ASSESSMENT OF LOW COST TERMINAL LOCATION AND CONFIGURATION IN AIRPORT Batari Saraswati Graduate School of Science Engineering Tokyo Institute of Technology 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8550, Japan Email: [email protected]

Shinya Hanaoka Associate Professor Graduate School of Science Engineering Tokyo Institute of Technology 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8550, Japan Email: [email protected]

ABSTRACT The growth of low-cost carriers (LCCs) industry has been coincident with the growth of airport industry. One of airports’ attempts to attract LCCs is by constructing dedicated terminal for LCCs, called a low-cost terminal (LCT). The ability to accommodate fast aircraft operational time in airfield is identified as one of the important requirements of LCT, and it is heavily influenced by the configuration of the LCT in the airport. The purpose of this paper is to quantitatively evaluate the location and configuration of LCT considering aircraft taxiing distance and passenger walking distance. The result provided is useful for determining efficient configuration and location of a new LCT in airports. The paper starts with comparison of wellknown LCTs operated worldwide. The mathematical model is developed to help in finding the most efficient location and configuration of LCT in airports. Keywords: Low cost terminal, terminal location, terminal configuration

1.

INTRODUCTION

The entry of low-cost carriers (LCCs) brings competition to the air transportation industry. In many instances, the entry has been influential because LCCs offer low prices to the market thus affecting almost all aspects of the business. Airport is one of the elements that is influenced heavily by LCCs. One of the reasons is because LCCs have distinct business model that requires different airport services than the ones usually offered to full-service airlines. Barret (2004) identified seven airport requirements needed to serve the low-cost carriers: (1) low airport charges, (2) quick 24-minute turnaround time, (3) single story airport terminal, (4) quick checkin, (5) good catering and shopping at the airport, (6) good facilities for ground transport, and (7) no executive/business lounge. Several airports have constructed low-cost terminal (LCT) to address the issues. LCT is an airport terminal specially designed to accommodate LCCs, the concept of which emphasizes on

2 Saraswati and Hanaoka

cost and time reduction. Developing a specialized terminal for LCCs is a considerable alternative for airports to avoid conflicting needs between full-service airlines and LCCs (Graham, 2008). Besides, construction of LCT in the airport is believed to be powerful to attract LCCs and produce a strong, positive impact on traffic volume for the airport (Zhang et al., 2008). This paper addresses two main subjects: configuration and location of LCT in airport. These are considered as two of main factors that would influence the success of LCT. The configuration of LCT affects passenger walking distance, while the location of LCT towards runways affects aircraft taxiing distance. Both passenger walking distance and aircraft taxiing distance influence time spent by aircrafts and passengers in airport, thus affecting efficiency of LCC operations. Since most of LCTs around the world are in the initial stage of operation, researches and studies about LCT are still on a limited basis. The previous researches mainly focused on the relationship between LCCs and airport, and on how LCCs entry affects airport development and planning. The pioneering works include excellent studies by de Neufville (2006) and de Neufville (2008). De Neufville (2006) examined the issues of accommodating LCCs at the main airport. De Neufville (2008) also proposed a flexible design strategy for low cost airports. The study used Portugal case to illustrate how flexible design could manage uncertainties caused by the dynamics of LCCs. Section 2 of this paper explains the current condition of LCTs worldwide, focusing on the configuration and location of LCT inside an airport. Section 3 examines suitable configuration for LCT concerning passenger walking distance and construction area. In section 4, mathematical model is presented to find the best terminal site and configuration for LCT that can minimize the distance travelled by passengers and aircrafts. Finally, section 5 draws conclusion on the results of research. 2.

PRESENT STATE OF LOW COST TERMINAL

2.1

Current LCT Growth

The number of LCTs throughout the world is increasing from time to time. It shows that airports were keen to see the growth of LCCs and recognized that the current facilities provided are not appropriate for LCCs. The airports have responded by either redeveloping existing facilities (old passenger terminal and old cargo terminal) or building new facilities. Currently, there are about eleven LCTs built in several airports worldwide. Seven of them are located in Europe, three are located in Asia and one is located in United States of America. The number of LCTs keeps growing as there are other three LCTs currently in the process to be opened to accommodate growing traffic from LCCs. According to CAPA (2009), CPH Swift Terminal in Copenhagen Airport will be opened before the end of 2010 and LCT in Brussels

3 Saraswati and Hanaoka

International Airport is planned to be opened in April 2011. LCT in Kiev Borispool International Airport, Ukraine will be opened in 2020. Table 1. List of LCT worldwide Low Cost Terminal

Opening Year

Description

Terminal 2 in Tampere Pirkkala (Finland)

2003

Conversion of cargo terminal

Terminal 1 in Budapest Ferihegy Airport (Hungary)

2005

Refurbished old terminal

Pier H & M in Schiphol Airport (Netherlands)

2005

Piers off existing terminal

Terminal 2 in Marseille Provence Airport (France)

2006

Conversion of cargo terminal

Terminal 2 in Milan Malpensa Airport (Italy)

2006

Refurbished old terminal

Low Cost Carrier Terminal in Kuala Lumpur Airport (Malaysia)

2006

Newly built terminal

Budget Terminal in Changi Airport (Singapore)

2006

Newly built terminal

Terminal 3 in Lyon Saint Exupery Airport (France)

2008

Conversion of old passenger terminal

Terminal 5 in John F. Kennedy Airport (United States)

2008

Newly built terminal focusing on old TWA terminal

Budget Terminal in Zhengzhou Airport (China)

2008

Renovated temporary international hall

Bordeux Illico in Bordeaux Airport (France)

2010

Newly built terminal

Source: Graham (2008), CAPA (2009), and airport websites

LCTs are constructed with an intention to obtain traffic volume from LCC segments. Figure 1 and 2 show numbers of passengers and aircraft movements of four LCTs. The LCT in Kuala Lumpur International Airport (KLIA) and LCT in John F. Kennedy Airport (JFK) have attracted more than 10 million passengers in 2008 and 2009.

Source: Respective Airports Figure 1. Number of Passengers in LCTs

4 Saraswati and Hanaoka

LCT in KLIA is mainly used by AirAsia, the leading LCC in Asia. LCT in KLIA is proved to be beneficial for AirAsia. It has contributed to AirAsia’s cost reduction and output expansion (Zhang et al., 2008). LCT in JFK is dedicated for JetBlue Airways. The number of passengers for LCTs in Schiphol and Budapest is currently around one fifth of passengers in LCTs in KLIA and JFK. This is clearly influenced by the capacity of the LCT itself. The construction area for LCT in KLIA and JFK is 35,290 m2 and 58,000 m2 respectively. On the other hand, the construction area for LCT in Schiphol is 6,150 m2, while it is 7,990 m2 for LCT in Budapest. The trend of traffic is slightly decreasing from 2008 to 2009 for LCT in Europe and America, while traffic in LCT in KLIA keeps increasing.

Source: Respective Airports Figure 2. Aircraft Movements in LCTs

2.2 Passenger Walking Distance Passenger walking distance is a major consideration in determining the configuration of an airport terminal. This aspect is important to maintain passengers’ convenient level. In LCT, walking distance aspect is becoming more important since one of LCC requirements is to have fast passenger embarking and disembarking process. Table 2. Average walking distance Terminal Configuration

Average Walking Distance from Entrance Door to Gates (meters)

Terminal 1 Budapest Airport

Open Apron

35.21

LCCT Kuala Lumpur Airport

Multiple Piers

144.84

Low Cost Terminal

Terminal 2 Milan Malpensa Airport Terminal 2 Marseille Provence Airport Source: Google Earth

Linear

40.23

Single Pier

112.65

Walking distance for LCTs in Table 2 was calculated as the average rectilinear distance between the gate and the midpoint of departure waiting area. In LCCT KLIA, the walking distance from

5 Saraswati and Hanaoka

midpoint of international departure hall and domestic departure hall to aircraft gate is 144.84 meters. In Terminal 1 Budapest, the average passenger walking distance from waiting hall to gate is around 35.21 meters. In this terminal, passengers embarked from gate 1 – gate 6 will be transported to aircrafts by bus, while the others are required to walk to the aircraft. In Terminal 2 Milan Malpensa Airport, departure gates are divided into two parts, for flights to Schengen and non-Schengen region. The walking distance required to travel from waiting area to gate for Schengen flights is around 48.28 meters, while for non-Schengen flights the walking distance is about 32.18 meters. In Terminal 2 Marseille Provence Airport, the average walking distance is around 112.65 meters for the 6 gates available. Figure 3 shows the location of aircraft gates and aircraft stands in several LCTs. The figure is adopted from each LCT map of respective airport.

Terminal 1 Budapest Airport

LCCT Kuala Lumpur Airport

Terminal 2 Marseille Provence Airport

Terminal 2 Milan Malpensa Airport

Figure 3. Aircraft gate and aircraft stand location in LCTs

6 Saraswati and Hanaoka

2.3 Aircraft Taxiing Distance Taxiways provide connection between runways and terminal/apron areas. The location of the terminal relative to the runways influences the taxiing time required by aircraft. The location of LCT is commonly far from the main terminal, such as the LCT in KLIA, which is located 19.7 km from the main terminal. The characteristic of LCT as an additional terminal potentially cause longer taxiing distance to and from runways.

Airports Marseille Provence Changi Airport Budapest Ferihegy KLIA Milan Malpensa

Runway Direction 14L/32R 14R/32L 02L/20R 02C/20C 13L/31R 13R/31L

Table 3. Average taxiway distance Average Taxiway Runway Runway distance from LCT Length (m) Width (m) to runways (m) 3500 45 2108.24 2370 45 4000 60 3287.08 4000 60 3707 60 3544.58 3010 60

14L/32R

4124

60

14R/32L

4056

60

17L/35R

3920

60

17R/35L

3920

60

3850.35

2401.945

Average Taxiway distance from Main Terminal to Runways (m) 2002.73 Terminal 2 3154.31

Terminal 3 3134.19

2651.39 Satellite Terminal 2800.26 Satellite Terminal 3033.61

Main Terminal 3190.52 Main Terminal 2965.22

Source: Google Earth, airport websites

3. SUITABLE CONFIGURATION FOR LOW COST TERMINAL Passenger walking distance in airport terminal building has been a tremendous research area. Subsequent works has been very analytical. Wirasinghe, et al., (1987) proposed a method to determine the optimal number of parallel equal-length pier fingers to minimize walking distance. Robuste (1991) analyzed several centralized hub terminal layouts, determining walking distance for each of them and using it to determine best geometry of the terminal. Bandara and Wirasinghe (1992) developed a procedure to obtain best terminal configuration with respect to walking distance by comparing optimal geometries for various configuration. De Neufville, et al. (2002) used excel spreadsheet models to analyze several configuration types as a function of walking distances and the amount of transfer passengers. In this study, suitable configuration for LCT will be examined. Parameters considered are passenger walking distance and construction area. These two parameters are chosen because they highly affect time and cost performance in LCT, for both aircraft and airport.

7 Saraswati and Hanaoka

1. Passenger walking distance affects time needed by passengers to embark to aircraft, thus tends to increase turnaround time of the aircraft and also affect passenger disutility 2. LCT, generally, is an additional facility that is built after the airport started operation, thus the available area is limited. Therefore, it is important to choose the terminal configuration that minimizes construction area. Construction area affects investment cost. There are 4 terminal configurations discussed: (1) linear; (2) single pier, (3) T-shaped pier, and (4) Y-shaped pier. They are simple configurations that are most likely used for LCT. Satellite and transporter configurations require highly-cost automatic passenger mover (APM) that is not preferable for LCT. The average walking distance is calculated as total walking distance required to travel from end point of waiting area to each gates (from the most distant and the closest gate) divided by the number of gates. It is important to remember that in LCC business process, transferring passenger is treated in similar way with arriving and departing passengers since most LCCs serve point-to-point flights. To make connection with LCC, two separate tickets are needed and they will be counted as separate contracts. The connection point will be treated as final destination and the transfer passengers need to check in again as if they depart from that airport. As a result, transfer passengers cannot directly travel from one gate to another. The construction area and average passenger walking distance can be calculated using the formulas provided in Table 4 and Table 5. The distance between gates is assumed similar. This assumption is reasonable since most LCCs are using one type of aircraft, thus the space needed for gates and aircraft stand is similar. Formulas in Table 4 and Table 5 are applicable for even number of gates (symmetrical configuration). In Y-shaped pier configuration, the angles between arms are assumed to be 120o. The formulas can be changed easily to suit the gate configuration problem in the real world, for instance, the unequal length of the arm piers. Although passenger walking distance formulation can be easily found from previous researches, the equation for average walking distance is provided in this paper to meet the needs of the proposed mathematical model. Table 4. Construction area Terminal configuration Linear 1 Pier T-shaped Pier Y-shaped pier

Construction Area

8 Saraswati and Hanaoka Table 5. Total and average walking distance Terminal configuration

Average

Condition

Linear

Number of entrance points =

1 Pier

Pier has even number of gates Piers in arm position have the equal length and number of gates N = N1 + N2 N1 = number of gate in main concourse. N2 = number of gate in arm piers

T-shaped Pier

Y-shaped pier

where: N = number of gates in the terminal (i = 1, …, N) d = distance between gates w = width of the piers y = clearance between main concourse and arm piers on the inner side To find the suitable configuration for LCT, an example is given to illustrate the walking distance and construction area for each terminal configuration. Consider terminal configurations with 12 gates. The gates are 50 meters apart. The width of the pier is assumed equal with the distance between gates. Table 6. Walking Distance and Construction Area Linear Single T-shaped Pier (Entrance points = 6) (y = 30) Pier Total Walking Distance 900 2100 2380

Y-shaped pier (y = 30) 2410.88

Average

75

175

198.33

200.91

Construction Area

30000

15000

19500

20582.53

Based on the calculation, the linear configuration gives shortest walking distance, but requires largest area among others and also multiple facilities for departing passengers. Single pier seems to be the preferable configuration since it gives shorter walking distance than T-shaped and Yshaped pier, and also requires smallest construction area. For better understanding, the comparison of average walking distance and construction area for each terminal configuration for various numbers of gates is presented below. The width of the pier is also assumed equal with the distance between gates (w = d = 50 meters). When number of gates become larger, T-shaped or Y-shaped is more preferable than single pier because its

9 Saraswati and Hanaoka

walking distance is about half from single pier’s and its construction area is just slightly larger. This result is aligned with previous works where pier configurations were preferable from the perspective of minimizing walking distance. Figure 4 below shows the location of gates and geometry condition in each terminal configuration. Table 7. Walking distance and construction area for various number of gates Criteria

Average Walking Distance

Construction Area

Number of Gates

Linear (N/2 entrance points)

Single Pier

T Pier (y = 30, w = 50)

Y Pier (y = 30, w = 50 )

6

75

100

140

142.58

12

75

175

198.33

200.91

18

75

500

256.67

259.25

6

15000

7500

12000

13082.5

12

30000

15000

19500

20582.53

18

45000

22500

27000

28082.5

Figure 4. Terminal configuration

10 Saraswati and Hanaoka

4. MODEL DEVELOPMENT 4.1 Model Formulation Research on determining location of airport facilities has been done before. The pioneering works mostly emphasized on airport location problem, including tremendous studies by Saatcioglu (1982) that developed mathematical model for airport site location in developing countries. Min, et al. (1997) proposed dynamic, multi-objective linear programming model that aims to determine the optimal airport site under capacity and budgetary restrictions. In this study, mathematical model is developed to solve the problem of terminal site and terminal configuration determination for LCT more systematically. The main idea is to find the best terminal site and configuration for LCT that minimize the distance travelled by passengers and aircrafts according to the number of aircraft gates desired. The mathematical model has two objectives. Objective (1) minimizes average passenger walking distance from waiting point to aircraft gates. Objective (2) minimizes average aircraft taxiing distance required from runways to apron area and vice versa. With the above discussion in mind, the following notation and model formulation are presented below. Consider an airport in a network that has a potential growth of LCCs and the decision maker wants to build a new terminal to serve LCCs. It is required to find the location and configuration such that the total distance travelled by passengers and aircraft is minimized. i = 1, 2, …, l

index for alternative sites for the new terminal

j = 1, 2, ..., m

index for terminal configurations

k = 1, 2, …, n

index for runway points that can be used for departing and arriving aircrafts

xij

number of aircraft gates that can be accommodated in new terminal site i with configuration j

f(xij)

passenger walking distance as a function of the number of aircraft gates xij in terminal site i with configuration j

dik

taxi-out distance required to travel by aircraft from terminal site i to runway point k

dki

taxi-in distance required to travel by aircraft from runway point k to terminal site i

Aij

capacity available for aircraft gates in terminal site i with configuration type j; each terminal site has different area thus has different capacity for accommodating aircraft gates

zij

equals to 1 if new terminal opens in site i with configuration j, 0 otherwise

11 Saraswati and Hanaoka

Model formulation Min DW =

(1)

Min DT =

(2) Subject to: (3) (4) (5)

Both objective functions use (0,1) multipliers zij. The role of zij is to assure the choice of one site and one configuration for the new terminal. The model is built based on the assumption that the decision maker has decided how many airport gates will be built in the new terminal. The value of xij will be set according to the decision. In objective (1), walking distance f(xij) is calculated based on the determined xij by using average passenger walking distance equations provided in Section 3 (Table 5). Objective (2) aims to minimize average total taxi-out and taxi-in distance from n available runways in airport. Constraint (3) guarantees that the number of aircraft gates desired in new terminal site i with configuration j does not exceed the capacity of the new terminal site, Aij. Constraint (4) requires that only one new site with one configuration should be chosen as a solution. Constraint (5) requires that a new terminal with certain configuration is either opened at a given site or not. 4.2 Model Application and Results To demonstrate how the model works and to verify its usefulness, the hypothetical problem regarding location and configuration of LCT in airport is presented. There are 4 possible sites (i = 1, 2, 3, 4) that are suitable for LCT construction in an airport. Each site has rectangular shape with different area width and length. Areas described in Table 8 are areas for aircraft gates construction. The airport has two runways. In each runway, there is one point for departing and another point for arriving aircrafts (k = 2). There are 4 choices of terminal configurations: linear (j = 1), single pier (j = 2), T-shaped pier (j = 3) and Y-shaped pier (j = 4). The decision maker decided that the new LCT should have at least 10 aircraft gates. Table 8. Input data for LCT location and configuration problem Site (i)

Site 1

Site 2

Site 3

Site 4

Area (m2)

17,500

24,000

18,000

22,000

Taxiing distance (m) (dik, dki)

155 x 112.9

210 x 114.3

200 x 90

200 x 110

Runway 1

(2000, 3200)

(4000, 3700)

(4500, 4500)

(3000, 4300)

Runway 2

(2500, 3800)

(2000, 3400)

(3500, 4500)

(3800, 2700)

12 Saraswati and Hanaoka

The types of aircraft that will be accommodated in new terminal are Boeing 737 series and Airbus 320 (ICAO and FAA airplane design group III), therefore 40-meter distance between aircraft gates is reasonable. The width of all configuration building is 18 meters. In T-shaped and Y-shaped pier, a clearance of main concourse on the inner sides is 15 meters. The weighted sum of the objective method can be applied to solve bi-objective optimization of terminal location and configuration problem. The basic idea of weighted sum method is to combine both objective functions in one single functional form. It entails selecting scalar weights (wi) and minimizing the following composite objective function: . If all the weights are positive, then minimizing U function provides a sufficient condition for Pareto optimality, which means the minimum of U is always Pareto optimal (Zadeh, 1963). There are many other advanced methods available to solve multi-objective optimization problem. However, there are only two objectives with similar unit (distance) in the terminal location and configuration problem, and we also have enough knowledge to compare both objective functions. The weighted sum method is considered sufficient to obtain optimal solution. The paired comparison method is chosen to set the weights because it provides systematic means to rate objective functions by comparing them. One function is treated as a reference function. Weight wi represents the tradeoff between Fi and the reference function at the solution point to the weighted sum problem. Considering the solution to the weighted sum problem is always Pareto optimal, the slope of the Pareto optimal curve in Figure below is determined as . The left side can be approximated as

(Marler and Arora, 2009). With

knowledge of the objectives and careful selection of the weights, the final solution may reflect the intended preferences that are incorporated in the weight. In terminal location and configuration problem, F1 refers to walking distance (DW) function and F2 refers to taxiing distance (DT) function. In order to obtain the weights (w1 and w2), passenger value of time per unit distance will be compared to aircraft value of time per unit distance. Table below shows the value of passenger and aircraft time of general aviation specified in US Federal Aviation Administration (2007). The information about average passenger walking speed and average aircraft taxiing speed are also available, therefore, we can obtain passenger time value and aircraft time value per unit distance. Table 9. Passenger and aircraft time value

Passenger time value (all-purpose passenger) Aircraft variable cost Average passenger walking speed Average aircraft taxiing speed Passenger time value Aircraft time value

$37.20/ hour $362.00 / hour 4.32 km/hour 30 km/hour $0.00861 / meter $0.01207 /meter

13 Saraswati and Hanaoka

Afterward, we can get the weights for the objective functions by setting F1 as reference function (theoretically any function can be used as reference), thus w1 = 1 and w2 = 1.40129. The decrease of 1 meter in walking distance is a substitute of 1.40129 meter increase in taxiing distance. The solution of single objective function can be generated using spreadsheet (for small number of data), LINGO or any linear programming software (large number of data). Graphic can also be very helpful to examine the non-inferior solutions. Figure 6 below shows that site 1 with T-shaped pier configuration (z13) and site 2 with single pier configuration (z22) are the non-inferior solutions.

Zij = new terminal in site i with configuration j Figure 6. Solution map

In calculating the areas that are needed, we have to consider the shapes of the terminal configuration whether they are suitable for the sites’ shapes. It is important to consider the gross area, instead of the effective construction area. Gross area means the outer rectangular area that can be obtained by multiplying the longest length with the longest width of the configuration. In this example, the gross areas for 10 aircraft gates for linear, single pier, T-shaped pier, and Yshaped pier are as following: (400 x 18) m2, (200 x 18) m2, (153 x 110) m2 and (199.19 x 51.53) m2. Linear configuration cannot be implemented in any sites because linear configuration requires more area than all sites can provide. Terminal site 1 also cannot accommodate single pier and Yshaped configuration because of the same reason. All the alternatives mentioned above do not satisfy area constraint in the model formulation (constraint 3), therefore they are not included in the solution set. As a result, we have nine possible solutions where two of them are non-inferior solutions (z13 and z22). After applying the single objective function with the determined weights, we can decide which one of these non-inferiors is the optimal solution of the problem. It is best to build a new

14 Saraswati and Hanaoka

terminal in site 1 with a single pier configuration (z13) and with the objective function value equal to 8174,019. According to the weights in this example, solution with shorter aircraft taxiing distance is more preferable than passenger walking distance. However, the value of weights can be vary. A decision maker’s decision can change depending on the point at which the different objective functions are evaluated. For example, when walking distance is too large, passenger inconvenience tends to increase, as a result, comparison between passenger time value and aircraft time value can be different. Hart (1985) have stated that an acceptable range for the maximum walking distance within airport terminal is around 300 – 400 meters, not including the walk between parking/curb and ticketing or baggage claim area.

5. CONCLUSION This paper has presented a method to determine location and configuration of LCT in an airport. The proposed model was designed to find optimal terminal site and configuration by considering aircraft taxiing distance and passenger walking distance. LCCs, as the main clients of LCT, care about passenger walking distance and aircraft taxiing distance because it affects their operational time and cost. Small savings in time may appear insignificant, but when cumulated over a day they can have a major impact. The solution of location and configuration problem is found by solving the linear integer programming model. The weighted sum method is used to combine the two objective functions into single objective function. The weight of each objective function is determined using pair comparison method. Although such model presented involves considerable simplification of the real world, it yields results that can be helpful in making some judgments regarding the solution to the problem. This paper also gives insight about LCT industry worldwide and can be categorized as a pioneer in this topic area. Despite the merits, the proposed model points a number of directions for future work. The model can be expanded to include other elements such as construction cost. Future works can pay closer attention in defining passenger and aircraft time value. The model can be tested using more realistic data for the expanded area.

6. REFERENCES Bandara, S. and Wirasinghe, S.C. (1992). “Walking Distance Minimization for Airport Terminal Configurations”, Journal of Transportation Research A, Vol 26A, No. 1, pp. 59-74 Barret, S.D. (2004), ”How Do the Demands for Airport Services Differ between Full-Service Carriers and Low-Cost Carriers?”, Journal of Air Transport Management, Vol 10, pp. 33-39

15 Saraswati and Hanaoka

Centre of Asia Pacific Aviation (CAPA). (2009), Low Cost Airport Terminals Report. de Neufville, R., de Barros, A.G. and Belin, S. (2002), “Optimal Configurations of Airport Passenger Buildings for Travelers”, ASCE Journal of Transportation Engineering, Vol 121, No. 3, pp. 211- 217 de Neufville, R. and Odoni, A. (2003), Airport Systems : Planning, Design and Management, McGraw-Hill, 1st Edition de Neufville, R. (2006), “Accommodating Low-Cost Airlines at Main Airports”, International Airport Review, No. 1, pp. 62 - 65 de Neufville, R. (2008), ”Low-Cost Airports for Low-Cost Airlines : Flexible Design to Manage the Risk”, Transportation Planning and Technology, Vol. 31, No. 1, pp. 35 - 68 Graham, A. (2008), Managing Airport: An International Perspective Third Edition, Elsevier, Burlington Hart, W. (1985), Airport Passenger Terminal, John Wiley, New York Marler R.T., Arora J.S. (2009). “The Weighted Sum Method for Multi-objective Optimization: New Insights”, Struct Multidiscipl Optim 26, pp. 369 - 395 Min, H., Melachrinoudis, E., Wu, X. (1997), Dynamic Expansion and Location of an Airport : A Multiple Objective Approach, Journal of Transportation Research A, Vol 31, No. 5, pp. 403-417 Robuste, F. (1991). “Centralized Hub-Terminal Geometric Concepts, I: Walking Distance”, Transportation Engineering Journal, 117(2), pp. 143 - 158 Saatcioglu, O. (1982), “Mathematical Programming Models for Airport Site Selection”, Journal of Transportation Research B, Vol 16 B, No. 6, pp. 435 - 447 US Federal Aviation Administration. (2007). “Economic Values for Evaluation of FAA Investment and Regulatory Program”. Office of Aviation Policy and Plans, US Department of Transportation, Washington DC Wirasinghe, S.C., Bandara, S., Vandebona U. (1987). “Airport Terminal Geometries for Minimal Walking Distances”, Transportation and Traffic Theory, pp. 483 -502 Zadeh L.A. (1963), “Optimality and Non-scalar-valued Performance Criteria”, IEEE Trans Automat Contr AC 8, pp. 59 - 60 Zhang, A., Hanaoka, S., Inamura, H., Ishikura, T. (2008), “Low Cost Carriers in Asia: Deregulation, Regional Liberalization, and Secondary Airports”, Research in Transportation Economics, 24, pp 36 - 50