*Article* **What Car for Car-Sharing? Conventional, Electric, Hybrid or Hydrogen Fleet? Analysis of the Vehicle Selection Criteria for Car-Sharing Systems**

**Katarzyna Turo ´n 1,\*, Andrzej Kubik 1,\* and Feng Chen <sup>2</sup>**


**Abstract:** Short-term car rental services called "car-sharing" or "carsharing" are systems that in recent years have been an alternative form of transport by individual car in an increasing number of cities around the world. With the growing popularity of services, new decision-making problems have arisen among system operators. Among the challenges faced by operators, due to the constantly growing environmental requirements, is the fleet of vehicles for car-sharing systems-appropriate selection. Noticing this research gap, this article was dedicated to determining the criteria that are important when choosing a fleet of vehicles for car-sharing and to indicate the best suited to the needs of car-sharing vehicles. Own research was proposed, considering desk research, expert research and analyses using the multi-criteria decision support method (ELECTRE III). This research was carried out for the Polish market of car-sharing services. Studying the Polish market is appropriate due to the occurrence of significant difficulties with the fleet incorrectly adjusted to the needs of urban conditions. This study covers vehicles with conventional, electric, hybrid and hydrogen propulsion. The analyses allowed for the determination of the vehicles best suited to the needs of car-sharing. The results show the dominance of hydrogen-powered vehicles over conventional, hybrid and electric vehicles. What is more, it was determined that the most important criteria are the purchase price of the vehicle and energy/fuel consumption per 100 km. The obtained results are a guide to proceeding when making decisions regarding the implementation or modernization of the fleet in car-sharing systems. The results also support achieving more sustainable urban mobility in the zero-emission trend through hydrogen mobility.

**Keywords:** car-sharing; carsharing; shared mobility; multi-criteria decision analysis; ELECTRE III; MCDA; decision making in transport; transportation engineering; hydrogen mobility; zero emission; green energy

#### **1. Introduction**

Short-term car rental services called "car-sharing" or "carsharing" are systems that in recent years have been an alternative form of transport by individual car in an increasing number of cities around the world. The car-sharing market size surpassed USD 2 billion in 2020 [1], and it is expected to grow in 2022 with a Compound Annual Growth Rate (CAGR) of 17.4% [2]. Car-sharing systems have undergone many modifications along with their development. Over the years, system management, the location and relocation of vehicles, price lists and service packages, infrastructure and vehicles have changed [3–6]. These changes resulted from many different factors due to the gradual adaptation of society to new forms of transport [7], and thus changing demand [8,9], due to new legislation or municipal car-sharing regulations being implemented [10,11], changes in environmental requirements [12–14], etc.

**Citation:** Turo ´n, K.; Kubik, A.; Chen, F. What Car for Car-Sharing? Conventional, Electric, Hybrid or Hydrogen Fleet? Analysis of the Vehicle Selection Criteria for Car-Sharing Systems. *Energies* **2022**, *15*, 4344. https://doi.org/10.3390/ en15124344

Academic Editor: Byoung Kuk Lee

Received: 19 May 2022 Accepted: 13 June 2022 Published: 14 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Evaluating car-sharing systems has become an interesting topic for scientists. A frequently studied topic is the car-sharing fleet. However, there are several leading topics in the field of fleet research. The first of the leading themes is the size of the fleet in car-sharing systems. For example, fleet size considerations are among the main motives; Xu et al. dedicated their research to electric vehicle fleet size for car-sharing services considering on-demand charging strategy and battery degradation [15]. In comparison, Monteiro et al. optimized car-sharing fleet size to maximize the number of clients served [16]. In turn, Hu and Liu analyzed the joint design of parking capacities and fleet size for one-way station-based car-sharing systems with road congestion constraints [4]. The second leading topics regarding the car fleet are aspects of vehicle location and relocation. For example, the Chang et al. dealt with the subject of location design and relocation of a mixed car-sharing fleet with a CO2 emission constraint [17]. Yoon et al. investigated car-sharing demand estimation and fleet simulation with electric vehicle adoption [18]. In turn, Fan et al. dealt with car-sharing dynamic decision-making obstacles for vehicle allocation [19]. There is, however, a literature gap in the research on the car-sharing fleet. The gap concerns analyses directly related to the type of vehicles used in the systems and their use. In our previous works, we dealt with the determination of the fleet which is most often used in car-sharing systems [20] and we analyzed the operational factors of vehicle use [21]. Receiving signals from operators of Polish car-sharing systems that they need to make changes to vehicle fleets, we have dedicated this article to the selection of vehicles for car-sharing systems.

From the point of view of common mobility services, the Polish market is a very interesting field. Although vehicle sharing services appeared relatively late to other European countries, e.g., bike-sharing in 2008 [22–26], car-sharing in 2016 [27], moped-sharing in 2017 [28], and scooter-sharing in 2018 [28], this market is characterized as dynamic and valuable [28]. A significant development of shared mobility services in Poland has been observed since 2017, when more and more car-sharing service operators appeared on the market [29]. At the peak of shared mobility systems development, there were 17 car-sharing operators available in 250 cities [29]. Revenues from car-sharing services in Poland in 2019 amounted to over 50 million PLN, and they achieved a double increase in 2021, reaching over 100 million PLN [30]. However, the market boom of new car-sharing operators has not lasted long. After the opening of many systems, the rapid disappearance of many systems from the market has occurred. The most spectacular closures included closure of the Vozilla electric car-sharing system with a fleet of 240 cars [31], closure of the InnogyGo system! with a fleet of 500 electric cars [32], and a few other operators who had pilot schemes have withdrawn from offering short-term car rental to long-term rental. It is worth mentioning that during the boom, the offered vehicle fleets and rental regulations were very chaotic and contradictory. For example, operators have implemented electric cars without having to consider the presence of infrastructure for electric vehicles in a given area [21]. Moreover, many system regulations prevented the efficient use of electric cars. For example, it was necessary to terminate the rental of a vehicle with an energy level in the car's battery that would allow another user to drive a further 30 km, where in practice there was no charging station in the area of the rental zone up to 30 km. What is more, many regulations forbid the user to connect cars to chargers by themselves, while others ordered the vehicle to be returned only under the charger. In practice, the idea of free-floating electric car-sharing did not take place at that time. The difficulties were not only with electric vehicles fleets. Conventionally fueled cars, on the other hand, were targeted at one car model, which discouraged some users from using the cars [20]. Despite the many challenges that Polish car-sharing has had to face in recent years, it is predicted that Polish car-sharing revenues will reach a value of over 265 million PLN in 2025 [30]. Currently, many Polish cities, striving to limit transport by individual cars [33–35], are implementing new transport policies, leading to a significant development of car-sharing service systems, which will result in the creation of new systems and modernization of existing systems

over several years; therefore, it is particularly important to determine the appropriate fleet to supply car-sharing systems to meet the expectations of stakeholders.

The aim of this study is to determine the criteria that are important when choosing a fleet of vehicles for car-sharing and to indicate the best suited to the needs of car-sharing vehicles. This research was carried out for the Polish market. This study covers vehicles with conventional, electric and hydrogen propulsion.

This article consists of four main sections. The first chapter presents a general description of the research problem and characterizes the Polish car-sharing market along with a historical outline. The second chapter presents information on the methods of multicriteria decision support, as well as a detailed description of the ELECTRE III method used, together with a test plan. The third chapter shows a detailed analysis and the obtained results. The fourth chapter discusses the obtained results and confronts them with the research of other authors. This article is a guide when making decisions regarding the implementation or modernization of the fleet in car-sharing systems. The results also support achieving more sustainable urban mobility in the zero-emission trend through hydrogen mobility.

#### **2. Materials and Methods**

Deciding which vehicle fleet to choose is a problem that requires consideration of many different criteria. In this case, we use multi-criteria decision support methods. Multicriteria decision making (MCDM), multi-criteria decision analysis (MCDA) or multi-criteria data analysis methods are a sub-discipline of operations research [36]. Their task is to provide a wide range of mathematical tools that can be used in the analytical process of decision making. MCDM means the process of determining the best feasible solution according to established criteria and problems that are common occurrences in everyday life [37]. Their specificity enables defining criteria, their weights and actors appearing in the decision-making process, i.e., stakeholders [38]. With their use, it is possible to obtain the final rankings of scenarios for the analyzed research questions [36–39]. MCDM is used to solve decision-making problems at the strategic, tactical, and operational levels [40].

MCDA is widely used to solve various transport problems, including for solving the problem of selecting projects to build the Paris metro [41], choosing the best transport connection between the city of Pittsburgh and international airport [42], or assessing transport solutions for the metropolitan area of Istanbul [43]. Moreover, these methods have also been applied to car-sharing systems. For example, they were used to determine the location of base stations of the EVCARD car-sharing system operator in the area of Shanghai [44], to analyze the selection of the location of car-sharing stations in Beijing [45] and to determine the location of car-sharing stations in the French city of La Rochelle [46]. Since MCDA is commonly used in decision-making aspects, one of the methods was included in the research process.

The research process considered secondary research on vehicles used in car-sharing systems, expert research among car-sharing service operators and the performance of mathematical analyses, considering the multi-criteria decision support method. The detailed procedure of the procedure is presented in Figure 1.

**Figure 1.** Research process.

Secondary research was carried out on a group of operators functioning in Poland in May 2022.They concerned the analysis of vehicle fleets in car-sharing systems in order to identify the most frequently used cars. Successively, the most popular and commercially available hydrogen-powered vehicles were added to conventional, electric and hybrid vehicles. Secondary research allowed the building of a database of vehicles that were considered in the calculations.

The next step was to conduct expert research on a group of car-sharing operators present in Poland. The aim of the research was to indicate the importance of individual criteria considered when selecting a vehicle for car-sharing systems. In accordance with the MCDA methodology, the respondents made pairwise comparisons of individual criteria on a scale from 1 to 9, where 1—same meaning; 2—very weak advantage; 3—weak advantage; 4—more than weak advantage, less than strong; 5—strong advantage; 6—more than a strong advantage, less than very strong; 7—a very strong advantage; 8—more than a very strong advantage, less than an extreme; 9—extreme, total advantage. The weights obtained were included in the analyses using the MCDA method.

The last step was to perform analyses using one of the MCDA methods. Among the group of methods frequently used in transport problems is the set of ELECTRE [47]. ELECTRE is an acronym for *Elimination Et Choix Traduisant la Realitè* and represents a set of multi-criteria decision support methods (ELECTRE I, II, III, IV, IS, and TRI), which are based on partial preference aggregation by overrun [47–53]. Different types of ELECTRE methods have different approaches to decision-making problems. The first method produces elections, the others provide ranking [47–53]. The ELECTRE III method is the most popular of the ELECTRE family methods [39]. The ELECTRE III method is most often combined with expert research (e.g., Delphi method) [54]. The method introduces a two-level preference for a given criteria. This means that they may be strongly or slightly better than each other, which means situations when the decision variants differ very or little from each other [39].

The algorithm in the ELECTRE III method includes 3 stages [24]:


The first stage of the analysis begins with the definition of a set of criteria that will be used to evaluate the set of decision variants [47–53]. Each criterion from the set is assigned an appropriate weight. Subsequently, by comparing the two decision variants, the exceedance index is calculated [47–53].

In the second step, based on the exceedance index, the answer is whether the first variant is not worse than the second one due to a given criterion. Subsequently, a computation of the compliance rate is performed to be able to obtain a response with the advantage of one option over the other in terms of all criteria [47–53]. The compliance rate is the sum of the weights of the criteria for which the evaluation value of one variant is greater than or equal to the evaluation value of the other variant [47–53].

In the second stage, based on the exceedance index, the answer is whether the first option is not worse than the second one due to the given criterion. Subsequently, a computation of the compliance rate is performed to be able to obtain a response with the advantage of one option over the other in terms of all criteria [47–53]. The compliance rate is the sum of the weights of the criteria for which the evaluation value of one variant is greater than or equal to the evaluation value of the other variant [47–53].

The third stage is based on creating an altitude difference matrix. The scenarios should be ranked sequentially, which begins with their initial ordering by means of the classification procedures: ascend distillation and descend distillation [47–53]. Both distillations rank the best to worst scenarios [47–53].

Ascend distillation is a scheduling process that begins with selecting the best scenario and placing it at the top of the classification [47–53]. The best scenario is then selected again from among the remaining scenarios and placed in the next position in the classification. This procedure is repeated until the set of scenarios is exhausted [47–53].

For descend distillation, the scheduling process begins with the worst-case selection and placement at the end of the ranking. The sequence is the same as in the ascend distillation procedure, with the difference that in subsequent iterations of the remaining scenarios to be considered, the worst scenario is always selected and placed on the next positions "from the bottom" [47–53].

Then, we create the final ranking based on the top-down and bottom-up ordering. The result is a final ranking of the scenarios. The results are presented in the next chapter.

#### **3. Results**

When determining which car-sharing vehicle to choose, in the first step, the most frequently used vehicle models on the Polish market, valid as of May 2022, were determined. The most frequently repeated cars are marked in green. The summary is presented in Table 1.

**Table 1.** Vehicles used in Polish car-sharing systems.


The most common vehicle models with conventional, electric and hybrid drive were selected successively. In line with global trends in reducing transport emissions, hydrogenpowered vehicles were also included. A total of 12 different vehicle models were included, representing a diverse set of vehicle classes. A detailed list of vehicles considered in the analysis is presented in Table 2.


**Table 2.** Cars included in the analysis that can be used in car-sharing systems.

The criteria for selecting the vehicles that have been considered are successively defined. The list of critics is presented in Table 3.



The preferences of experts were directed towards vehicles with the highest possible comfort of movement, with relatively high engine power, luggage compartment capacity, and the lowest possible exhaust emissions due to possible restrictions on access to city centers in the future. The values of individual criteria have been presented in sequence for a selected fleet of vehicles that can be implemented in car-sharing systems. The results are presented in Table 4.

**Table 4.** Adopted values for particular criteria.


Then, in accordance with the guidelines of the ELECTRE III method, the equivalence, preference, and veto thresholds were determined for each of the criteria, which are presented in Table 5.


**Table 5.** Values of the equivalence, preference, and veto thresholds in light of the considered criteria.

In the next step, the values of the concordance matrix were determined, which are presented in Table 6.

**Table 6.** Concordance matrix C values.


The non-compliance indicators were successively determined for each of the seven considered criteria, which are presented in Tables A1–A7. Based on the noncompliance indicators, the values of reliability indicators were determined, which are presented in Table A8.

The next step was to perform the ascend and descend distillation. The results are presented in the form of a dominance matrix in Table 7.


**Table 7.** Dominance matrix L.


**Table 7.** *Cont*.

Where I—a pair of variants is equivalent; P+—the first option is better than the second option; P−—the first option is worse than the second option.

Based on the value of the exceedance relation matrix, the final ranking of decision values was created, depending on the type of distillation, which are presented in Table 8. The final ranking presented in Table 8 defines which of the considered scenarios is the most optimal in terms of the assumed criteria and the assessment of the preferences of experts.

**Table 8.** Final ranking.


Based on the value of the exceedance relation matrix, the final ranking of decision values was created, depending on the type of distillation, which are presented in Figure 2.

**Figure 2.** The final ranking of the best vehicles to implement in the car-sharing fleet.

#### **4. Discussion and Conclusions**

The research carried out with the use of the ELECTRE III multi-criteria decision support method was used to determine the best selection of vehicles for car-sharing systems based on the criteria established and assessed by experts. The obtained results indicate that of the analyzed car models, Honda Clarity achieved the top ranking and is the optimal vehicle that meets the expectations of experts.

Moreover, the conducted research shows that hydrogen-powered vehicles are on the podium in the obtained ranking. When analyzing the obtained results in detail, it can be noticed that electric vehicles occupy the last places in the ranking. An interesting finding is that conventionally powered vehicles rank better than electric vehicles. This result is mainly caused by a large disproportion between the purchase prices of an electric vehicle and a vehicle with a conventional drive.

Based on the obtained results, it was found that the most important criteria are the purchase price of the vehicle, energy/fuel consumption per 100 km and the time of refueling/charging the vehicle's battery. The results, therefore, indicate that it is the economic and operational criteria that are of greatest importance for shared mobility cars.

It is worth mentioning that vehicles in car-sharing systems generate profits in terms of traffic. Unfortunately, all vehicles whose battery charging process requires a large amount of time have limited transport availability for users, reducing the often-insufficient vehicle fleet in car-sharing systems.

Therefore, despite the widespread interest in electric vehicles for car-sharing, if the fleet of vehicles is not so large that cars that are being charged cannot be replaced with ready-to-use vehicles, and the infrastructure will not allow the charging time to be reduced to the level of conventional or hydrogen vehicles, electric cars in a car-sharing model will not be the optimal choice.

When translating the obtained results into business practices of car-sharing systems, it is worth emphasizing that hydrogen-powered vehicles are not currently used in systems both in Poland and Europe, and the current trends are directed towards electric vehicles. Unfortunately, the analysis of the market activities of companies shows that most companies with a fleet of electric vehicles in Poland failed, and the systems were closed after several months of operation. This type of practice was also visible in the case of the Paris carsharing system and the American system in San Diego. Currently, especially in the Polish market, infrastructure for servicing electric vehicles is still too little for individual cars, let alone for servicing car-sharing systems [55]. As Poland is looking for solutions for the development of low-emission transport, more and more hopes are placed on hydrogen. Currently, hydrogen refueling stations are already being created with plans to expand by 2025, when the number of stations will increase by 3200% [56]. Therefore, the dissemination of a hydrogen-powered car for Polish car-sharing is a future-proof scenario.

Due to the area character of the research and the results being limited to the Polish market, the authors plan to expand future research to a larger scale and conduct research considering other European countries. Due to the lack of scientific research on the selection of the vehicle fleet, no direct reference was made in the discussion to the results of other authors' research.

The obtained results support the operators of car-sharing systems in the decisionmaking process when selecting vehicles for the fleet of their systems.

**Author Contributions:** Conceptualization, K.T.; methodology, K.T. and A.K.; validation, F.C.; resources, K.T., A.K., and F.C.; data curation, K.T. and A.K.; writing—original draft preparation, K.T., and A.K.; writing—review and editing, K.T. and A.K.; supervision, F.C.; project administration, K.T.; funding acquisition, K.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding. Research conducted as part of the Silesian University of Technology subsidy for the maintenance and development of the research potential of early career researchers—BKM-692/RT1/2022, 12/010/BKM22/1058 and BKM-693/RT1/2022 12/010/BKM22/1059.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the authors.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

**Table A1.** Values of the non-compliance indicators according to the 1st criterion.


**Table A2.** Values of the non-compliance indicators according to the 2nd criterion.


**Table A3.** Values of the non-compliance indicators according to the 3rd criterion.



**Table A4.** Values of the non-compliance indicators according to the 4th criterion.

**Table A5.** Values of the non-compliance indicators according to the 5th criterion.


**Table A6.** Values of the non-compliance indicators according to the 6th criterion.



**Table A7.** Values of the non-compliance indicators according to the 7th criterion.

**Table A8.** Credibility matrix values D.


#### **References**


### *Article* **Compromise Multi-Criteria Selection of E-Scooters for the Vehicle Sharing System in Poland**

**Paweł Ziemba 1,\* and Izabela Gago <sup>2</sup>**


**Abstract:** In Poland, there is a high ratio of private transport and unfavorable patterns of daily commuting. These patterns can be changed by introducing comfortable and eco-friendly vehicles, such as e-scooters and e-bikes. At the same time, the development of the e-micromobility-based vehicle sharing services market is developing. The aim of the article is to analyze selected e-scooters available on the Polish market and to identify the most useful vehicles from two opposing perspectives, i.e., the potential customer and owner of the vehicle sharing system. The PROSA GDSS (PROMETHEE for Sustainability Assessment—Group Decision Support System) method and the graphical representation of GAIA (Geometrical Analysis for Interactive Assistance) were used to search for a compromise and balance between the needs of the indicated stakeholders. The results of the methods used were compared with the results of the PROMETHEE GDSS method, which does not take into account the balance between the stakeholders and allows for a strong compensation of the assessments of decision makers. The conducted research allowed indicating the optimal e-scooter to meet the needs of both decision makers, and it is the JEEP 2xe Urban Camou. Both the sensitivity analysis and the solution obtained with the use of the PROMETHEE GDSS method confirmed that it is the optimal alternative, the least sensitive to changes in criteria weights and changes in the decision makers' compensation coefficients.

**Keywords:** e-micromobility; e-scooters; electric vehicles; PROSA GDSS; multi-criteria decision aid; MCDA; MCDM; compromise solution

#### **1. Introduction**

The development of the automotive industry significantly affects not only the comfort of travel for motorists, but also has a significant impact on the Earth. The amount of exhaust fumes emitted into the environment is constantly increasing. This is a huge problem in the further progress of civilization, having a destructive influence on the air, soil, and atmosphere. The Organization of the Petroleum Exporting Countries [1] estimates that the number of passenger cars will increase from nearly 870 million in 2009 to 1.76 billion in 2035. These data show how important it is to popularize alternative means of transportation. In the case of Poland, it is most important in the case of larger cities such as Warsaw, Pozna ´n, Gda ´nsk, Szczecin, Katowice, Kraków, or Wrocław. These cities are particularly vulnerable to vehicle exhaust fumes and the associated environmental pollution. Due to the structure of the Polish energy mix, in the winter season the pollution is additionally combined with the burning of coal in order to heat houses and generate energy in coal-fired power plants. All these factors create smog which is harmful to both human health and the environment.

The reduction of greenhouse gas (GHG) emissions resulting from the combustion of crude oil and coal is the first major step in meeting the requirements imposed by the European Union (EU) to combat climate change. Continuous automotive progress and the related greater demand for crude oil and gas, until a few years ago, accounted for approx. 60% of total energy consumption and GHG emissions in transport [2].

**Citation:** Ziemba, P.; Gago, I. Compromise Multi-Criteria Selection of E-Scooters for the Vehicle Sharing System in Poland. *Energies* **2022**, *15*, 5048. https://doi.org/10.3390/ en15145048

Academic Editor: Katarzyna Turo ´n

Received: 5 June 2022 Accepted: 6 July 2022 Published: 11 July 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Passenger transport requires decisive steps to meet the requirements imposed by the EU. In the light of the European Green Deal, the key task is to make Europe the first climate neutral continent [3]. The decarbonization of the transport sector is expected to contribute to the achievement of an 80% reduction in GHG emissions by 2050 [4]. EU countries impose restrictions on the movement of older, substandard means of transport in designated zones, e.g., in cities such as Berlin or Paris. In Poland, these standards are also beginning to take shape and are described in the Act on electromobility and alternative fuels [5]. This is important due to the fact that Poland has the second largest percentage of cars over 10 years of age (approx. 73%) in the EU [6]. This impacts significantly the number of exhaust gases emitted into the environment [7]. Older cars, which do not have a Diesel Particulate Filter, emit much more soot, i.e., solid particles, and if they enter the circulatory or respiratory system, they can cause cancer.

Another important factor determining the need to use alternative means of transport is the difficult access to raw materials necessary in the transport process, such as crude oil, its derivatives, and gas. So far, Russia has been the main supplier of both gas and crude oil to Poland. Due to the war in Ukraine and the sanctions imposed on Russia as a result of the war, the prices of oil and gas have increased significantly. The lack of independence in obtaining such important raw materials has caused unstable fuel prices throughout Europe, including in Poland.

The context mentioned above indicates the need to change the structure of means of transport used in urban traffic. A potential direction for such a change is the development of electro-micromobility (e-micromobility) and the use of e-scooters and e-bikes, implementing the idea of green cities [8]. Interest in such vehicles on the part of individual users has been growing dynamically in recent years [9]. There are also more and more rental companies of this type of vehicles in larger cities [10]. The advantage of using e-bikes and e-scooters is the relief of the vehicle user compared to a traditional bicycle or scooter. Thanks to the use of electric means of microtransport, you can cover a longer distance without losing your strength and park these vehicles practically anywhere (which is not possible in the case of cars) [11]. The use of e-vehicles additionally allows you to travel from several to several dozen kilometers on a single battery charge. E-bikes and e-scooters are also an ideal choice for people with health problems who want to gradually start playing sports [9,12]. The hybrid ability to ride these vehicles allows you to switch to assist mode in order to regain strength. Manufacturers of vehicles such as e-bikes and e-scooters are releasing newer and newer functionalities on the market, e.g., the ability to synchronize with a smartphone and share the distances covered with others. Using e-bikes and e-scooters to travel to and from work, users do not waste time waiting in traffic jams. In large, crowded cities, this is a particularly important advantage. In addition, people using e-bikes and e-scooters to travel to work do not waste energy to cover the distance, which would be the case with a traditional bicycle or scooter. Although the environmental benefits of using e-micromobility are debatable, there is a consensus among researchers that switching from cars and motorcycles to e-micromobility would result in an overall reduction in GHG emissions [13]. A natural way of introducing e-micromobility in cities are, in turn, sharing stations, which make vehicles available to users quickly and cheaply, while also providing other benefits, such as creating jobs, stimulating economic growth, etc. [14]. In the case of such sharing systems, it is important to respect the points of view of all stakeholders, i.e., both investors and users [15]. Due to the wide availability on the Polish market and frequent use by sharing stations in Poland [16], this study considers e-scooters as the primary means of e-microtransport in cities.

The aim of the article is to analyze selected e-scooters available on the Polish market and to recommend the most useful vehicles of this type. The practical contribution of the article is to consider e-scooters both from the perspective of the individual user as well as the owner of the vehicle sharing system (VSS). Each of these stakeholders has different preferences when choosing a vehicle (fleet of vehicles). It is important to find a compromise between the owner and the VSS customer so that each is satisfied with the vehicle they use. The multi-criteria decision aid (MCDA) method called PROMETHEE for Sustainability Assessment—Group Decision Support System (PROSA GDSS) was used to identify e-scooters taking into account the preferences of both stakeholders [17]. PROSA GDSS supports groups decisions by rewarding compromise solutions and punishing unbalanced solutions between stakeholders. The use of this method in the context of seeking a compromise between two contradictory perspectives (VSS customer and owner) is a methodological contribution of the article. The resulting compromise was visualized graphically using the PROSA Geometrical Analysis for Interactive Assistance (GAIA) plane. The article is divided into seven sections. Section 2 provides an overview of contemporary research related to micromobility and e-micromobility. Section 3 discusses the research procedure and the methods used. The research results are presented in Section 4 and the results are discussed in Section 5. Section 6 contains managerial and environmental implications, and the article's conclusions are presented in Section 7.

#### **2. Literature Review**

In recent years, the interest in research on micromobility and e-micromobility has grown significantly. This is confirmed by the dynamically growing number of research papers on this subject [18,19]. MCDA methods are also used more and more often in such studies, both in the case of classic micromobility and e-micromobility. Tian et al. [20] developed a decision support framework for bike-sharing programs. The framework is based on the fuzzy BWM and MDM methods, which were used to weigh the criteria, and the MULTI-MOORA method, with which preferences were aggregated. Karolemeas et al. [21] prepared an index based on the AHP method for the planning of bicycle routes in the existing road network. In turn, in the studies by Kurniadhini and Roychansyah [22], Kabak et al. [23], Eren and Katanalp [24], and Guler and Yomralioglu [25], the potential locations of the bike-sharing system stations were considered and assessed. In the aforementioned studies, the aggregation of multi-criteria preferences was carried out using various MCDA methods, which were, respectively: SMCA, MULTIMOORA, VIKOR, and TOPSIS. In each of these studies, the AHP method was used to obtain the criteria weights, and in the publication by Guler and Yomralioglu [25], the criteria were additionally weighed using the BWM and fuzzy AHP methods. The above-mentioned articles dealt with decision-making problems related to conventional bikes, while the following papers mainly related to e-micromobility. Fazio et al. [26] used the SMCA method to study the adjustment of the road network to the needs of e-scooters. Kalakoni et al. [27] developed an environment matching index for different types of micromobility based on the AHP method. Using the developed index, they adjusted the appropriate micromobility and e-micromobility vehicles for individual areas. Torkayesh and Deveci [28] proposed a TRUST-based location assessment framework for battery swapping stations for e-scooters. Tang and Yang [29] used the interval-valued Pythagorean fuzzy preference relation to choose a supplier of e-bikes recycling. Deveci et al. [30] dealt with the issue of safety assessment of e-scooters using the fuzzy LAAW and qROFS Einstein WASPAS methods. Bajec et al. [15] using a set of DAHP and DEA methods selected the supplier of the e-bike-sharing system. Wankmüller et al. [31] used the BWM method to identify criteria relevant to the selection of e-micromobility vehicles for mountain rescue. Finally, Sałabun et al. [32] using the COMET method chose e-bikes for sustainable urban transport. More detailed applications of MCDA methods in research on micromobility and e-micromobility are presented in Table 1.


**Table 1.** Applications of MCDA methods in decision problems related to microtransport.

BSP—Bike-Sharing Program, BSS—Bicycle/Bike-Sharing System, BL—Bicycle Lane, BSST—Battery Swapping Stations, DSS—Decision Support System, GIS—Geographic Information System, BWM- Best-Worst Method, MDM—Maximizing Deviation Method, MULTIMOORA—Multi-Objective Optimization by Ratio Analysis plus the Full Multiplicative Form, AHP—Analytic Hierarchy Process, SMCA—Spatial Multi-Criteria Analysis, VIKOR— Multicriteria Optimization and Compromise Solution (Visekriterijumska Optimizacija i Kompromisno Resenje), TOPSIS—Technique for Order of Preference by Similarity to Ideal Solution, TRUST—Multi-Normalization Multi-Distance Assessment, IVPFIDM—Interval-Valued Pythagorean Fuzzy Information Decision-Making Approach, DAHP—Distance-based AHP, DEA—Data Envelopment Analysis, LAAW—Logarithmic Additive Assessment of the Weight Coefficients, qROFS—q-Rung Orthopair Fuzzy Sets, WASPAS—Weighted Aggregated Sum Product Assessment, COMET—Characteristic Objects Method.

When analyzing Table 1, it is easy to notice that there are few publications in which vehicles belonging to the e-micromobility category were assessed in a multi-criteria evaluation. Such issues appear only in the work of Sałabun et al. [32], where e-bikes were considered. The works of Wankmüller et al. [31] and Bajec et al. [15] are also partially similar to this topic. The first of these articles analyzed the potential criteria for selecting various e-micromobility solutions. In the second article, in the context of choosing the e-bike-sharing system supplier, there were also criteria that directly refer to the vehicles offered by the suppliers. Therefore, a research gap is visible in the topic of selecting e-scooters for the needs of individual users or VSSs. The second research gap is that few studies take into account the different perspectives represented by individual stakeholders. Only in the articles by Tian et al. [20], Tang and Yang [29], and Deveci et al. [30] was a group assessment approach used. Nevertheless, in each of these articles, the decision was the result of the views of field experts (entrepreneurs, academic professors, officials), and to the best of our knowledge, in the context of e-micromobility, no study has been conducted so far taking into account the contrary views of VSSs customers and owners.

#### **3. Materials and Methods**

*3.1. Research Approach*

The research scheme was based on the PROSA GDSS method, consisting of three stages [17]:


In the first stage, a set of decision alternatives *A* = {*a*1, *a*2,..., *am*} is defined, containing the acceptable alternatives (variants), from among which the alternative that best satisfies the decision makers is selected. This stage also specifies a set of criteria for evaluating alternatives *C* = {*c*1, *c*2,..., *cn*}. On the basis of the sets *A* and *C*, the performance table *E* = *C*(*A*) is built, containing the performance of alternatives based on criteria. This matrix is the basis for the assessment for each of the *K* decision-makers (stakeholders, experts) belonging to the set *DM* = {*dm*1, *dm*2,..., *dmK*}.

The second stage is an individual assessment of the various alternatives by each of the decision makers. In this stage, you can use one of the methods belonging to the PROMETHEE / PROSA families. In the e-scooters study, the PROMETHEE II method was used due to the fact that it is computationally simple and, at the same time, sufficient to aggregate the criteria for the purposes of this study. The result of this stage are the values of *ςdmk* (*ai*) obtained for each alternative, separately for individual decision makers.

In the third stage, the values *ςdmk* (*ai*) are aggregated into a group assessment taking into account a compromise or balance between decision makers. An aggregation is made using the PROSA-C method, and the result are the *PSVnet*(*ai*) values obtained for each of the alternatives considered. Both in the second and third stage, numerical studies can be supported by graphical analyses using the PROMETHEE GAIA method in the second stage and PROSA GAIA in the third stage. The diagram of the research procedure is presented in Figure 1.

**Figure 1.** Diagram of the research procedure based on the PROSA GDSS method.

#### *3.2. PROMETHEE II Method*

The second stage of the proposed research approach is based on the PROMETHEE II method [33,34] in the variant using the single criterion net flow. Four steps are performed in this stage.

#### **Step 1. Calculating the deviations based on pair-wise comparisons.**

In this step, all alternatives from set *A* are compared in pairs in terms of successive criteria *ck* and for each such comparison the deviation *dk ai*, *aj* is determined, according to Formula (1):

$$d\_k(a\_i, a\_j) = c\_k(a\_i) - c\_k(a\_j), \ \forall i, j = 1, \dots, m, \ \forall k = 1, \dots, n \tag{1}$$

where *ck*(*ai*) is the rating/performance of the alternative *ai* in terms of the *ck* criterion.

#### **Step 2. Applying the preference functions.**

For each *k*-th criterion, preference functions *Fk* are selected, allowing for the conversion of the deviation *dk* to the normalized preference value *Pk* ∈ [0, 1], according to Formula (2):

$$P\_k(a\_i, a\_j) = F\_k\left[d\_k(a\_i, a\_j) \mid , \forall i, j = 1, \dots, m, \ \forall k = 1, \dots, n\right] \tag{2}$$

Six different preference functions as shown in Figure 2 can be used in this step.

**Figure 2.** Preference functions used in the PROMETHEE method.

These functions are described by Formulas (3)–(8), where the following thresholds are used in selected functions: *qk*—indifference, *pk*—preference, *rk*—Gaussian.

1. Usual criterion (true criterion) (3):

$$P\_k(a\_{i\prime}a\_j) = \begin{cases} 0, \text{ for } d\_k(a\_{i\prime}a\_j) \le 0\\ 1, \text{ for } d\_k(a\_{i\prime}a\_j) > 0 \end{cases} \tag{3}$$

2. U-shape criterion (semi-criterion) (4):

$$P\_k(a\_i, a\_j) = \begin{cases} 0, \text{ for } d\_k(a\_i, a\_j) \le q\_k \\ 1, \text{ for } d\_k(a\_i, a\_j) > q\_k \end{cases} \tag{4}$$

3. V-shape criterion (pre-criterion) (5):

$$P\_k(a\_i, a\_j) = \begin{cases} 0, & \text{for } d\_k(a\_i, a\_j) \le 0 \\ \frac{d\_k(a\_i, a\_j)}{p\_k}, & \text{for } 0 < d\_k(a\_i, a\_j) \le p\_k \\ 1, & \text{for } d\_k(a\_i, a\_j) > p\_k \end{cases} \tag{5}$$

4. Level criterion (6):

$$P\_k(a\_{i\prime}, a\_j) = \begin{cases} 0, & \text{for } d\_k(a\_{i\prime}, a\_j) \le q\_j \\ \frac{1}{2}, & \text{for } q\_k < d\_k(a\_{i\prime}, a\_j) \le p\_k \\ 1, & \text{for } d\_k(a\_{i\prime}, a\_j) > p\_k \end{cases} \tag{6}$$

5. V-shape with indifference criterion (pseudo-criterion) (7):

$$P\_k(a\_i, a\_j) = \begin{cases} 0, & \text{for } d\_k(a\_i, a\_j) \le q\_k \\ \frac{d\_k(a\_i, a\_j) - q\_k}{p\_k - q\_k}, & \text{for } q\_k < d\_k(a\_i, a\_j) \le p\_k \\ 1, & \text{for } d\_k(a\_i, a\_j) > p\_k \end{cases} \tag{7}$$

6. Gaussian Criterion (8):

$$P\_k(a\_i, a\_j) = \begin{cases} 0, & \text{for } d\_k(a\_i, a\_j) \le 0 \\\ 1 - \exp\left(\frac{-d\_k\left(a\_i a\_j\right)^2}{2r\_k^2}\right), & \text{for } d\_k(a\_i, a\_j) > 0 \end{cases} \tag{8}$$

#### **Step 3. Calculating net outranking flows for individual criteria.**

Based on the preference value *Pk*, the net outranking flow of alternative *ai* over each other alternative for the *k*-th criterion is calculated, using Formula (9):

$$\phi\_k(a\_i) = \frac{1}{m-1} \sum\_{j=1}^{m} \left[ P\_k(a\_i, a\_j) - P\_k(a\_j, a\_i) \right], \forall i = 1, \dots, m, \ \forall k = 1, \dots, n \tag{9}$$

The *φ<sup>k</sup>* values allow you to order the alternatives separately for each criterion.

#### **Step 4. Calculating the global net outranking flow.**

The global net outranking flow for each of the alternatives is determined on the basis of Formula (10):

$$\phi\_{n\ell}(a\_i) = \sum\_{k=1}^{n} \phi\_k(a\_i) \ w\_{k\prime} \forall i = 1, \dots, m \tag{10}$$

where *wk* is the weight of the *<sup>k</sup>*-th criterion. Weights are normalized ( *<sup>n</sup>* ∑ *k*=1 *wk* = 1). The obtained values of *φnet* are the final solution according to the PROMETHEE II method, and in the PROSA GDSS method they are the results obtained by each of the decision makers separately. Therefore, for each *k*-th decision-maker and for the *i*-th alternative, there is an
