Ranking conceptual locations for a park-and-ride parking lot using EDAS method

The decision regarding the location of a Park-and-Ride (P&R) parking lot must be taken by simultaneously considering economic, social, urban, environmental, and other factors. This task is simplified by selection of three conceptual locations for P&R parking lots near the Vilnius western bypass. The aim of the paper is to identify key criteria that promote successful functioning of the private and public transport systems, and to rank these conceptual locations using the Evaluation based on Distance from Average Solution (EDAS) multiple-criteria decision-making method.


Introduction
The recent decade saw a high increase in concerns regarding sustainability of transport. As congestion ranks among the most important problems in medium-sized and large cities [1], various means of mass transit, and especially the Park-and-Ride (P&R) system, have become one of the most discussed sustainable transport alternatives. Park and ride facilities offer the possibility to access a city centre using public transport. It can be a bus, bus rapid transit, rail, or metro. It is regarded as a new travel mode that attracts private car users to use public transport. The P&R system can be explained in very simple terms: people use their private cars to drive from their place of residence to a P&R facility, park there, switch to public transport, and reach their destination. This system allows users to avoid highly congested city areas, and to use private car in the least congested areas only, which results in travel time and cost savings. Besides, the city becomes less congested as transit systems become more popular. In their research covering 25 European cities, Dutch researchers confirm that the development of the P&R option has reduced congestion and air pollution [2]. Locating P&R facilities is a complex task involving multiple factors, such as characteristics of a P&R facility, the level of service (LOS) of the transit system, the expected demand and market potential, the impact on the surrounding area, the transport policy (tolls, restricted private car areas, etc.), and user benefits and costs. Also, urban areas have limited location alternatives. The research reported in this paper relies on the expert-based Kendall evaluation for identifying priority of the most important criteria that determine the P&R location near city bypasses/ ring roads, and uses the multiple-criteria decision-making methodology for suggesting the ranking of potential locations for the P&R parking lot. This paper is organised as follows. Section 2 discusses literature on different approaches for finding an optimum location for a P&R facility. Section 3 gives an overview of well-known P&R facilities located near city bypasses\ring roads. The research methodology is described in Section 4. Section 5 presents the multiple-criteria evaluation of parking lot locations. Finally, research conclusions are discussed in Section 6.

Literature review
Many researchers have studied decisions determining P&R locations and have identified various possible approaches in the process. for example, Horner and Grubesic [3] suggest the Geographic Information System (GIS), Faghri et al. [4] propose an expert system, Abdul Hamid et al. [5] use the demand analysis, and Farhan and Murray [6] offer optimisation formulations. In addition, Holguí n-Veras et al. [7] use the maximization of potential demand, Khakbaz et al. [8], Horner and Groves [9] place emphasis on maximization of flow capture, while Wang et al. [10] focus on profit maximization and social cost minimization.
However, one of the most important tasks in determining P&R locations is to find the most rational way to maximize potential benefits of the P&R system [11][12][13]. P&R is a part of the mode choice, and it is one of the most complex decision processes in demand modelling, as it depends on the attributes of travellers, trip characteristics, and performance of competing modes [14]. A lot of P&R location guidelines, formulated based on case studies, are also available, but their suggestions are rather confusing and inconsistent. It is therefore not surprising that most P&R facilities are inadequately positioned [2]. Also, all these location-based methods lead to a potentially large number of alternative sites. Having in mind that in most cases the public sector investment in P&R facilities is relatively restricted, it is better to use practical rules to make location decisions. It appears that the expert-based method is the most suitable. It is relatively fast and accurate compared to other methods. Farhan and Murray [6] highlight at least three important concerns: user coverage, the effect of existing facilities, and closeness to major roadways. Based on location criterion, Mingardo [15] distinguishes three categories of P&R: -Remote P&R, which aim to intercept drivers at the beginning of their commutes, usually located in suburban areas near user homes.
-Peripheral P&R, which aim to intercept drivers just before their final destination, usually located on the urban periphery. This model is typical of the UK and the US.
-Local P&R, which aim to intercept drivers somewhere along their trip, usually located along main transport corridors.
City bypasses/ring roads are common locations of P&R parking lots. Because city bypasses are located outside of a city and act as borders between uncongested and congested roads, they are in a perfect location to intercept drivers who want to access or leave the city quickly and without any stress. In this research, the authors consider the most important criteria that determine the location for a P&R parking lot near city bypasses/ring roads. The research is based on the idea that a group of experts evaluates all the criteria, after which the results are processed and are then used for multiple criteria decision-making.

Overview of foreign practice
Good practice can be found in the UK, where the P&R system has been in use for more than half a century, and where new P&R parking lots are still being created. A park-and-ride solution was first applied some 50 years ago by small historic cities. Back then, the UK saw a significant growth in the busbased P&R schemes. Now, there are over 130 interchange sites operating and serving more than 60 towns and cities across the Great Britain. All P&R sites share some similarities. First, they are implemented by local authorities and are considered to be a part of local transport strategies. Second, sites are found around four km from the urban core, close to main access corridors. Third, sites are designed to attract the motorists with pleasant GRAĐEVINAR 70 (2018) 11, 975-983 Ranking conceptual locations for a park-and-ride parking lot using EDAS method surroundings and often on-site facilities, such as waiting areas. Fourth, high-quality buses operate at high frequency, generally every 8 and 15 minutes in peak periods. Finally, only one or two stops are usually made to minimise journey time [16]. The UK started experimenting with P&R facilities since the 1960s. The majority of P&R schemes failed. Success was enjoyed mostly by historic cities, such as Bath, Chester, Shrewsbury, and Oxford which, by their nature, had constrained opportunities for road access and parking development in city centres [17]. Parking lots were usually located in urban periphery and corresponded to the Mingardo's second category of the P&R location. There are six P&R sites in Oxford: four of them are located around the Oxford ring road, one of them is at a close distance, and the last one is far from the Oxford ring road, near the Bicester Shopping Village, which is not included in our research as a ring road P&R. The remaining five locations of P&R parking lots are presented in Figure 1. The Thornhill P&R is located off the A40 and M40, to the east of Oxford, about 15 minutes away from the Oxford city centre. The Redbridge P&R is located in the south, near the A34, about 10 minutes from the city centre. Seacourt P&R is located off the A420 from Swindon and Bath, to the west of Oxford on the Botley Road. The Pear Tree P&R is located in the north of Oxford accessed from the A40, A4260, A34 and M40. It is 12 minutes away from the city centre. The Water Eaton P&R is located to the north of Oxford, 12 minutes away from the town, and can be accessed from the A40, A34 and A4260. They provide regular bus services to the city centre, and from some car parks, there are bus routes to hospitals.

Figure 1. Locations of P&R parking lots in Oxford
The website of the Oxfordshire county council [18] shows realtime occupancy of Oxford's P&R parking lots. Based on this information, data were gathered to make the diagram of the daily use of P&R parking lots during workdays. The occupancy of P&R parking lots was registered on 7 June 2016 (Tuesday), from 5:30 to 21:30. The data are presented in Figure 2. The most successful P&R site is Seacourt with the occupancy ratio of 93%. There are multiple factors that could determine these good results, and one of them could be that P&R sites located on the inner ring road tend to get better results than others as they represent the breaking points. At these breaking points, the private transport starts to lose its benefits to public transport. In other words, the private transport becomes unattractive as the travel speed decreases compared to that of the public transport; besides, fewer parking places are available. It was established that P&R parking lots located near city ring roads or bypasses definitely bring good results in terms of the occupancy ratio.

Evaluation of development of P&R parking lots near a city bypass: the multiple-criteria decision-making method
Multi-criteria decision-making methods are designed for complex tasks. Such methods are helpful when there is a need to combine objective information with subjective preferences, and when multiple decision makers are to be involved. There are two categories of evaluation methods: quantitative and qualitative. Quantitative analysis methods require large quantities of reliable data that are, in most cases, hard or even impossible to obtain due to various reasons. A qualitative analysis is based on expert judgement. In Vilnius, obtaining large quantities of data is a difficult task. Also, the number of alternatives is rather limited. Multi-criteria methods cope very well with cases of this kind. Multiple criteria problems can be solved using various single decision-making methods, both in crisp form or with fuzzy numbers [19][20][21]. A combination of several methods [22] or hybrid methods can also be used [23]. One of the most common complex decision-making methods is the Analytic Hierarchy Process (AHP). This method was adopted by Lanović and Krasić in their P&R facility planning work [24]. In this research, the authors use a new method introduced in 2015 by Keshavarz Ghorabaee et al. This new method is known as the Evaluation based on Distance from Average Solution (EDAS) [25]. The GRAĐEVINAR 70 (2018) 11, 975-983 Andrius Barauskas, Konstantinas Jakovlevas-Mateckis, Vytautas Palevičius, Jurgita Antuchevičienė EDAS method uses an average solution for appraising the alternatives, considering the positive distance from the average, and the negative distance from the average. This method is very useful when conflicting criteria have to be considered. As has been found and claimed by the authors of the method [25], the EDAS method is stable when various criteria weights are used, and is consistent with other methods. In addition, the simplicity and faster computation are advantages of the proposed method, especially as these advantages do not affect accuracy of calculation. The steps for using the proposed method are presented as follows [25]: STEP 1: Construction of the decision-making matrix (X). The method evaluates the decision matrix X, which refers to n alternatives and m criteria: where X ij is the performance of the i-th alternative with regard to the j-th criterion.
If j-th criterion is beneficial: If j-th criterion is non-beneficial: STEP 4: Determination of weighted sum for PDA and NDA: where w j is the weight of the j-th criterion.
STEP 5: Normalization of SP and SN: STEP 6: Calculation of appraisal score (AS): where 0 ≤ AS i ≤ 1. Information system at the P&R parking lot (number of free spaces, schedule of public transport, other information) x 10 Free parking lot places at the city centre Ranking conceptual locations for a park-and-ride parking lot using EDAS method STEP 7: Ranking alternatives by decreasing values of appraisal score (AS).
The alternative with the highest appraisal score is the best choice. To use the EDAS multi-criteria decision-making method, necessary tasks are completed in the next section, and the evaluation results are presented in the final subsection.

Sampling criteria for selecting P&R parking lot location near city bypass
Several criteria of crucial significance for defining location of the P&R parking lot near the city bypass are determined in the first phase. Various criteria such as the economics, urban planning, social, environmental, and other aspects, are considered in this phase. The final list of criteria was created by the group of three experts in consultation with the local community and authorities concerned with this development. The criteria are presented in Table 1. Out of these ten criteria, six are quantitative (x 1-5 , x 10 ) and the remaining four are qualitative (x 6-9 ). Qualitative criteria are estimated using the Likert item scale ( Table 2). Detailed information about the criteria: x 1 : land price (€/1000 m 2 ) and construction cost (€/1000 m 2 ) [thousands of €], average prices in 2016, considered for the conceptual locations of the P&R parking lot x 2 -number of public transport routes near the P&R parking lot during the peak hour [number of routes/peak hour] x 3 : traffic flow volume at the bypass next to the Park & Ride parking lot (at the moment, some parts of the Vilnius western bypass are not finished, therefore, the modelled traffic volume data were used) [veh/peak hour] x 4 : distance to the city centre (central post office) from the P&R parking lot [km] x 5 : total price when using the combination of the P&R parking lot and public transport [€], assuming that leaving a car at the P&R parking lot costs EUR 2 for 11 hours and the public transport ticket is EUR 1; In the 1 st and 2 nd concept, P&R parking lots are considered to be free for of charge, and so the public transport ticket is charged only, and the 3 rd concept is fully priced, i.e. EUR 2 for a parking space and EUR 1 for public transport (this price policy is inversely proportional to the price of each concept of the P&R parking lot, including the price of the land). , assuming that the conceptual P&R parking lots located near the main roads connecting major Lithuania cities would be more attractive for users coming from other major cities (the 3 rd concept of the P&R parking lot is the most attractive because it is located near the road connecting the main Lithuanian cities, i.e. Klaipėda, Kaunas and Vilnius; the 1 st concept is less attractive than the 3 rd as it connects Šiauliai, Panevėžys, and Vilnius; and the 2 nd concept does not connect any of the main cities, and so it is considered unattractive) x 8 : benefits of the public transport compared to the use of a private car (faster, more comfortable, no need to search for a parking space, etc.) [1 -very weak, 2 -weak, 3 -medium, 4 -strong, 5 -very strong], assuming that the best results are delivered by the fastest public transport arteries, -the 1 st and 2 nd concepts for the location of the P&R parking lot are on rapid public transport arteries; and the 3 rd one is on an average speed artery x 9 :information system at the P&R parking lot, and attractiveness of its location from the visibility perspective from the bypass or/and the main road [1 -very unattractive, 2 -unattractive, 3 -neutral, 4attractive, 5 -very attractive], all locations are considered attractive in terms of visibility from the bypass or/and the main road x 10 : free parking lot places driving to the city centre [the number], assuming that the lack of free parking at the city centre would encourage choosing the P&R instead.

Determining significance of criteria for location of &R parking lot near city bypass
The final ten criteria were presented to experts. The expert group consisted of 21 experts, out of which six from Slovenia and the remaining ones from Lithuania. The experts were construction engineering, transport engineering, civil engineering, road engineering, energy and environmental engineering scientists and specialists. They were selected according to their work experience (no less than ten years), position at their workplace (division level and higher), and degree (Master's degree and higher). A variety of methods have been developed for expert evaluations. The criteria ranking method, proposed by Kendall [26], ranks among the most popular methods. Ranking is a procedure in which an expert gives the highest ranking and the score of one to the most important criteria; the second highest criteria are given the score of 2; the third the score of 3, and so on. This method is very simple and can easily be applied in practical calculations [27]. According to Kendall, the concordance coefficient is associated with ratings of all respondents or experts for criteria in respect j = 1, 2, ..., m (15) and the average for criteria is obtained by dividing the sum of all ranks by the number of criteria: (16) where: R ij -the i-th expert for the j-th criteria given rank, n -the number of experts (i = 1, 2, ... , n), m -the number of criteria (j = 1, 2, ..., m).
Relative significance of criteria is calculated as follows: The concordance coefficient W is calculated according to the formula: The sum of the square R j of the deviation from the total average of values S is defined by the formula: The significance of the concordance coefficient and the compatibility of the group of criteria by expert evaluation is determined by : The minimum value of the concordance coefficient W min is calculated using the formula: (21) where: is the critical Pearson statistics, the value of which is found in Table [ 26], taking into account the degree of freedom v = m -1 and the level of significance α. The concordance coefficient W=0.404 was calculated. Based on Eq. (20), the calculated value = 90.81 exceeds the critical value with the significance level amounting to α =0.05. Results obtained by calculation show that expert judgements are consistent. The expert survey analysis showed that the first criterion (the highest importance criterion) was "Frequency of public transport at the Park & Ride parking lot" (0.150). The second most important criterion was "Benefits of the public transport compared to the use of a private car (faster, more comfortable, no need to search for a parking space, etc.)" (0.136), and the third one was "Free parking lot places at the city centre" (0.129). The least important criteria were: "Private investments in the Park & Ride parking lot (shopping malls, services, etc.)" (0.046) and "Promotion by public institutions to use the Park & Ride parking lot when taking intercity travels" (0.061). All criteria weights are presented in Figure 3.

Alternatives and criteria for solving problem of P&R location near city bypass
Three conceptual alternatives for the location of the P&R parking lot near the Vilnius western bypass have been chosen (Figure 4). The first conceptual location is near the main road A2 (Ukmergės Street) and at the end of the western bypass. The second conceptual location is at the intersection of Pilaitės Avenue and the western bypass, and the third one -at the intersection of Oslo  Table 3.

Calculation results
According to calculation results received by applying the EDAS method (Eq. 1-14), the conceptual alternatives for the P&R parking lot are ranked as follows: α 2 first place, α 3 -second place, and α 1 -third place (Table 4 and Table 5).
In this case study, the best alternative is the second one (conceptual location of the P&R parking lot, at the intersection of Pilaitės Avenue and the Western Bypass) and the least favourable alternative is the first one (conceptual location of the P&R parking lot, near the main road A2 (Ukmergės Street) and the end of the western bypass). The best alternative is way ahead of the other two suggested conceptual locations of the P&R parking lot, as appraisal scores were 0.833 > 0.569 > 0.378.

Conclusions
The analysis of scientific literature shows that different approaches are currently used for finding the most suitable locations for P&R parking lots. However, most of them could lead to a potentially large number of alternative locations. Practical rules, working examples of P&R parking lots, expert experience and knowledge, and multi-criteria approach, could be adopted to make a quick decision regarding the location of a P&R parking lot. A multiple criteria approach is proposed in the paper for analysing the problem regarding location of a P&R parking lot. The problem was evaluated using the multiple-criteria decisionmaking method EDAS to rank conceptual locations for P&R parking lots, as suggested by experts. The results show that this method for the development of P&R parking lots is easy to perform, does not require a lot of hardly accessible data, and could be easily adopted in other cities despite the differences in size, sprawl, population, or transport systems. The most import guidelines for applying this method involve the use of practical rules for selecting locations for a P&R parking lot, collection of main criteria regarding these concept locations, and processing all data by means of the multiple criteria approach. The analysis of the expert survey shows that the most important criterion was "Frequency of public transport at the Park & Ride parking lot" (0.150), the second important criterion was "Benefits of the public transport compared to the use of a private car (faster, more comfortable, no need to search for a parking space, etc.)" (0.136), and the third one was "Free parking lot places at the city centre" (0.129). It is clear that, to be attractive to its users, the transit system must offer better quality than that offered by a private car, and the city must be unattractive to private cars. To increase the modal split of the transit system, these main criteria must be considered during development of the urban transport system, starting from the political level to engineering. In the case study, conceptual locations/alternatives of the P&R parking lot were ranked as follows (starting from the most favourable one): the second concept (at the intersection of Pilaitės Avenue and the western bypass), the third concept (at the intersection of Oslo Street and the beginning of the western bypass), the first concept (near the main road A2 (Ukmergės Street) and the end of the western bypass). Appraisal scores (AS) were as follow: 0.883 (2 nd alternative), 0.569 (3 rd alternative), and 0.378 (1 st alternative). A large difference between AS shows that the multiple-criteria approach is much more reliable than the basic expert decision, which was performed in the first place to choose between equally possible conceptual alternatives of the P&R parking lot near the city bypass. The same concept of the location decision method could be used for solving other urban or transport planning problems involving location decisions.