An Application Using ELECTRE and MOORA Methods in the Selection of International Airport Transfer Center (Hub) in Türkiye

Abstract

In today’s world, air transport has become a favored choice for enhancing the value of a national economy, driven by advancing technology, escalating volumes of national and international trade, and population growth. The proliferation of airport transfer centers, particularly within air transport, plays a pivotal role in fostering the advancement of the aviation sector. Therefore, the selection of these hubs is of great importance. This study evaluated the New Çukurova, Antalya, Sivas Nuri Demirağ, Erzurum and Muğla Airports in Türkiye for the selection of a new airport transfer center in terms of criteria such as airport costs, airport terminal and apron facilities, airport passenger transportation services, airport operating capacity, airport location, demand factors in the service region and other factors. The study employed three methods for evaluating alternative international airports: AHP (Analytic Hierarchy Process), MOORA (Multi-Objective Optimization by Ratio Analysis) and ELECTRE (Elimination and Choice Translating Reality). In the initial phase, the priority ranking of criteria was established based on expert opinions. Subsequently, Antalya Airport was the most suitable airport transfer center according to the ELECTRE method, while New Çukurova Airport emerged as the preferred choice according to the MOORA method. Both airports secured top rankings in both evaluation methods.

1. Introduction

Air transport has emerged as the foremost choice for travel, both domestically and internationally, in recent times. Enhancements in air transport not only foster the expansion of a country’s industry and international trade but also facilitate convenient, swift and comfortable travel for individuals. Türkiye has notably risen as a pivotal hub for air transport, serving as a strategic crossroads connecting Asia, Africa, America and Europe. It stands as a highly developed nation in both domestic and international passenger and cargo transportation. The significance of an airport is directly proportional to the number of destinations it serves and the frequency of flights it offers. Second airports, particularly those facilitating transfers outside of central airports (hubs), hold significant importance in this regard. These airports are often regarded as transfer hubs, serving as vital conduits for the movement of people and cargo between various locations. Hence, the selection of an airport as a hub holds paramount importance.

In this study, the initial step involved reviewing prior research on the selection of hubs. In their 2002 study, Janic and Reggiani tackled the issue of selecting hub airports for a European Union airline. They employed three distinct methods to address the problem. Seven alternative airport terminals underwent evaluation across nine criteria utilizing Simple Additive Weighting (SAW), Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) and Analytic Hierarchy Process (AHP) methods [1]. In 2013, Jantachalobon and Vanichkobchinda conducted a study aimed at determining the optimal alternative for selecting a regional air transport hub in Southeast Asia, as well as identifying the factors influencing this selection. In their study, the primary objectives were to minimize total transportation costs and reduce travel times [2]. In 2014, I Casas et al. devised a simulation model to design a new terminal for Barcelona International Airport. Agent-based simulation techniques were also integrated into the simulation methodology. This endeavor has resulted in the system evolving into a decision-making tool for terminal management and dynamic optimization of operations [3]. In 2015, Çiftçi and Şevkli put forward a mathematical model for the selection of a new hub airport. The model, encompassing over 90 cities in Europe and the Middle East, was applied to three prominent cities in Türkiye: Ankara, Antalya and İzmir. The model considers various factors such as unit passenger revenues, operating costs, distances, occupancy factors and flight times. As a result, Antalya has been identified as a viable airport for the establishment of the new hub [4]. In the 2018 study by Korani et al., significant factors and criteria were initially identified to evaluate all alternatives for hub airport selection. The criteria were weighted using the AHP method, after which the alternatives were evaluated through Data Envelopment Analysis. This hybrid method has been implemented for Iranian airports [5]. In Dozic’s 2019 study, 166 articles in the field of aviation published between 2000 and 2018 were analyzed. In this research, studies addressing topics related to airlines, airports, air traffic, etc., and resolved through multi-criteria decision-making methods were categorized. The analysis of the articles revealed that Taiwan leads in studies conducted within the aviation sector, with the AHP method being predominantly utilized among multi-criteria decision-making methods [6]. In the study conducted by Görçün in the same year, transportation alternatives between Silivri Airport and Sabiha Gökçen Airport were evaluated using the AHP and TOPSIS methods. As a result of the study, four transit transportation alternatives were selected, with speed identified as the most crucial evaluation criterion in this selection process [7]. In another study by Soylu and Katip in the same year, they focused on the capacity-free multiple P-hub median problem for the location allocation of new hub airports. They put forth exact and heuristic algorithms aimed at determining Pareto bounds for new investment strategies and for minimizing transportation costs within airline networks. Their study revealed that opting for routes with fewer stops is not always cost-effective, as it may lead to an increase in the overall costs within the airport network [8]. In another study in 2019, Sugianto et al. employed the AHP method to identify parameters for the establishment of a new hub airport. They conducted their study for four international airports in Indonesia. Five criteria were evaluated, and the order of importance was determined for the selection of an airport that could serve as a hub airport. As a result, they found that airport charges and airport costs are the most significant criteria [9]. In 2020, Aydın and Şeker conducted a study in which they evaluated five airports for the selection of a new hub airport aimed at ensuring low costs. In their study, they employed the Interval-valued Intuitionistic Fuzzy (IVIF) method, the Weighted Aggregated Sum Product (WASPAS) method and the Multi-Objective Optimization (MULTIMOORA) method. Based on twelve evaluation criteria, Antalya Airport was determined to be the most suitable alternative [10]. In 2022, Zargini’s study also tackled the vector-optimization problem, considering variable dominant and non-dominant solution structures for the selection of a new hub airport. He utilized his proposed approach to evaluate seven airports across seven criteria. As a result of the study, Brussels Airport was identified as the most suitable airport [11]. In 2023, Badi et al. sought to determine the hub airport and evaluated five airports in North African countries, considering five criteria. A hybrid gray-CODAS (Combined Distance-Based Assessment) method was employed in the assessment. As a result of the evaluation, Morocco International Airport Mohammed V, was identified as the best alternative [12]. In another study conducted by Taçoğlu et al. in the same year, they undertook an evaluation considering six criteria for the selection of a hub airport. The evaluation encompassed twelve airports in Türkiye, utilizing the Fuzzy AHP method. Following the study, İzmir Airport emerged as the most suitable choice for a hub airport [13].

In summary, numerous studies have been conducted, employing a variety of methods in the literature for the selection of airport transfer centers. However, there has not been a study that combines the AHP method with MOORA and ELECTRE techniques for airport hub selection, particularly focusing on international airports in Türkiye. MCDM methods offer very effective solutions because they represent a collective mind rather than a single person and are comparable between methods. Using methods together or in a hybrid way plays a major role in increasing the accuracy of the decision. This study can be characterized as novel, as it explores a unique combination of methodologies and considers diverse criteria for airport hub selection. Other parts of the study can be summarized as follows: Section 2 delves into the methodologies employed and the data utilized, while Section 3 evaluates the findings, presents a sensitivity analysis and underscores the study’s contributions. Finally, Section 4 discusses the outcomes of the evaluation and provides recommendations for future research directions.

2. Materials and Methods

In this study, as shown in Figure 1, the criteria weights are determined by using the AHP method as a first step. Then, in the second and third steps, ELECTRE and MOORA methods are used to rank the alternatives.

2.1. Analytic Hierarchy Process

The AHP technique, utilized in multi-criteria decision-making problems, was pioneered by Thomas Saaty in 1977 [14].

In this technique, characterized by its hierarchical structure, criteria are first established in alignment with the problem’s objectives. Subsequently, alternatives are evaluated based on these criteria. The steps outlined in Figure 2 are followed throughout the evaluation process.

Define problem: For effective decision-making, it is essential to clearly and comprehensively define the problem and reveal the purpose.

Identify criteria: Criteria suitable for the defined purpose should be identified for assessing the alternatives.

Identify alternatives: Once the criteria are determined, the decision alternatives to be evaluated should be identified.

Perform pairwise comparisons: Comparison matrices should be constructed by establishing the importance values of the determined criteria relative to each other. The importance scale defined by Saaty is utilized to ascertain the degree of importance for each criterion. The importance scale is shown in

Table 1. Importance scale used in criteria comparisons.

Intensity of Importance	Description
1	Equal Importance
3	Moderate Importance
5	Strong Importance
7	Very Strong Importance
9	Extreme Importance
2-4-6-8	Average (Intermediate) Values

.

The matrix formed by comparing the criteria in pairs is referred to as matrix A, where each element signifies their importance values relative to each other. Matrix A is a square matrix of size n $\times$ n, where n represents the number of criteria used in the evaluation. Its diagonal elements are all equal to 1. The matrix A is shown in Equation (1).

$$\mathrm{A}=a_{ij}=\left[\begin{array}{c}{a}_{11} a_{12}\dots a_{1n}\\ a_{21} a_{22}\dots a_{2n}\\ .\\ .\\ a_{n1} a_{n2}\dots a_{nn}\end{array}\right]$$

(1)

Each element of matrix A, denoted as a_ij, represents the importance value of criterion i relative to criterion j. In pairwise comparisons, a_ji is calculated as specified in Equation (2).

$$a_{ji}={\displaystyle \frac{1}{a_{ij}}}$$

(2)

Once the importance values of each criterion relative to each other are determined, the weights of the criteria among all criteria are calculated. The formula in Equation (3) is used for this calculation.

$$c_{ij}={\displaystyle \frac{a_{ij}}{{\sum _{i=1}^{n}a}_{ij}}}$$

(3)

The resulting matrix, known as the normalized matrix, is denoted as matrix C. Subsequently, the average of each row in the normalized matrix C is calculated, leading to the determination of the importance weights of the criteria ($W_i$). The formula in Equation (4) is used at this stage.

$$W_i={\displaystyle \frac{{\sum _{i=1}^{n}c}_{ij}}{n}}$$

(4)

Calculation of consistency ratios: Once the importance weights of each criterion are determined, consistency ratios are calculated to assess the consistency of the decision makers’ personal evaluations. For this calculation, the largest eigenvalue (${\lambda }_{max}$) should be determined first. The formula in Equation (5) is used to determine this value.

$${\lambda }_{max}={\displaystyle \frac{\sum _{j=1}^n{a}_{ij} . w_j}{w_i}}\mathrm{i}=\mathrm{1,2},\dots .,\mathrm{n}$$

(5)

After determining the largest eigenvalue, the “Consistency Index (CI)” and “Consistency Ratio (CR)” are then calculated. The Consistency Index is calculated by the formula in Equation (6) and the Consistency Ratio is calculated by the formula in Equation (7).

$$\mathrm{CI}={\displaystyle \frac{{\lambda }_{max}-n}{n-1}}$$

(6)

$$\mathrm{CR}={\displaystyle \frac{CI}{RI}}$$

(7)

The RI value in Equation (7) represents the random index value. These values, set by Saaty as fixed numbers, vary depending on the value of n. The index values corresponding to different n values are presented in

Table 2. Random index values.

N Value	1	2	3	4	5	6	7	8	9	10	11	12
RI Value	0	0	0.58	0.89	1.12	1.24	1.32	1.41	1.45	1.49	1.51	1.48

[15].

If the Consistency Ratio (CR) is less than 0.10, the pairwise comparison matrix is considered consistent. If there is a discrepancy, it is necessary to re-evaluate the pairwise comparison matrices.

2.2. ELECTRE Method

The ELECTRE method, developed by Bernard Roy in 1968, is a technique utilized for multi-criteria decision-making problems. It was introduced to address the shortcomings observed in other decision-making approaches [16]. While initially employed for selecting the best alternative, this method has been subsequently adapted for use in ranking and classification problems.

There are six variations of the ELECTRE method documented in the literature. The ELECTRE I method, initially proposed, has been applied in selection problems. Later, the ELECTRE II method was devised for ranking periodicals in media planning. The ELECTRE III method was developed to address the adverse impacts of incomplete information and to manage uncertainty by incorporating threshold values. The ELECTRE IV method was subsequently developed for ranking without factoring in the importance coefficients, specifically for a problem concerning the metro network.

The method incorporates outranking relations. In the ELECTRE IS method, artificial criteria are considered in cases where the data are imprecise, facilitating the selection of alternatives. The ELECTRE TRI method was developed to address the loan request challenges encountered by a bank. It was developed to tackle the challenge of assigning alternatives to predefined categories.

Table 3. Development of ELECTRE methods.

ELECTRE Method	Proposed by	Emerged in	Purpose
ELECTRE I	Bernard Roy	1968	Selection
ELECTRE II	Bernard Roy and Patrice Bertier	1971	Ranking
ELECTRE III	Bernard Roy	1978	Ranking
ELECTRE IV	Bernard Roy and Jean-Christophe Hugonnard	1982	Ranking
ELECTRE IS	Bernard Roy and Jean Michel Skalka	1985	Selection
ELECTRE TRI	Bernard Roy, Denis Bouyssou and Wei Yu	1991–1992	Assignment

summarizes information about these methods [17].

At the core of all ELECTRE methods lies the preference of alternatives based on outranking. The ELECTRE method generally consists of the steps shown in Figure 3 [18].

Create decision matrix: First, the decision matrix for the problem is constructed. In this matrix, the rows represent alternatives, while the columns represent criteria. This matrix is referred to as matrix A. The initial matrix A contains the scores of each alternative for each criterion as determined by the decision maker. Matrix A is shown in Equation (8).

$$A_{ij}=\left[\begin{array}{c}\begin{array}{cccc}{a}_{11}& a_{12}& \dots & a_{1n}\\ a_{21}& a_{22}& \dots & a_{2n}\\ .& & & .\\ .& & & .\\ & & & \\ .& & & .\\ a_{m1}& a_{m2}& \dots & a_{mn}\end{array}\end{array}\right]$$

(8)

In matrix A, as represented in Equation (8), m is the number of alternatives, n is the number of criteria and $a_{ij}$ is the importance score of alternative m for criterion n.

Normalize decision matrix: Each element of the decision matrix is normalized within the range [0,1]. It is important to note that separate equations are employed for criteria to be maximized and those to be minimized. Equation (9) is used for the criteria to be maximized, while Equation (10) is used for the criteria to be minimized.

$$x_{ij}={\displaystyle \frac{a_{ij}}{\sqrt{\sum _{k=1}^m{a}_{kj}^2}}}$$

(9)

$$x_{ij}={\displaystyle \frac{\frac{1}{a_{ij}}}{\sqrt{\sum _{k=1}^m{\left(\frac{1}{a_{kj}}\right)}^2}}}$$

(10)

The normalized decision matrix is shown in Equation (11).

$$X_{ij}=\left[\begin{array}{c}\begin{array}{cccc}{x}_{11}& x_{12}& \dots & x_{1n}\\ x_{21}& x_{22}& \dots & x_{2n}\\ .& & & .\\ .& & & .\\ & & & \\ .& & & .\\ x_{m1}& x_{m2}& \dots & x_{mn}\end{array}\end{array}\right]$$

(11)

Construct weighted normalized decision matrix: In the normalized decision matrix, the values in each column are multiplied by the importance values assigned to that criterion. Here, the weights are shown as $w_i$. The normalized decision matrix, $Y_{ij}$, is calculated as shown in Equation (12) and presented as shown in Equation (13).

$$Y_{ij}=w_i \ast X_{ij}$$

(12)

$$Y_{ij}=\left[\begin{array}{c}{w_{1}x}_{11} w_2{x}_{12}\dots w_n{x}_{1n}\\ w_1{x}_{21} {w_{2}x}_{22}\dots {w_{n}x}_{2n}\\ .\\ .\\ {w_{1}x}_{m1} {w_{2}x}_{m2}\dots {w_{n}x}_{mn}\end{array}\right]$$

(13)

Determine concordance and discordance sets: Once the weighted normalized decision matrix is established, the alternatives are compared with each other pairwise. These comparisons lead to the formation of concordance and discordance sets. The concordance set is denoted as C (a,b), while the discordance set is denoted as D (a,b), with the number of concordance sets being equal to the number of discordance sets. The number of matching sets is of size (m × m − m) due to the condition that the values of a and b are not equal to each other. The set C (a,b) is calculated as shown in Equation (14) and the set D (a,b) is calculated as shown in Equation (15).

$$C(a,b)=\left\{j, Y_{aj}\ge Y_{bj}\right\}$$

(14)

$$D(a,b)=\left\{j, Y_{aj}<Y_{bj}\right\}$$

(15)

Determine concordance and discordance indexes: Concordance and discordance matrices are formulated by considering the concordance and discordance sets obtained in the preceding step along with their respective weights. The concordance matrix C is derived by summing the weights of the criteria within the concordance set. The calculation of indexes for the concordance matrix is shown in Equation (16).

$$c_{ab}={\sum }_{j\in C (a,b)}^{ }{w}_j$$

(16)

The discordance matrix D is calculated by the formula provided in Equation (17).

$$d_{ab=}{\displaystyle \frac{\mathit{max}\left|{ Y}_{aj}-Y_{bj} (j \in D\left(a,b\right))\right|}{\mathit{max}\left|{ Y}_{aj}-Y_{bj}\right|}}$$

(17)

Conduct outranking comparison: Once the concordance and discordance matrices are determined, concordance and discordance outranking matrices are established.

When constructing the concordance outranking matrix, the threshold value, which is the average of the concordance indexes, is initially determined. The calculation of this value is shown in Equation (18).

$${\mathrm{C}}_{\mathrm{e}\mathrm{ş}\mathrm{i}\mathrm{k}}={\displaystyle \frac{1}{\mathrm{m}(\mathrm{m}-1)}}{\sum }_{\mathrm{a}=1}^{\mathrm{m}}{\sum }_{\mathrm{b}=1}^{\mathrm{m}}{\mathrm{c}}_{\mathrm{a}\mathrm{b}}$$

(18)

Subsequently, by comparing the values in the concordance matrix with these threshold values, the elements of the $\mathrm{E}(\mathrm{a},\mathrm{b})$ concordance outranking matrix are determined. Equation (19) shows this comparison.

$${\mathrm{e}}_{\mathrm{a}\mathrm{b}}=\left\{\begin{array}{c}{\mathrm{e}}_{\mathrm{a}\mathrm{b}}=1 {\mathrm{c}}_{\mathrm{a}\mathrm{b}}\ge {\mathrm{C}}_{\mathrm{e}\mathrm{ş}\mathrm{i}\mathrm{k}}\\ {\mathrm{e}}_{\mathrm{a}\mathrm{b}}=0 {\mathrm{c}}_{\mathrm{a}\mathrm{b}}<{\mathrm{C}}_{\mathrm{e}\mathrm{ş}\mathrm{i}\mathrm{k}}\end{array}\right.$$

(19)

When constructing the discordance outranking matrix, the threshold value, which is the average of the discordance indexes, is initially determined. The calculation of this value is shown in Equation (20).

$${\mathrm{D}}_{\mathrm{e}\mathrm{ş}\mathrm{i}\mathrm{k}}={\displaystyle \frac{1}{\mathrm{m}(\mathrm{m}-1)}}{\sum }_{\mathrm{a}=1}^{\mathrm{m}}{\sum }_{\mathrm{b}=1}^{\mathrm{m}}{\mathrm{d}}_{\mathrm{a}\mathrm{b}}$$

(20)

Subsequently, by comparing the values in the concordance matrix with these threshold values, the elements of the $\mathrm{F}(\mathrm{a},\mathrm{b})$ concordance outranking matrix are determined. Equation (21) shows this comparison.

$${\mathrm{f}}_{\mathrm{a}\mathrm{b}}=\left\{\begin{array}{c}{\mathrm{f}}_{\mathrm{a}\mathrm{b}}=1 {\mathrm{d}}_{\mathrm{a}\mathrm{b}}<{\mathrm{D}}_{\mathrm{e}\mathrm{ş}\mathrm{i}\mathrm{k}}\\ {\mathrm{f}}_{\mathrm{a}\mathrm{b}}=0 {\mathrm{d}}_{\mathrm{a}\mathrm{b}}\ge {\mathrm{D}}_{\mathrm{e}\mathrm{ş}\mathrm{i}\mathrm{k}}\end{array}\right.$$

(21)

Rank alternatives by net indexes: The outranking values are computed by multiplying the values in matrix E, the concordance outranking matrix, by the values in matrix F, the discordance outranking matrix. These outranking values consist of values of 1 and 0, similar to the concordance and discordance matrices. Finally, the alternatives are ranked based on these calculated values.

If multiple alternatives outrank a given alternative, net index values are calculated for ranking purposes. This calculation is shown in Equations (22) and (23).

$$C_a={\sum }_{\begin{array}{c}k=1\\ k\ne b\end{array}}^m{C}_{ab}-{\sum }_{\begin{array}{c}k=1\\ k\ne b\end{array}}^m{C}_{ba}$$

(22)

$$D_a={\sum }_{\begin{array}{c}k=1\\ k\ne b\end{array}}^m{D}_{ab}-{\sum }_{\begin{array}{c}k=1\\ k\ne b\end{array}}^m{D}_{ba}$$

(23)

When ranking, $C_a$ values are sorted from largest to smallest, while $D_a$ values are sorted from smallest to largest. Finally, in the final ranking, the alternative with the largest $C_a$ value and the smallest $D_a$ value is deemed more important.

2.3. MOORA Method

The MOORA method is a multi-criteria decision-making technique based on proportional analysis, developed by Willem Karel M. Brauers and Edmundas Kazimieras Zavadskas in 2006 [19]. This method finds application in various fields owing to its reliability, simultaneous evaluation of alternatives and criteria, swift solution attainment, user-friendliness and minimal mathematical calculations [18].

Despite its recent emergence in the literature, the MOORA method has found application in diverse fields including transition economics, road design, facility location, regional development, project management and real estate procurement [20]. The method is applied in the literature in four different ways: MOORA Ratio Approach, MOORA Importance Coefficient Approach, MOORA Reference Point Approach and MOORA Full Multiplicative Form.

2.3.1. Ratio Approach

In this approach, initially, a decision matrix A is constructed, wherein m alternatives are evaluated based on n criteria. Each element of matrix A, x_ij, denotes the value of criterion i for alternative j. Here, i = 1,2,…,n and j = 1,2,…,m, and the decision matrix A is represented as shown in Equation (24).

$$\mathrm{A}=\left[\begin{array}{c}\begin{array}{cccc}{x}_{11}& x_{12}& \dots & x_{1n}\\ x_{21}& x_{22}& \dots & x_{2n}\\ .& & & .\\ .& & & .\\ & & & \\ .& & & .\\ x_{m1}& x_{m2}& \dots & x_{mn}\end{array}\end{array}\right]$$

(24)

In the second step, each x_ij value in the decision matrix is normalized within the range [0,1]. Normalization is performed using the formula provided in Equation (25). During the normalization process, each x_ij value is divided by the square root of the sum of the squares of the alternatives in the i. column, resulting in the normalized value x_ij^*.

$$x_{\mathrm{ij}}*={\displaystyle \frac{x_{ij}}{\sqrt{\sum _{j=1}^m{x}_{ij}^2}}}$$

(25)

In the subsequent step, the values of criteria to be maximized and those to be minimized are separately summed for each alternative in the decision matrix comprising normalized x_ij*’s. The sum of the maximized values is then subtracted from the sum of the minimized values to derive the ranking values for each alternative. The calculation of the resulting ranking values, $y_j$, is illustrated in Equation (26) [21].

$$y_j={\sum }_{i=1}^g{x}_{ij}^{*}-{\sum }_{i=g+1}^n{x}_{ij}^{*}$$

(26)

Here, the values of the criteria to be maximized are expressed as i = 1,2,…,g, and the values to be minimized as i = g + 1,g + 2,…,n.

Finally, ranking is performed based on the obtained $y_j$ values. The alternative with the highest $y_j$ value is considered as the best alternative, while the alternative with the lowest $y_j$ value is considered as the worst alternative.

2.3.2. Importance Coefficient Approach

In the ratio approach, it is assumed that the criteria are of equal importance during the evaluation of alternatives. However, there are instances where the criteria may not hold equal importance. In such cases, different importance values are assigned to the criteria. The initial step in implementing this approach is to construct the decision matrix A, which delineates the values of the alternatives concerning the criteria. Next, each element of matrix A is normalized within the interval [0,1]. Unlike the ratio approach, in this approach, the column values of each criterion of matrix A are multiplied by the importance values of the criterion. In the final step, the values of criteria to be maximized and those to be minimized are summed separately for each alternative. Subsequently, the sum of the maximized values is subtracted from the sum of the minimized values to derive the ranking values, $y_j^{*}$ of each alternative. Finally, ranking is performed based on the obtained $y_j^{*}$ values. The calculation of $y_j^{*}$ values obtained with importance values is illustrated in Equation (27) [21].

$$y_j^{*}={\sum }_{i=1}^g{w_{i }x}_{ij}^{*}-{\sum }_{i=g+1}^n{w_{i }x}_{ij}^{*}$$

(27)

Here, $w_i$ denotes the importance value of the criterion i.

2.3.3. Reference Point Approach

In this approach, initially, the maximum value in the column of each maximized criterion and the minimum value in the column of each minimized criterion in the normalized matrix A are designated as the “reference point”. Subsequently, these values for each column are subtracted from all values in that column, and the differences are calculated. These differences represent the distances to the reference points and are depicted as shown in Equation (28).

$$r_i-x_{ij}^{*}$$

(28)

Here, $r_i$ is defined as the reference point of criterion i.

The ranking of the alternatives is determined by applying the “Tchebycheff Min-Max Metric” formula to the new matrix containing the distances to the reference point. Equation (29) contains the Tchebycheff Min-Max Metric formula [22].

$$\underset{j}{\min }\left\{\underset{i}{\max }\left(\left|r_i-x_{ij}^{*}\right|\right)\right\}$$

(29)

According to this formula, the maximum values in each row of the matrix of differences are determined, and the alternative with the smallest value among these maximum values is ranked as the best alternative, while the alternative with the highest value is ranked as the worst alternative.

2.3.4. Full Multiplicative Form Approach

In this approach, alternatives are assessed utilizing the formulation devised by Brauers and Zavadskas. To conduct an evaluation, the values of criteria to be maximized are initially multiplied together, followed by the multiplication of values to be minimized. Following these calculations, the maximized product is divided by the minimized product to ascertain the evaluation ranking of each alternative [23]. This formula is provided in Equation (30).

$$U_j={\displaystyle \frac{A_j}{B_j}}$$

(30)

Here, $U_j$ represents the ranking value of each alternative, $A_j$ represents the product of the criteria to be maximized in the decision matrix and $B_j$ represents the product of the criteria to be minimized in the decision matrix. The calculation of $A_j$ and $B_j$ is shown in Equation (31).

$$A_j={\prod }_{i=1}^g{x}_{ij}\hspace{1em}{B}_j={\prod }_{i=g+1}^n{x}_{ij}$$

(31)

Lastly, the ranking of alternatives is determined based on their $U_j$ values. The alternative with the highest $U_j$ value is considered the best, while the one with the lowest $U_j$ value is deemed the worst.

3. Findings and Discussion

3.1. Identification of Criteria for Airport Transfer Center Selection

In this section of the study, we first outline the criteria employed in evaluating alternatives for airport hub selection using the AHP method. Subsequently, the priority values of these criteria are calculated. Seven criteria are identified based on the existing literature [24]. The determined criteria are evaluated by the insights of three experts boasting over fifteen years of professional experience. One of the designated experts is the operational team leader, one is the refueling expert and the other evaluator is the support services expert. The selected experts are selected from airports other than the airports evaluated in the study. Evaluations obtained from experts are used in the analysis by taking the arithmetic average. The identified criteria are shown in

Table 4. Identified criteria for airport hub selection.

Criteria Order	Symbol	Identified Criteria
1	CRI1	Airport costs
2	CRI2	Airport terminal and apron facilities
3	CRI3	Airport passenger transportation services, maintenance facilities
4	CRI4	Airport operating capacity
5	CRI5	Airport large and small scale location
6	CRI6	Demand drivers in the service area
7	CRI7	Other (social facilities, etc.)

.

The first criterion pertains to cost considerations, while the subsequent criteria encompass factors aimed at enhancing airline operations. Facilities such as the quantity of aprons and runways within the airport terminal are instrumental in streamlining flight planning activities for airline companies.

Similarly, passenger transportation services and aircraft maintenance facilities play a pivotal role in the planning activities of airline companies. Airport operating capacity stands as another crucial criterion favored by airline companies. Airports with limited capacity are likely to receive lower preference rates. The availability of both in-city and out-of-city transportation and transfer options enhances the attractiveness of airports. The ease with which passengers can reach their ultimate destinations post-flight serves as a significant factor influencing their preference for this mode of transportation in the future. The strategically advantageous location of the terminal for cultural or leisure trips was deemed as another criterion with potential benefits. Finally, social facilities within the terminal are expected to impact the flight plans of airlines, particularly benefiting connecting passengers [24].

3.2. Pairwise Comparison of Criteria

During the pairwise comparison of criteria, a comparison matrix was formed by considering Saaty’s recommended scale of importance. The comparison matrix is shown in

Table 5. Pairwise comparison matrix for criteria.

Pairwise Comparison Matrix for Criteria	CRI1	CRI2	CRI3	CRI4	CRI5	CRI6	CRI7
CRI1	1.000	4.000	3.000	6.000	4.000	7.000	7.000
CRI2	0.250	1.000	3.000	1.000	2.000	3.000	4.000
CRI3	0.333	0.333	1.000	2.000	0.500	3.000	3.000
CRI4	0.167	1.000	0.500	1.000	1.000	0.500	2.000
CRI5	0.250	0.500	2.000	1.000	1.000	3.000	3.000
CRI6	0.143	0.333	0.333	2.000	0.333	1.000	4.000
CRI7	0.143	0.250	0.333	0.500	0.333	0.250	1.000
Total	2.286	7.417	10.167	13.500	9.167	17.750	24.000

.

Subsequently, the values within this comparison matrix were normalized, and the priority values of the criteria were calculated. Normalized values and priority values of the criteria are shown in

Table 6. Inter-criteria normalization matrix and priority values.

Pairwise Comparison Matrix for Criteria	CRI1	CRI2	CRI3	CRI4	CRI5	CRI6	CRI7	Priorities
CRI1	0.438	0.539	0.295	0.444	0.436	0.394	0.292	0.406
CRI2	0.109	0.135	0.295	0.074	0.218	0.169	0.167	0.167
CRI3	0.146	0.045	0.098	0.148	0.055	0.169	0.125	0.112
CRI4	0.073	0.135	0.049	0.074	0.109	0.028	0.083	0.079
CRI5	0.109	0.067	0.197	0.074	0.109	0.169	0.125	0.122
CRI6	0.063	0.045	0.033	0.148	0.036	0.056	0.167	0.078
CRI7	0.063	0.034	0.033	0.037	0.036	0.014	0.042	0.037
Total	1.000	1.000	1.000	1.000	1.000	1.000	1.000

.

Upon examining the importance values of the criteria, it becomes evident that the most crucial criterion for determining the transfer center is “airport costs,” with a value of 0.406, followed by “airport terminal and apron facilities” with a value of 0.167. The criterion deemed least important is the “other” category, encompassing social facilities, with a value of 0.037.

Following the calculation of the priority values for the criteria, the inconsistency ratio among them was determined. This ratio is shown in

Table 7. Consistency ratio calculation between criteria.

All Priorities Matrix	λ	λmax	CI	CR
3.174	7.827	7.668	0.111	0.084
1.309	7.850
0.867	7.721
0.604	7.661
0.955	7.858
0.575	7.346
0.273	7.413

. Based on this calculation, the CR value is 0.084. Since it is less than 0.10, it can be concluded that the pairwise comparison matrix among the criteria is consistent.

3.3. Evaluation of Alternatives for Airport Transfer Center Selection

3.3.1. Identification of Alternative Airport Terminals

Türkiye boasts numerous airport terminals. Several airport terminals, including İstanbul Airport, Ankara Airport and İzmir Airport, serve as significant transfer centers. The hubs considered for the selection of the new airport hub were chosen based on criteria such as maximizing geographical coverage across Türkiye and ensuring extensive utilization within their respective regions. The chosen hub airports for the new transfer center are listed in

Table 8. Alternative airports for new airport hub selection.

Symbol	Airport Terminals Evaluated
AP1	New Çukurova Airport
AP2	Antalya Airport
AP3	Sivas Nuri Demirağ Airport
AP4	Erzurum Airport
AP5	Muğla Airport

and shown in Figure 4.

The new Çukurova Airport, constructed to enhance the capacity of Adana Airport in Adana, is scheduled to commence operations in 2024. The terminal, situated in the Tarsus district between Adana and Mersin provinces, is intended to alleviate the load on the existing Adana Airport [25].

Antalya and Muğla Airports experience a surge in flight volume, particularly during peak holiday seasons, making them pivotal hubs for increased air traffic. The remaining selected terminals, Erzurum and Sivas Nuri Demirağ Airports, are strategically positioned in the Eastern Anatolia Region of Türkiye, offering opportunities for further development. Two of the alternative airports were chosen from the Mediterranean region, one from the Aegean and two from the Eastern Anatolia Region. No selection was made for airports from the Marmara and Central Anatolia regions as they are already being used as transfer terminals.

3.3.2. Evaluation of Alternative Airport Terminals According to ELECTRE Method

The decision matrix created for the evaluation of airport terminals according to the identified criteria is shown in

Table 9. Decision matrix created for the evaluated airports.

	CRI1	CRI2	CRI3	CRI4	CRI5	CRI6	CRI7
AP1	7	7	8	8	9	9	7
AP2	9	8	7	9	9	9	8
AP3	3	5	4	6	4	3	3
AP4	2	3	2	4	3	5	3
AP5	8	6	5	7	7	8	8

. In the practical implementation, experts provided scores ranging from 1 to 10 for each criterion of every alternative.

The values in the decision matrix presented in Table 9 represent the criteria to be maximized, as all criteria are assessed based on their importance values and then normalized according to the formula outlined in Equation (9). The normalized decision matrix is shown in

Table 10. Normalized decision matrix created for the evaluated airports.

	CRI1	CRI2	CRI3	CRI4	CRI5	CRI6	CRI7
AP1	0.4865	0.5175	0.6364	0.5101	0.5859	0.5582	0.5013
AP2	0.6255	0.5914	0.5569	0.5738	0.5859	0.5582	0.5729
AP3	0.2085	0.3696	0.3182	0.3825	0.2604	0.1861	0.2148
AP4	0.1390	0.2218	0.1591	0.2550	0.1953	0.3101	0.2148
AP5	0.5560	0.4435	0.3978	0.4463	0.4557	0.4961	0.5729
Weights	0.406	0.167	0.112	0.079	0.122	0.078	0.037

.

Subsequently, the weighted normalized decision matrix was derived by multiplying the values in each criterion’s column of the normalized decision matrix by the respective weights of the criteria calculated using the AHP method, as outlined in Table 6. The weighted normalized decision matrix is shown in

Table 11. Weighted normalized decision matrix created for the evaluated airports.

	CRI1	CRI2	CRI3	CRI4	CRI5	CRI6	CRI7
AP1	0.1975	0.0864	0.0713	0.0403	0.0715	0.0435	0.0185
AP2	0.2540	0.0988	0.0624	0.0453	0.0715	0.0435	0.0212
AP3	0.0847	0.0617	0.0356	0.0302	0.0318	0.0145	0.0079
AP4	0.0564	0.0370	0.0178	0.0201	0.0238	0.0242	0.0079
AP5	0.2258	0.0741	0.0446	0.0353	0.0556	0.0387	0.0212
Weights	0.406	0.167	0.112	0.079	0.122	0.078	0.037

.

Following the establishment of the weighted normalized decision matrix, pairwise comparisons of the alternatives were conducted to form concordance and discordance sets. Once the concordance (C (X,Y)) and discordance (D (X,Y)) sets were identified, concordance (C) and discordance (D) indexes were computed. Subsequently, outranking comparisons were made for all pairs of alternative airports based on these indexes. The results of the concordance and discordance indexes and outranking matrices are shown in

Table 12. Comparison of the outrankings for the evaluated airports.

C (X,Y)	C	C > CAVG	D (X,Y)	D	D < DAVG	RESULT
1,2	0.312	0	1,2	0.8956	0	0	NO
1,3	1.001	1	1,3	0.0000	1	1	YES
1,4	1.001	1	1,4	0.0000	1	1	YES
1,5	0.558	1	1,5	0.3226	1	1	YES
2,1	0.889	1	2,1	0.1044	1	1	YES
2,3	1.001	1	2,3	0.0000	1	1	YES
2,4	1.001	1	2,4	0.0000	1	1	YES
2,5	1.001	1	2,5	0.0000	1	1	YES
3,1	0	0	3,1	1.0000	0	0	NO
3,2	0	0	3,2	1.0000	0	0	NO
3,4	0.923	1	3,4	0.0983	1	1	YES
3,5	0	0	3,5	1.0000	0	0	NO
4,1	0	0	4,1	1.0000	0	0	NO
4,2	0	0	4,2	1.0000	0	0	NO
4,3	0.115	0	4,3	0.9017	0	0	NO
4,5	0	0	4,5	1.0000	0	0	NO
5,1	0.443	0	5,1	0.6774	0	0	NO
5,2	0.037	0	5,2	1.0000	0	0	NO
5,3	1.001	1	5,3	0.0000	1	1	YES
5,4	1.001	1	5,4	0.0000	1	1	YES

.

In the last step, net concordance and discordance indexes were calculated to establish the ranking of alternatives, from the most favorable to the least favorable, according to these index values.

Table 13. Net concordance and discordance values and ranking for evaluated airports.

	CP	DP	CP	DP
AP1	1.54	−1.5636	2	2
AP2	3.543	−3.7913	1	1
AP3	−2.195	2.1966	4	4
AP4	−3.811	3.8034	5	5
AP5	0.923	−0.6451	3	3

presents the net concordance (CP) and discordance (DP) values, as well as the ranking of alternative airports based on the ELECTRE method.

Based on the ELECTRE method, the prioritization is as follows: Antalya Airport is ranked first, followed by Çukurova Airport (to be built in Tarsus district of Mersin province) in second place, Muğla Airport in third place, Sivas Nuri Demirağ Airport in fourth place and finally Erzurum Airport in fifth place.

3.3.3. Evaluation of Alternative Airport Terminals According to MOORA Method

The decision matrix for evaluating alternative airports according to the MOORA Ratio Approach is displayed in

Table 14. Decision matrix created for airport hub selection.

Criteria		CRI2	CRI3	CRI4	CRI5	CRI6	CRI7	CRI1
Criteria Directions		Max	Max	Max	Max	Max	Max	Min
Importance Coefficients (wj)		0.167	0.112	0.079	0.122	0.078	0.037	0.406
Airports	AP1	7	8	8	9	9	7	7
	AP2	8	7	9	9	9	8	9
	AP3	5	4	6	4	3	3	3
	AP4	3	2	4	3	5	3	2
	AP5	6	5	7	7	8	8	8

.

Following normalization using the formula in Equation (25), the values in

Table 15. Normalized decision matrix for airport hub selection.

Criteria		CRI2	CRI3	CRI4	CRI5	CRI6	CRI7	CRI1
Criteria Directions		Max	Max	Max	Max	Max	Max	Min
Airports	AP1	0.5175	0.6364	0.5101	0.5859	0.5582	0.5013	0.4865
	AP2	0.5914	0.5569	0.5738	0.5859	0.5582	0.5729	0.6255
	AP3	0.3696	0.3182	0.3825	0.2604	0.1861	0.2148	0.2085
	AP4	0.2218	0.1591	0.2550	0.1953	0.3101	0.2148	0.1390
	AP5	0.4435	0.3978	0.4463	0.4557	0.4961	0.5729	0.5560

were derived, ensuring they fall within the range of [0,1].

Ranking of Alternatives According to the MOORA Ratio Approach

After establishing the normalized decision matrix, the y_j values for each alternative airport were computed using the formula in Equation (26). These calculated values are shown in

Table 16. Ranking of airport hubs according to MOORA-Ratio Approach.

		yj	Ranking
Airports	AP1	2.8227	1
	AP2	2.8134	2
	AP3	1.5231	4
	AP4	1.2171	5
	AP5	2.2563	3

.

When arranging the obtained y_j values in descending order, it becomes apparent that according to the MOORA Ratio Approach, Çukurova Airport to be constructed in Tarsus district of Mersin province is given priority, with Antalya Airport following in second place. Muğla Airport ranks third, Sivas Nuri Demirağ Airport ranks fourth and Erzurum Airport ranks last.

Ranking of Alternatives According to MOORA Importance Coefficient Approach

In this approach, the importance coefficients (w_i) of the criteria determined by the AHP method were initially multiplied by the normalized decision matrix values in Table 15. This process yielded the values presented in

Table 17. Weighted normalized decision matrix for alternatives.

Criteria		CRI2	CRI3	CRI4	CRI5	CRI6	CRI7	CRI1
Criteria Directions		Max	Max	Max	Max	Max	Max	Min
Importance Coefficients (wi)		0.167	0.112	0.079	0.122	0.078	0.037	0.406
Airports	AP1	0.0863	0.0714	0.0402	0.0712	0.0437	0.0185	0.1973
	AP2	0.0986	0.0625	0.0452	0.0712	0.0437	0.0211	0.2537
	AP3	0.0616	0.0357	0.0301	0.0316	0.0146	0.0079	0.0846
	AP4	0.0370	0.0179	0.0201	0.0237	0.0243	0.0079	0.0564
	AP5	0.0740	0.0447	0.0352	0.0554	0.0388	0.0211	0.2255

.

The calculation of the y_j values for the airports was conducted using the weighted normalized decision matrix, as depicted in

Table 18. Ranking of airport hubs according to MOORA Importance Coefficient Approach.

		yj	Ranking
Airports	AP1	0.1340	1
	AP2	0.0887	3
	AP3	0.0971	2
	AP4	0.0745	4
	AP5	0.0436	5

. When arranging the obtained y_j values in line with the Importance Coefficient approach, it becomes apparent that according to the MOORA Ratio Approach, Çukurova Airport is given priority, with Sivas Nuri Demirağ Airport following in second place. Antalya Airport ranks third, Erzurum Airport ranks fourth and Muğla Airport ranks last.

Ranking of Alternatives According to MOORA Reference Point Approach

Using this approach, the R_j values were determined for each column in the normalized decision matrix values, as presented in Table 15. Next, the differences between the R_j values for each value in its respective column were calculated as presented in

Table 19. Matrix of distances to the reference point for alternatives.

Criteria		CRI2	CRI3	CRI4	CRI5	CRI6	CRI7	CRI1
Criteria Directions		Max	Max	Max	Max	Max	Max	Min
Rj Values		0.5914	0.6364	0.5738	0.5859	0.5582	0.5729	0.1390
Airports	AP1	0.0739	0.0000	0.0638	0.0000	0.0000	0.0716	0.3475
	AP2	0.0000	0.0796	0.0000	0.0000	0.0000	0.0000	0.4865
	AP3	0.2218	0.3182	0.1913	0.3255	0.3721	0.3581	0.0695
	AP4	0.3696	0.4773	0.3188	0.3906	0.2481	0.3581	0.0000
	AP5	0.1478	0.2387	0.1275	0.1302	0.0620	0.0000	0.4170

.

In the matrix from Table 19, the maximum values (p_j) in each row, corresponding to each alternative, were identified and ranked from smallest to largest. The ranking is shown in

Table 20. Ranking of airport hubs according to the MOORA Reference Point Approach.

		pj	Ranking
Airports	AP1	0.3475	1
	AP2	0.4865	5
	AP3	0.3721	2
	AP4	0.4773	4
	AP5	0.4170	3

.

When arranging the obtained y_j values in line with the Reference Point Approach, it becomes apparent that according to the MOORA Ratio Approach, Çukurova Airport is given priority, with Sivas Nuri Demirağ Airport following in second place. Muğla Airport ranks third, Erzurum Airport ranks fourth and finally Antalya Airport ranks last.

Ranking of Alternatives According to MOORA Full Multiplicative Form Approach

Using this approach, the U_j values were calculated based on the decision matrix presented in Table 15, as illustrated in

Table 21. Ranking of airport hubs according to MOORA Full Multiplicative Form Approach.

		Aj	Bj	Uj	Ranking
Airports	AP1	254,016	7	36,288	1
	AP2	326,592	9	36,288	1
	AP3	4320	3	1440	4
	AP4	1080	2	540	5
	AP5	94,080	8	11,760	3

. The alternative with the highest U_j value was identified as both Çukurova Airport and Antalya Airport. Muğla Airport secured the third position, followed by Sivas Nuri Demirağ Airport in fourth place. Erzurum Airport, with the smallest U_j value, was ranked last.

In addition, when making pairwise comparisons between criteria, evaluations may differ from person to person or the opinions of the evaluators may change over time. For this reason, examining the changes in the weights of the criteria and the possible effects from the perspective of decision makers enables decisions to be made more effectively. By evaluating different scenarios, one can be sure about the stability of the methods according to the ranking changes. For this reason, in the last part of the study, sensitivity analysis was carried out to see how the ranking results changed with the change of criterion weights in airport transfer center selection. The aim is to handle variability in rankings according to methods. The current weights and the weights of seven different scenarios (Scen1, Scen2, Scen3, Scen4, Scen5, Scen6 and Scen7) are shown in

Table 22. Scenarios determined for sensitivity analysis.

	CRI1	CRI2	CRI3	CRI4	CRI5	CRI6	CRI7
Current Weights	0.406	0.167	0.112	0.079	0.122	0.078	0.037
Scen1	0.037	0.078	0.122	0.079	0.112	0.167	0.406
Scen2	0.143	0.143	0.143	0.143	0.143	0.143	0.143
Scen3	0.037	0.167	0.112	0.079	0.122	0.078	0.406
Scen4	0.037	0.078	0.022	0.012	0.079	0.067	0.706
Scen5	0.037	0.078	0.022	0.012	0.379	0.267	0.206
Scen6	0.179	0.167	0.106	0.137	0.178	0.122	0.112
Scen7	0	0.167	0.106	0.137	0.178	0.233	0.179

. In line with each scenario, the change in the ranking of alternative airports was observed for both the ELECTRE method and the MOORA Importance Coefficient method, which takes into account the criterion weights.

The change in the rankings as a result of the evaluation made for the ELECTRE method according to seven different scenarios is shown in Figure 5.

When the sensitivity analysis for the ELECTRE method was examined, it was observed that the rankings of the alternatives did not change much with the change in criterion weights.

The change in the rankings as a result of the evaluation made for the MOORA Importance Coefficient Approach according to seven different scenarios is shown in Figure 6.

When the sensitivity analysis for the MOORA Importance Coefficient Approach is examined, it is observed that both Çukurova Airport and Antalya Airport are in the first two ranks with the change in criterion weights, and the ranks of the other airports have not changed much. With the sensitivity analysis, it was observed that the results of both methods were stable.

As the demand for air transportation is increasing day by day, we think that it would be beneficial to determine a new hub for Türkiye and we think that the results of this study are satisfactory. Çukurova Airport is located in a location with land, air and sea transportation connections. Çukurova Airport, which has more capacity than the airport that previously served the region, ranks first in the study results. It is thought that Çukurova Airport will make significant contributions to the region’s foreign trade and logistics sector and will play a very important role in regional development. Antalya province is important for the tourism policies of the country where supply and demand are developed in terms of accommodation sector. For this reason, the fact that Antalya Airport is another airport that can be preferred as a transit center in the results strengthens the results of our study.

4. Conclusions and Recommendations

In recent years, there has been a significant surge in air transport, driven by advancements in transportation technology and shifts in travel preferences. Türkiye in particular holds a prominent position in both national and international air transport, owing to its strategic geographical location and cutting-edge aviation technologies. Türkiye boasts numerous airport terminals, facilitating the transportation of passengers and cargo to various destinations worldwide. This not only fuels Türkiye’s economic growth but also fosters trade development, thereby contributing to the advancement of other nations. Türkiye stands as an advanced nation in short-, medium- and long-distance transportation, particularly with its array of transfer centers interconnected with various airports. The selection of new hubs holds paramount importance in advancing the overall transportation network, aiming to gain a competitive edge and foster economic development. Numerous studies have delved into pinpointing transfer centers, employing a variety of methodologies to ascertain the most fitting alternative locations. This study aimed to ascertain the significance of existing airports in Türkiye and the forthcoming airport terminal set to function as a transfer center. The initial phase of the study employed the AHP method to prioritize the criteria pivotal in the selection of the airport transfer center, subsequently computing their respective priority weights. Subsequently, the ELECTRE and MOORA methods were applied to assess the significance of alternative airport terminals in their selection as hubs. The rankings obtained from the ELECTRE and MOORA methods are consolidated in

Table 23. Ranking of airport terminals according to ELECTRE and MOORA methods.

ELECTRE Method	MOORA Methods
	Ratio Approach	Importance Coefficient Approach	Reference Point Approach	Full Multiplicative Form Approach
AP2	AP1	AP1	AP1	AP1	AP2
AP1	AP2	AP3	AP5	AP2	AP1
AP4	AP4	AP2	AP2	AP4	AP4
AP5	AP5	AP4	AP4	AP5	AP5
AP3	AP3	AP5	AP3	AP3	AP3

and Figure 7.

Table 22 illustrates that both the New Çukurova Airport and Antalya Airport emerge as promising options for hub selection. Following these airports, Erzurum Airport is positioned as the next preferable choice, while Sivas Nuri Demirağ Airport concludes the evaluation in the last place among the terminals assessed. This study considered various criteria and employed two distinct evaluation methods to assess airport terminals, ultimately ranking the most suitable alternatives. Future research endeavors could explore diverse methodologies, alternatives and criteria for selecting airport hubs, shedding further light on their significance within Türkiye.