The Location Optimization of Urban Shared New Energy Vehicles Based on P-Median Model: The Example of Xuzhou City, China

Abstract

Sharing new energy vehicles is crucial for addressing the issue of traditional vehicles’ carbon emissions, reducing urban traffic congestion, safeguarding the environment, and promoting citizens’ use of green transportation. However, the parking lot’s drawbacks—poor location, challenging parking, and difficulty finding a car—lead to a low popularity rate, few users, and infrequent use. How to scientifically choose parking outlets and maximize the advantages of sharing new energy vehicles has become an important topic in current urban traffic management. This paper constructed a “G-B-U” framework starting with quasi-public goods and stakeholders to analyze the factors influencing the location selection of these vehicles. On this basis, a three-stage location decision method of “market demand prediction—alternative network screening—location model solution” is proposed to optimize the location selection of shared new energy vehicles. The factors are analyzed, and numerical examples are studied, using the districts of Xuzhou City in China as examples: Gulou, Yunlong, and Quanshan. The findings indicate that the main variables influencing how frequently Xuzhou residents use shared new energy cars are network dispersion, rental and return convenience, and usage experience. After site selection optimization, the journey distance is nearly cut in half, saving users a significant amount of travel time. It may meet the travel needs of residents better based on the same number of parking lots.

1. Introduction

Most cities currently experience the “high-density gathering” phenomenon. Infrastructure for urban transportation cannot be developed at the same rate as the rising demand for transportation. According to the Ministry of Public Security, China had owned 319 million cars by the end of December 2022, accounting for 76.59% of all motor vehicles, with 464 million people driving those cars (http://www.gov.cn/shuju/2023-01/11/content_5736278.htm accessed on 12 April 2023). Congestion in traffic is becoming worse and worse. The average road network peak travel delay index for China’s 50 largest cities was 1.609 in 2020, and it had climbed to 60% by 2021. Only 2% of locations affected by the new coronavirus epidemic saw an increase in traffic in 2022 (https://report.amap.com/index.do accessed on 12 April 2023). Overall, though, traffic congestion will only become worse as pandemic control becomes more commonplace. A driver’s license has grown more rapidly than automobile ownership, so car ownership will likely expand more rapidly in the future. In addition, the growing fleet of urban fuel-powered vehicles significantly threatens both traffic congestion and urban air pollution due to the high exhaust emissions. It is crucial that shared new energy vehicles are made more widely available for the city’s development to be healthy and sustainable. Shared new energy vehicles are electric vehicles that consumers use on the appropriate platform over the course of a brief rental period. These vehicles not only promote environmentally friendly transportation for locals and address the long-standing issue of traditional vehicles’ carbon emissions, but they can also effectively utilize human resources, conserve urban land resources, and reduce traffic congestion. It has the inherent benefits of significant development and is one of the primary fields of the future sharing economy against the backdrop of the internet.

Shared new energy vehicles are becoming increasingly integrated into people’s daily mobility, from short-term rental sharing services such as Airbnb and Uber to the introduction of apps such as Panda Car and EVCARD. Although the market for shared new energy electric vehicles has grown quickly recently, there are unavoidably challenging operational issues. Due to issues with high startup and operating costs as well as limited land resources, pertinent businesses must plan their development. Consumers’ unsatisfied with the user experience caused by “difficult to find, park, and charge a car” (Wei et al., 2020 [1]) decreases their inclination to use, which inhibits the development of shared new energy cars. A few businesses, including EZZY and others, have shut down, Car2Go has left the Chinese market, and Panda is experiencing operational issues.

Therefore, the “G-B-U” triangle theoretical framework based on the “stakeholder—quasi-public goods” perspective is constructed in this study. This paper analyzes the factors that influence the location choice of shared new energy vehicles based on geographic location, enterprise cost and benefit, user demand, government policy, and other factors, which corrects the shortcomings of previous studies that are primarily based on the single dimension of economic benefits. In addition to providing a comprehensive integration of urban transportation, low-carbon environmental protection, and resident convenience, this paper offers a systematic analysis framework for the location deconstruction and industrial layout of urban shared new energy vehicles and theoretically advances the research on the integration of the sharing economy, urban economy, logistics site selection, and public management. Based on this, this paper uses the P-median model to optimize the location selection of urban shared electric vehicles in three districts, which are Gulou, Yunlong, and Quanshan in Xuzhou, a city in the Jiangsu province of China. It hopes to serve as a reference for the scientific location selection of urban shared new energy electric vehicles.

2. Research Status

Now, the creation of location models and influencing factors, as well as the use intention of consumers of shared new energy vehicles, are the key areas of domestic and international researchers’ research on shared new energy vehicles.

Based on social and economic aspects, researchers examined the variables influencing the demand for car-sharing. Sun Lishan et al. (2020) [2] developed the demand prediction model of the land use index and extensively examined the relationship between the demand for car-sharing and the distribution traits of the nearby land use. Pucci Paola (2021) [3] researched how metropolitan spatial patterns affect consumer demand for electric vehicles. Using 13 variables, including journey features, personal attributes, and perceived attributes, Zhang Ronghua et al. (2022) [4] conducted a quantitative analysis of the factors influencing consumers’ use of shared electric vehicles based on the theory of planned behavior. The empirical findings revealed that the most significant aspects influencing travelers’ attitudes toward electric vehicles were whether the charge was affordable, the trip was comfortable, and the process of renting was simple. Their intention to use is significantly influenced by subjective norms, attitudes, perceived risk, and control practices. The extended technology acceptance model “belief-attitudinal-intention” was used by Jaiswal et al. (2021) [5] and other researchers to demonstrate how attitudes, perceptions of utility, usability, and risk can either directly or indirectly influence an individual’s decision to adopt an electric car. The one with the biggest effect on consumers’ use intentions is perceived usefulness. Lee et al. (2020) [6] optimized the design of the SAEV system with the goal of minimizing total cost and satisfying customer waiting time. Young individuals, those with low and intermediate incomes, those with higher education levels, and others are more likely to join car-sharing, as noted in studies by Huang Youlin (2021) [7]. These articles have made great contributions to guiding the operation and development of car-sharing.

Urban sharing new energy cars are a very convenient new form of transportation for locals. However, as it grows, it also encountered a few issues, such as a lack of parking spaces, insufficient charging piles, hazards to automobile safety, high operating costs of businesses, a lack of funding, and ambiguous legal obligations (Wei et al., 2020 [8]; Gao, 2020 [9]; Wang et al., 2020 [10]). According to Zhang Jingwen (2020) [11], to understand the operation and management style of local car-sharing enterprises and compare the degree of compatibility between the operation strategy of the enterprises and the current situation of the city, comprehensive factors such as urban traffic, residents, and enterprises should be investigated and analyzed. Researchers are looking at the placement choice of shared automobiles from a variety of angles, including those of businesses, users, power grids, and overall social benefits, in an effort to break the shared new energy vehicle development bottleneck. From an enterprise’s perspective, building a location model is primarily focused on minimizing costs and maximizing profits. To calculate the ideal quantity, capacity, location, and development time of EV parking lots in a distribution system, Mirzaei M. J. and Kazemi A. (2020) [12] used a dynamic programming approach. Their findings demonstrate that the dynamic programming strategy they provide can assist businesses in maximizing revenues while taking system and financial constraints into account. Based on the user utility maximization perspective, Guo et al.’s (2018) [13] study on the location of charging stations considered consumers’ distance anxiety and deviation. A model created by Zenginis et al. (2016) [14] and validated through a simulation enabled parking charging stations to service more clients without lengthening queue wait times. Based on the grid level, Li Ran et al. (2021) [15] created an optimal planning model and solved it with the goal of reducing network losses and voltage shifts in the distribution system. From the perspective of system management, the key requirement of a location optimization system is high energy efficiency and low cost (Ghorpade et al., 2021) [16]. To reduce the overall societal cost, Liu Haoxiang et al. (2021) [17] developed a “two-layer model” based on the charging procedure, driving route, and charging behavior. They also concentrated on the ideal location of EV charging facilities.

Previous research has examined customer demand preferences, network selection modeling techniques, and the effect factors related to the location selection of shared new energy electric vehicles. However, most current studies concentrate on one-dimensional scattered analysis, and there is currently no theoretical direction or organized framework for shared automobile placement (Mirzaei et al., 2020 [12]; Li et al., 2021 [18]). As a result, this study develops an analytical framework of the “G-B-U” theory based on the “stakeholder—quasi-public goods” perspective, which provides a theoretical framework for the examination of the variables that affect the market demand for shared new energy vehicles and the selection of the parking lots. In addition, the earlier models for predicting market demand were primarily built using straightforward demographic data. An in-depth investigation is used in this work to attempt to assess the elements that affect the consumption and use of urban shared new energy vehicles. A market demand prediction model for these vehicles is built based on two dimensions of social demographic characteristics and user intentions, particularly the relationship between shared new energy vehicle usage experience on subsequent use intentions. This approach significantly increases the accuracy of market demand prediction for urban shared new energy vehicles when compared with the prior prediction model based solely on population size and density.

3. Construction of Theoretical Analysis Framework for Shared New Energy Vehicles

3.1. The Quasi-Public Product Attributes of Shared New Energy Vehicles

There are three ways to supply modern public goods: independent supply by the government, purchase of public services by the government, and supply by multiple subjects (Yang, 2013 [19]). However, whether the supplier is the public sector or the private sector cannot be used as a basis for judging quasi-public goods. It is about whether the product on offer has the characteristics of non-competitive and non-exclusive, which are the most prominent feature of public goods. However, competitive and segmented quasi-public products are not exclusive. Shared transportation is a vital component of contemporary urban transportation and includes typical features of quasi-public goods. Consumption of shared bikes is non-competitive if the availability of shared bikes in a given area of a city is sufficient to meet user demand, the use of shared bikes by one consumer does not reduce the use of shared bikes by other consumers, and the marginal cost of the shared bike business enterprise to provide shared bikes to another user is zero (Guo, 2019 [20]). Nonetheless, shared bikes exhibit some exclusivity and competitiveness of private goods when market demand exceeds supply quantity in a particular area of the city, which is considered to be limited and non-competitive for shared bikes.

Drawing on the above analysis, shared new energy vehicles have the typical indivisibility but also have the limited non-competitiveness of quasi-public goods. Therefore, this paper considers that urban shared new energy vehicles are quasi-public transportation products under the background of sharing economy.

3.2. Analysis of Stakeholders of Sharing New Energy Vehicles

As a quasi-public good, the site selection of urban shared electric vehicles is affected by many factors from the perspective of the main stakeholders involved, including the government, businesses, and users. Any urban transportation facility’s modernized development is inextricably linked to the government’s strategic planning and direction. The cost and profitability of new energy vehicle companies, which serve as the operating unit for urban sharing new energy cars, are also major determinants of how these vehicles will be developed and how they will be laid out in the city. Urban sharing new energy vehicle development planning and network layout, whether motivated by governmental governance or corporate profit, ultimately satisfy market demand, notably the travel requirements of urban residents and users. Thus, the above three stakeholders must be prioritized in order to examine the network layout and placement of urban sharing new energy vehicles.

3.3. Construction of Theoretical Analysis Framework for Location Selection of Urban Shared New Energy Vehicles

The aforementioned section summarizes the quasi-public product characteristic of city-sharing new-energy cars and its important constituents, namely governments, business, and users. This serves as the theoretical foundation for this paper’s construction of the “G-B-U” analysis framework, which is used to analyze the variables influencing the supply and layout of these vehicles. The government (location characteristics and policy regulations), operation platform enterprises (costs and advantages), and users (demands) will all be thoroughly examined in this framework as elements impacting the location choice of new urban energy vehicles.

4. Influencing Factors of Choosing the Location of New Energy Electric Vehicles

Based on the above theoretical analysis, this paper mainly considers the main factors affecting the location selection of urban shared new energy electric vehicles from the perspectives of government, businesses, and users.

4.1. Location Factors

Yang Dingwu (2017) [21], Luo Dong (2018) [22], and Liu Guohua (2021) [23] found in their research that location characteristics are very important influencing factors in the location selection of logistics centers, commercial hotels, and shared bikes, which mainly have the following specific influences.

First, the location of the parking lots of shared new energy vehicles for urban transportation should be customized to the overall design and development of the city. The network’s location should be convenient for switching between different modes of transportation on the one hand while avoiding encroaching on public transportation road resources and the already congested urban parking lots on the other.

Second, the placement of the shared new-energy cars directly affects the company’s operating profit. In densely populated places, such as shopping malls, schools, office buildings, and residential communities, the demand and frequency of sharing new energy vehicles are higher (China Journal of Highway and Transport, 2017 [24]).

Thirdly, for users, if the location of shared new energy cars is in a flat and open area that is simple to discover, then it will be easier for users to locate outlets quickly and offer convenience.

Therefore, this paper preliminary determined the placement of alternative locations near 10 different types of areas, as illustrated in Figure 1, based on careful consideration of the aforementioned factors.

4.2. Enterprise Factors

The shared new energy automotive sector has not had a long development period. It lacks sophisticated theoretical support and management techniques as compared with the conventional vehicle sector. Enterprise management is the coordination of a company’s employees, vehicles, equipment and charging stations, parking spaces, and other elements. Enterprises must pay for software development and repair, operating personnel salaries, and other expenses in addition to the cost of purchasing vehicles, charging infrastructure, and equipment. Enterprises have comparatively less income, and a number of enterprises closed down because their income does not offset their expenses.

When choosing the location, the operating and construction costs should be as low as possible because of the large initial investment capital. In order to reap benefits, businesses should include as many users as possible in the network layout and boost vehicle usage.

4.3. Demand Factors

Through the Wenjuanxing platform, the link (https://www.wjx.cn/vm/t7v9U0o.aspx accessed on 12 April 2023) to the survey was distributed to the inhabitants of Xuzhou, and the factors influencing their use of shared new energy cars were examined. In total, 420 questionnaires were given out for this study, and 395 of them were valid, accounting for 96.58%. SPSS was used to process the data, and the following findings were obtained (considering the length of the paper, the detailed data are not revealed here):

(1) Among the participants, just 17.47 percent had used shared new energy vehicles. They think sharing new energy vehicles can address issues, such as a shortage of personal vehicles, parking challenges, the inconvenience of public transit, and lengthy commutes. The findings also demonstrate that shared new energy vehicle penetration is modest. A total of 80% of the participants possessed a driver’s license, while 12% did not. A total of 11% of the subjects did not have a car at home, according to the statistics on family automobile ownership, indicating that there are more potential users of these vehicles and broad market potential.

(2) According to the survey, the participants using experience primarily care about the convenience of renting and returning, and the secondary influencing factors are travel comfort, expense, and urgency. The ease of renting and returning remained the top concern for individuals without using experience. However, different from those with user experience, for the individuals who had never driven such shared cars, the potential risk, along with the urgency, cost, and comfort of the trip, was the second most crucial factor. Simply said, whether a shared new energy vehicle is used or not, the ease of rental and return should be a priority. Uneven distribution, too few parking lots, a small number of cars, and not being as convenient as other modes of transportation are the reasons that hinder many participants from using shared new energy vehicles.

(3) The respondents evaluated their likelihood and frequency of utilizing shared new energy cars in the future after being asked about the elements considered when using such vehicles. Respondents who have used such vehicles before are more likely to do so in the future. This shows that most electric shared vehicle users are content with their experiences and have high expectations for their continued use. Further cross-tabulations and model regressions were carried out to assess future use willingness with other characteristics. The findings indicated that individuals’ future readiness to utilize shared new energy cars was highly influenced by criteria such as age, low carbon, environmental protection, risk of use, and prevalence.

4.4. Policy Factors

The environmental problem is represented by the “haze” that is seriously plaguing China’s urban air quality. Existing studies show that traditional cars burning fossil energy are one of the sources of haze, which causes pollution to the environment. Second, the gradual increase in personal car ownership has put great pressure on urban traffic. Based on environmental protection and alleviating traffic congestion, the state has introduced several policies to support the development of shared new energy electric vehicles (Cao et al., 2021 [25]). All over the country have also responded to the call to refine relevant policies to help green travel. Some of the policy documents are shown in

Table 1. National and local policies on new energy vehicles.

Time	Name	Network Requirements
November 2018	<Guangzhou…Guidance>	Interchange (P + R) parking lots, public parking lots at parks and attractions, transportation hubs, government centers, hotels, etc.
August 2019	<Green travel…the future will increase charging facilities investment (2019–2022)>	The future will increase investment in charging facilities to solve the problem of the range of new energy vehicles.
…….
February 2021	<General Office of the Ministry of Commerce … notice>	Facilitate the replacement of new energy vehicles, encourage places that are in a position to do so to introduce subsidies for the construction and operation of replacement infrastructure, support the construction of replacement infrastructure relying on gas stations, highway service areas, street lights, etc.
January 2022	<Implementation Opinions on Further…Energy Regulation (2022) No. 53)>	Charging facilities for residential communities, central urban areas, enterprises and institutions, commercial buildings, transportation hubs, public parking lots, and industrial parks.

.

As seen in Table 1, the Chinese government and local authorities have released policies to direct the location categories for shared electric vehicles, including transfer parking lots, parks and scenic areas, etc. This clarifies the essential role of these vehicles in urban transportation, effectively guaranteeing and promoting the development of shared new-energy vehicles.

5. Case Analysis

Xuzhou is the sub-central city of Jiangsu province, known as the “transportation center in five provinces”. Xuzhou has actively responded to the national call and issued a series of policies to promote the development of the new energy vehicle industry. For example, “Notice on Local financial subsidies for the promotion and application of new energy vehicles in 2018–2020” provides subsidies to operating units. Data from the Xuzhou Statistical Yearbook in 2021 and the 7th national census in November 2020 show that the populations of Gulou District, Quanshan District, and Yunlong District are 638,534, 619,653, and 421,157, respectively. Jiawang District and Tongshan District have comparatively low population densities. Therefore, this study takes the Gulou District, Yunlong District, and Quanshan District of Xuzhou City as examples to demonstrate the location selection model of new energy-sharing vehicles.

5.1. Determining the Requirement Points

According to the daily travel range of residents, Yunlong District, Gulou District, and Quanshan District of Xuzhou was divided into 140 research areas with 2500 m × 2500 m on the map as the minimum research unit. The demand for this incomplete area is transferred to the closest research unit because some grids are partial, and the study area is tiny. A total of 110 demand points was eventually confirmed by using the geometric center point of each research unit as the demand point designating the research unit. Demand points were labeled, and their longitude and latitude were calculated using the ArcGIS10.2 program. The longitude and latitude of demand points are shown in

Table 2. Longitude and latitude information of demand points (part).

Demand Point	Longitude	Latitude	Demand Point	Longitude	Latitude
1	117.163346	34.26166	56	117.298093	34.18705
2	117.185804	34.28031	57	117.253151	34.16842
3	117.140888	34.26166	58	117.298093	34.20571
4	117.185804	34.26166	59	117.097928	34.30008
5	117.140888	34.28031	60	117.298093	34.22436
6	117.185804	34.29895	61	117.275635	34.31759
7	117.163346	34.28031	62	117.231963	34.18614
……
50	117.275635	34.22436	105	117.094685	34.34986
51	117.298093	34.24302	106	117.318300	34.15089
52	117.253178	34.20571	107	117.299975	34.15268
53	117.275635	34.18705	108	117.364403	34.31246
54	117.275635	34.20571	109	117.364403	34.31246
55	117.253178	34.18705	110	117.298527	34.3359

, and the network distribution diagram of demand points is shown in Figure 2.

5.2. Determining Alternative Network Points

This study uses 10 types of locations as potential selection sites, including schools, hospitals, large shopping malls, public institutions, tourist attractions, transportation hubs, business office buildings, residential communities, large business hotels, and industrial parks. These locations were chosen based on survey results, policy planning, and practical site selection factors. The crawler tool was used to find the latitude and longitude of these 250 locations on the Baidu Map and Amap Map, and the data were imported into ArcGIS10.2 software and displayed on the maps of Gulou District, Quanshan District, and Yunlong District. The coordinates of alternative dots are shown in

Table 3. Coordinates of alternative dots (part).

SN	LON	LAT	SN	LON	LAT	SN	LON	LAT	SN	LON	LAT
1	117.2	34.21	19	117.17	34.27	215	117.29	34.22	233	117.2	34.23
2	117.19	34.26	20	117.17	34.23	216	117.19	34.28	234	117.27	34.16
3	117.23	34.29	21	117.2	34.23	217	117.19	34.27	235	117.25	34.26
4	117.29	34.19	22	117.2	34.24	218	117.15	34.27	236	117.22	34.27
5	117.21	34.26	23	117.29	34.21	219	117.25	34.26	237	117.31	34.27
6	117.27	34.25	24	117.18	34.27	220	117.19	34.27	238	117.17	34.27
7	117.16	34.24	25	117.19	34.29	221	117.19	34.27	239	117.3	34.21
8	117.27	34.3	26	117.26	34.26	222	117.19	34.27	240	117.31	34.27
9	117.19	34.26	27	117.21	34.22	223	117.29	34.21	241	117.15	34.28
10	117.14	34.31	28	117.29	34.21	224	117.2	34.27	242	117.2	34.27
11	117.2	34.23	29	117.25	34.21	225	117.19	34.27	243	117.21	34.27
12	117.19	34.25	30	117.29	34.21	226	117.25	34.25	244	117.23	34.23
13	117.2	34.26	31	117.3	34.21	227	117.26	34.26	245	117.23	34.21
14	117.12	34.33	32	117.29	34.21	228	117.2	34.29	246	117.32	34.3
15	117.14	34.31	33	117.17	34.25	229	117.19	34.28	247	117.33	34.35
16	117.19	34.26	34	117.29	34.21	230	117.18	34.27	248	117.29	34.21
17	117.14	34.32	35	117.2	34.28	231	117.19	34.27	249	117.3	34.34
18	117.21	34.24	36	117.37	34.27	232	117.25	34.26	250	117.31	34.27

Note: SN: serial number; LON: longitude; and LAT: latitude.

, and the dot distribution is shown in Figure 2.

5.3. Population Prediction for Each Point

5.3.1. Calculate the Number of Potential User Groups of Each Street in Yunlong District, Quanshan District, and Gulou District

According to the population data provided by the Xuzhou Municipal Bureau of Statistics in May 2022, the total population of each district can be known (

Table 4. Total population, number of customers by district and subdistrict.

District	Street	Total Population	Potential Customer Group	Target Customer Group
Gulou District	Tongpei	60,565	37,350	25,370
	Pipa	57,193	35,271	23,957
	Pailou	42,981	26,506	18,004
	Jiuli	72,529	44,729	30,381
	Jinshanqiao	38,212	23,565	16,007
	Huanglou	32,773	20,211	13,728
	Huancheng	82,742	51,027	34,660
	Fengcai	76,360	47,091	31,986
	Donghuan	23,051	14,216	9656
	Damiao	82,314	50,763	34,480
	Dahuangshan	69,814	43,054	29,244
Quanshan District	Yongan	76,304	45,859	31,149
	XZ Quanshan Economic Development Zone	126	76	51
	Wangling	40,522	24,354	16,542
	Taoyuan	9266	5569	3783
	Taishan	50,553	30,382	20,637
	Sushan	23,559	14,159	9617
	Qiligou	13,188	7926	5384
	Pangzhuang	21,845	13,129	8918
	Kuishan	30,177	18,136	12,319
	Jinshan	24,912	14,972	10,170
	Huohua	40,900	24,581	16,696
	Hubin	59,889	35,993	24,448
	Heping	95,998	57,695	39,189
	Duanzhuang	54,475	32,739	22,238
	Zhaishan	77,939	46,841	31,817
Yunlong District	Zifang	52,337	32,428	22,026
	Pengcheng	43,934	27,222	18,490
	Pantang	12,326	7637	5187
	Luotuoshan	70,066	43,413	29,488
	Huangshan	89,266	55,309	37,568
	Huaihai	1461	905	615
	Dalonghu	92,519	57,325	38,937
	Daguozhuang	17,850	11,060	7512
	Cuipingshan	41,398	25,650	17,423

). According to the seventh national census data in 2020, the proportions of the population aged 15–59 in each district of Xuzhou are 64.84% in Gulou District, 63.11% in Quanshan District, and 64.97% in Yunlong District. Considering the legal age to obtain a driver’s license and residents’ regular driving age, this study takes individuals aged between 20 and 59 as potential customers. According to the news report of Xuzhou, the proportion of high school students is about 3.01% of the total citizens, which can be subtracted to estimate the proportion of potential customers in the three districts: 61.67% in Gulou District; 60.10% in Quanshan District; and 61.96% in Yunlong District.

According to Formula (1), the number of potential customers on each street can be calculated as follows:

$$H_{ij}=T_{ij}\times W_i$$

(1)

where i is the district number as i = {1, 2, 3}; j is the street number as j = {1, 2, 3, …, 15}; H_ij represents the number of potential customers of Street j in District i; T_ij represents the total population of street j in District i; and W_i represents the proportion of potential customers in District i. The results are shown in Table 4.

5.3.2. Calculate the Number of Target User Groups of Each Street in Yunlong District, Quanshan District, and Gulou District

According to the results of the questionnaire, user experience has a significant impact on the possibility of customers using shared electric vehicles in the future. This study employs the usage ratio in a questionnaire survey as the basis for calculating the number of users with experience in each district of Xuzhou. Similarly, 83% is used as the proportion of people who have no experience in sharing new energy vehicles in Xuzhou, and the target number of people on each street in each region is calculated based on the two data above. The calculation formula is as follows:

$$R_{ij}=H_{ij}\times \left(P_1\times K_1+P_2\times K_2\right)$$

(2)

where, i is the district number as i = {1, 2, 3}; j is the street number as j = {1, 2, 3, …, 15}; R_ij represents the number of target users of Street j in District i; H_ij represents the number of potential customers at Street j in District i; P₁ represents the proportion of users with shared travel experience in Xuzhou (P₁ = 17%); P₂ represents the proportion of people without shared travel experience in Xuzhou (P₂ = 83%); K₁ represents the proportion of users with sharing the experience of new energy vehicles who have the intention to use them in the future (K₁ = 82.2%); and K₂ represents the proportion of people who have no experience in using electric sharing new energy vehicles and are willing to use them in the future (K₂ = 65%). P₁, P₂, K₁, and K₂ are derived from the results of the preliminary questionnaire survey. The final target population of each street is shown in Table 4.

5.3.3. Calculate the Demand of Each Demand Point

Firstly, identify which streets are included in each research unit, then the proportion of the street area falling in the study area to the total street area is calculated, which is the reference standard for assigning the total target population of the street to the study area, and finally, the total population of the study area is added together as the target population of corresponding demand points. The study region, taking demand point 1 as an example, contains 0.34% of Duanzhuang Street, 29.6% of Heping Street, 37.22% of Wangling Street, and 64.49% of Yongan Street. The research area of demand point 1 is allotted to the target population of Duanzhuang Street, 22,238, at a ratio of 0.34%. That is, the target number of customers assigned to demand point 1 in Duanzhuang Street is 75. Similar to this, the objective number of people in Heping Street assigned to the demand point 1 research area is 11,599, Wangling Street 6157, and Yong’an Street 20,089. Hence, the target population at demand point 1 is 37,920.

Table 5. The demand quantity corresponds to each demand point.

Demand Point	Target Population	Demand Point	Target Population	Demand Point	Target Population	Demand Point	Target Population	Demand Point	Target Population
1	37,919	23	10,147	45	5523	67	2255	89	3273
2	32,327	24	9740	46	4045	68	2249	90	1638
3	31,500	25	8928	47	3572	69	2447	91	1638
4	27,327	26	8722	48	3834	70	2220	92	1638
5	20,686	27	7509	49	3551	71	2507	93	1638
6	19,662	28	7464	50	3065	72	2095	94	1887
7	18,506	29	7117	51	3017	73	2299	95	1617
8	18,373	30	7078	52	2986	74	1986	96	1915
9	17,439	31	7025	53	2976	75	2287	97	1428
10	17,408	32	6711	54	2976	76	1846	98	1391
11	17,234	33	6550	55	2976	77	1822	99	1339
12	16,164	34	6448	56	2975	78	1792	100	1323
13	15,745	35	6122	57	3227	79	1773	101	1342
14	15,739	36	5911	58	2900	80	1757	102	1375
15	14,010	37	5822	59	2876	81	2348	103	1136
16	13,502	38	5339	60	2803	82	1688	104	1124
17	13,181	39	5337	61	3621	83	1686	105	1740
18	12,428	40	5183	62	2774	84	1663	106	847
19	12,176	41	5098	63	2271	85	1660	107	713
20	12,091	42	4970	64	2271	86	1643	108	458
21	11,411	43	5996	65	2269	87	1638	109	348
22	11,333	44	4858	66	2264	88	1638	110	32

shows the target population of each demand point.

5.4. Calculate the Travel Distance between Each Demand Point and the Alternative Outlets

In the process of this research, the distance between the demand point a and the alternative point b is used as the main index to measure the travel cost. According to the coordinates of 110 demand points and 250 alternative points, the point spacing between different sites is calculated through the domain analysis module of ArcToolbox in ArcGIS10.2 software, and part of the point spacing is shown in

Table 6. Partial point spacing between demand points and alternative points.

	a	1	2	3	4	5	6	7
b
1		8553.79	10,134.35	9995.79	7675.85	11,987.59	12,609.18	10,814.51
2		2839.23	2441.27	5338.67	347.90	5860.05	4929.36	3728.27
3		7794.22	4633.69	10,115.79	5627.66	9598.43	4875.62	7109.88
4		16,630.43	16,398.10	18,730.56	14,656.35	20,122.70	18,316.29	18,184.07
5		8351.93	7825.21	7675.97	9648.26	5202.44	6471.38	6156.27
6		6036.34	8281.86	7197.69	5793.80	9318.15	10,775.45	8453.31
7		9306.81	10,448.37	10,970.83	8085.43	12,812.24	12,863.24	11,419.76
8		15,817.25	13,545.15	18,317.25	13,317.26	18,483.60	14,215.94	16,009.60
9		4163.52	5528.39	6159.27	3038.62	7698.47	8024.52	6218.10
10		8145.99	9911.71	9488.68	7428.26	11,536.81	12,401.81	10,460.40
11		9741.53	6417.39	11,989.51	7651.58	11,242.09	6026.54	8805.33
12		5714.06	2339.26	7909.02	3904.93	7264.72	2862.82	4782.69

(a represents the demand point and b represents the alternative point).

5.5. Construct and Solve the Location Model

This study aims to optimize the original 38 site selection points in the three districts and select 38 more suitable points among the alternative sites so as to minimize the total travel cost of all residents within the study scope.

Table 7. Information on existing charging stations.

SN	LON	LAT	SN	LON	LAT
1	117.209	34.317	20	117.267	34.271
2	117.129	34.306	21	117.18	34.234
3	117.227	34.206	22	117.231	34.241
4	117.276	34.204	23	117.359	34.265
5	117.234	34.271	24	117.243	34.288
6	117.184	34.276	25	117.252	34.255
7	117.153	34.259	26	117.28	34.22
8	117.255	34.289	27	117.151	34.268
9	117.182	34.252	28	117.283	34.27
10	117.294	34.209	29	117.289	34.203
11	117.23	34.242	30	117.274	34.204
12	117.218	34.225	31	117.171	34.256
13	117.181	34.288	32	117.205	34.296
14	117.125	34.259	33	117.154	34.265
15	117.166	34.238	34	117.144	34.293
16	117.234	34.242	35	117.282	34.204
17	117.2	34.288	36	117.217	34.202
18	117.133	34.257	37	117.285	34.259
19	117.147	34.277	38	117.221	34.246

Note: SN: serial number; LON: longitude; and LAT: latitude.

shows the information on existing charging stations.

In order to reduce the overall weighted travel distance between each demand point and the associated unique site selection point, the P-median model was utilized in this study to identify P ideal site selection locations among n alternative points. The journey distance of users to 110 demand sites is minimized by choosing 38 optimal site selection points from 250 alternative places in this study. The equation reads as follows:

$$\mathrm{MinZ}={\sum }_{a\in M}{\sum }_{b\in N}{D}_a{d}_{ab}{Y}_{ab}$$

(3)

$$\mathrm{s}.\mathrm{t}.\sum _{b\in N}{X}_b=P$$

(4)

$$\sum _{a\in N}{Y}_{ab}=1,a\in M$$

(5)

$$Y_{ab}\le X_b,a\in M,b\in N$$

(6)

$$X_a\in \left\{0,1\right\},b\in N$$

(7)

$$Y_{ab}\in \left\{0,1\right\},a\in M.b\in N$$

(8)

where Z represents the objective function; N is the set of 250 alternative nodes (N = {1, 2, ⋯, 250}); let M denote the set of 110 demand points (M = {1, 2, ⋯, 110}); a represents the sequence number of each demand point; b denotes the serial number of each alternative dot; P represents the number of sites (P = 38); D_a denotes the quantity demanded at demand point a; d_ab denotes the distance between the demand point a and the alternative point b; and X_b is 0/1 decision variable, where 0 represents that the location point is not established at point b, and 1 represents that the location point is established at point b. Y_ab is a 0/1 decision variable, where 0 means that location b does not serve demand a, and 1 means that location b provides service to demand a.

The site selection model of the charging pile and parking space alternative network in Yunlong District, Quanshan District, and Gulou District is designed according to D_a, d_ab, and P, which have been identified in the earlier stage, and in accordance with Formulas (3)–(8). The model is created first by calling Matlab’s yalmip solution toolbox. The final step in the approach is to apply the CPLEX 12.8 optimization solver. Finally, the optimal layout scheme of charging facilities in the three districts of Xuzhou is calculated, as shown in Figure 3. The corresponding demand points of all site selection points are shown in

Table 8. Information of location points and corresponding demand points after optimization.

SSO	LON	LAT	CDP	SSO	LON	LAT	CDP
3	117.227	34.286	14	143	117.183	34.281	2
4	117.286	34.19	53,54,56,58,75	145	117.211	34.299	19
			100,102	152	117.192	34.208	24
6	117.18	34.219	17,36,39	153	117.163	34.26	1
7	117.213	34.206	18	160	117.131	34.272	3,5
9	117.189	34.239	8	161	117.197	34.303	6,34,47
16	117.208	34.261	12	165	117.252	34.296	25,26,77
17	117.27	34.255	28,42,80	166	117.215	34.308	37,45,49
18	117.16	34.244	16,29,72,99	202	117.296	34.228	50,51,60,79,91
27	117.124	34.333	68,76,73,81,98				92,101
			104,105	203	117.163	34.282	7,22
28	117.144	34.314	30,33,35,40	207	117.266	34.16	57,69,95,106,107
31	117.206	34.242	10	208	117.251	34.26	21
51	117.207	34.224	27	213	117.203	34.279	11
65	117.371	34.27	85,90,94,96,103	221	117.243	34.221	31,44
73	117.117	34.271	15,23,38,43	233	117.235	34.208	41,52,55,62
			46,59	234	117.322	34.298	64,65,70,78,86
83	117.187	34.263	4	236	117.234	34.264	13
87	117.231	34.298	20	237	117.326	34.35	71,84,97,110
109	117.243	34.246	9,32	241	117.362	34.297	67,74,89,108,109
128	117.332	34.266	82,83,87,88,93	249	117.277	34.315	48,61,63,66

Note: SSO: site selection after optimization; LON: longitude; CDP: corresponding demand point; and LAT: latitude.

.

Before optimization, consumers traveled a total of 1,356,537,148.7 m (Z_front = 1,356,537,148.7 m) to the alternative network. After optimization, the total distance is 916,902,356.24 m (Zafter = 916,902,356.24 m). It is discovered that while the number of current charging piles stays the same, the weighted distance difference between the before and after optimization is 439,634,792.5 m. With a combined target population of 696,146 individuals across Gulou, Quanshan, and Yunlong Districts, the travel time per person has been cut by 631.5267 m. At an average walking speed of 5 km/h, this translates to a reduction of about 7.58 min.

6. Conclusions and Recommendations

6.1. Conclusions

This paper considers that shared new energy vehicles are quasi-public goods with indivisibility, limited competitiveness, and non-exclusivity and analyzes the three core stakeholders that affect the location of shared new energy vehicles, namely, the government, enterprises, and users. On this basis, the theoretical framework of “G-B-U “is constructed to guide the layout of such vehicles. Government plays a leading, coordinating, and supervising role in the layout planning process of new energy vehicles. Enterprises are builders and directly responsible parties. The user is the service terminal and is the direct decider in promoting the development.

Location and policy are the two key determining elements at the government level. Shared new energy vehicle locations should be chosen in accordance with urban development planning and in response to governmental directives. Cost and income are the two key variables at the enterprise level. This study primarily discusses the impact of user demand in terms of the user dimension and performs a questionnaire survey among Xuzhou locals.

This study develops a three-stage site selection optimization model that includes “market demand projection-alternative point selection-model construction and solution”. Xuzhou City’s Yunlong, Gulou, and Quanshan districts are chosen as the case study’s subjects. According to the findings, the overall weighted journey distance for users under the current architecture of 38 site selection points is 1,356,537,148.7 m, compared to 916,902,356.24 m for the improved data. The per capita travel distance is cut by 631 m, and the difference between the two is 439,634,792.5 m, demonstrating the logic and efficacy of the location selection approach.

6.2. Recommendations

The government can plan the development layout of the new energy automobile industry in cities through policies and coordinate land and financial resources to support enterprises. The government should develop supervision and planning schemes as soon as possible in accordance with the current situation, as the sharing of new energy vehicle industry has only recently emerged, and the corresponding laws and regulations have not yet been perfected. Additionally, the majority of users access shared new energy vehicles through the download and registration of apps. Relevant government entities should closely oversee the privacy leakage issue to protect user information security.

The location layout and quantity distribution of urban shared new energy vehicles should be paid attention to by the government and enterprises. It should avoid the situation that there are not enough vehicles in areas with high demand or it is inconvenient to borrow and return cars and that the number of places with low demand is distributed more and the land area is divided into large areas.

Before site selection, enterprises should screen schools, tourist attractions, large shopping plazas, hospitals, and other places with a high frequency of resident activities in the region to determine the location of alternative outlets and try to understand the number of residents in the network radiation area, frequency of use, and other information to predict the demand of the network. At the same time, in order to reduce construction and operation costs, enterprises should estimate the cost of each suitable alternative point and select the most suitable construction site.

The government should help businesses advertise shared new energy cars by enhancing publicity and developing subsidy programs to draw in more users. In order to collectively create a green and civilized city, education and training should be given to residents to help them use and park in a civilized manner.