand to economic activity, population, and tourism. Demand Forecasting In the next step, we would like to estimate the future demand of each airport. Thus, a specific model for each airport should be developed. First, the best model structure which has high predicting power should be selected. The criteria used in the selection is using Mean Average Percentage Error (MAPE). Table 4.4. MAPE of the models. Source: own work Stepwise Full Airport Code Pax Dom Line Log-Log Line Log-Log Pax Int Total Pax Pax Dom Pax Int Total Pax Pax Dom Pax Int Total Pax Pax Dom Pax Int Total Pax CGK 10% 5% 8% 18% 10% 11% 9% 4% 7% 16% 11% 10% KNO 8% 6% 7% 16% 10% 10% 8% 5% 7% 28% 16% 18% HLP 90% 138% 88% 74% 70% 72% 92% 147% 88% 60% 194% 58% PLM 15% 60% 15% 27% 190% 27% 18% 45% 18% 25% 374% 32% PNK 15% 28% 15% 19% 30% 18% 15% 23% 15% 18% 26% 18% PDG 12% 32% 11% 19% 56% 19% 14% 22% 14% 20% 68% 22% PKU 10% 28% 10% 23% 59% 24% 12% 24% 12% 21% 99% 23% BDO 40% 7998% 43% 30% 6730% 20% 25% 3618% 23% 40% 2507% 34% BTJ 19% 53% 17% 51% 38% 45% 16% 36% 15% 58% 41% 52% TNJ 76% 76% 66% 66% 21% 21% 26% 26% PGK 4% 4% 5% 5% 5% 5% 4% 4%

46 Application of the Proposed Method 30 DJB 7% 7% 6% 6% 5% 5% 5% 5% DTB 27% 27% 26% 26% 18% 18% 171% 171% Average 26% 927% 25% 29% 799% 27% 20% 436% 19% 38% 371% 36% From the MAPE result in Table 4.4, the linear model with full variables has lowest average MAPE. Since the goal of forecasting is to have the model with the smallest error, the linear model with full variables is used in the next step. The selected demand model of each airport and passenger type is presented in Table 4.5 and Table 4.6. Airport CGK KNO HLP PLM PNK PDG PKU BDO BTJ TNJ PGK DJB DTB Table 4.5. Domestic Passenger Model of AP2 Airports Source: own work Domestic Passenger Demand Model PaxDom = GDRP Pop DomTour (0.01) (0.25) (0.01) (0.28) R 2 = PaxDom = GDRP Pop DomTour (0.02) (0.00) (0.02) (0.64) R 2 = PaxDom = GDRP Pop DomTour (0.88) (0.05) (0.75) (0.00) R 2 = PaxDom = GDRP Pop DomTour (0.28) (0.27) (0.25) (0.18) R 2 = PaxDom = GDRP Pop DomTour (0.90) (0.00) (0.80) (0.12) R 2 = PaxDom = GDRP Pop DomTour (0.01) (0.24) (0.01) (0.34) R 2 = PaxDom = GDRP Pop DomTour (0.06) (0.26) (0.03) (0.44) R 2 = PaxDom = GDRP Pop DomTour (0.95) (0.69) (0.88) (0.00) R 2 = PaxDom = GDRP Pop DomTour (0.22) (0.00) (0.21) (0.01) R 2 = PaxDom = GDRP Pop DomTour (0.85) (0.04) (0.98) (0.43) R 2 = PaxDom = GDRP Pop DomTour (0.12) (0.44) (0.09) (0.95) R 2 = PaxDom = GDRP Pop DomTour (0.02) (0.18) (0.01) (0.08) R 2 = PaxDom = GDRP Pop DomTour (NaN) (NaN) (NaN) (NaN) R 2 = 1 1 Notes: since DTB is open from 2013, the data is not sufficient to produce statistical analysis

47 Application of the Proposed Method 31 Airport CGK KNO HLP PLM PNK PDG PKU BDO BTJ TNJ Table 4.6. International Passenger Model of AP2 Airports Source: own work International Passenger Demand Model PaxInt = GDP Pop Exp Imp IntTour (0.81) (0.21) (0.73) (0.93) (0.32) (0.10) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.78) (0.03) (0.60) (0.86) (0.46) (0.01) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.43) (0.99) (0.51) (0.84) (0.64) (0.07) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.28) (0.62) (0.26) (0.36) (0.69) (0.52) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.24) (0.11) (0.27) (0.93) (0.75) (0.72) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.06) (0.94) (0.07) (0.60) (0.16) (0.33) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.14) (0.48) (0.05) (0.30) (0.13) (0.24) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.94) (0.30) (0.96) (0.89) (0.14) (0.06) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.19) (0.09) (0.10) (0.49) (0.05) (0.56) R 2 = PaxInt = GDP Pop Exp Imp IntTour (0.24) (0.01) (0.95) (0.08) (0.02) (0.18) R 2 = Notes: There is no international flight in PGK, DJB, and DTB Future Airport Demand The successfully obtained models are used to forecast the future airport demand. The growth of the input variables is calculated using exponential smoothing method that embedded in Ms. Excel as we assume that all input has a trend. Next, these future inputs are employed in the model to calculate the future demand for each, domestic and international passenger. Then, the total number of passengers handled by each airport is summation between domestic and international. The total demand, total growth and cumulative annual growth rate (CAGR) AP2 Airports are shown in Table 4.7. The CAGR of the passenger is ranging between -0.24% (HLP) and 18.53% (DTB). In cumulative, AP2 is expected to handle a total of 174,757,647 passengers in 2030 with CAGR of 5.08%. The air traffic movements (ATM) have been projected based on the ratios of passengers to a number of aircraft movement in The result is presented in Table 4.8 below. The CAGR of ATM is ranging between -0.2% (HLP) and 11.16% (BDO). In cumulative, AP2 is expected to handle a total of 1,346,779 ATM in 2030 with CAGR of 5.1%.

48 Application of the Proposed Method 32 Airport Table 4.7. Total Passenger in 2015 and 2030 Source: own work Total Pax (2015) (1) Total Pax (2030) (2) Cumulative Growth CAGR (3) CGK 52,306, ,230, % 5.03% KNO 8,004,791 16,050, % 4.75% HLP 3,059,153 2,949,258 96% -0.24% PLM 4,237,834 6,708, % 3.11% PNK 2,713,088 4,866, % 3.97% PDG 3,162,245 5,882, % 4.23% PKU 2,670,046 5,621, % 5.09% BDO 3,146,807 15,371, % 11.15% BTJ 747,495 1,291, % 3.71% TNJ 258, , % 5.62% PGK 1,658,393 3,575, % 5.26% DJB 1,168,219 2,394, % 4.90% DTB 17, , % 18.53% AP2 83,151, ,757, % 5.08% Notes: 1) Actual number of Passenger in ) Forecasted number of Passenger in ) Cumulative Annual Growth Rate Table 4.8. Total ATM in 2015 and 2030 Source: own work Airport Total ATM (2015) (1) Total ATM (2030) (2) Cumulative Growth CAGR (3) CGK 386, , % 5.10% KNO 63, , % 4.76% HLP 30,235 29,344 97% -0.20% PLM 33,164 52, % 3.15% PNK 25,183 44, % 3.93% PDG 21,757 40, % 4.23% PKU 19,206 40, % 5.04% BDO 25, , % 11.16% BTJ 6,047 10, % 3.76% TNJ 2,542 5, % 5.64% PGK 13,985 30, % 5.26% DJB 9,486 19, % 4.90% DTB 953 3, % 9.93% AP2 638,624 1,346, % 5.10% Notes: 1) Actual number of ATM in ) Forecasted number of ATM in ) Cumulative Annual Growth Rate

49 Application of the Proposed Method 33 Sizing the Infrastructure As proposed in the method, three important infrastructures are calculated and evaluated: Runway, Apron- Gate Complex, and Passenger Terminal. The design parameter that is estimated in runway system is the ideal number of runways. In the Apron-Gate Complex, the design parameters that are estimated is the number of aircraft parking stands. In the passenger terminal, the design parameter that is estimated is the overall size of passenger terminal. Input Data The capacity data is represented by the existing infrastructures (2015) and the projected infrastructure that had been already planned in the master plan. The important variables considered in the project are the number of runways, the number aircraft parking stands in apron-gate complex, and the size of the passenger terminal. Peak Hour Passenger, Operational Procedure of arrival and departure in the runway, Turn Around Time and Fleet Mix of the airport are using data of AP2 in Data requirement for infrastructure calculation is presented in Table 4.9. These data are the input needed by the model to calculate the infrastructure requirement. Table 4.9. Input for Facility model Source: own work No Facilities Infrastructure estimated Variables 1 Runway System Number of runways Fleet mix Separation rules Final approach speed Runway occupancy time Length of final approach 2 Apron-Gate Complex Number of aircraft parking stands Average Turn Around Time Fleet Mix ATM in Peak Hour 3 Passenger Terminal Size of Passenger Terminal Passenger in Peak Hour Passenger Terminal Standard The data of available infrastructure in 2015 is collected from the internal data, and the master plans for each airport is collected from the website of Directorate General of Civil Aviation, Ministry of Transportation Republic of Indonesia (link: Two problems are distinguished by comparing the estimated infrastructure and the master plan. First, if the infrastructure required is higher than the existing infrastructure, it means AP2 should build it. Further, it is indicated by yellow mark. Second, if the infrastructure required is higher than the maximum size or number than planned infrastructure in the master plans, it means AP2 should make a revision to the existing master plan. This problem is indicated by red mark. Runway System The first infrastructure to be calculated is the number of the runway in airport system. Under an assumption that the number of ATM will grow proportionally with passenger number, the fleet mix and operational procedure in 2015 used in the calculation. The fleet mix in each airport is presented in Appendix C. Most of the aircraft landed in AP2 airports are large aircraft such as B737 and A320. This happens since most of the domestic airlines use this type of aircraft in their operation. Higher small-aircraft proportion presents in smaller airports such as PGK and DTB because usually, the demand is very low. Thus, many airlines utilize small turboprop aircraft. Arrival and departure procedure in AP2 airports are the same for all airports since the operation is handled by Indonesian Air Navigation (AirNav). The operational data regarding landing and taking-off operation can be seen in Appendix C.

50 Application of the Proposed Method 34 Using the input, we calculated the number of runways required for each airport in AP2 (see Table 4.10). All airports, except CGK and BDO, do not need an additional runway in An additional one runway is needed for CGK and BDO by The extra runway for CGK is already mentioned in the current master plan but not in BDO. Therefore, AP2 need to check whether adding new runway in BDO is feasible. Table Required Number of Runways Source: own work Airport Existing (2015) Master Plan CGK KNO HLP PLM PNK PDG PKU BDO BTJ TNJ PGK DJB DTB Notes: yellow color means the recommendation is already mentioned in the master plan, red color means it is not in the master plan yet. Apron-Gate Complex Next, the required number of gate and or parking stands at each airport are calculated using Turn Around Time (TAT) or gate occupancy time of each aircraft type. It is assumed that the utilization of the stands is 70% and extra 10 minutes is added as the taxing and parking time or the time to dock in and dock off from the gate. The average TAT and Fleet Mix of the airports in AP2 is shown in Appendix C. The assumption that there is no regular RON aircraft using stands in the airport. The calculation of required parking stands for each airport in 2030 is presented in Table Most of the airports have enough number of parking stands for serving demand in Since there is an indication that the number of parking stands in 2015 is over the required number of parking stands, it may be useful in the further study to also take into account number of RON in the estimation. From the result, the airports that need additional parking stands are BDO and PDG. Since additional parking stands are already listed in the master plan of PDG, AP2 only needs to update master plan of BDO. Table Required parking stands in Apron-Gate Complex Source: own work Airport Existing (2015) Master Plan CGK KNO HLP PLM PNK

51 Application of the Proposed Method 35 PDG PKU BDO BTJ TNJ PGK DJB DTB Notes: yellow color means the recommendation is already mentioned in the master plan, red color means it is not in the master plan yet. Passenger Terminal Since AP2 is an airport operator in Indonesia, a guideline from Indonesian Government which is Ministerial Decree No. 178 of 2015 on Service Standards of Airport Service Users or (PM178/2015), is used. Approaches from (Whitford, 2003) and (Horonjeff, McKelvey, Sproule, & Young, 2010) are used in order to compare the difference between Indonesian standard with other approaches. Table 4.12 shows the terminal size requirement in The result indicates that some of the passenger terminals need to be expanded. Table Required Passenger Terminal Size in 2015 in m 2 Source: own work Airport Existing (2015) Master Plan PM178/2015 Whitford (2003) Horronjeff (2010) CGK 334, , , , ,124 KNO 118, , ,992 71,984 96,374 HLP 21,108 64,019 39,095 26,007 37,799 PLM 34,000 43,572 54,074 34,557 50,455 PNK 32,000 32,000 34,625 23,161 33,796 PDG 20,568 49,950 40,750 27,963 39,757 PKU 16,700 88,221 34,326 23,739 33,955 BDO 5,000 17,000 41,701 31,155 41,645 BTJ 14,742 15,075 15,757 8,367 11,317 TNJ 8,210 8,210 8,533 2,679 3,986 PGK 12,170 37,916 21,041 14,261 21,222 DJB 13,000 35,000 14,822 10,404 15,483 DTB 1,700 10, Notes: Yellow color means the recommendation is already mentioned in the master plan, Red color means it is not in the master plan yet. Before going through the analysis, a validation check was made to see if the result shows a plausible with the capacity problem stated in Chapter The comparison of occupancy level of the airports using stated capacity and calculated capacity using PM178/2015 is provided in Table 4.13 below. An interesting insight can be captured from this information.

52 Application of the Proposed Method 36 In the problem in the description, the occupancy of CGK, the most severe airport in AP2, is 2.5 times than its designed capacity. However, from the theoretical calculation using guideline from PM178/2015, it has only surpassed 1.5 times than its designed capacity. It makes more sense because even though the delay sometimes happened in CGK, they were still manageable. It is also confirmed by the corporate planning manager of AP2. She said that it is very suspicious that passenger terminals in AP2 look like so terrible in number, but actually, they can still manage it in daily activity. However, for the other airports, the passenger terminal capacity from the existing infrastructure is way more overstated. It looks that most of the airports suffer facility shortage. From the perspective of the regulator, there is no problem about overdesign of infrastructure because it makes the airports have a better level of service. However, in a business perspective, unnecessary expansion (e.g., building infrastructure) should be avoided. No Table Occupancy level comparison Source: own work IATA Code Stated Occupancy Level 1 Hypothetical Occupancy Level 2 Difference 1 CGK 247% 146% +110% 2 KNO 89% 89% 0 3 HLP 161% 185% -24% 4 PLM 113% 159% -46% 5 PNK 113% 108% +5% 6 PDG 106% 198% -93% 7 PKU 76% 206% -130% 8 BDO 131% 834% -703% 9 BTJ 75% 107% -32% 10 TNJ 22% 104% -82% 11 PGK 332% 173% +159% 12 DJB 78% 114% -35% 13 DTB 18% 34% -16% Notes: 1. The passenger traffic of 2015 divided by stated capacity of AP2. 2. The passenger traffic of 2015 divided by hypothetical capacity from PM178/2015 Since the goal of this method is initially to find the general method to assess facility requirement, it is better to locate the problem of this estimation by looking more detail in each step of calculating the terminal size. There are two possible sources of problems. The first is in the process of translating the annual passenger into the peak hour passenger. This relation can be analyzed empirically such as in Jones & Pitfield (2007). Therefore, the adjustment of Peak Hour Coefficient using empirical data is necessary to be conducted. The second is the incorrect overall passenger terminal size standard. Most of the literature that evaluate the standard size of gross terminal size focus in the microscopic analysis of specific terminal area such as in (Solak, Clarke, & Johnson, 2009). Thus, the standard gross terminal size used is not analyzed further. After collecting Peak Hour Passenger data of each airport and compared them to the Peak Hour Passenger estimation, it turned out that the Peak Hour Coefficient in PM178/2015 and FAA guideline, are not accurate. All the conversions are off-point and overestimate (see Table 4.14). The result is in line with the conclusion of a study by (Jones & Pitfield, 2007) which stated that the US FAA s TPHP method consistently over-designs passenger terminal area. Thus, using the observed result of the current relation of peak hour and annual passenger of the airports in AP2 seems more logic and reasonable.

53 Passenger Hour Coefficient (%) Application of the Proposed Method 37 Table Peak Hour Coefficient of Empirical Data and PM178/2015 Source: own work No Notes: IATA Code Empirical data 1 PM178/2015 & FAA in Horrenjeff (2010) 2 Ashford (2011) 3 1 CGK 0.021% 0.035% 0.036% 2 KNO 0.028% 0.050% 0.043% 3 HLP 0.027% 0.050% 0.045% 4 PLM 0.031% 0.050% 0.047% 5 PNK 0.027% 0.050% 0.048% 6 PDG 0.031% 0.050% 0.047% 7 PKU 0.033% 0.050% 0.048% 8 BDO 0.032% 0.050% 0.048% 9 BTJ 0.046% 0.080% 0.054% 10 TNJ 0.110% 0.130% 0.138% 11 PGK 0.035% 0.050% 0.050% 12 DJB 0.045% 0.050% 0.051% 13 DTB 0.103% 0.130% 0.146% 1. From the collected data. 2. From Table Using formula in Chapter 2.2 The PHC relationship comparison between FAA&PM178, Whitford (2003) and empirical data in AP2 can be seen in Figure 4.1. The power function is the best fit with the data with R 2 of The relationship between Peak Hour Coefficient and Annual Passenger is PHC = Pax (23) 0,250% 0,200% 0,150% 0,100% FAA & PM178 y = x R² = Annual Passenger vs PHC AP2 y = x R² = Whitford y = 0.004x -0.1 R² = 1 FAA & PM178 AP2 Whitford Power (FAA & PM178) Power (AP2) Power (Whitford) 0,050% 0,000% Annual Passenger (Pax) Figure 4.1. Relationship between Annual Passengers and Peak Hour Coefficient Source: own work After calibrating the relationship between Annual Passenger and Peak Hour Passenger with empirical data found, the terminal size is then calculated. The required passenger terminal size of airports in 2030 is shown in Table The result indicates that all the airports in AP2 still need to be expanded. Some of them have already mentioned in the master plan (yellow color), but some of them are still not planned yet (red color). When it is already referred to in the master plan, the recommendation is to execute the plan

54 Application of the Proposed Method 38 according to the estimated size. However, if the result shows a red color, AP2 should start planning the next master plan and proposed it to the MoT. From this result, the PM178/2015 method with modification needs much smaller passenger terminal compare to others, especially for CGK and KNO which have an enormous annual traffic. This result is good news for AP2 since smaller passenger terminal means less capital investment should be invested. Table Passenger terminal size requirement in 2030 (base scenario) Source: own work Airport Existing (2015) Master Plan Adjusted PM178/2015 PM178/2015 Whitford (2011) Horronjeff (2010) CGK 334, , ,951 1,103, ,349 1,011,467 KNO 118, , , , , ,399 HLP 21,108 64,019 26,938 37,637 25,611 36,892 PLM 34,000 43,572 55,109 82,621 52,387 76,387 PNK 32,000 32,000 45,876 64,327 38,890 56,960 PDG 20,568 49,950 53,616 73,175 49,601 69,983 PKU 16,700 88,221 53,887 72,086 47,112 66,846 BDO 5,000 17, , , , ,416 BTJ 14,742 15,075 18,920 29,135 13,967 18,677 TNJ 8,210 8,210 11,500 22,849 5,681 8,406 PGK 12,170 37,916 39,119 36,509 28,474 42,372 DJB 13,000 35,000 29,227 24,654 19,851 29,540 DTB 1,700 10,000 10,246 3,889 2,385 3,550 Notes: Passenger terminal size is in m 2 Yellow color means the recommendation is already mentioned in the master plan, Red color means it is not in the master plan yet.

55 Application of the Proposed Method 39 Summary of infrastructure needs From the calculation and analysis in the previous sub-chapter, the base infrastructure recommendation to be constructed by AP2 is shown in Table All airports need infrastructure expansion by 2030 except KNO. A total of 2 runways, 17 parking stands, and 411,980 m 2 of passenger terminal should be built. Airport Table Base infrastructure recommendation for AP2 Source: own work Additional runway 1 Additional parking stands 2 Additional passenger terminal size 3 CGK 1 105,892 KNO HLP 5,830 PLM 21,109 PNK 13,876 PDG 2 33,048 PKU 37,187 BDO ,848 BTJ 4,178 TNJ 3,290 PGK 26,949 DJB 16,227 DTB 8,546 Total ,980 Notes: 1) The additional runway is in parallel and have enough distance to achieve maximum capacity 2) The parking stands can be in gate or remote 3) The passenger terminal size is in m 2 In the following parts, the uncertainty is analyzed and included to improve the recommendation. Uncertainty Analysis The demand of AP2 experienced a promising growth for the past years. In previous sub-chapters, the model has been developed for such purpose. The required facilities have been calculated for the year of In this part, the uncertainty in AP2 is explored and analyzed. Uncertainty Identification There is no statement related to uncertainty being addressed or discussed in the master plan of AP2 airports. Thus, in this section, the uncertainty analysis presented in (Kwakkel, Walker, & Wijnen, 2008) is taken as the point of departure in identifying the uncertainties in AP2. Several adjustments to the list are conducted due to two reasons: 1. The scale of operation: AP2 is a system of airports while Schiphol Airport is a single airport, and 2. The location: AP2 manages the airports which are located in a developing country while Schiphol Airport is an airport in a developed country. The adjustment of the considered uncertainties is derived from the understanding of how AP2 manages and operates the airports and the dynamic situation in Indonesia. The modification of uncertainty from reference study is explained in Table 4.17 below. The characteristic and description of each uncertainty are explained in Table The uncertainty is characterized using the uncertainties framework (Walker, et al., 2003).

56 Application of the Proposed Method 40 Table Uncertainty modification to fit the condition of AP2. Source: own work. Sources Uncertainty in AP2 Status Explanation AP2 itself Strategy for each airport Added Management of AP2 can assign a strategic role for a specific airport, e.g., to be a hub. This type of decision can create uncertainty to other airports. Airport network Added The number of airports in Indonesia in general and AP2 in specific can still grow. Government keen to maximize the benefit of having airport infrastructure across the country. Airlines Number of airlines Added Route development Added Demand Driving Forces Fleet mix As-is - GDRP Adjusted Trade Adjusted Population Adjusted Tourism Adjusted Model Demand relationship Adjusted Regulations System operations As-is - Technology Aviation Technology As-is - ATM Technology As-is - Policy National Added Local Added Others Evaluation of Outcomes Absolute levels of impacts Rules for calculation of impacts Land use developments Ignored Ignored Ignored Ignored The location of most airports in AP2 is in suburb. The land use problem will not happen in the near future. Landside accessibility Ignored The urgency in using air transport in Indonesia is quite irreplaceable as many cities are separated by the sea. Thus, landside accessibility would not bother the airport demand. Weather pattern Ignored The weather in Indonesia is quite stable over time. Thus, this type of uncertainty has an insignificant impact on the operation of AP2. Modal choice Ignored This type of uncertainty is not yet important in Indonesia because, first, the cities are located far apart and second, other transport modes are in general underdeveloped Notes: As-is : the same as the reference study Adjusted : the term was changed but the general idea remains same Added : the uncertainty is appropriate to the case study but not listed in the reference study Ignored : the uncertainty is not relevant to the case study

57 Sources AP2 itself Airlines Demand Driving Forces Uncertaintie s Strategy for each airport Airport network Number of airlines Route development Fleet mix GDRP Trade Population Table List of Uncertainties in AP2 Source: own work Location 1 Level 2 Nature 3 Description Controllable input Controllable input uncontrollable input uncontrollable input uncontrollable input uncontrollable input uncontrollable input uncontrollable input Scenario Recognized Ignorance Scenario Scenario Scenario Scenario Scenario Scenario Epistemic Epistemic Epistemic Epistemic Epistemic Epistemic Epistemic Epistemic In the future, the company can change the strategic position of several airports in the network. Currently, CGK is appointed to be a primary hub airport, and KNO is a secondary hub airport. The result of demand model shows that these two airports get benefits due to this strategic position. In the future, the management of AP2 may introduce more hub or change the hub location within the network. The number of airports currently handled by AP2 is 13. The latest one is DTB which is still 2 years in operation under the management of AP2. The introduction of new airports may influence the performance of AP2, but the relation is not clear whether it will benefit the system or not. The existence of new airports may not only generate new passengers but also reduce the passenger from the other airports. As the number of airports increases, indicating more options for passengers which increase probability of competition between airport which do not present or cannot be observed in current situation. The number of airlines in Indonesia is dynamically changed in the past decade in the sense that several companies are collapsed (e.g., Merpati Air, Adam Air, and Batavia Air) but, on the other hand, there are several new airlines popping up in the market (e.g., Batik Air and Citilink). Since the domestic passenger is dominating the traffic in AP2, this dynamic situation creates uncertainties as well. Airlines still try to optimize their revenue by exploring the new possible routes. No one can guarantee which route will still remain, which route will be disappeared, and which route will be opened. Hence, this dynamic situation can disrupt the demand in AP2 which the result can be positive or negative. The airlines can deliberate choose on what kind of aircraft used in their operation. Previously, the full carriers tend to change their narrow-body aircraft to wide body aircraft. However, the entry of Low-Cost Carriers which use only narrow-body aircraft forces the full-service airlines to keep using the narrow body aircraft to maintain their competitiveness. The fleet mix in an airport influence the capacity of the airside infrastructure. Thus, the fleet mix of airlines create capacity uncertainty for the airports. GDP growth of Indonesia as a country is predicted to be sustained. However, as an indicator of economic activity, the future of GDP can deviate in the future. Also, whether the growth will sustain for all regions is uncertain. The trade is an effect of imbalance sources between two locations. Export and import is a form of trade that involved with other countries. Along with the trade that occurred, people movements through air transport are also expected. Indonesia is expected to have a demographic bonus in Economists predict this phenomenon will result in positive impact of economic growth and many of productive population belong to the middle- Development of

58 Application of the Proposed Method 42 Model Regulation Technology Policy Tourism Demand relationship System operations Implementati on ncertainty Aviation Technology ATM Technology National Local uncontrollable input Model System Uncontrollable input uncontrollable input uncontrollable input uncontrollable input uncontrollable input Scenario Recognized ignorance Recognized ignorance Recognized ignorance Scenario Scenario Recognized ignorance Recognized ignorance Epistemic Epistemic Epistemic Epistemic Epistemic Epistemic Epistemic Epistemic Notes: 1) Location: controllable input, uncontrollable input, model structure uncertainty, model technical uncertainty and parameter uncertainty 2) Level: determinism, statistical uncertainty, scenario uncertainty, recognized ignorance, total ignorance 3) Nature: can be studied further (Epistemic) or hard to be studied (natural variability) class population. This info can be a good sign for any industry but the magnitude impact of this phenomena especially to aviation industry is still uncertain. The tourism sector is now on the focus of national government. They want to boost the tourism industry and attract more tourist, both local and international. This is the good news for the airport as the gateway of most of the tourist. The relationship of passenger demand and the explanatory variables may evolve over time. Since there is no general model suitable for all airport, the model itself may change in the future. The modeled system is designed as the knowledge in the current situation. However, the procedure in the future airport is not necessarily the same as now. The changes in procedures may impact on the requirement of the infrastructure. Also, the current stakeholders in Indonesia do not concern much about the side effect of airport operation. This condition may change in the future when people have realized or felt the negative impacts. This uncertainty is common in big projects and even worst in developing country. Especially in this case, AP2 should allocate their expense to 13 airports. This will be difficult to make sure the implementation of infrastructure project is on schedule. The aircraft technology is continuous to grow. Regarding the weight, it may be lighter or heavier, due to the material used or the size of the aircraft. It may also make the runway system change, either it needs a longer runway, or even no runway at all if vertical take-off and landing are applied to the commercial jet. Technology advancement in ATM has increased the capacity of ATC in managing airspace. It can extend the capacity in handling aircraft take-off and landing process and result in higher runway capacity. The political situation in Indonesia is indirectly playing a role in determining the future of AP2. The last airport owned by AP2 is DTB, an example of an airport that is an order of the national government through Ministry of Transportation to support tourism in the Toba Lake area. Perhaps in coming years, more airports will be managed by AP2? Decentralization in local government creates an opportunity for each region to maximize their economic potential to their fullest. However, the succession of a leader sometimes changes the ongoing policy and create instability to the economic development.

59 From this analysis, we found that most uncertainties experienced by Schiphol Airport are relevant to AP2. It indicates that benchmark on the uncertainty condition in other airports can help decision makers to understand the unforeseen future. Several uncertainties that are expected to be experienced by other airport operators are technological changes and fleet mix. However, several uncertainties in AP2 are different from SA because of the differences in characteristics between both systems. 1. The scale of operation: Single airport or system of airports? Different from SA which manages only Schiphol Airport, AP2 with 13 airports have more uncertainty in managing each airport within the network due to many reasons e.g., political and economic situation in each region. Also, the number of airports itself can grow as Indonesia still seek to capture the benefit of airports to the economic development across the country. The strategy and the number of airports in the system also result in resources distribution problem. Managing many airports within the large market (half nation) in AP2 creates unique uncertainties which will not be the case in SA. 2. The location: in developed or developing country? In developing country, especially in Indonesia, the use of air transport is favored because the integration of ground transportation and sea transportation is still underdeveloped. Thus, the uncertainty about mode choice is inessential. Several uncertainties are also less or not relevant yet for AP2, for example, the evaluation of outcome, at least in the planning horizon (2030). The environmental impact of airport operation has not been in the focus of the government in developing country since ensuring economic development is most crucial From the uncertainty analysis for both airport operators, it can be concluded that in general, the uncertainties influence the demand as well as the capacity. The demand uncertainty is a result of many (uncertainty) factors such as the economic development. The capacity is also uncertain as it is strongly influenced by the operation performed on the infrastructure. New technology or new procedure of the airport operation may extend the capacity of airport infrastructure without expansion. However, different characteristics between airport operators produce different uncertainties. From the case of SA and AP2, such characteristics include the scale of operation and the location of the airport operator. From this point, it can be a good direction to explore all possibilities of uncertainties in Airport System and if possible to model all possible uncertainties within this industry. Moreover, in this analysis, the impact magnitude of the uncertainty is not assessed further. A more comprehensive study on the various uncertainty which has materialized in the past can be another room for the research improvement. Scenario Development The list of uncertainties identified in previous subchapter is used to construct some plausible future scenarios as input for AP2 in managing their business. The future scenario developed refers to the year 2030 which is the same as the time horizon used in forecasting. Many plausible futures can be synthesized from the list of uncertainties. However, in this part, we only derive the scenarios that can be tested in our developed model. Two conditions are considered for each uncertainty: high scenario and low scenario. Several uncertain conditions that can be accommodated in the model is developed. 1. The demand driving factors. The Economic and social condition shows different situation than the prediction. The average growth for GDRP, population, Export, Import, Number of International Tourist and number of domestic tourist is 5.2%, 1.5%, 6.9%, 7.5%, 5.5%, and 5.6% respectively. In the high scenario, the economic condition in all regions in Indonesia is much better than expected. Population grows higher than the prediction. By considering the stable growth in the previous years, all the explanatory variables are set to be grown +1% than the base scenario. In the low scenario, the macro condition of Indonesia is lower than expected. All indicator is set to grow -1% than the base scenario. Development of

60 Application of the Proposed Method Peak Hour Coefficient. The result in Figure 4.1 shows that PHC in AP2 is much lower than the reference guidelines. Hence, the PHC might also change in the future. The average PHC in AP2 is %. In the high scenario, the traffic in peak hour is reduced due to the more distributed traffic. In the high scenario, reduction of 0.05% is applied to the PHC. In the low scenario, the traffic is more concentrated to the peak hour and increase 0.01% than the normal PHC. 3. Airline s decisions. In the high scenario, the airline can optimize their fleet and replace the aircraft to the higher seat-capacity. So, the average number of passenger per aircraft (APA) is increased by 20% and the share of Large Aircraft increase by 10%. On the other hand, in the low scenario, airlines replace the higher seat-capacity aircraft to the smaller aircraft. This initiative decreases the APA by 10% and share of Large Aircraft decrease by 10%. Result and Discussion After simulating the model for all the scenarios, the result is summarized as shown in Table The infrastructure recommendation deviates significantly from the base scenario except the number of runways. Scenario Table 4.19 Total infrastructure required in 2030 Source: own work Total Number of Runway Total Number of Parking Stands Existing ,187 Base ,055,056 1-L ,957 1-H ,365,729 2-L ,513,819 2-H ,674 3-L ,055,056 3-H ,055,056 Notes: 1 Demand; 2 PHC; 3 ALF; Fleet Mix; H High; L Low Total Size of Passenger Terminal For the number of runways, AP2 should at least invest a new runway (scenario 2-L or 1-L). The base case recommends AP2 to invest two additional runways in their system. However, AP2 should be careful if scenario 2-L happens since it means AP2 should invest five additional runways before The result of parking stands number in Apron-Gate Complex is a bit mixed. In the base case, current infrastructure is still sufficient to handle the demand. Serious attention needs to be given if the condition as stated in scenario 1-H or 2-L becomes apparent. The scenario 1-H is a condition when the demand grows higher than the base scenario, and the scenario 2-L is the condition when the Peak Hour Coefficient in 2030 is higher than the base case. Deficiency of parking stands will occur if one out of the two conditions happens. For passenger terminal, there is a clear sign that AP2 should expand their terminal. Even the lowest value which is in scenario 1-L (low demand scenario), AP2 still need 789,957 m 2 compared to the existing situation with 632,187 m 2. The highest estimation is in the scenario 1-H in which AP2 should double the size of the current passenger terminal. Knowing that the uncertainty can significantly affect the infrastructure estimation, we should consider giving a range of plausible outcome rather than a single number. In the further stage, the decision maker can decide how much the infrastructure will be built. AP2 can also consider making a flexible design in the master plan. The recommendation from this model can be a starting point or input to examine the current master plan available.

61 Application of the Proposed Method 45 The minimum value and maximum value from all scenarios become the lower and upper limit of the recommendation. The final recommendation of additional infrastructure in AP2 is presented in Table This table shows that AP2 should give the attention to CGK, KNO, PDG, BDO, and DJB as they might need improvements on both airside and landside. While the others only need an expansion in the landside infrastructure. Table Final infrastructure recommendation for AP2 by 2030 Source: own work Airport Additional runway 1 Additional parking stands 2 Additional passenger terminal size 3 CGK ,939 KNO ,858 HLP - - 2,865 14,209 PLM ,016 39,944 PNK - - 8,846 27,661 PDG ,468 55,591 PKU ,830 53,306 BDO , ,020 BTJ - - 2,948 8,118 TNJ - - 1,975 5,441 PGK ,212 39,405 DJB ,173 31,854 DTB - - 7,406 10,273 Total , ,619 Notes: 1) The additional runway is in parallel and have enough distance to achieve maximum capacity 2) The parking stands can be in gate or remote 3) The passenger terminal size is in m 2 This range of recommendation shows that the best value is hard to determine when uncertainty analysis is added in estimating airport infrastructure. Therefore, the decision makers can use this information to make a detailed plan, but they should also include flexibility in the design within the master plan. For example, the development can take place in huge area but conducted in several modular infrastructures (gradual construction). Thus, in the execution, before the next module is constructed, the decision makers can reevaluate current demand situation and budget availability to decide whether the construction project can proceed at that year or not.

62

63 Conclusions and Recommendations 47 5 Conclusions and Recommendations This chapter concludes the study performed. The first section provides conclusions based on the research question formulated before. The second section discusses the policy implication and the practical recommendation for the case study. The third section presents the contribution of this thesis. The last section explains the limitation and future research. Conclusions The main research question formulated in the first chapter is: What is the convenient method for an airport system owner in developing countries to assess the infrastructure requirement of the airports for balancing the future demand considering the uncertainties in the future? This research question was reformulated into several sub-questions. By answering the sub-questions, the main research question can be concluded to be answered. The sub-questions are answered in the following list. 1. What is the best method to explain the development of demand and forecast the future demand of airports? 1.1. What are the determinants of airport demand in an airport system? The literature review in Chapter 2.1 showed that economic activity, demographic structure, and tourism are the main drivers of airport demand. Several variables or indicators that represent those factors are GDP, Population, and number of tourists Specific for AP2, the examined indicators are GDP, Population, Import, Export, number of domestic tourist and number of international tourist. The detail formulas can be seen in What is the best model to forecast airport demand in an airport system? The econometric model structures that were used in the literature are a linear model and log-log model. The explanatory variables can be used entirely or only the significance one. Thus, there are four possible models that can be estimated to be the demand model which are presented in Chapter In the case of AP2, a linear model structure using entire variables is the most appropriate for forecast the future demand based on MAPE result. The comparison of MAPE for the case study is presented in How to estimate the required infrastructure to balance the future demand? 2.1. How to estimate the required infrastructure to serve the demand? The analytical method which uses limited input is employed in this study. Despite many simplifications in this approach, it is found that this approach is still relevant to the case study. As already presented in Chapter 0, the method consists of two main models: the demand model and the infrastructure model. The goal of demand model is to find the future annual passenger traffic. This passenger traffic is translated to peak hour passenger using Peak Hour Coefficient. This passenger in peak hour is then used to estimate the infrastructure What are the infrastructures to be expanded in the network? The infrastructure considered are the number of runways in the runway system, the aircraft parking stands in the apron-gate complex and the total size of the passenger terminal. The method to estimate the value is presented in 3.3.

64 Conclusions and Recommendations 48 In chapter 4.2, the required facility for balancing future demand for the case study are estimated. Among the other infrastructure, passenger terminal is the crucial one since most the airports need this kind of expansion before What are the uncertainties that need to be considered in developing the airport system and how these uncertainties can be included in the decision-making process? A list of uncertainties that might influence the success of the plan has been identified and classified (see Chapter 4.3). The airport system owner and the operator can recheck these uncertainties when the development in the future deviates from the plan. Scenario analysis is used to assess the impact of the uncertainty to the result. From this result, it is suggested that the recommendation for infrastructure expansion given in range rather than a single number. Based on the aforementioned answers, the research question can be answered. By following the method presented in this report, AP2 as one example of airport operator in the developing country can have a plausible recommendation to evaluate infrastructure plan of its airports in their master plan. Therefore, this research shows that for the airport operator in the developing countries with a lack of detailed data can use a relatively simple method to estimate the required airport infrastructures. This method consists of two models: the demand model using a multiple linear regression, infrastructure sizing model which calculate the number of runway, the number of aircraft parking stands, and the size of passenger terminal. It should be kept in mind that the method takes uncertainties into account by using scenario analysis. The aforementioned method suffices to give the airport planners a plausible range of infrastructure to design the new airport layouts. Additionally, the authorities can use the outcomes (ranges) to formulate flexible policies which is currently not available. The method developed in this thesis is in principle the same as the one in a developed country. However, its simplicity allows airport operators in developing countries which usually lack of data availability, to have a rough estimation of their future infrastructures. Discussion Policy Implication The base calculation estimates the required infrastructure when all the assumptions still hold. However, the uncertainty in future can alter the recommendations, especially when the demand deviates much higher or lower than the base condition. Thus, they should evaluate the situation regularly, and if the situation does not meet the expectation, they should review the calculation. Therefore, having a range of recommendation is much better rather than a single number. This range of recommendation shows that the best value is hard to determine when uncertainty analysis is added in estimating airport infrastructure. Therefore, the decision makers can use this information to make a detailed plan, but they should also include flexibility in the design within the master plan. For example, the terminal design in the master plan is designed in the huge area but in several modular infrastructures. Thus, in the execution, before the next module is constructed, the decision makers can reevaluate current demand situation and budget availability to decide whether the construction project can proceed at that year or not. Also, they can use it to make a prioritization in the implementation process by considering the financial condition of the company and in each airport. Another important message from this result is AP2 should encourage the Indonesian DGCA to evaluate the regulation about airport planning in Indonesia. The empirical data from AP2 showed that the PHC stated in the PM178/2015 is not applicable to the situation of Indonesia. It is assumed that the conversion in PM178/2015 is adapted from FAA. Thus, it is good for DGCA of Indonesia to conduct an empirical study to estimate the Peak Hour Conversion. Hence, the airport owner can be spared from investing unnecessary infrastructure.

65 Conclusions and Recommendations 49 Currently, uncertainty is not addressed in the master plan of Indonesian Airports. The government should revise the guideline of conducting a master plan for airport owner to force the airport owner to include uncertainty analysis in the master plan. Practical Recommendation Based on the case study, several recommendations can be proposed to AP2: Focus on documenting operational performance indicator. The AP2 should think more about measuring and documenting performance indicators to have a better infrastructure planning. This study mainly uses modest methods in finding the suitable infrastructure needs due to the lack of performance data. When more detail input is available, employing a more advanced model would produce a better estimation result. Performance data such as delay information for each process in the passenger terminal and delay in the runway is recommended to be collected in the future. Update the demand model input periodically. As a nature of forecasting process, the model should be adjusted by new input to make the model being updated. Increase human capability to analyze all aspects of airport operation internally. As one of the main airport operator in Indonesia, AP2 could be the reference of other airports in Indonesia and even in Southeast Asia. Unfortunately, this decent chance is not well captured by the management team. The experience of managing many airports is not well documented and transferred to employees, especially the knowledge to plan the infrastructure. Propose new regulation standard to DGCA-MoT of Indonesia. There is a gap between the existing regulation of Airport operation in Indonesia with the characteristic of Indonesian airports. The only regulation related to standard infrastructure available is the passenger terminal size in PM178/2015. It looks promising as a guideline for managing airport facilities in Indonesia, unfortunately, it is a very similar with FAA standard. This study confirms that PM178/2015 is not suitable or applicable to Indonesian Airports. Specific study of ideal passenger terminal size for airports in Indonesia can be a good start to have an own guideline that fit with Indonesia characteristics. Contribution Scientific Contribution There are several methods available to address the problem of estimating future infrastructure needs. However, on one hand, there was no specific study that offers the method can be easily implemented by practitioners in developing countries because the existing models are too advanced. On the other hand, existing studies rarely include the comprehensive analysis of plausible uncertainties that may affect the recommendation result. So, this research offers a method that uses a combination of simple methods which are still sufficient to give plausible recommendation on required infrastructure for airport operator. Further, it also addresses the uncertainty that might affect the recommendation. Practical Contribution In general, this method can be used by airport operators who suffer from the dependency of the external party in developing the long-term infrastructure planning. Also, for the airport operator who has limited availability of data regarding their operation performance (e.g., the delay data in passenger terminal). Therefore, this method can be a reference method before they can adopt the more sophisticated model once all the data of operation performance have been well-documented. In particular for AP, this method can be used as a new standard in evaluating future infrastructure, considering the evaluation process of infrastructure availability in AP2 has not been conducted

66 Conclusions and Recommendations 50 periodically. Moreover, they do not have the capability to do it yet. Thus, this proposed method gives AP2 the ability to evaluate their airport infrastructure plan internally. Limitation and Future Research There are several limitations in this research that can be an idea for future research. Limitation in the case study. In applying the proposed method, only data from AP2 as the case study is used. Applying in more airport system owner in the developing countries can increase the validity of this proposed method. Limitation in an econometric model. The functional relationship in the demand model is only limited to linear-linear and log-log model. The other structural relationship can also be explored such as log-linear, linear-log or non-linear model, non-linear model or other complex models. Limitation in the data availability. The proposed method is designed by considering available data in the case study. It is assumed that data availability in airport operator in AP2 represents the problem in the airport operators in developing countries. Further research can analyze the data availability in airport operators in developing countries, suggest various data collection needed and explore the potential benefit if they adopt advance their planning process. Limitation in the detailed facilities estimation. There are only three main infrastructures design parameter calculated in this research. Whereas, there are many other facilities and infrastructures which need to be estimated in Airport Planning. For example, in this project, taxiway design is omitted because of its complexity. Thus, a study to estimate other facilities and infrastructure such as the taxiway design parameter can extend the efficacy of this approach. If multiple design decisions can be estimated at once, it can be used to do a simple feasibility analysis method. Limitation in decision process covered in the method. This study only focuses on giving ideal infrastructure recommendations to balance the demand in the planning horizon. The next step is to decide when and how much the infrastructures should be planned so the development of infrastructures can be more systematic. From the uncertainty analysis in the case study, it is found that the uncertainty in one airport is not completely the same as the other airport. It can be a good direction to explore all possible uncertainties in airport system and if possible to model all possible uncertainties within this industry. Moreover, the impact of the uncertainty can also be assessed further. A more comprehensive study on the various uncertainty impact on infrastructure capacity can be another room for the research improvement.

67 References 51 6 References Abbas, K. (2004). Conceptual and Regression Models for Passenger Demand Prediction: A case study of Cairo Airport and Egyptair. Aerlines Magazines e-zine edition(26), 1-4. Retrieved from Aerlines Magazine e-zine edition, Issue 26: Abed, S., Ba-Fail, A., & Jasimuddin, S. (2001). An Econometric Analysis of International Air Travel Demand in Saudi Arabia. Journal of Air Transport Management, 7, ACI. (2011). ACI Airport Traffic Forecasting Manual. Airport Council Internationals. Airbus. (2017). Airbus Global Market Forecast France: Airbus. Andreoni, A., & Postorino, M. (2006). Time Series Models to Forecast Air Transport Demand: A Study about A Regional Airport. 11th IFAC Symposium on Control in Transportation Systems, (pp ). Delft, The Netherlands. Ashford, N., Mumayiz, S., & Wright, P. (2011). Airport Engineering: Planning, Design, and Development of 21st-Century Airports (4th ed.). John Wiley & Sons. ATAG. (2005). Air Transport Drives Economic and Social Progress. Geneva: Air Transport Action Group. ATAG. (2016). Aviation Benefits Beyond Borders. Geneva: Air Transport Action Group. BaFail, A. (2004). Applying Data Mining Techniques to Forecast number of Airline Passengers in Saudi Arabia (Domestic and International Travels). Journal of Air Transportation, BaFail, A., Abed, S., & Jasimuddin, S. (2000). The Determinants of Domestic Air Travel Demand in The Kingdom of Saudi Arabia. Journal of Air Transportation World Wide, 5(2), Baikgaki, O., & Daw, O. (2013). The Determinants of Domestic Air Passenger Demand in the Republic of South Africa. Mediterranean Journal of Social Sciences, 4(13), Boeing. (2017). Current Market Outlook Seattle: Boeing. Bubalo, B., & Daduna, J. (2011). Airport capacity and demand calculations by simulation the case of Berlin-Brandenburg International Airport. Netnomics, 12, Burghouwt, G., Boonekamp, T., Suau-Sanchez, P., Volta, N., Pagliari, R., & Mason, K. (2017). The impact of airport capacity constraints on air fares. Amsterdam: SEO Amsterdam. Burghouwt, G., & Huys, M. (2003). Deregulation and the Consequences for Airport Planning in Europe. The Planning Review, 39(154), Dantas, T., Oliveira, F., & Repolho, H. (2017). Air transportation demand forecast through Bagging Holt Winters methods. Journal of Air Transport Management, 59, Devoto, R., Farci, C., & Lilliu, F. (2002). Analysis and Forecast of Air Transport Demand in Sardinia's Airports as a function of Tourism Variables. In L. Sucharov, C. Brebbia, & F. Benitez, Urban Transport VIII (pp ). Southampton: WIT Press, cop. European Commission. (2002). Study on Competition between Airports and the Application of State Aid Rules. Brussels: European Commission. FAA. (1988). Planning and Design Guidelines for Airport Terminal Facilities. Advisory Circular AC 150/ , Washington D.C. FAA. (2005). Advisory Circular 150/5070-6B: Airport Master Plans. Washington DC: US Department of Transportation. FAA. (2015). FACT3: Airport Capacity Needs in the National Airspace System. FAA.

68 References 52 Grosche, T., Rothlauf, F., & Heinzl, A. (2007). Gravity models for airline passenger volume estimation. Journal of Air Transport Management, 13, Hansman, R., Magee, C., De Neufville, R., Robins, R., & Roos, D. (2006). Research agenda for an integrated approach to infrastructure planning, design, and management. International Journal of Critical Infrastructures, 2(2-3), Horonjeff, R., McKelvey, F., Sproule, W., & Young, S. (2010). Planning & Design of Airports (5th ed.). Mc Graw Hill. IATA. (2004). Airport Development Reference Manual. Montreal: IATA. ICAO. (1987). Airport Planning Manual: Part I. International Civil Aviation Organization. ICAO. (2006). Manual on Air Traffic Forecasting. International Civil Aviation Organization. Indonesia Ministry of Transportation. (2015). Ministerial Decree No. 178 of 2015 on Service Standards of Airport Service Users. Jakarta. Indonesia Ministry of Transportation. (2005). Cetak Biru Transportasi Udara Retrieved January 2017, from Indonesia Ministry of Transportation. (2014). DG Decree No. 590 Year 2014 on Technical Guidelines for Making Airport Master Plan. Jakarta. Ishutkina, M., & Hansman, R. (2009). Analysis of Interaction between Air Transportation and Economic Activity. The 26th Congress of ICAS and 8th AIAA ATIO. American Institute of Aeronautics and Astronautics, Inc.,. Janic, M. (2013). Airport analysis, planning, and design: Demand, capacity, and congestion. New York: Nova. Jones, D., & Pitfield, D. (2007). The Effectiveness of Conceptual Airport Terminal Designs. Transportation Planning and Technology, 30(5), Kopsch, F. (2012). A demand model for domestic air travel in Sweden. Journal of Air Transport Management, 20, Kwakkel, J., Walker, W., & Marchau, V. (2010). Adaptive Airport Strategic Planning. European Journal of Transport and Infrastructure Research, 10(3), Kwakkel, J., Walker, W., & Wijnen, R. (2008). The Treatment of Uncertainty in Airport Strategic Planning: The Case of Schiphol Airport's Long Term Vision. Delft, Netherlands: Delft University of Technology. Mirkovic, B., & Tosic, V. (2014). Airport apron capacity: estimation, representation, and flexibility. Journal of Advanced Transportation, 48, Scala, P., Mota, M., & Delahaye, D. (2016). No more conflict: the development of a generic airport model in a sequence-optimization framework. 28th European Modeling & Simulation Symposium. Larnaca. Shepard, W. (2017). forbes. Retrieved from Singh, D., Dalei, N., & Raju, T. (2016). Forecasting investment and capacity addition in Indian airport infrastructure: Analysis from post-privatization and post-economic regulation era. Journal of Air Transport Management, 53,

69 References 53 Sismanidou, A., & Tarradellas, J. (2017). Traffic demand forecasting and flexible planning in airport capacity expansions: Lessons from the Madrid-Barajas new terminal area master plan. Case Studies on Transport Policy, 5, Sivrikaya, O., & Tunç, E. (2013). Demand Forecasting for Domestic Air Transportation in Turkey. The Open Transportation Journal, 7, Solak, S., Clarke, J.-P. B., & Johnson, E. L. (2009). Airport terminal capacity planning. Transportation Research Part B: Methodological, 43(6), Sölveling, G., & Clarke, J.-P. (2014). Scheduling of airport runway operations using stochastic branch and bound methods. Transportation Research Part C, 45, Uehara, K., Hiraishi, K., & Kobayashi, K. (2015). Mesoscopic Modeling of Airport Surface by Object Petri Nets. International Conference on Automation Science and Engineering (CASE). Gothenburg. Valdes, V. (2015). Determinants of air travel demand in Middle Income Countries. Journal of Air Transport Management, 42, Wadud, Z. (2011). Modeling and Forecasting Passenger Demand for a New Domestic Airport with Limited Data. Transportation Research Record: Journal of the Transportation Research Board, 2214, Wadud, Z. (2014). Simultaneous Modeling of Passenger and Cargo Demand at an Airport. Transportation Research Record: Journal of the Transportation Research Board, 2336, Walker, W., Harremoës, P., Rotmans, J., van der Sluijs, J., van Asselt, M., Janssen, P., & Krayer von Krauss, M. (2003). Defining Uncertainty: A Conceptual Basis for Uncertainty Management in Model-Based Decision Support. Integrated Assessment, 4(1), Whitford, R. (2003). Airport Planning and Design. In The Civil Engineering Handbook (2nd ed.). CRC Press LLC. Yao, S., & Yang, X. (2012). Air transport and regional economic growth in China. Asia-Pacific Journal of Accounting & Economics, 19(3),

70

71 Appendix 55 Appendix A. Brief introduction to AP2 Airports Angkasa Pura II (AP2) is one of two state-owned enterprises of Indonesia who manages airports in Indonesia. Currently, there are 13 airports which are spread across 11 provinces in the west part of Indonesia. In this section, each airport is described briefly followed by the presentation of airport layout and the traffic figures. 1. Soekarno-Hatta International Airport (IATA: CGK) Soekarno-Hatta International Airport (SHIA) is the first airport, together with Halim Perdanakusuma Airport (ICAO: HLP), managed by AP2. Nowadays, CGK plays a role as primary hub airport of Indonesia and main gateway of international passengers. It is located in Cengkareng a small district outside Jakarta, approximately 20 km from the center of Jakarta. Since the opening in 1985, SHIA evolved into one of the busiest airports in Asia. Currently, it has two runways and three terminals with a total capacity of 22 million passengers. The expansion of the third terminal will be finished this year and will elevate the capacity to 29 million passengers. 2. Kualanamu International Airport (IATA: KNO) Kualanamu International Airport is an international airport serving North Sumatera. It is located 26 km north-east of Medan, the capital city of the province. It is the second largest airport in Indonesia and the fourth busiest airport in Indonesia. This airport is replacing Polonia Airport (IATA: MES) on 25 July 2013 since there is no room for expansion of this airport to balance the demand. 3. Halim Perdanakusuma Airport (IATA: HLP) Halim Perdanakusuma Airport, or usually called Halim, is located in East Jakarta. The airport is now home to many turboprops, charter, and general aviation companies. It also becomes a major air force base of the Indonesian Air Force. As a civilian airport, Halim was one the city's main airports, along with Kemayoran Airport. Halim served all international routes bound for Jakarta, while Kemayoran handled domestic flights. After the opening of Soekarno Hatta International Airport in Tangerang in 1985, Kemayoran Airport is shut down, and the Halim served only charter flight. To relieve congested traffic of SHIA, Halim started to serve domestic scheduled commercial flights again on January 10, Sultan Mahmud Badaruddin II Airport (IATA: PLM) Sultan Mahmud Badaruddin II Airport is international airport serving Palembang City and South Sumatra Province. It is located 10 km from the center of Palembang. The name comes after the last Sultan of Palembang, Sultan Mahmud Badaruddin. Effective 1 April 1991, this airport is managed by APII. It has become an international airport as 27th September 2005 after extended its runway along 300 meters x 60 meters to 3000 meters x 60 meters and built a new terminal building. 5. Supadio International Airport (IATA: PNK) Supadio International Airport is an Airport in Pontianak, West Kalimantan. It is located 17 km to the southeast of the city center. This airport has been done some development projects to have a world-class standard such as widened and lengthened the runway as well as built new terminal. 6. Minangkabau International Airport (IATA: PDG) Minangkabau International Airport is a major airport serving West Sumatera Province. It is located 23 km northwest of Padang. The airport opened in 2005, replacing the previously Tabing Airport which has been operating for 34 years. The construction cost approximately 9,4 billion yen which came from Japan in the form of a soft loan from Japan International Corporation Bank.

72 Appendix Sultan Syarif Kasim II International Airport (IATA: PKU) Sultan Syarif Kasim II International Airport is an airport in Pekanbaru, Riau. This airport is often referred as SSK and formerly known as Simpang Tiga Airport. The airport area, especially the runway is shared with Indonesian Air Force Base. On July 2012, a new terminal was opened to accommodate 1,5 million passengers a year to replace the old terminal building. 8. Husein Sastranegara Airport (IATA: BDO) Husein Sastranegara Airport is an airport in Bandung, the capital city of the most populated province in Indonesia, Jawa Barat. The airport is located 5 km away from the city center and 2,4 km from Bandung Central Train Station. In 2016, the expansion of the terminal has finished which fly the capacity from 1 million passengers per year to 3,5 million passengers per year. 9. Sultan Iskandar Muda International Airport (IATA: BTJ) Sultan Iskandar Muda International Airport is an airport serving Nangroe Aceh Darussalam Province. It is located 13,5 km southeast of the capital city, Banda Aceh. The recent development of this airport was in 2009 when the runway was increased to 3000 meters long and 45 meters width also a new terminal building replaced the old one. 10. Raja Haji Fisabilillah International Airport (IATA: TNJ) Located in Tanjung Pinang, Raja Haji Fisabilillah International Airport is the second largest airport in Riau Islands after Hang Nadim International Airport in Batam. The new passenger terminal was opened in June 2013 with a capacity of one million passengers, ten times larger than the old airport. 11. Depati Amir Airport (IATA: PGK) Depati Amir Airport is an airport in Pangkal Pinang, Bangka Island and the capital city of Bangka-Belitung. The other airport in this province is the other being H.A.S. Hanandjoeddin International Airport in Belitung Island. The new terminal with a capacity of 1,5 million passengers has been opened early this year (11 January 2017) from previously passengers. 12. Sultan Thaha Airport (IATA: DJB) Sultan Thaha Airport is the main airport in Jambi City and located 5 km from the downtown. The airport was built by Dutch and Japanese by the name of Paalmerah Airport. After the Independence Day, this airport is managed by Technical Implementation Unit of the Department of Transportation until the ministry of transportation handed over the operation to APII in January Developments are continuously performed such as the expansion of runway and terminal to make this airport an International Airport. 13. Silangit Airport (IATA: DTB) Silangit Airport is located in Siborong-borong, North Sumatera, 260 km from its capital city, Medan. Silangit is the newest airport managed by AP2 due to the instruction of Ministry of Transportation Although Kualanamu Airport is much greater than Silangit Airport, but this airport is located closer to the Lake Toba, the main tourist attractions in the province. Silangit airport is 77 km (2 hours driving) away from Prapat Meanwhile Kualanamu is 164 km (4 hours driving) away from Lake Toba. Therefore, this airport is expected to be the main gateway for tourists to visit Danau Toba. A.1. Airport Layouts In Table A.1 below, the layout of AP2 s Airports is presented. Existing runway is indicated in Black, existing Apron and Taxiway are indicated in gray, existing terminal is in yellow, and future development is indicated in red. Most of the airports are using single runway configuration except CGK. CGK has utilized the runway capacity potential by applying parallel taxiway and multiple (rapid) exit to the runway.

73 Appendix 57 Table A.1. Airport Layouts. Source: adapted from AP2 s documents CGK KNO HLP PLM PNK PDG PKU BDO BTJ TNJ PGK DJB Notes: The airport layouts are not in correct direction DTB

74 Number of ATM/Year Number of ATM/Year Number of ATM/Year Appendix 58 A.2 Traffic Data AP2 ( ) Domestic ATM Time CGK /10 KNO /10 HLP PLM PNK PDG PKU BDO Figure A. 1. Domestic ATM. Source: processed from AP2 s documents International ATM Time CGK /10 KNO /10 HLP PLM PNK PDG PKU BDO Figure A. 2. International ATM. Source: processed from AP2 s documents Total ATM Time CGK /10 KNO /10 HLP PLM PNK PDG PKU BDO Figure A. 3. Total ATM. Source: processed from AP2 s documents

75 Number of Passanger/Year Number of Passenger/Year Number of Passenger/Year Appendix 59 Domestic Passengers Time CGK /10 KNO /10 HLP PLM PNK PDG PKU BDO Figure A. 4. Domestic Passengers. Source: processed from AP2 s documents International Passengers Axis Title CGK /10 KNO /10 HLP PLM PNK PDG PKU BDO Figure A. 5. International Passengers. Source: processed from AP2 s documents Total Passengers Time CGK /10 KNO /10 HLP PLM PNK PDG PKU BDO Figure A. 6. Total Passengers. Source: processed from AP2 s documents

76 Appendix 60 Figure A. 7. Passenger Traffic Forecast in AP2 ( ). Source: own work

77 Appendix 61 B. Determinant of airport demand The list of literature that develops econometric model for air travel demand is presented in the following table. Table B.0.1. Airport Demand Model in Literature. Source: own work N o Study Location Model Structure Method Data Data Type (Passenger Type) Explanatory Variables BaFail et.al. Saudi Multiple National Population and Total 1 (2000) Arabia Linear Regression 24 years (Domestic) Expenditure* National Abed et.al. Saudi Multiple (Internation Population and Total 2 (2001) Arabia Linear Regression 24 years al) Expenditure* No of Bed, Population, Devoto et Multiple Airport Tourist Arrival, No of Bed 3 al. (2002) Sardinia Linear Regression 20 years (Total) per Capita* Cities/Airpo Domestic: Oil GNP, IPC, rt Import, Population, CPI, (Domestic PCE, GDP* 18 years and International: Oil GNP, IPC, BaFail Saudi Neural (annual) x Internation Population, GDP, 4 (2004) Arabia Unknown Network 5 cities al) Expenditure* Airport Abbas Cairo Multiple (Internation Population and Foreign 5 (2004) (Egypt) Linear Regression 10 years al) Tourists Cities- 8 months x Pair/Airport GDP, Population, Distance, Grosche et Multiple 1228 city- -Pair Catchment, Travel Time, 6 al. (2007) Germany Log-log Regression pairs (Domestic) Buying Power Index Wadud Banglade Multiple 36 years x Airport GDPpc, Population, Time 7 (2011) sh Log-log Regression 2 airports (Total) Ratio 12 years (annual) x GDP, Population Density, Yao and Multiple 31 Provinces Trade/GDP, Services, 8 Yang (2012) China Log-log Regression provinces (Total) Ground Passengers GDP, Population, Fares, Train Fares, Dummies Kopsch Multiple 27 years National (special events and 9 (2012) Sweden Log-log Regression (month) (Domestic) seasonality) Total consumption, Baikgaki South Multiple 42 years National population size, airfares, 10 (2013) Africa Log-log Regression (annual) (Domestic) and oil prices

78 Appendix Sivriyaka (2013) Turkey Lin-log Wadud (2014) Priyadarsha na Fernando & Banglade sh log-log (2015) Sri Lanka log-log Valdes (2015) Erraitab Middle Income Countries (MICs) Log-log (2016) Morocco Linear Multiple Regression Multiple Regression Multiple Regression Multiple Regression Multiple Regression 1 year (annual) 18 years (annual) 24 years 7 years (annual) x 32 countries 10 years (annual) Pair/Airport -Pair (Domestic) National (Total) Cities- Airport- Airport- National (Total) National (Total) National (Total) Travel match, Travel time, Connection type, Number of airlines, Ticket price, Schedule Consistency, Bed Capacity, Cities population, and Distance GDP, Crude Oil Price, National Price Level GDP, Jet Fuel Price, Tourist Traffic Growth Index, Terrorist Activity GDPpc, FDI, LCC, CPI, Fuel Price, (regulation) Dummies CPI, GNI, HFCEPC, ITNA Notes: GNP: Gross National Product; IPC: Income Per Capita; PCE: Private Consumption expenditure; GDP: Gross Domestic Product, FDI: Foreign Domestic Investment, LCC: seat capacity offered by Low Cost Carrier, CPI: Consumer Price Index * explanatory variables in the best model

79 Appendix 63 C. Forecasting model development ICAO published a guideline for analysts to develop an econometric model to forecast air traffic. The schematic flow is presented in Figure C.0.1. The main steps in the development of a forecast using an econometric model are: a) Define the problem. The type of traffic to be modeled must be clear since it becomes the dependent variables of the model. It includes the time horizon used for the analysis. b) Select the relevant causal or explanatory variables. Traffic demand is usually affected by many factors. The variables should be chosen so that together they explain full picture of the demand development. They are also must be measurable, quantifiable, continuous and predictable. In identifying the possible variables to include in the model, it can be from a list of possible causal variables or a small number of most relevant or suspected variables. c) After the relevant variables are selected, based on judgment or prior analysis, establish the availability of data or the selection of substitutes or proxy variables if such data are not available. The data of potential variables are collected. Some variables may be rejected due to lack of available data. Some variables may need to be calculated or adjusted first if they cannot be obtained directly. However, this process should be done carefully. d) Once the data availability is established, formulate the model specifying the type of functional relationship between the dependent variable and the selected explanatory (causal) variables. e) Carry out an analysis to test the relationship being hypothesized, including the estimation of the model coefficients, their magnitudes, and signs and statistical measures. f) When the foregoing criteria are achieved, establish the model in final form. g) Develop forecasts of future scenarios for the explanatory variables from which the traffic forecast is subsequently derived. Figure C.0.1. Econometric Model Development Flowchart. Source: (ICAO, 2006)

80 Appendix 64 D. Forecasting model development D.1. Runway System The proportion of aircraft size in each airport is presented in Table D.0.2 below. Airport Code Table D.0.2. Fleet Mix in AP2. Source: processed from AP2 s documents Small AC proportion Large AC proportion CGK 0% 90% 10% KNO 19% 78% 3% HLP 17% 83% 0% PLM 26% 74% 0% PNK 35% 65% 0% PDG 14% 81% 5% PKU 20% 80% 0% BDO 18% 82% 0% BTJ 7% 80% 13% TNJ 40% 60% 0% PGK 49% 51% 0% DJB 28% 72% 0% DTB 50% 50% 0% Heavy AC proportion Information regarding the arrival and departure procedure to approximate the runway capacity is presented in Table D.0.3, Table D.0.4, Table D.0.5, and Table D.0.6. Table D.0.3. Characteristics of Arrival Operation Aircraft type Vi (knots) tai (s) da (NM) Small Large Heavy Table D.0.4. Separation rules for arrival (in NM). Small Large Heavy Small Large Heavy 6 5 4

81 Appendix 65 Table D.0.5. Characteristics of departure operation. Aircraft type Vi (knots) Tdi (s) dd (NM) Small Large Heavy Table D.0.6. Separation rules for departure (in sec) Small Large Heavy Small Large Heavy D.2. Apron-Gate Complex The average turn-around time and aircraft proportion in each airport are presented in the table below. Airport Code Average TAT Small AC (min) Table D.0.7. Average Turn Around Time and Fleet Mix. Source: Processed from AP2 s documents. Average TAT Large AC (min) Average TAT Heavy AC (min) Small AC proportion Large AC proportion CGK % 90% 10% KNO % 78% 3% HLP % 83% 0% PLM % 74% 0% PNK % 65% 0% PDG % 81% 5% PKU % 80% 0% BDO % 82% 0% BTJ % 60% 0% TNJ % 80% 13% PGK % 51% 0% DJB % 72% 0% DTB % 50% 0% Heavy AC proportion

82