Vehicle sharing systems: A review and a holistic management framework

Ataç, Selin; Obrenović, Nikola; Bierlaire, Michel Article Vehicle sharing systems: A review and a holistic management framework EURO Journal on Transportation and Logistics (EJTL) Provided in Cooperation with: Association of European Operational Research Societies (EURO), Fribourg Suggested Citation: Ataç, Selin; Obrenović, Nikola; Bierlaire, Michel (2021) : Vehicle sharing systems: A review and a holistic management framework, EURO Journal on Transportation and Logistics (EJTL), ISSN 2192-4384, Elsevier, Amsterdam, Vol. 10, Iss. 1, pp. 1-19, https://doi.org/10.1016/j.ejtl.2021.100033 This Version is available at: https://hdl.handle.net/10419/325142 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ Vehicle sharing systems: A review and a holistic management framework Selin Ataç a , * , Nikola Obrenovi c b , Michel Bierlaire a a  Ecole Polytechnique F ed erale de Lausanne, Transport and Mobility Laboratory (TRANSP-OR), GC B2 385, 1015, Lausanne, Switzerland b University of Novi Sad, BioSense Institute, Dr Zorana Đinđi ca 1, 21000, Novi Sad, Serbia ARTICLE INFO Keywords: Vehicle sharing system Holistic management framework Literature review ABSTRACT Although different vehicle sharing systems (VSSs) use different vehicle types, the management challenges and optimization problems to be solved are similar or even the same. This observation led us to create a generalized and holistic VSS management framework, which aims to be applicable to the system using any vehicle type. The framework components, their mutual relationships, and framework tasks have been identified through a thorough and systematic literature review. Furthermore, the literature is positioned in the line with the framework. Finally, the framework and systematic literature review allowed us to identify gaps in the literature, and interesting research avenues. 1. Introduction A vehicle sharing system (VSS) offers users to rent vehicles for a short period of time. The price of the trip is generally determined according to its duration and length. The majority of existing applications of this service include car sharing systems (CSSs), electric CSSs (eCSSs), and bike sharing systems (BSSs). Some recent applications also include autonomous connected electric vehicle (ACEV)-based CSSs and electric light vehicles, i.e., personal intelligent city accessible vehicle (PICAV) sharing systems, moped sharing systems (MSSs), kick scooter sharing systems (SSSs), and light electric vehicle sharing systems (LEVSSs). This type of shared mobility is becoming more and more popular due to both financial and environmental effects. On the other hand, they face many challenges, such as inventory management of vehicles and parking spots, imbalance of vehicles, pricing strategies, and demand forecasting. If these are not addressed properly, the system may experience asignificant loss of users and therefore revenue. The idea of sharing vehicles arose in the late 1940s with cars. The first known CSS, Selbstfahrergemeinschaft, was initiated in Zurich, Switzerland, in 1948 (Shaheen et al., 1998). The idea came up again in the early 1970s. Some examples were Procotip, which was initiated in France, in 1971, and Witkar, initiated in 1973, in Amsterdam. However, these particular applications did not last long and disappeared. As for CSSs, Amsterdam also pioneered the BSSs. The first BSS was introduced in 1965 by the organization Provo. The organization distributed 50 bikes throughout the inner city and left them for free use. These bikes were identified with their white color and they were left unlocked. Despite the environmental objectives of the organization, the system was abused and the thefts and damages could not be prevented (Shaheen et al., 2010). With the technological and operational improvements, now it is easier to identify users and secure bikes. Midgley (2011) analyzes these improvements in five different generations that caused change in the operation of BSSs. Although these first initiations were mostly based on the economic and environmental reasons, profit-based companies saw the opportunity in these systems and invested money. The first sharing system that uses a vehicle different from cars and bikes started with mopeds. The company Scoot launched an MSS at the beginning of 2012 in San Francisco with 10 vehicles and expanded to 50 through the end of the year (Lawler, 2012). In 2017, the first sharing system using electric kick scooters was launched in Santa Monica, California. The company Bird announced that 10 million rides were done in the first year of application (Bird, 2018). In 2018, a start-up named ENUU launched a new type of VSS that uses light electric vehicles in Switzerland. Later, another company named Getaround deployed similar vehicles in Rotterdam (Getaround, 2019). VSSs are not only profitable from the operator perspective but also for the users since it is less costly than owning the vehicle. For instance, the costs of parking, fuel, insurance, and maintenance are all included in the price of the usage. It also allows to use the system occasionally rather than on a daily basis. According to Bates and Leibling (2012), a study concerning the UK shows that an average car is in use only 3–4% of the time. It is parked the rest of the time, i.e., 96–97%. Another study conducted by Shoup (2011) also points out that this proportion is 95%. These * Corresponding author. E-mail addresses: selin.atac@epfl.ch (S. Ataç), [email protected] (N. Obrenovi c), michel.bierlaire@epfl.ch (M. Bierlaire). Contents lists available at ScienceDirect EURO Journal on Transportation and Logistics journal homepage: www.journals.elsevier.com/euro-journal-on-transportation-and-logistics https://doi.org/10.1016/j.ejtl.2021.100033 Received 9 June 2020; Received in revised form 22 February 2021; Accepted 24 February 2021 2192-4376/©2021 Association of European Operational Research Societies (EURO). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). EURO Journal on Transportation and Logistics 10 (2021) 100033 findings further encourage users to share a vehicle for both environmental and convenience perspectives. A VSS has several kinds of configurations. These can be characterized by the type of trips offered: return trip or one-way, the imbalance management strategy: user-based, static staff-based, or dynamic staff-based, the pricing strategy: fixed or dynamic, the parking organization: station-based or free-floating. In return-trip configuration, the user is requested to return the vehicle to the pick-up location. It is not imposed in a one-way system, which is therefore more flexible for the users. One-way systems suffer from imbalance, since vehicles tend to accumulate in some popular destinations. Consequently, it generates shortage of vehicles in other locations. The operator must therefore rebalance the fleet. It can be done by providing incentives to users, i.e., user-based rebalancing, in order to reduce the imbalance. It can also be done by the staff, who can either drive each vehicle to a desired location or use trucks to move vehicles around. We refer to these trucks as rebalancing trucks and to shared vehicles as vehicles throughout this paper. Staff-based rebalancing can be further divided into two categories: static and dynamic. Static rebalancing is done when the system is closed, generally during the night, every day. Dynamic rebalancing, on the other hand, is flexible and executed throughout the day. We see two applications of dynamic rebalancing operations, i.e., offline and online. In the former, the demand knowledge is assumed to be known in advance and the rebalancing operations that will take place throughout the day is calculated once at the beginning of the day. It is not updated during the day even though more information might become available. On the other hand, online rebalancing is reactive to the updates during the day. The routing decisions can change. The pricing incentives are proposed to the users, to promote userbased rebalancing. These incentives are usually implemented using a dynamic pricing strategy. A pricing strategy is called dynamic when the price of the trip does not only depend on the characteristics of the trip (origin, destination, trip length and duration), but also on the context (time of day, status of the fleet, congestion level, etc.). If the price is based only on the trip characteristic, it is called static. Finally, a system is station-based if the vehicles must be picked up and dropped off in some designated parking areas for the VSS in particular. If it is not the case, i.e., if a vehicle can be dropped off at any public parking place, it is called free-floating. The free-floating configuration is more flexible from the user point of view yet it introduces complexity to the operator. This paper analyzes the previous works that either deal with a particular problem isolated from the other ones, or represent reviews of such papers. In the light of this extensive review, we propose a general framework characterizing the operations of a VSS. The proposed framework is built from the decision maker’s point of view and aimed to be applicable for any kind of VSS. In other words, we approach all problems in a holistic manner by identifying their mutual dependencies and relations. The literature review is then positioned according to the proposed framework to be able to identify the research gaps and future research directions in the field. In addition to that, we perform a thorough analysis of the future work suggested by the reviewed papers. For each proposed further research path, we strive to identify other works which address it. By that, additional open research questions are identified. The structure of the paper is as follows: In the light of the literature review, the proposed framework is introduced in Section 2. Then, Section 3presents an extensive literature review under framework structure and Section 4discusses the future work analysis, identified research gaps and research directions. In Section 5, we conclude the paper. 2. Proposed framework This section illustrates the proposed framework that is designed to be applicable to a VSS with any type of vehicle. The challenges faced in VSSs can be classified by their decision levels. Another relevant classification deals with the type of actor that is under analysis: the suppliers (that is, the operators who supply service) or the users (that is, the people who generate demand). Finally, the solution methods to analyze challenges involve the following three layers, i.e., data, models, and actions, which are identified as the third dimension of the framework. Therefore, we take them as further criteria for classification. As a result, we introduce three different dimensions for our framework. The main terminology used and the framework are presented in Section 2.1. Following Sections 2.2, 2.3, and 2.4 give the details about the decision levels of the framework. 2.1. Terminology and the framework The framework consists of three dimensions: decision levels, actors, and layers. The components refer to the challenges faced, data obtained, models developed and actions taken in general. The modules are either a combination of one decision level, one actor, and one layer (either models or actions), or one decision level and data layer. The components are placed in modules. There are 15 modules in this framework, illustrated in Fig. 1. The first dimension deals with the horizon of the decision levels: strategic, tactical, and operational. The strategic level corresponds to the long-term planning horizon typically equal to more than a year. The scope of the system is defined at this level. The tactical level refers to the mid-term planning horizon. The decisions made here are subject to change within 4–6 months. This level can be considered as an intermediate level which connects the strategic and operational levels. The operational level monitors the current status of the system and defines the physical actions to be taken in the short term (a day, or a couple of hours). The second dimension, actors, deals with the type of actors that are under consideration. The operators, on the supply side, organize the system and try to generate profit. The users, on the demand side, want to perform their trip in the best possible conditions in terms of travel time, cost, convenience, arrival time, etc. The third and last dimension, layers, deals with the specific methodological aspects, and has three levels: data, models, and actions. Data provide information about the system and its context. The models quantify the objectives and constraints of the system. They are necessary to handle quantities for which no data is available, in particular about the future (forecasting). Finally, the quantities provided by the data and the models are used to perform specific actions. We define two types of relations between the modules: intra-level interactions represented with white arrows and inter-level interactions represented with dashed arrows. The intra-level interaction represents the information flow between the modules within the same decision level. The inter-level interaction passes the outputs of a higher decision level to the lower one’s models layer as inputs. We discuss the various parts of the framework in the following sections. 2.2. Strategic level The data used at this level is typically aggregate, and usually static. It includes for example, the geographical location and characteristics of the city, the network of the parking spots, the economic status of the area and the distribution of welfare. An important piece of data is the aggregate historical travel demand typically in the form of origin-destination matrices, organized by mode, and time of day. Demand Demand models at the strategic level involve mode and destination choice models. They are used to take business decisions S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 2 Fig. 1. Vehicle sharing system framework. S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 3 related to (i) user segmentation, that is the identification of the segments of the population that are likely to adopt the system, (ii) the advertisement investment, and (iii) the market placement. Supply Supply models at the strategic level involve mainly a network design optimization model, possibly combined with a simulation of the demand/supply interactions. They support the decisions related to the network topology, the type of vehicles used, the type for rebalancing strategy, and the level of service offered. As part of the network design, deciding the optimal location and the size of the parking facilities is an important decision to be made to prevent both overstocking and understocking of the vehicles. These decisions are done at the beginning of the system installment in the case of station-based configurations. The situation does not change for the free-floating configuration since each parking spot can be considered as a station with only one vehicle capacity. In this layer, the simulation component helps the operator to evaluate the system and take actions accordingly. The type of rebalancing refers to the strategy used in the rebalancing operations and strongly relates to the type of vehicles used in the system. For instance, it is not efficient to use big trucks to rebalance a CSS since it is not possible to carry dozens of cars on a truck. On the other hand, for a BSS, it is generally not practical and desirable to rebalance bikes using human power since it is exhaustive. Also, a staff member cannot relocate another one. Moreover, the user-based strategies can be used to improve the balance in the system or a combination of staff-based and user-based approaches might be applied. Lastly, the level of service that the operator wants to provide to the users should be decided at this level. If the operator wants to place importance on the users, then the strategy might be increasing the number of vehicles and/or parking spots in the city. On the other hand, if the operator is more capital oriented, the investment on the physical components of the system (vehicles, parking spot, etc.) is more crucial than the level of service, which makes it a secondary objective for the operator. 2.3. Tactical level The data used at this level corresponds to mid-term time horizon and seasonal characteristics. It includes the seasonal weather forecast, precipitation, important events/festivals, and cost of the workforce. Contrarily to the strategic level, this level incorporates the disaggregate historical demand data. The disaggregate data helps the operator to develop a model that forecasts the demand in mid-term and take action for the mid-term time horizon. One should note that this disaggregate demand data can be aggregated at any level to come up with the actions. Aggregations can be performed according to any criteria such as the selected time period, origins, and destinations. Demand The inputs regarding the budget and the target audience passed from the strategic level are used as attributes for the mid-term demand forecasting. They are used to take actions related to (i) pricing strategy (fixed or dynamic) and (ii) the offers/campaigns for the users. In the case of fixed pricing, the values for pricing is also determined at this level. Supply Supply models at the tactical level involve a vehicle fleet sizing model, possibly combined with a simulation of the demand/supply interactions. They support the decisions related to the fleet size of the system, the number of staff, the time of rebalancing strategy (static or dynamic), and possible temporary stations. The operator monitors the system and adjusts the fleet size. The fleet sizing is determined at this decision level since it is not practical to change this for just a short period of time. The operator should be able to monitor and analyze such a need in a reasonable amount of time. In order to evaluate the current and/or possible future applications of strategies, such as rebalancing, reservations, and pricing, we also include a component that simulates the system. In the staticrebalancing case,theoperator should decide at whattimeof the day the rebalancing will be done whilst in dynamic case the operator also needs to decide the frequency of the rebalancing operations as well as doing it online or offline. Accordingly, the number of staff should be determined to be able to rebalance the system. Lastly, locating temporary stations for possible important events/festivals is also done here. 2.4. Operational level The data used at this level is disaggregate, and usually dynamic. It includes for example, daily weather forecast, the current parking availability, and current reservations (if applicable). The disaggregate historical travel demand data is detailed in location, time, and mode. Demand Demand models at the operational level involve short-term demand forecasting per station/zone. It is used to decide the actual pricing in the case of dynamic pricing. Supply Supply models at the operational level involve the models for parking availability and the location of the vehicles in the next time window. They support routing decisions related to the rebalancing operations and maintenance. Routing for rebalancing operations can be examined under two cases: routing for rebalancing trucks and staff relocation. Simulation component at this level helps the operator to analyze changes in the routing strategies by using the short-term data. 3. Literature review In this section, we review the literature on the VSSs from a decision making point of view. We start with some review papers providing a general idea about the main aspects discussed so far. Then, we organize our review according to the different horizons related to the planning and management of a VSS. Section 3.1 covers the strategic level, corresponding to the long-term decisions (typically more than a year). Section 3.2 deals with the tactical level, corresponding to the mid-term decisions (typically, between a month and a year). Finally, Section 3.3 covers the operational level, corresponding to the short-term decisions (typically, between a couple of hours and a day). As in any classification task, the boundaries between classes may be fuzzy. For certain papers, we identify that they span more than one decision level. Such papers are presented in multiple sections, where in each of them, the aspects of the corresponding level are discussed. We first discuss five review papers. Jorge and Correia (2013) provide a thorough review about CSSs, with special emphasis on demand estimation. In particular, they observe that linear regression is the most popular technique for demand estimation, as opposed to choice models. They also point out the fact that the simulation models developed for one-way systems are too context specific and cannot be generalized. The articles that they have considered did not include any free-floating system. Moreover, they claim that the inclusion of the other modes of transportation, such as public transportation and private car, is also necessary for the realistic scenarios. Laporte et al. (2018) propose a general literature review on sharing systems. They also include the BSSs, and different configurations of the sharing systems in the review, i.e., free-floating and station-based, and static and dynamic rebalancing. They classify the former studies by problem type and decision level and stress the fact that some combinatorial questions, such as optimal inventory level within a theoretical framework, remain to be answered. They also state that the dynamic and stochastic characteristics of these systems bring complexity which has not been addressed in the literature. Although they include both car and bike sharing systems in the review, most of the work is on the BSSs. Therefore, the latest studies for CSSs are not comprehensively discussed. This gap has been partially addressed by the review paper of Illgen and H€ ock (2019). They focus on the rebalancing problem in one-way CSSs. They organize the paper by classifying the works in terms of the methodology used to solve the rebalancing problem. They take the idea of Laporte et al. (2018) about the decision levels, and improve it by mapping the problem types to the decision levels. However, they do not map the works themselves to the decision levels. As in Jorge and Correia S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 4 (2013),Illgen and H€ ock (2019) also suggest that the stochasticity in VSSs should be examined further. In Nath and Rambha (2019), the authors focus on the BSSs of any configuration. As in Laporte et al. (2018), the review is done by classifying the former works under problem type and decision level. However, this study considers only the strategic and operational decision levels. The most recent studies are discussed and suggestions for future work are provided. These include the consideration of the effects of elastic demand for the rebalancing operations and identifying the interactions between the supply and demand rather than regarding them as two independent actors. Another recent work by He et al. (2019) considers the problem types and identify the corresponding decision level for each problem type. However, as in Nath and Rambha (2019), they do not consider the tactical level decisions. The focus is on the free-floating systems that use electric vehicles in particular, which form the latest trends in VSSs. They point out the fact that the literature still lacks some concepts for the electric vehicle usage in VSSs, such as the placement of charging stations, integration of real-time data, and the charging schedules. Finally, they reckon that with more detailed operational data, different aspects of the origin-destination (O-D) demand pattern can be observed and can lead to conclusions on the impacts on the ridership. We provide first an overview to the configurations used in the works in Table 1. The works are ordered in alphabetical order and the vehicle Table 1 Summary of the configurations. Author Year Vehicle type Objective function Trips Parking Rebalancing operation Pricing RC LOS Other N/A RT OW SB FF N S D UB F D Aguilera-García et al. 2020 MSS ✓✓✓✓ ✓ Ashqar et al. 2019 BSS ✓✓✓✓ ✓ Balac et al. 2019 CSS ✓✓✓✓✓ Barth and Todd 1999 eCSS ✓✓ ✓✓ ✓✓ ✓ Boyacıand Zografos 2019 eCSS ✓✓✓✓✓ Boyacıet al. 2015 eCSS ✓✓✓✓✓ Boyacıet al. 2017 eCSS ✓✓ ✓✓ ✓ ✓ Bruglieri et al. 2019 eCSS ✓✓✓✓✓ Caggiani et al. 2019 BSS ✓✓✓✓ ✓ Caggiani et al. 2020 BSS ✓✓✓✓ ✓ Campbell et al. 2016 BSS ✓✓✓✓ ✓ Çelebi et al. 2018 BSS ✓✓✓✓✓ Cepolina and Farina 2012 PICAV ✓✓✓✓✓ Chemla et al. 2013 BSS ✓✓✓✓✓ Chiariotti et al. 2018 BSS ✓✓✓✓✓ Ciari et al. 2013 CSS ✓✓ ✓ ✓ ✓ Clemente et al. 2017 CSS ✓✓✓✓✓ Degele et al. 2018 MSS ✓✓✓✓ ✓ Dell’Amico et al. 2014 BSS ✓✓✓✓✓ Deng and Cardin 2018 VSS ✓✓ ✓✓ ✓ ✓ Faghih-Imani et al. 2017 BSS ✓✓✓ ✓✓ Febbraro et al. 2012 CSS ✓✓✓✓✓ George and Xia 2011 VSS ✓✓✓✓✓ Ghosh et al. 2017 BSS ✓✓ ✓✓ ✓ ✓ Ghosh et al. 2016 BSS ✓✓✓✓✓ Hansen and Pantuso 2018 CSS ✓✓✓✓✓ Huang et al. 2020 eCSS ✓✓ ✓✓ ✓ ✓ Illgen and H€ ock 2018 CSS ✓✓✓✓ ✓ Jin et al. 2020 eCSS ✓✓✓✓ ✓ Jorge et al. 2014 CSS ✓✓✓✓✓ Jorge et al. 2015 CSS ✓✓✓ ✓✓ Kaspi et al. 2014 BSS ✓✓✓✓✓ Kaspi et al. 2016 BSS ✓✓✓✓✓ Kek et al. 2006 CSS ✓✓ ✓✓ ✓ ✓ Kek et al. 2009 CSS ✓✓ ✓✓ ✓ ✓ Kumar and Bierlaire 2012 eCSS ✓✓✓✓ ✓ Kutela and Teng 2019 BSS ✓✓✓✓ ✓ (continued on next page) S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 5 type used and the objective of the system is included under the columns Vehicle type and Objective function, respectively. The objectives considered in each study are characterized as (i) cost-related (RC) such as maximizing revenue/profit and minimizing the cost, (ii) maximizing the level of service (LOS), (iii) other objective types (Other), and (iv) not applicable (N/A), which denotes that the study does not define an optimization problem, nor an objective function. As mentioned in Section 1,we characterize the types of VSSs by (i) the type of trips, (ii) the parking organization, (iii) the type of rebalancing operations, and (iv) the pricing strategy. These are placed in columns of Table 1 in the same order. The Trips column is further divided into return-trip (RT) and one-way (OW). The Parking, i.e., parking organization, has two sub categories: stationbased (SB) and free-floating (FF). Under the Rebalancing operations column we have four classes: not mentioned or not applied (N), static (S), dynamic (D), and user-based (UB). Lastly, the Pricing is categorized as either fixed (F) or dynamic (D). This table is provided to the reader to give her an idea about the reviewed works distribution among different configurations as another insight into the state of the art literature. Among the papers we review, we were able to include at least one work for each type of VSS. Moreover, in some papers the work is concerned with VSSs in general, which means that they are compatible with both BSSs and CSSs. We see that works on one-way systems are more popular than the return-trip systems. The number of papers which study station-based configuration is much higher than the free-floating configuration in the study that we reviewed. We observe that the userbased rebalancing operations are fewer compared to the other strategies. This also implies the fact that dynamic pricing is not studied as much as fixed pricing. We further organize the reviewed articles according to the proposed framework (Section 1), and present this organization in Table 2. The first and the second columns give the author names and the year the study is published, respectively. The next six columns represent the two dimensions of the framework, i.e., decision levels and actors. If a paper belongs to a combination of decision level and actor, the information regarding the third dimension, i.e., layers, is provided at the intersection with the corresponding study. Here “M”represents the models module and “A”represents the actions module. Papers which contain “M”develop a model, and papers which contain “A”discuss the actions considering a case study. We have seen from the literature that the data can be collected through surveys, obtained from private companies or open source repositories, and created synthetically. Since all analyzed approaches use the data obtained via one of those aspects, the data module is not illustrated in Table 2. The last row of the table shows the total number of studies reviewed and their distribution among the two dimensions of the framework. 3.1. Strategic level The strategic level corresponds to long-term decisions, with a horizon longer than a year, say. The most salient issues discussed in the literature are related to (i) data, (ii) business model, and (iii) system design. Two types of data are crucial for the design and operations of a VSS: Table 1 (continued ) Author Year Vehicle type Objective function Trips Parking Rebalancing operation Pricing RC LOS Other N/A RT OW SB FF N S D UB F D Li et al. 2018 CSS ✓✓✓✓ ✓ Lin et al. 2018 BSS ✓✓✓✓ ✓ Liu et al. 2016 BSS ✓✓✓✓✓ Masoud et al. 2019 SSS ✓✓✓✓✓ Miao et al. 2019 ACEV-CSS ✓✓ ✓✓✓ ✓ ✓ Morton 2020 BSS ✓✓✓✓ ✓ Nair and Miller-Hooks 2011 VSS ✓✓ ✓✓ ✓ ✓ Nourinejad and Roorda 2014 CSS ✓✓✓✓✓ Nourinejad et al. 2015 CSS ✓✓✓✓✓ Pal and Zhang 2017 BSS ✓✓✓✓✓ Pfrommer et al. 2014 BSS ✓✓✓✓✓✓ Raviv et al. 2013 BSS ✓✓✓✓✓ Repoux et al. 2019 CSS ✓✓✓✓✓ Rossi et al. 2016 CSS ✓✓ ✓ ✓ ✓ ✓ Schuijbroek et al. 2017 BSS ✓✓✓✓✓ Scott and Ciuro 2019 BSS ✓✓✓✓ ✓ Shu et al. 2013 BSS ✓✓ ✓✓ ✓✓ ✓ Soriguera and Jim enez 2020 BSS ✓✓✓✓✓ Warrington and Ruchti 2019 VSS ✓✓ ✓✓✓ ✓ ✓✓ Waserhole and Jost 2012 VSS ✓✓✓✓✓ Weikl and Bogenberger 2015 CSS ✓✓✓✓✓ Wu et al. 2019a CSS ✓✓✓✓ ✓ Wu et al. 2019b BSS ✓✓ ✓ ✓ ✓ ✓ Yuan et al. 2019 BSS ✓✓✓✓✓ Zhang et al. 2019 eCSS ✓✓✓✓✓✓ Zhao et al. 2018 CSS ✓✓✓✓✓ S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 6 Table 2 Summary of the presented works. Author Year Strategic level Tactical level Operation level Supply Demand Supply Demand Supply Demand Aguilera-García et al. 2020 –MA –––– Ashqar et al. 2019 –––––M Balac et al. 2019 ––MA MA –– Barth and Todd 1999 MA ––––– Boyacıand Zografos 2019 ––––MA – Boyacıet al. 2015 MA –MA ––– Boyacıet al. 2017 ––––MA A Bruglieri et al. 2019 ––––MA – Caggiani et al. 2019 MA ––––– Caggiani et al. 2020 MA ––––– Campbell et al. 2016 –M–––– Çelebi et al. 2018 MA ––––– Cepolina and Farina 2012 ––MA ––– Chemla et al. 2013 ––––MA A Chiariotti et al. 2018 ––MA ––– Ciari et al. 2013 –M–––– Clemente et al. 2017 ––MA ––– Degele et al. 2018 –A–––– Dell’Amico et al. 2014 ––––MA – Deng and Cardin 2018 MA ––––– Faghih-Imani et al. 2017 ––––MA MA Febbraro et al. 2012 –––––MA George and Xia 2011 ––MA ––– Ghosh et al. 2017 ––A–MA – Ghosh et al. 2016 ––––MA – Hansen and Pantuso 2018 ––MA ––– Huang et al. 2020 MA –MA –MA – Illgen and H€ ock 2018 MA ––––– Jin et al. 2020 –M–––– Jorge et al. 2014 ––––M– Jorge et al. 2015 –––––MA Kaspi et al. 2014 ––MA ––– Kaspi et al. 2016 ––MA ––– Kek et al. 2006 ––––MA – Kek et al. 2009 ––––MA – Kumar and Bierlaire 2012 ––A––– Kutela and Teng 2019 –––M–– Li et al. 2018 ––AMA–– Lin et al. 2018 –––––MA Liu et al. 2016 ––––MA – Masoud et al. 2019 ––––MA – Miao et al. 2019 MA ––––– Morton 2020 –MA –––– Nair and Miller-Hooks 2011 ––––MA – Nourinejad and Roorda 2014 ––MA ––– Nourinejad et al. 2015 ––––MA – Pal and Zhang 2017 ––––MA – (continued on next page) S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 7 demand data, such as VSS trip history, socio-economic and demographic information, mobility and travel-related variables, personal attitudes and preferences, and perceptions of VSS, and data describing the VSS and geographical location the VSS is operating at, such as VSS configuration, city terrain characteristics and elevation, parking locations, and weather conditions. Demand data is recorded during the previous VSS operations, e.g. trip history, or collected using stated preferences surveys (Campbell et al., 2016;Aguilera-García et al., 2020;Wu et al., 2019a;Çelebi et al., 2018; Jin et al., 2020), where the respondents make decisions in the context of hypothetical situations. Additionally, demand data can be extracted from mobile phone trajectory of the users (Miao et al., 2019). 3.1.1. Demand Using the demand data, we can build different types of demand forecasting models. Firstly, the mode-choice models can be constructed. Such models estimate whether the users will accept and use the offered VSS service, in the presence of the other transportation systems (Aguilera-García et al., 2020;Wu et al., 2019a;Jin et al., 2020). Also, users may have a choice between different types of shared vehicles (Campbell et al., 2016;Illgen and H€ ock, 2018), such as between conventional and electric bikes, or between different VSS configurations (Wu et al., 2019a). The methodologies for building such models include logit models (Campbell et al., 2016;Aguilera-García et al., 2020;Wu et al., 2019a;Jin et al., 2020), risky-choice behavior models (Wu et al., 2019a), or simulation (Illgen and H€ ock, 2018). Demand data, together with location characteristics and weather data, may also be used to model and forecast the average usage of a VSS, for a certain period, such as hourly or daily. Here, the forecast may be at the station level (Kutela and Teng, 2019;Ciari et al., 2013), or at the zone level, which is usually done in case of free-floating VSSs. The forecast is performed by using binomial models (Kutela and Teng, 2019), microsimulation (Ciari et al., 2013), or queuing theory (Çelebi et al., 2018). Furthermore, the environmental conditions such as air quality can also be considered to examine the variation in demand for cycling (Morton, 2020). Long term decisions are also related to the business model for the VSS. The literature deals with issues such as market placement, user segmentation, and budget allocation, in the ways respectively presented in the following paragraphs. By simulating a VSS and analyzing the ratio of total number of vehicles to the product of the total number of trips per day and the number of stations, we can select vehicle types that are the best fit for an observed market (Illgen and H€ ock, 2018). This approach is also suitable for determining the appropriate time for the transition from conventional CSS to electric CSS. In Illgen and H€ ock (2018), the authors concluded that this transition should be rather gradual than immediate. Regarding the market research, Degele et al. (2018) also put effort on the analysis of user segmentation in an MSS. The authors apply hierarchical clustering and identify four different user segments, i.e., geographic, demographic, psychographic, and behavioral. In order to assess the impact of each segment for business development, the authors provide a growth-share matrix and determine which segments contribute to the total revenue most. These results show the segments to target through advertisement and marketing. Morton (2020) analyzes two different types of users in BSSs. He shows that regular and casual users tend to differently react to changes in weather and air quality measures. Budget allocation can be performed through network design. I.e., while determining the optimal number of docks and bikes to allocate to each station that will minimize the lost demand, the budget limit can be regarded as a constraint (Caggiani et al., 2019). 3.1.2. Supply The last set of issues at the strategic level relates to the network design. This includes works on station location and size, vehicle allocation, fleet composition, and type of rebalancing. All these decisions may be made to favor and focus on either the operator performance or user satisfaction. In the former case, system operations are evaluated with operator-oriented performance indicators, such as operations costs, number of vehicle relocations (NoR), or average state of battery charge or fuel tank (Deng and Cardin, 2018;Barth and Todd, 1999;Soriguera and Jim enez, 2020). In the latter case, the performance indicators are user-oriented and may include lost demand, zero-vehicle-time (ZVT), full-port-time (FPT), average and total user waiting times, number of users waiting, NoR, or waiting time vs. NoR analysis (Caggiani et al., 2019;Barth and Todd, 1999;Çelebi et al., 2018). The user-oriented performance indicators, or a subset of them, are often jointly denoted Table 2 (continued ) Author Year Strategic level Tactical level Operation level Supply Demand Supply Demand Supply Demand Pfrommer et al. 2014 ––––MA MA Raviv et al. 2013 ––––MA – Repoux et al. 2019 A–MA ––– Rossi et al. 2016 ––––MA – Schuijbroek et al. 2017 ––––MA M Scott and Ciuro 2019 –––M–– Shu et al. 2013 ––MA –MA – Soriguera and Jim enez 2020 MA –MA ––– Warrington and Ruchti 2019 ––––MA – Waserhole and Jost 2012 ––––MA – Weikl and Bogenberger 2015 ––––MA – Wu et al. 2019a –MA –––– Wu et al. 2019b –––––MA Yuan et al. 2019 MA –MA –MA – Zhang et al. 2019 ––––MA – Zhao et al. 2018 ––––MA – Total number of works ¼63 12 7 18 4 28 10 S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 8 Table 3 Suggested future work directions. Author Year Suggested future work directions Addressed by Aguilera-García et al. 2020 Considering heterogeneity in individuals’preferences Li et al. (2018),Jin et al. (2020),Aguilera-García et al. (2020) Quantifying the impacts of a wider adoption of MSSs – Ashqar et al. 2019 Investigating variables such as bikes coming from other stations and relative locations of each station – Balac et al. 2019 A real-life case study which involves competition of sharing systems – Developing a destination-choice model – Boyacıand Zografos 2019 Considering last-minute reservations – Improving clustering by adding a feedback loop between the rebalancing optimization and station clustering models – Integrating behavioral models to operational decisions – Boyacıet al. 2015 Incorporating a simulation model to better represent rebalancing operations Boyacıet al. (2017) Modeling the operational level with battery requirement constraints Boyacıet al. (2017) Boyacıet al. 2017 Extending the models to the systems with no reservations – Combining rebalancing operations and dynamic pricing Boyacıand Zografos (2019),Chemla et al. (2013),Pfrommer et al. (2014) Caggiani et al. 2019 Performing a temporal clustering of stations with respect to the days of the week Lin et al. (2018) a Caggiani et al. 2020 Developing an equality-based bi-level programming where the second level is a demand model – Considering vertical equity indices – Campbell et al. 2016 Analyzing the effect of weather and air quality in relation to objective measures Morton (2020) Çelebi et al. 2018 Integrating cost structures and rebalancing operations – Developing a demand model that takes weather, LOS, availability, and user behavior into account – Chiariotti et al. 2018 Considering multiple rebalancing trucks and their capacity – Clustering stations to reduce the rebalancing problem size Schuijbroek et al. (2017) Modeling user satisfaction – Ciari et al. 2013 Incorporating a reservation system Kaspi et al. (2016),Huang et al. (2020),Repoux et al. (2019) Clemente et al. 2017 Dynamic determination of incentives during the trip Pfrommer et al. (2014),Waserhole and Jost (2012) Degele et al. 2018 Refining the user segments using further information and expanding to a multifactor classification – Considering the spatial distribution of the rentals Ataç et al. (2020) Deng and Cardin 2018 Considering flexibility in system design to respond to the changing demand Cardin et al. (2017) Febbraro et al. 2012 Introducing electric vehicles into the simulation-based optimization model for vehicle rebalancing Boyacıet al. (2017),Boyacıand Zografos (2019) George and Xia 2011 Utilizing queueing network model to derive efficient rebalancing strategies – Optimizing the capacity of stations Huang et al. (2020),Yuan et al. (2019),Caggiani et al. (2020) Ghosh et al. 2017 Extending with a robust optimization technique to account for demand realizations – Ghosh et al. 2016 Considering multi-step planning to better account for the future demand stochasticity – Applying a decomposition technique to solve the rebalancing operations Ghosh et al. (2017) Huang et al. 2020 Optimizing the number of charging piles per station – Considering the stochastic demand in mode-choice forecasting Aguilera-García et al. (2020),Jin et al. (2020) Finding an appropriate pricing strategy that balances demand and vehicle supply – Illgen and H€ ock 2018 Investigating battery degeneration over time and time-dependent electricity costs in eCSSs – Introducing dynamic pricing models into simulation of eCSSs – Assessing the life-cycle of an eCSS – Jin et al. 2020 Short-term/real-time demand estimation – Inducing trip demand shift from private vehicles to VSSs using incentives – Jorge et al. 2014 Including stochastic trip variability and the travel time Yuan et al. (2019),Ataç et al. (2020) (continued on next page) S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 15 Table 3 (continued ) Author Year Suggested future work directions Addressed by Jorge et al. 2015 Studying different principles to determine the demand zones, i.e., station clusters Caggiani et al. (2019),Schuijbroek et al. (2017),Boyacıet al. (2017) Investigating the efficiency of dynamic pricing in terms of imbalance Pfrommer et al. (2014) Analyzing the influence of price on the demand variation – Kaspi et al. 2014 Considering the misuse of the system by the users Kaspi et al. (2016) Predicting the near future of the system using the information received via reservations – Kaspi et al. 2016 Developing a measure, which combines several objectives for itinerary evaluation – Examining the effects of parking reservation policies on the profit– Kumar and Bierlaire 2012 Introducing one-way trips in a VSS, while retaining the return trips – Kutela and Teng 2019 Exploring the effect of station vicinity to the ridership to determine station locations Boyacıet al. (2015) Exploring network density influence on the ridership – Li et al. 2018 Modeling users’choice behavior and incorporating socio-economic characteristics Aguilera-García et al. (2020),Jin et al. (2020) Linking the activity-based model to operational level supply policies – Considering competition among multiple sharing systems Balac et al. (2019) Lin et al. 2018 Including weather and social events in the demand model Morton (2020),Kutela and Teng (2019) Incorporating this model in dynamic rebalancing operations – Masoud et al. 2019 Considering the inaccuracy of the vehicle positions – Using fuzzy logic to address the uncertainty – Miao et al. 2019 Considering non-overlapping and equal size zones, such as hexagon cells Weikl and Bogenberger (2015) Incorporating and clustering the charging infrastructure – Morton 2020 Examining different categories of cyclists (e.g. by sex or age) Kutela and Teng (2019) Nair and Miller-Hooks 2011 Incorporating staff relocation operations to rebalancing operations Boyacıand Zografos (2019),Zhao et al. (2018),Boyacıet al. (2017),Nourinejad et al. (2015) Nourinejad and Roorda 2014 Assigning each user to a station to reduce rebalancing costs – Nourinejad et al. 2015 Incorporating stochastic demand into the vehicle rebalancing and staff relocating problem Nourinejad and Roorda (2014),Ataç et al. (2020) b Including user waiting time as a cost component – Considering the parking availability Repoux et al. (2019),Kaspi et al. (2014) Pal and Zhang 2017 Extending the proposed valid inequalities – Column generation approach to overcome computational complexity – Studying partial rebalancing operations, which will not serve all expected demand Boyacıet al. (2015) Pfrommer et al. 2014 Relaxing deterministic user arrival assumption Yuan et al. (2019),Ataç et al. (2020) Relaxing linearized user reaction to incentives assumption – Raviv et al. 2013 Including several depots in the system – Improving the demand model using data mining models Ashqar et al. (2019),Lin et al. (2018),Faghih-Imani et al. (2017) Designing robust methods, with respect to unplanned events for rebalancing operations – Repoux et al. 2019 Considering both dynamic rebalancing and dynamic pricing Pfrommer et al. (2014),Chemla et al. (2013) Rossi et al. 2016 Incorporating stochastic demand and travel information for rebalancing operations Nourinejad and Roorda (2014) c ,Ataç et al. (2020) c Exploring the integration with public transit – Scott and Ciuro 2019 Exploring the influence of the geographical location and context on the ridership Faghih-Imani et al. (2017) Shu et al. 2013 Exploring incentive schemes – Including demand endogenously to be dependent on the given incentives – Soriguera and Jim enez 2020 Considering mixed system type, i.e. free-floating with a small number of stations – (continued on next page) S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 16 management and operations optimization of VSSs. The optimization problems which appear in these systems repeat themselves, i.e., with a rather small variations the same management approaches can be used in different VSSs. Therefore, with the framework, we aim to address all possible VSS configurations and vehicle types. In the light of the framework, a thorough literature review on VSS management and optimization has been presented. Simultaneously, the conclusions drawn from the review have contributed to produce the framework architecture, components, and required tasks. From the literature review, we have also concluded that a vast number of methodologies for solving most planning problems in VSSs exist. Nevertheless, by mapping the existing works to the framework, we have also identified unanswered research questions, such as the lack of literature dealing with users’destination-choice analysis or competing VSSs. Also, certain vehicle types, such as moped and e-scooters, are not thoroughly studied although they are very popular in practice. The open research questions have been discovered additionally by a thorough analysis of the further research paths proposed by the reviewed papers. Here, we have tried to match each of the suggested research paths with other papers which tackle it. The analysis is presented in Section 4 and Table 3, and reveals many uncovered research paths. Therefrom, the further research of VSS management and optimization seems justified. In our future work, we intend to further extend the framework through analysis of other VSSs or other types of such systems which are yet to be introduced in the practice. Although we find the proposed framework to be the most comprehensive work of such kind on the topic of VSSs, we must allow possibility that not all management and optimization tasks have been covered by it. Additionally, we believe that the general framework is not unique to the VSSs. Although it is designed for the VSSs, the framework could be applicable for other transportation or logistic settings. Therefore, the application of the framework to the other systems will be considered in our future research agenda. Also, driven by the presented framework, we have identified the first concrete research question we intend to tackle. Although there are many works on the demand forecasting models, none of them tries to determine the exact value of such models and justify the investment in building them. Therefore, our next research step will be to quantify the improvement of system operations when demand forecasting models are applied. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. References Aguilera-García, A., Gomez, J., Sobrino, N., 2020. Exploring the adoption of moped scooter-sharing systems in Spanish urban areas. Cities 96, 102424. https://doi.org/ 10.1016/j.cities.2019.102424. Ashqar, H.I., Elhenawy, M., Rakha, H.A., 2019. Modeling bike counts in a bike-sharing system considering the effect of weather conditions. Case Stud. Transp. Policy 7, 261–268. https://doi.org/10.1016/j.cstp.2019.02.011. Ataç, S., Obrenovi c, N., Bierlaire, M., 2020. Vehicle sharing systems: does demand forecasting yield a better service?. In: 20th Swiss Transport Research Conference. Balac, M., Becker, H., Ciari, F., Axhausen, K.W., 2019. Modeling competing free-floating carsharing operators a case study for Zurich, Switzerland. Transport. Res. C Emerg. Technol. 98, 101–117. https://doi.org/10.1016/j.trc.2018.11.011. Barth, M., Todd, M., 1999. Simulation model performance analysis of a multiple station shared vehicle system. Transport. Res. C Emerg. Technol. 7, 237–259. https:// doi.org/10.1016/S0968-090X(99)00021-2. Bates, J., Leibling, D., 2012. Spaced Out: Perspectives on Parking Policy. Technical Report. RAC Foundation. Bird, 2018. Bird marks one year anniversary with 10 millionth environmentally-friendly ride. URL. https://www.bird.co/press/bird-marks-one-year-anniversary-with-10-m illionth-environmentally-friendly-ride/. accessed: 2020-02-26. Boyacı, B., Zografos, K.G., 2019. Investigating the effect of temporal and spatial flexibility on the performance of one-way electric carsharing systems. Transp. Res. Part B Methodol. 129, 244–272. https://doi.org/10.1016/j.trb.2019.09.003. Boyacı, B., Zografos, K.G., Geroliminis, N., 2015. An optimization framework for the development of efficient one-way car-sharing systems. Eur. J. Oper. Res. 240, 718–733. https://doi.org/10.1016/j.ejor.2014.07.020. Boyacı, B., Zografos, K.G., Geroliminis, N., 2017. An integrated optimization-simulation framework for vehicle and personnel relocations of electric carsharing systems with reservations. Transp. Res. Part B Methodol. 95, 214–237. https://doi.org/10.1016/ j.trb.2016.10.007. Bruglieri, M., Pezzella, F., Pisacane, O., 2019. An adaptive large neighborhood search for relocating vehicles in electric carsharing services. Discrete Appl. Math. 253, 185–200. https://doi.org/10.1016/j.dam.2018.03.067 (CTW 2016). Table 3 (continued ) Author Year Suggested future work directions Addressed by Warrington and Ruchti 2019 Considering FFBS Pal and Zhang (2017),Wu et al. (2019b) Weikl and Bogenberger 2015 Creating a task assignment model for rebalancing staff to increase the number of service trips in the given time window – Estimating rebalancing model sensitivity to the variation in zoning of the service area – Wu et al. 2019a Investigating the optimality of different market-making mechanisms from different perspectives – Analyzing the difference between myopic and strategic behavior of users – Wu et al. 2019b Consideration of non-stationary Poisson process for demand modeling – Introducing environmental performance measures for system evaluation Miao et al. (2019) Yuan et al. 2019 Considering preventative and repair activities for station maintenance – Zhang et al. 2019 Introducing stochastic demand into the EV allocation model – Jointly solving the staff scheduling problem Zhao et al. (2018),Boyacıet al. (2017),Nourinejad et al. (2015) Zhao et al. 2018 Incorporating a detailed charging process into rebalancing operations Soriguera and Jim enez (2020) Considering the stochasticity of travel time and battery charging – Introducing uncertainty in the demand, in the form of last minute trip cancellation – a The analysis on the days is done. b Stochastic demand is incorporated only in vehicle rebalancing problem. c Only stochastic demand is covered. S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 17 Caggiani, L., Camporeale, R., Marinelli, M., Ottomanelli, M., 2019. User satisfaction based model for resource allocation in bike-sharing systems. Transport Pol. 80, 117–126. https://doi.org/10.1016/j.tranpol.2018.03.003. Caggiani, L., Colovic, A., Ottomanelli, M., 2020. An equality-based model for bike-sharing stations location in bicycle-public transport multimodal mobility. Transport. Res. Part A Policy and Practice. https://doi.org/10.1016/j.tra.2020.08.015. Campbell, A.A., Cherry, C.R., Ryerson, M.S., Yang, X., 2016. Factors influencing the choice of shared bicycles and shared electric bikes in Beijing. Transport. Res. C Emerg. Technol. 67, 399–414. https://doi.org/10.1016/j.trc.2016.03.004. Cardin, M.A., Deng, Y., Sun, C., 2017. Real options and flexibility analysis in design and management of one-way mobility on-demand systems using decision rules. Transport. Res. C Emerg. Technol. 84, 265–287. https://doi.org/10.1016/ j.trc.2017.08.006. URL. http://www.sciencedirect.com/science/article/pii/S09680 90X17302073. Çelebi, D., Y€ orüsün, A., Is ¸ık, H., 2018. Bicycle sharing system design with capacity allocations. Transp. Res. Part B Methodol. 114, 86–98. https://doi.org/10.1016/ j.trb.2018.05.018. Cepolina, E.M., Farina, A., 2012. A new shared vehicle system for urban areas. Transport. Res. C Emerg. Technol. 21, 230–243. https://doi.org/10.1016/j.trc.2011.10.005. Chemla, D., Meunier, F., Pradeau, T., Calvo, R., Yahiaoui, H., 2013. Self-service Bike Sharing Systems: Simulation, Repositioning, Pricing. Technical Report hal-00824078. Chiariotti, F., Pielli, C., Zanella, A., Zorzi, M., 2018. A dynamic approach to rebalancing bike-sharing systems. Sensors 18, 512. https://doi.org/10.3390/s18020512. Ciari, F., Schuessler, N., Axhausen, K.W., 2013. Estimation of carsharing demand using an activity-based microsimulation approach: model discussion and some results. Int. J. Sustain. Transport. 7, 70–84. https://doi.org/10.1080/15568318.2012.660113 arXiv. Clemente, M., Fanti, M.P., Iacobellis, G., Nolich, M., Ukovich, W., 2017. A decision support system for user-based vehicle relocation in car sharing systems. IEEE Trans. Syst, Man, Cybernet.: Systems 1–14. https://doi.org/10.1109/TSMC.2017.2665881. Degele, J., Gorr, A., Haas, K., Kormann, D., Krauss, S., Lipinski, P., Tenbih, M., Koppenhoefer, C., Fauser, J., Hertweck, D., 2018. Identifying e-scooter sharing customer segments using clustering. In: 2018 IEEE International Conference on Engineering, Technology and Innovation (ICE/ITMC), pp. 1–8. https://doi.org/ 10.1109/ICE.2018.8436288. Dell’Amico, M., Hadjicostantinou, E., Iori, M., Novellani, S., 2014. The bike sharing rebalancing problem: mathematical formulations and benchmark instances. Omega 45, 7–19. https://doi.org/10.1016/j.omega.2013.12.001. Deng, Y., Cardin, M.A., 2018. Integrating operational decisions into the planning of oneway vehicle-sharing systems under uncertainty. Transport. Res. C Emerg. Technol. 86, 407–424. https://doi.org/10.1016/j.trc.2017.11.018. Faghih-Imani, A., Hampshire, R., Marla, L., Eluru, N., 2017. An empirical analysis of bike sharing usage and rebalancing: evidence from Barcelona and Seville. Transport. Res. Pol. Pract. 97, 177–191. https://doi.org/10.1016/j.tra.2016.12.007. Febbraro, A.D., Sacco, N., Saeednia, M., 2012. One-way carsharing: solving the relocation problem. Transport. Res. Rec. 2319, 113–120. https://doi.org/10.3141/2319-13 arXiv. George, D.K., Xia, C.H., 2011. Fleet-sizing and service availability for a vehicle rental system via closed queueing networks. Eur. J. Oper. Res. 211, 198–207. https:// doi.org/10.1016/j.ejor.2010.12.015. Getaround, 2019. Lev - how it works? URL. https://www.getlev.nl/en/how-it-works. accessed: 2020-03-02. Ghosh, S., Trick, M., Varakantham, P., 2016. Robust repositioning to counter unpredictable demand in bike sharing systems. In: Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence. AAAI Press, pp. 3096–3102. Ghosh, S., Varakantham, P., Adulyasak, Y., Jaillet, P., 2017. Dynamic repositioning to reduce lost demand in bike sharing systems. J. Artif. Intell. Res. 58, 387–430. Hansen, R., Pantuso, G., 2018. Pricing car-sharing services in multi-modal transportation systems: an analysis of the cases of Copenhagen and Milan. In: 9th International Conference on Computational Logistics, ICCL 2018, pp. 344–359. https://doi.org/ 10.1007/978-3-030-00898-7. He, L., Mak, H.Y., Rong, Y., 2019. Operations Management of Vehicle Sharing Systems. Springer International Publishing, Cham, pp. 461–484. https://doi.org/10.1007/ 978-3-030-01863-4_19. Huang, K., An, K., de Almeida Correia, G.H., 2020. Planning station capacity and fleet size of one-way electric carsharing systems with continuous state of charge functions. Eur. J. Oper. Res. 287, 1075–1091. https://doi.org/10.1016/j.ejor.2020.05.001. URL. http://www.sciencedirect.com/science/article/pii/S0377221720304112. Illgen, S., H€ ock, M., 2018. Electric vehicles in car sharing networks - challenges and simulation model analysis. Transport. Res. Transport Environ. 63, 377–387. https:// doi.org/10.1016/j.trd.2018.06.011. Illgen, S., H€ ock, M., 2019. Literature review of the vehicle relocation problem in one-way car sharing networks. Transp. Res. Part B Methodol. 120, 193–204. https://doi.org/ 10.1016/j.trb.2018.12.006. Jin, F., Yao, E., An, K., 2020. Analysis of the potential demand for battery electric vehicle sharing: mode share and spatiotemporal distribution. J. Transport Geogr. 82, 102630. https://doi.org/10.1016/j.jtrangeo.2019.102630. URL. http://www.sciencedirect.co m/science/article/pii/S0966692319305149. Jorge, D., Correia, G.H.d.A., 2013. Carsharing systems demand estimation and defined operations: a literature review. Eur. J. Transport Infrastruct. Res. 13, 201–220. https://doi.org/10.18757/ejtir.2013.13.3.2999. Jorge, D., Correia, G.H.d.A., Barnhart, C., 2014. Comparing optimal relocation operations with simulated relocation policies in one-way carsharing systems. IEEE Trans. Intell. Transport. Syst. 15, 1667–1675. https://doi.org/10.1109/TITS.2014.2304358. Jorge, D., Molnar, G., de Almeida Correia, G.H., 2015. Trip pricing of one-way stationbased carsharing networks with zone and time of day price variations. Transp. Res. Part B Methodol. 81, 461–482. https://doi.org/10.1016/j.trb.2015.06.003. Kaspi, M., Raviv, T., Tzur, M., 2014. Parking reservation policies in one-way vehicle sharing systems. Transp. Res. Part B Methodol. 62, 35–50. https://doi.org/10.1016/ j.trb.2014.01.006. Kaspi, M., Raviv, T., Tzur, M., Galili, H., 2016. Regulating vehicle sharing systems through parking reservation policies: analysis and performance bounds. Eur. J. Oper. Res. 251, 969–987. https://doi.org/10.1016/j.ejor.2015.12.015. Kek, A.G., Cheu, R.L., Meng, Q., Fung, C.H., 2009. A decision support system for vehicle relocation operations in carsharing systems. Transport. Res. E Logist. Transport. Rev. 45, 149–158. https://doi.org/10.1016/j.tre.2008.02.008. Kek, A.G.H., Cheu, R.L., Chor, M., 2006. Relocation simulation model for multiple-station shared-use vehicle systems. Transport. Res. Rec. 1986, 81–88. https://doi.org/ 10.3141/1986-13. Kumar, P., Bierlaire, M., 2012. Optimizing Locations for a Vehicle Sharing System. Swiss Transportation Research Conference (STRC). Kutela, B., Teng, H., 2019. The influence of campus characteristics, temporal factors, and weather events on campuses-related daily bike-share trips. J. Transport Geogr. 78, 160–169. https://doi.org/10.1016/j.jtrangeo.2019.06.002. Laporte, G., Meunier, F., Calvo, R., 2018. Shared mobility systems: an updated survey. Ann. Oper. Res. 271 https://doi.org/10.1007/s10479-018-3076-8. Lawler, R., 2012. “Zipcar for scooters”startup Scoot networks launches to the public in san Francisco. URL. https://techcrunch.com/2012/09/26/scoot-sf-launch/. accessed: 2020-02-26. Li, Q., Liao, F., Timmermans, H.J., Huang, H., Zhou, J., 2018. Incorporating free-floating car-sharing into an activity-based dynamic user equilibrium model: a demand-side model. Transp. Res. Part B Methodol. 107, 102–123. https://doi.org/10.1016/ j.trb.2017.11.011. Lin, L., He, Z., Peeta, S., 2018. Predicting station-level hourly demand in a large-scale bike-sharing network: a graph convolutional neural network approach. Transport. Res. C Emerg. Technol. 97, 258–276. https://doi.org/10.1016/j.trc.2018.10.011. Liu, J., Sun, L., Chen, W., Xiong, H., 2016. Rebalancing bike sharing systems: a multisource data smart optimization. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1005–1014. https://doi.org/10.1145/2939672.2939776. Masoud, M., Elhenawy, M., Almannaa, M., Liu, S.Q.S., Glaser, S., Rakotonirainy, A., 2019. Heuristic Approaches to Solve E-Scooter Assignment Problem. IEEE Access. https:// doi.org/10.1109/ACCESS.2019.2957303,1–1. Miao, H., Jia, H., Li, J., Qiu, T.Z., 2019. Autonomous connected electric vehicle (ACEV)- based car-sharing system modeling and optimal planning: a unified two-stage multiobjective optimization methodology. Energy 169, 797–818. https://doi.org/ 10.1016/j.energy.2018.12.066. Midgley, P., 2011. Bicycle-sharing Schemes: Enhancing Sustainable Mobility in Urban Areas, vol. 8. United Nations, Department of Economic and Social Affairs, pp. 1–12. Morton, C., 2020. The demand for cycle sharing: examining the links between weather conditions, air quality levels, and cycling demand for regular and casual users. J. Transport Geogr. 88, 102854. https://doi.org/10.1016/j.jtrangeo.2020.102854. URL. http://www.sciencedirect.com/science/article/pii/S0966692319306167. Nair, R., Miller-Hooks, E., 2011. Fleet management for vehicle sharing operations. Transport. Sci. 45, 524–540. Nath, R., Rambha, T., 2019. Modelling methods for planning and operation of bikesharing systems. J. Indian Inst. Sci. 1–26. https://doi.org/10.1007/s41745-019- 00134-8. Nourinejad, M., Roorda, M.J., 2014. A dynamic carsharing decision support system. Transport. Res. E Logist. Transport. Rev. 66, 36–50. https://doi.org/10.1016/ j.tre.2014.03.003. Nourinejad, M., Zhu, S., Bahrami, S., Roorda, M.J., 2015. Vehicle relocation and staff rebalancing in one-way carsharing systems. Transport. Res. E Logist. Transport. Rev. 81, 98–113. https://doi.org/10.1016/j.tre.2015.06.012. Pal, A., Zhang, Y., 2017. Free-floating bike sharing: solving real-life large-scale static rebalancing problems. Transport. Res. C Emerg. Technol. 80, 92–116. https:// doi.org/10.1016/j.trc.2017.03.016. Pfrommer, J., Warrington, J., Schildbach, G., Morari, M., 2014. Dynamic vehicle redistribution and online price incentives in shared mobility systems. IEEE Trans. Intell. Transport. Syst. 15, 1567–1578. https://doi.org/10.1109/ TITS.2014.2303986. Raviv, T., Tzur, M., Forma, I.A., 2013. Static repositioning in a bike-sharing system: models and solution approaches. EURO J. Transport. Logist. 2, 187–229. Repoux, M., Kaspi, M., Boyacı, B., Geroliminis, N., 2019. Dynamic prediction-based relocation policies in one-way station-based carsharing systems with complete journey reservations. Transp. Res. Part B Methodol. 130, 82–104. https://doi.org/ 10.1016/j.trb.2019.10.004. Rossi, F., Zhang, R., Pavone, M., 2016. Autonomous vehicle routing in congested road networks. CoRR abs/1603.00939. URL. http://arxiv.org/abs/1603.00939. arXiv: 1603.00939. Schuijbroek, J., Hampshire, R.C., van Hoeve, W.J., 2017. Inventory rebalancing and vehicle routing in bike sharing systems. Eur. J. Oper. Res. 257, 992–1004. https:// doi.org/10.1016/j.ejor.2016.08.029. Scott, D.M., Ciuro, C., 2019. What factors influence bike share ridership? an investigation of Hamilton, Ontario’s bike share hubs. Travel Behav Society 16, 50–58. https:// doi.org/10.1016/j.tbs.2019.04.003. Shaheen, S., Guzman, S., Zhang, H., 2010. Bikesharing in Europe, the Americas, and Asia: Past, Present, and Future. Institute of Transportation Studies, UC Davis, Institute of Transportation Studies. https://doi.org/10.3141/2143-20. Working Paper Series 2143. S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 18 Shaheen, S., Sperling, D., Wagner, C., 1998. Carsharing in europe and north America: past, present, and future. Transport. Q. 52. Shoup, D.C., 2011. The High Cost of Free Parking, Updated [edition]. American Planning Association Planners Press, Chicago. Shu, J., Chou, M., Liu, Q., Teo, C., Wang, I.L., 2013. Models for effective deployment and redistribution of bicycles within public bicycle-sharing systems. Oper. Res. 61, 1346–1359. https://doi.org/10.1287/opre.2013.1215. Soriguera, F., Jim enez, E., 2020. A continuous approximation model for the optimal design of public bike-sharing systems. Sustain. Cities Society 52, 101826. https:// doi.org/10.1016/j.scs.2019.101826. Warrington, J., Ruchti, D., 2019. Two-stage stochastic approximation for dynamic rebalancing of shared mobility systems. Transport. Res. C Emerg. Technol. 104, 110–134. https://doi.org/10.1016/j.trc.2019.04.021. Waserhole, A., Jost, V., 2012. Vehicle Sharing System Pricing Regulation: A Fluid Approximation. Technical report hal-00727041vol. 4. Weikl, S., Bogenberger, K., 2015. A practice-ready relocation model for free-floating carsharing systems with electric vehicles - mesoscopic approach and field trial results. Transport. Res. C Emerg. Technol. 57, 206–223. https://doi.org/10.1016/ j.trc.2015.06.024. Wu, C., Vine, S.L., Sivakumar, A., Polak, J., 2019a. Traveller preferences for free-floating carsharing vehicle allocation mechanisms. Transport. Res. C Emerg. Technol. 102, 1–19. https://doi.org/10.1016/j.trc.2019.02.019. Wu, R., Liu, S., Shi, Z., 2019b. Customer incentive rebalancing plan in free-float bikesharing system with limited information. Sustainability 11, 3088. Yuan, M., Zhang, Q., Wang, B., Liang, Y., Zhang, H., 2019. A mixed integer linear programming model for optimal planning of bicycle sharing systems: a case study in Beijing. Sustain. Cities Society 47, 101515. https://doi.org/10.1016/ j.scs.2019.101515. Zhang, D., Liu, Y., He, S., 2019. Vehicle assignment and relays for one-way electric carsharing systems. Transp. Res. Part B Methodol. 120, 125–146. https://doi.org/ 10.1016/j.trb.2018.12.004. Zhao, M., Li, X., Yin, J., Cui, J., Yang, L., An, S., 2018. An integrated framework for electric vehicle rebalancing and staff relocation in one-way carsharing systems: model formulation and Lagrangian relaxation-based solution approach. Transp. Res. Part B Methodol. 117, 542–572. https://doi.org/10.1016/j.trb.2018.09.014. S. Ataç et al. EURO Journal on Transportation and Logistics 10 (2021) 100033 19