Bike sharing station site selection for Gaziantep

1.1. Literature Review

Although the bike sharing program is a well-known topic nowadays, there is relatively little literature published on the strategic design of bicycle sharing systems. In this section we examine the “Bike Sharing Station Siting” papers that apply different solution techniques. The deployment of an interactive campus bicycle-parking map is proposed by Leigh et al. [10] for application on the Monash University campuses. Later, Lin and Yang [11] address the strategic planning of public bicycle sharing systems with service level considerations. The proposed model attempts to determine the number and locations of bike stations. A preference survey for estimating potential users and a model to determine the optimal area for locating bike rental stations by using GIS is proposed by Dell'Olio et al. [12]. Another GIS-based method is proposed by Palomares et al. [13] to calculate the spatial distribution of potential demand for public bicycle trips and to determine the potential area for locating bike stations. Then Larsen et al. [14] propose a GIS-based grid cell method for bicycle facility prioritization and location. A strategic design problem for bicycle sharing systems is proposed by Lin et al. [1] that also address incorporating bicycle stock considerations. In another interesting paper, Croci and Rossi [15] assess which attractors influence the use of bike sharing stations. This issue is very important when deciding the bike sharing station siting. Their main results suggest that the presence of metro and train stations, universities, museums, cinema and restricted traffic areas in correspondence of bike sharing stations significantly increase use. Afterwards, Liu et al. [16] propose a bike sharing network optimization approach by considering multiple influential factors. They use Artificial Neural Networks for predicting station demand and balance. An optimization method is proposed by Frade and Ribeiro [17] to design the bike sharing system such that it maximizes the demand covered and takes the budget as a constraint. The model determines the optimal location of the bicycle stations. In a recent study, Wang et al. [18] determine bike sharing station spots in Taiwan by using spatial-temporal analysis. As can be seen; in the limited literature of bike sharing station siting, mostly mathematical models or geographic information systems are used. To the best of our knowledge, MCDA techniques have not been used for this problem. Thus we apply a Fuzzy AHP-TOPSIS technique to determine the most suitable places for bike sharing stations. This paper contributes to the literature in two ways: (i) proposing a three step Fuzzy AHP-TOPSIS

approach –fuzzy AHP is used to priori tize the criteria, fuzzy TOPSIS is used to rank the alternatives –, (ii) application of proposed approach to evaluate bike sharing station siting availability of each pre-determined alternatives.

2.1. Evaluation Criteria

An advisory board including a transportation engineer of Metropolitan Municipality who is responsible for bike sharing station siting and experts from Bicycle Association is constructed to determine and assess the criteria affecting the location of a bike sharing station. Also the authors are entrusted to search the literature and find the related criteria. After the interview with the engineers, experts and literature review, 13 different sub-criteria are determined under three groups (Figure 2).

• Proximity to shopping malls • Proximity to prim/secondary schools • Proximity to commercial buildings • Proximity to tourism/recreation area • Proximity to campus • Proximity to cultural elements

• Proximity to bike lane • Proximity to public bus network • Proximity to tramway network • Proximity to minibus network • Proximity to transit hubs

Figure 2. Bike sharing station siting criteria Table 1 gives information about the criteria literature, the papers on the table either used the criteria in their study or defined a relation between criteria and bike sharing station siting. In addition, Table 2 defines the criteria briefly.

Table 1. Considered criteria and relation with literature Urban Life Related Criteria Autho rs [19] [20] [21] [22] [13] [23] [15] [17] [24] [PP]

Transportation Network Related Criteria C3. C3. C3. C3. C3. 1 2 3 4 5 √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ √

Authors: [19] The NACTO Bike Share Station Siting Guide [20] Pucher et al. (2011), [21] Pucher et al. (2010 a), [22] Pucher et al. (2010 b), [13] Palomares et al. (2012), [23] Kim et al. (2012), [15] Croci and Rossi (2014), [17] Frade and Ribeiro (2015), [24] Murphy and Usher (2015), [PP] Proposed paper. Below, you can see the descriptions of criteria that are used in this study. Table 2. Description of each criterion Main and Sub-criteria C1. Urban Life Related Criteria C1.1. Proximity to shopping malls C1.2. Proximity to prim/second schools C1.3. Proximity to commercial buildings C1.4. Proximity to tourism/recreation area C1.5. Proximity to campus C1.6. Proximity to cultural elements C2.User Count Related Criteria C2.1. Population density C2.2. Proximity to young population C3. Transportation Network Criteria C3.1. Proximity to bike lane C3.2. Proximity to bus transport network C3.3. Proximity to tramway network C3.4. Proximity to minibus network C3.5. Proximity to transit hubs

Description The bike station should be near to the shopping malls The bike station should be near to primary/secondary schools The bike station should be near to commercial buildings The bike station should be near to tourism/recreation area The bike station should be near to campus The bike station should be near to cultural elements (museum, theatre, cinema) The bike station should be near to high volume of population/crowded neighborhood The bike station should be near to places that contain high percentage of young population The bike station should be near to bike lane The bike station should be near to public bus transport network The bike station should be near to tramway network The bike station should be near to minibus network The bike station should be near to transit hubs

(Although it is hard to determine the places that mostly the young population lives, there are furnished apartments and residences for young people in specific areas of Gaziantep.) There are six alternatives for Gaziantep in this study. Alternatives are determined by Gaziantep Metropolitan Municipality after a small-scale survey performed by transportation

engineers. Alternatives are spread among the city, their common feature is that; they are all crowded and they have a strong transportation infrastructure when compared to other districts of the city. The alternatives are shown on the map in Figure 3.

A6 Karataş Male’s Dormitory Figure 3. Bike sharing station alternatives 539

2.2. Methodology

After determination of the criteria, the weights of each criterion that will be used in evaluation process are assigned by using the FAHP method. In this phase, above mentioned experts are asked for forming an individual pairwise comparison matrix according to the scale given in Table 3. Superdecision software is used to create the hierarchic structure of the evaluation criteria. It is noted that the inconsistency ratio, which means the user makes the evaluations consistently, is smaller than 0.1. Geometric means of these values are found to obtain the pairwise compassion matrix on which there is a consensus. Table 3. Fuzzy comparison scale, Kabak et al. [25] Linguistic scale for importance

Triangular fuzzy reciprocal scale (1, 1, 1) (1/4, 1/3, 1/2) (1/6, 1/5, 1/4) (1/8, 1/7, 1/6) (1/10, 1/9, 1/8)

Following questions are asked to the expert team: “Which criterion is considered more important and how much more important when selecting a bike sharing station site?” The experts select one criterion and then determine the degree of importance according to Table 3. The same questions are asked for each criterion. The pairwise comparison matrices for main criteria and urban life related sub-criteria are given in Tables 4 and Table 5 with their calculated weights. All calculated scores of the criteria weights are defuzzified through COA (center of area) defuzzification method [26]. (To decrease complexity, the other analyses for sub-criteria are obtained but they are not given in the text.) Table 4. Weights and pairwise comparison matrix of main criteria l 1.00 2.00 4.00

Table 5. Local weights and pairwise comparison matrix of urban life related sub-criteria C11 C12 C13 C14 C15 C16

Weights of sub-criteria are defuzzified by COA method and crisp weights are obtained after the normalization. Final weights obtained by FAHP are shown in Table 6. According the results,

the most important factor of bike sharing station siting is “proximity to bike lane” with 0.327 as expected. Other factors can be seen in Table 6 below. Table 6. Weights of criteria obtained by fuzzy AHP Main and Sub-criteria C1. Urban Life Related Criteria C1.1. Prox. to shopping malls C1.2. Prox. to prim/second school C1.3. Prox. to commercial building C1.4. Prox. to tourism/recreation C1.5. Prox. to campus C1.6. Prox. to cultural elements C2. User Count Related Criteria C2.1. Prox. to density C2.2. Prox. to young population

Main and Sub-criteria C3. Transport Network Related Criteria C3.1. Prox. to bike lane

After determining the weights, TOPSIS method is applied to rank the alternatives. The TOPSIS method is based on the concept that the most preferred alternative should have the shortest distance from the positive ideal solution and have the longest distance from the negative ideal solution [27], [28]. The readers can find the detail formulation of applied TOPSIS method in Hwang et al. [29]. Scores and ranking are determined by using TOPSIS as shown in Table 7. Table 7. Final ranking based on Fuzzy AHP-TOPSIS A1 A2 A3 A4 A5 A6 0.107 0.131 0.09 0.105 0.057 0.063 0.071 0.062 0.07 0.064 0.141 0.094 0.601 0.679 0.564 0.622 0.289 0.401 ∗ 3 1 4 2 6 5 Rank : The separation of each alternative from the ideal solution / : The separation from the negative ideal solution / ∗: The relative closeness of the alternative with respect to According to Table 7, the best location for a bike sharing station is A2, while the worst location is the A5. Possible reasons for A2’s high ranking are its “proximity to bike lane, and proximity young population. If one location is to be selected, then A2 should be chosen because it possesses the highest Ci value. Above table is calculated by taking the weights of each criterion into consideration. If it is assumed that each criterion has equal weights we find the results as shown in Table 8. Table 8. Final ranking based on TOPSIS (Equal Weights)

: The separation of each alternative from the ideal solution / : The separation from the negative ideal solution / ∗: The relative closeness of the alternative with respect to According to Table 8, the best location for a bike sharing station is A1, while the worst location is the A5. Possible reasons for A1’s high ranking are its compatibility to all types of criteria especially the urban related criteria group. If one location is to be selected, then A1 should be chosen because it possesses the highest Ci value.