Decision Making Support for Local Authorities Choosing the Method for Siting of In-City EV Charging Stations

Abstract

Development of electromobility in urban areas requires an appropriate level of vehicle charging infrastructure. Numerous methods for siting of charging stations have been developed to date, and they appear to be delivering diverse outcomes for the same area, which is why local authorities face the problem of choosing the right station layout. The solution proposed in this article is to use a travel planner to evaluate the distribution of charging stations over the area of a metropolis. The decision making support is achieved by determining optimal travel routes for electric vehicles according to their initial state of charge for the three selected station siting methods. The evaluation focused on the following three aspects: (1) number of travels that cannot be made (due to the lack of a charging station at a certain distance around the start point), (2) extension of the travel caused by the need to recharge the vehicle on-route, and (3) additional energy consumption by electric vehicles required to reach the charging station (necessity of departing from the optimal route). An analysis of the results has made it possible to determine a solution which is superior to others. For the case study analysed in the paper, i.e., the territory of the Metropolis of Upper Silesia and Dabrową Basin (Górnośląsko-Zagłębiowska Metropolia, GZM), the distribution of charging stations established in line with method I has returned the best results. What the method in question also makes possible is to indicate a safe minimum energy reserve to complete the travel by eliminating situations of unexpected vehicle immobilisation due to on-route energy depletion and by minimising the phenomenon referred to as range anxiety.

1. Introduction

Transport leaves a significant mark on the natural environment. The negative impact of transport takes diverse forms, primarily including the emission of harmful substances, noise, and the resulting traffic obstructions caused by congestion. Research explicitly shows that the world is largely dependent on petroleum (in terms of mobility and transport) [1,2]. At the same time, the right to mobility is considered as an important value, and in accordance with the relevant guidelines, minimisation of negative environmental impact of transport cannot be pursued through mobility limiting, but rather through efficient utilisation of natural resources [3]. Current trends pertaining to the sphere of alternative fuels highlight electricity as a viable option, among other solutions. It is possible to accelerate the desired changes in the share of alternative fuels in transport through infrastructure development, implementation of shared technical specifications, as well as changes in the image of the related technologies (consumer acceptance).

In order to effectively manage a municipality, numerous factors that are considerably related to transport and energy must be taken into account. What stands out in this respect is the need for proper planning of development in both these spheres as well as for mutual coexistence within an organism, such as a city.

The increase in the number of electric cars observed in municipal roads is determined by such factors as the existing support infrastructure. On the other hand, further development of electromobility requires decisions concerning the siting of new charging stations to be made on a continuous basis. The problem of optimised arrangement of charging stations within the given territory is often addressed in the literature of the subject (among others [4,5,6,7,8,9,10]). Authors take numerous criteria into consideration, while the actual application of the methods they propose brings different outcomes (due to differences in the infrastructure setup in individual cities). One of the subjects commonly addressed is the relations between vehicle transfers and the existing electrical network, different parking space policies, or use of the existing petrol stations as sites for charging stations, etc. Facing such circumstances, local authorities have a difficult choice to make, and not only with regard to the in-city locations where charging stations are to be installed, but also when considering the method assumed to deliver good results in the given case. This issue was the research gap which required an adequate method of support for the decision making process to be developed by comparing specific potential layouts of charging stations.

The method proposed in this article is intended to support local authorities in choosing between different arrangement variants for charging stations. This proves particularly important when considering areas of low electromobility development maturity. The method is based on the use of a tool which is one of the deliverables of an international project the “Electric travelling—platform to support the implementation of electromobility in Smart Cities based on ICT applications” within the scope of ERA-NET CoFund Electric Mobility Europe Programme financed among others from the means of the National Centre for Research and Development. Main goal of the project is to accelerate the development of electromobility in cities. It is an original travel planner, known as ETPlanner, which has been used to study a number of proposed spatial arrangements of charging stations. The territory analysed under the research is a metropolitan area in the south of Poland, characterised by a dense transport network. It was divided into ca. 300 zones in which locations corresponding to the origin and destination points of travel were subsequently established. Using a set compiled in such a manner, optimum travelling routes were then identified using ETPlanner, assuming that they were to be covered by means of an electric car with its battery charged to different levels. The results thus obtained have been analysed and discussed in this paper.

In light of the above goals, the article has received the following structure: Section 2 is a review of the existing methods for optimal siting of electric vehicle charging stations, and the selected ones have been implemented against a defined area (Section 3); details of the method proposed for evaluation of the layout of charging stations using the multimodal travel planner in question, i.e., ETPlanner, have been provided in Section 4; the layouts of charging stations were studied by the method proposed, while the selected research results have been provided in Section 5 and discussed in Section 6; and Section 7, which concludes the article, provides specific insights and defines future research plans for the body of problems addressed in the paper.

2. Optimal Siting of Electric Vehicle Charging Stations—Review of Selected Methods

The development of electromobility requires special attention to be paid to the infrastructure used for charging of electric vehicles (EV). The low number of stations, their inefficient siting and poor parameters, when combined, constitute a barrier to further popularisation of electric vehicles. There are also fundamental infrastructural problems connected with the charger type in use, the location of EV charging stations as well as their number that should be located within the given area. In light of these issues, attempts were made to identify the locations where electric vehicle charging stations should be established. To this end, the existing methods for siting of charging stations were reviewed, and three of them were selected for application in practice. The testing ground chosen for the studies was the GZM Metropolis (Górnośląsko-Zagłębiowska Metropolia) situated in the southern part of Poland. The studies on the optimal method for siting of charging stations constituted input data for a multimodal travel planner enabling evaluation of the arrangement of charging stations based on the results obtained for various travel scenarios.

2.1. Overview of the Methods Used for Optimal Siting of Charging Stations

The literature of the subject mentions numerous approaches to the problem of optimal siting of electric vehicle charging stations.

The authors of article [11] performed a three-dimensional analysis of a territory using a multi-criteria factor analysis and a method of hierarchical process analysis. The paper elaborates upon a method used to determine the weights of the factors; however, it disregards the way in which ratings are to be assigned. The greatest weight was attributed to the factors associated with the distance to green areas (as much as 32%), while the second most important factor was the one that which described the quality of the given area. The analyses employed in the method were performed on the basis of interpolation of pixel values, and the results were presented as a heat map.

A different concept of a method for determining the locations of charging stations was discussed in paper [12], with the potential locations being analysed with reference to the existing infrastructure. The Voronoi diagram was used to divide the area subject to analysis and search for new charging station sites. Additionally, the locations characterised by the longest time of waiting for vehicle charging were identified using the vehicles’ positioning data. A similar approach was presented in paper [13], where the matter of accessibility of the existing charging stations was analysed against the distance in the road network.

A charging station siting method based on input data connected with the number of travels made and the vehicle use intensity was proposed in [14]. What was also attempted under that study was a search for locations intended for public and private charging points. The results of the relevant analyses were presented for large administrative regions.

The authors of paper [15] proposed a two-stage method for charging station siting through an analysis first performed in a macro-scale (in the case described, the territory was entire Hungary), and in the micro-scale in the second stage (the Budapest district). The analysis they proposed was based on hexagon-shaped primary areas. The conclusions thus formulated imply that it is necessary to install the charging stations at Park and Ride facilities and near locations with highly concentrated service points, while the locations of the existing petrol stations are not good sites for charging stations.

In the study described in paper [16], a fuzzy hierarchical analysis of the process was applied. The outcome was a comparison of the identified potential sites with the existing infrastructure of charging stations. Based on 15 factors grouped into three categories (environment, economy, and urban planning), six weight assignment scenarios were proposed, combined with verification of changes in the utility of the parcels assumed for the charging station sites. The authors highlighted that the scenario should be ultimately chosen on account of the relevance of the factors, as perceived by the given decision maker.

Paper [17] proposes a different approach to the problem of siting of charging stations for electric vehicles, namely two-step linear programming. The purpose of this approach is minimisation of the costs incurred by the charging station owners. The method described in the said paper is based on using the electricity distribution network. Paper [18] provides an analysis of the sites chosen for hybrid stations using renewable energy sources. The charging stations subject to the study are designed for both electric vehicles and natural gas vehicles.

The method described in paper [19] is characterised by a large amount of input data, namely the factors subject to analysis, being as many as 34, divided into aspects that positively and negatively affect the potential location of charging stations. The factors were checked to establish if they occurred in the given squares, and then the layer of results was built—one which concerned the overall evaluation of the given square. On such a basis, the utility of the given space assumed as a potential charging station site was established.

The paper submitted under the Joint Research Centre of the European Commission [20], elaborating upon a search for an optimal site for electric vehicle charging stations, was divided into two parts addressing the national and municipal level. The former entailed searching for potential charging station locations along national roads, considering the directionality of the road network and the auxiliary infrastructure. This part was not taken into account in the selected methods, since it did not meet the pre-assumed conditions. The latter part, pertaining to the city level, included a division of the analysed area into a grid of 100 × 100 m, and introduced factors increasing the area’s potential for the siting of charging stations. Additionally, special buffers reflecting the impact of the given factor on the area were taken into account, while the factors themselves were expressed in point- and polygon-based ratings. The study in question covered the territory of the Italian town of Bolzano.

The method presented in paper [21] is based on a multi-criteria analysis of input factors combined with a fuzzy hierarchical analysis of the process. The authors proposed 10 factors affecting the charging station siting potential of the analysed areas. Unlike in the other methods, a factor was analysed not only for whether or not it was present in the area in question, but also in terms of the distance of the given area from the factor in a straight line, and that provided grounds for ratings. The AHP (analytic hierarchy process) analysis was divided into two scenarios concerning environmental impact and accessibility, followed by a fuzzy hierarchical analysis aimed to deliver the final weights for the selected factors.

Based on the foregoing review of the methods proposed in the literature of the subject as the means to search for optimal locations of electric vehicle charging stations, three were selected and applied in the metropolitan area subject to the research. The choice of the methods for further investigation was related to the assumptions previously adopted:

• The selected method should comprise a complete description enabling it to be applied in any area (other than that described in the literature).
• The method must not be based on the existing EV charging infrastructure, and the territories subject to analysis must not be larger than 100 m × 100 m (making it possible to establish the potential charging station location on the map with a relatively high accuracy).
• The selected methods should be universal in nature, highlighting areas of high potential from the EV charging perspective.

As discussed further on in this paper, the said methods for siting of charging stations were chosen with reference to papers [19,20,21], and then used in practice in the GZM Metropolis.

The chosen methods represent diversified approaches to the searching optimal sites for EV charging stations. Method [19] takes the largest number of station siting factors (24) into consideration, being the only one among those selected which also identifies negative factors that are unfavourable to the charging infrastructure. A single pre-defined rating is assigned when the given factor emerges in the primary area. On the other hand, method [20] takes the fewest factors (6) into account, but they are more extensive, e.g., a single factor comprises multiple public utility buildings. It is also the sole method which includes buffer zones of the given factor’s impact on its environment. The overall rating of the given factor under this method does not result directly from the ratings it has been assigned, but stems from the number of occurrences of the given factor in the primary area, which is subsequently normalised (Equation (1)). Method [21] comprises 10 factors, being the only one where ratings are based on a 1–5 grading scale and depend on degree to which the given condition has been met. The method is also the only one which takes into account the straight line distance between specific factors and centroids of primary areas.

2.2. Description of the Selected Methods

Three methods were selected for further research, and their detailed assumptions as well as the outcomes of their application in the area subject to analysis have been presented in the following section of the paper. The chosen methods share some common features, such as the identical base area of analysis forming a grid of 100 m × 100 m, and the application of vector-based geo-information recorded as polygons and points.

2.2.1. Method I

The first of the selected methods was described in article [19]. The authors originally assumed using 34 layers of factors associated with the territorial division of the analysed area. With such a division in mind, for the purpose of this article, 24 layers of factors were left, and they have been provided in

Table 1. Factors with ratings, affecting the potential of the primary area for siting of an EV charging station as per method I, based on [19].

Name	Layer Type	Rating
Population	Polygon	1/3/5
Airfields	Polygon	5
Golf courses	Polygon	5
Parks	Polygon	5
Petrol stations	Polygon	5
Administration’s buildings	Polygon	5
Healthcare facilities	Polygon	5
Libraries	Polygon	5
Main attractions	Polygon	5
Main places of work	Polygon	5
Parking areas	Polygon	5
Main roads	Line	5
Park & Ride facilities	Polygon	5
Places of worship	Polygon	5
Post offices	Polygon	5
Schools	Polygon	5
Senior citizen homes	Polygon	5
Large-format shopping centres	Polygon	5
Swamps	Polygon	−5
Rivers	Line	−1
Existing charging points	Point	−5
Residential areas	Polygon	−4
Agriculture	Polygon	−5
Protected areas	Polygon	−5
Lakes	Polygon	−1

.

The authors divided these factors on account of their either positive or negative impact on the primary area’s potential to host an EV charging station. Individual elements of the environment were given negative rating points to reduce the potential of low-utility areas. Moreover, the existing charging stations were also assigned negative rating points to avoid close distances between stations. The authors also decided to give negative rating points to residential areas, since they had assumed that private EV charging facilities would exist in these locations. The population’s rating for the given primary area was assigned on the basis of the mean value of all primary areas in the analysed territory [15].

Each layer with the data concerning the factors in question (Table 1) was transformed into a grid corresponding to an area of 100 m × 100 m. Where an element of the given factor intersected with the grid, an assessment rating was given as per Table 1. In the final step, the grids containing individual ratings were superimposed to obtain the final result with a sum of ratings for all primary areas of the territory subject to the analysis.

2.2.2. Method II

The authors of the second method put into practice, described in paper [20], used the 6 factors presented in

Table 2. Factors affecting the potential of the primary area for siting of an EV charging station as per Method II, based on [20].

Name	Layer Type
Electrical network	Line
Commercial buildings	Polygon
No. of parking spaces	Polygon
No. of public utility buildings (hospitals, museums, theatres, and higher education schools)	Polygon
Population	Polygon
No. of collective transport system components (airports, ports, railway, and bus stations)	Polygon

. The analysis was conducted for a territory divided into primary areas of 100 m × 100 m. All input data were buffered in order to account for the impact of the given factor on the surrounding area. An example of a buffer zone has been shown in Figure 1.

For the electrical network, the buffer was 50 m (it was assumed that longer transfers could be difficult to complete), while for other items in Table 2 a value of 100 m was assigned on account of the maximum acceptable walking distance of 300 m (example provided in Figure 2).

Figure 2 illustrates a geometrical analysis where the point buffer radius (100 m) combined with 100 × 100 m cells results in a maximum Euclidean distance of less than 300 m between the centre of a point and the outermost corner of the adjacent cell.

Once the factors have been buffered, the number of instances of the given factor in the primary area is counted. This procedure was identical to all the factors except the electrical network, for which only zeroes or ones were assigned depending on whether or not the infrastructure was present, as well as for the population, for which the actual value established for the given primary area is assigned. Then the number of instances of the given factor in the cell was normalised (Equation (1)).

SCORE(i) = value(i) × 10/max,

(1)

where

i—number of the given primary area,

value(i)—value of the given factor in the primary area, and

max—maximum value of the given factor for all primary areas of the given layer.

Once the results had been normalised, factor evaluation values were determined to range between 0 and 10. In the next step, the final evaluation was calculated for each basic area (Equation (2)) comprising the layers of individual factors and their weights.

Layer_electricty_network × (layer_file₁ × factor₁+layer_file₂ × factor₂+ …),

(2)

The authors of method [16] proposed that equal weights of 0.2 should be used for each layer; however, they also stressed that this method can be modified as needed to obtain results which account for the greater impact of the given factor on the choice of the charging station site. Furthermore, what was proposed in the article is that this factor is to be disregarded as needed and depending on how accurate the mapping of the electrical network layer is.

2.2.3. Method III

The authors of method [21] introduced 10 factors (C1–C10) in their analysis aimed at finding an optimal site for charging stations (

Table 3. Factors affecting the potential of the primary area for siting of an EV charging station as per method III, including the rating method [21].

Name	Sub-Criterion	Score
Population density (C1)	33< 23–32 17–22 10–16 <9	5 4 3 2 1
Shopping centres (C2) (m)	<250 250–500 500–750 750–1000 1000<	5 4 3 2 1
Roads (C3)	<250 250–500 500–750 750–1000 1000<	5 4 3 2 1
Income rates (C4) (TRY)	9025 6987 6577	5 3 1
Public transport stations (C5) (m)	<250 250–500 500–750 750–1000 1000<	5 4 3 2 1
Petrol stations (C6) (m)	<250 250–500 500–750 750–1000 1000<	5 4 3 2 1
Park areas (C7) (m)	<250 250–500 500–750 750–1000 1000<	5 4 3 2 1
Green areas (C8) (m)	300< 250–300 150–200 100–150 <100	5 4 3 2 1
Slope (C9) (°)	<5 5–10 10–15 15–20 20–25 25<	5 4 3 2 1 0
Land Values (C10) (TRY)	<677 678–1068 1069–1450 1451–2907 2908<	5 4 3 2 1

). A rating method was assigned to each factor. The territory of analysis was divided into primary areas of 100 m × 100 m each. The data obtained for the monetary factors were provided in the Turkish currency, as in the original article. It was then converted into the Polish currency to match the conditions of the territory subject to examination. For the road factor, the distance between centroids and national as well as regional roads was checked. With regard to public transport stations, the entire selection of the means of public transport was taken into account along with all places enabling passengers to travel.

According to this method, in the first stage, individual factors are assigned ratings to account for their occurrence in the given primary area and their distance from the centre of the primary area. Figure 3 provides an example of how a straight line distance from each primary area centroid to the nearest object located in the layer of the analysed factor was determined.

Next, the AHP analysis [21] was performed, broken down into two scenarios concerning the environmental impact and accessibility of charging stations. The final stage was a fuzzy hierarchical analysis of the process aimed at obtaining the final factor weights.

Table 4. Weights of the factors affecting the potential of the primary area for siting of an EV charging station as per method III [21].

Name	Weights for the Environmental Impact Scenario	Weights for the Accessibility Scenario	Final Values of Weights for AHP 1
Population density (C1)	0.0280	0.0605	0.0951
Shopping malls (C2) (m)	0.0486	0.1050	0.0887
Roads (C3)	0.0848	0.1835	0.0929
Income rates (C4) (TRY)	0.0341	0.0737	0.0983
Transportation stations (C5) (m)	0.0213	0.0459	0.0970
Petrol stations (C6) (m)	0.0613	0.1325	0.0862
Park areas (C7) (m)	0.0416	0.0898	0.0964
Green areas (C8) (m)	0.5583	0.0915	0.1590
Slope (C9) (°)	0.0305	0.0544	0.0921
Land Values (C10) (TRY)	0.0915	0.1632	0.0943

¹ AHP—analytic hierarchy process.

summarises the weights obtained for two scenarios as well as the final weights delivered by the fuzzy AHP analysis.

The highest value of the environmental impact scenario weights was obtained for the green areas factor (0.5583), which was due to the attempt to make transport more distant from green areas as well as the low utility value of the areas indicated. For the final weights, this factor gained the highest value; however, its impact on the results was lower, −0.1590. For the accessibility scenario, the maximum value was obtained for the distance between the centroid and the main roads.

3. Case Study

The area of analysis is the GZM Metropolis which currently consists of 26 municipalities (of which 13 have the rights of an administrative district—Polish “powiat”), 2 urban-rural communes, and 13 rural communes. It lies in the central part of the Silesian Province (Polish województwo śląskie) and an area of 2553 km², which accounts for ca. 20.70% of the entire province area. The territory of the GZM Metropolis is inhabited by 2,244,850 people. It is diversified in terms of the population density, ranging between communes with ca. 3000 persons per km² and those with a very low population density, barely reaching 100 (persons/km²). Figure 4 shows the location of the Metropolis against the Silesian Province. The territory is located in the south of Poland.

The polygon, point and raster input data required to apply the methods in question were obtained from the following sources:

• web-based portal: OpenStreetMap^®,
• open-source data from the GZM website: https://otwartedane.metropoliagzm.pl/dataset,
• data on public transport from the Board of Metropolitan Transport,
• inclinations established on the basis of the area’s numerical terrain model,
• terrain parameters based on the data of the Central Statistical Office,
• inhabitants’ income based on the data of the Central Statistical Office,
• population density based on the data of the Central Statistical Office, and
• data concerning the electricity distribution network and individual transformer sites according to the data made available by the electricity distributors operating in specific municipalities of the Metropolis.

3.1. Results of Application of the Charging Station Siting Methods in the Territory of the GZM Metropolis

Figure 5 illustrates the results obtained in respect of the potential of the primary areas for siting of electric vehicle charging stations according to the three aforementioned methods (and for two variants for the second method) with regard to the analysed territory of the GZM Metropolis.

An analysis of Figure 5 highlights evident differences between the potentials of primary areas depending on the method used. Method I is characterised by an even distribution of high-potential primary areas in space. The results of method III form large clusters of primary areas of similar potential. In method II, the fact of having included the electricity distribution network clearly affects the results and limits the potential sites for the charging stations to a considerable extent.

3.2. Number of Charging Stations Along with Potential Locations

In the next stage of the research, the number of charging stations expected to suffice to service the given area of analysis was to be defined. In this respect, one must take into account the Polish Act of 11 January 2018, on electromobility and alternative fuels, which defines the number of charging stations that should be built in municipalities by the end of 2020 depending on their population [22]. In accordance with this law, cities with more than 100,000 inhabitants should have charging stations. On such grounds, a statistical analysis of the number of inhabitants was performed for the GZM Metropolis cities, and the number of charging points was determined.

Table 5. Population, number of charging stations, and area of the municipalities of the GZM Metropolis with more than 100,000 inhabitants [22].

Municipality Name	Population	Area (km2)	Required Number of Charging Stations
Katowice	292,774	165	100
Sosnowiec	199,974	91	100
Gliwice	178,603	134	100
Zabrze	172,360	80	100
Bytom	165,263	69	100
Ruda Śląska	137,360	78	60
Tychy	127,590	82	60
Dąbrowa Górnicza	119,373	189	60
Chorzów	107,807	33	60

compares the municipalities of the GZM Metropolis with more than 100,000 inhabitants and states the number of chargers that should be installed in the respective cities.

The cities where charging stations should be installed occupy an area of ca. 36.07% of the total area analysed, while the inhabitants of these areas account for ca. 66.86% of the entire population of the GZM Metropolis. Based on the analysis of the relevant provisions of the said act, it was established that the total number of charging stations in the research area should be 740. Two variants were adopted on such grounds:

• siting of charging stations in accordance with the act (number and municipalities), and
• siting of charging stations without taking the municipality-related restrictions into account (the established number of 740 stations remains but applies to the entire territory of the GZM Metropolis).

Figure 6 provides a comparison of the results obtained by methods I and III for the two variants, indicating 740 primary areas with the highest potential: without any restrictions for the entire GZM Metropolis and with the restrictions imposed on the towns indicated in Table 5.

An analysis of Figure 6 clearly highlights the clusters of sites with the highest potential as per method III, while method I distribute the potential sites more evenly across the analysed area. In Figure 6b,d, where the entire area is analysed, only a small number of potential sites are outside the major towns listed in Table 5 (marked in dark grey). Under method I, 30% of all primary areas are to be found outside the major towns, and 20.4% of them under method III. For a large fraction of the area subject to analysis (63.93% of the whole), the methods have pointed at a small number of potential charging station sites due to the low values of the factors which strengthened the potential of the primary areas.

Figure 7 provides a comparison of the results obtained under method II for two scenarios (regarding and disregarding the electricity distribution network) and variants, identifying 740 primary areas of the highest potential in each, without restrictions for the entire GZM Metropolis and with those restrictions imposed on the municipalities indicated in Table 5.

Under the scenario which takes the electricity distribution network into consideration and disregards the municipality-related restrictions, 20.5% of the charging stations are to be found outside the major towns, while 19.86% of the primary areas have been identified under the scenario which disregard the electrical network. For all the results thus obtained, there are similarities in the distribution of the primary areas showing the highest potential, which is due to having taken into account similar factors affecting the area rating.

4. Using a Travel Planner to Assess the Distribution of Charging Stations over the Area of a Municipality

The siting of charging stations in a municipal area depends on their intended use. They may be designed for EV recharging during long-distance journeys, in which case charging stations are best located along major transit routes and at motorways. However, this article focuses on the stations installed in an urban development system, whose purpose is to enable ad hoc recharging on short routes or leaving the vehicle in the charging mode once it has arrived at the destination point. The choice of the right method for siting of charging stations can be supported by a numerical analysis of the impact of the charging station arrangement on travelling by means of electric vehicles. One of the applications developed under the Electric Travelling project was used to assess the distribution of charging stations over a specific area (the GZM Metropolis, in this case). ETPlanner is an elaborate and multimodal travel planner which extends the features of the tool described in paper [23]. The algorithms implemented in the program make it possible to find optimal travelling routes for a dozen or so modes of travelling, including the use of an electric vehicle. Moreover, the planner suggests to the travelling person which charging station would be optimal for the selected travel criterion when the vehicle’s energy level is too low to cover the entire distance of the journey. The latter property was used for purposes of the method proposed. The planner’s operating principle is a server-based query processing service. Queries can be submitted by the travelling person using a graphical user interface available on a webpage. In this case, a query is initially built via the interface by selecting the right options, such as the means of transport, the start and end point, the optimisation criterion, and optionally some additional parameters depending on the chosen means of transport, whereupon the query is submitted. An alternative is sending commands directly to the server via the tool’s software (ready-to-use query structure), which enhances the fast result processing and storage capability.

The planner’s road layout is described by a network graph. The shape of the graph depends on the Open Street Map (OSM) data [24]. Individual vertices and arcs of the graph are assigned parameters which enable different optimised route search criteria to be applied. In order for the origin-to-destination route optimisation algorithms to function properly, the travel planner must process a road network parameterised by information on individual edges of the graph, including data defining the road section length and permissible speed (the planner can also take into account some dynamic data of the average speed recorded in the road sections, provided that they can be retrieved from the town’s Smart Transport System) available in OSM, inclination data extracted from the numerical terrain model, as well as temperature (actual or seasonal average). Depending on the travelling mode examined, what the tool requires is either the timetables and locations of stops for individual means of public transport, or the current location of the available vehicles being dynamically uploaded for the car sharing services. Most importantly, the tool makes it possible to simulate different layouts of charging stations, as imported by the user into the planner’s network graph.

Consequently, the method in question proceeds in two stages:

• Stage I—collecting data on the potential travelling routes between defined points of origin and destination, independently for all the examined charging station layout variants, and
• Stage II—comparison and analysis of results.

The method used to evaluate the spacing of the charging stations required a large number of queries about optimal routes, taking different EV battery charge levels into account. The locations of the points of origin and destination were centroids of 304 transport districts (zones) into which the territory in question was divided, meaning that 92,112 queries were submitted to the travel planner in each series. These querying series comprised the following scenarios:

• Travelling with an electric vehicle assuming that there is no charging station,
• travelling with an electric vehicle assuming the layout of charging stations as per method I,
• travelling with an electric vehicle assuming the layout of charging stations as per method II in the variant which accounts for the electricity distribution network,
• travelling with an electric vehicle assuming the layout of charging stations as per method III,
• travelling with an electric vehicle assuming diverse initial battery charge level, and
• two vehicle segments.

For each method, the distribution of the charging station sites was assumed to disregard the restrictions imposed on the actual municipalities (740 stations scattered over the entire territory of the GZM Metropolis). What the authors also assumed was that the charging stations were identical, i.e., Type 2 with 7.4 kW of power capacity (mentioned as the most popular type, e.g., in [25,26,27]). Such a distribution is more natural, which has made it possible to present the method as a universal approach to this kind of analysis (with the prospect of the method implementation for other areas).

The possibility of completing a transfer by means of an electric vehicle was studied on different vehicle battery charge levels, starting from 50% and assuming a 1% increment/decrement. Additionally, a minimum charge level, below which the vehicle could not be started, was set at 5% (approx. 15 km).

The relevant tests were performed using vehicles of the two most popular segments:

• Segment B, including BMW i3, Hyundai Kona, Kia Niro, Kia Soul Electric, and Renault Zoe; and
• Segment C, including Hyundai Ioniq Electric, Nissan Leaf, and Volkswagen e-Golf.

The route optimisation criterion was assumed to be time, which is also the main criterion taken into account by the travelling population [28].

A total of 360 series of input data combinations were tested, resulting in more than 33 million queries submitted to ETPlanner. The submitted queries comprised data of a specific location, namely, geographical coordinates of the start and end points of the travel, as well as the date and time of the travel beginning. Defining a day and a specific time of that day for the query to be generated affects the characteristics of the transfers made in the simulated transport system. The fixed parameters assumed for the query were the means of transport, namely an electric vehicle, distinguishing between two popular segments, and the optimisation criterion based on the travel time minimisation, including electric vehicle charging as required. With regard to the characteristics of the electric vehicle’s power source and its charging at the simulated stations, what the queries submitted to the ETPlanner application defined was the initial vehicle battery level, the minimum permissible battery level for which the route seeking procedure was performed, as well as the plug type. Further parameters that were also defined, affecting the function of the vehicle charging and discharging, included the battery capacity, number of phases, vehicle weight, and ambient temperature.

Having read a route planning query from a file, ETPlanner generates an output file, and having received 33 million queries, it analogically provided 33 million replies. The output files contain information which describe the completed travels in a direct form, as well as information on the possibility of completing the travel, its duration, route, actual distance, energy consumption, need for charging and charging time, all with reference to individual travels. Through aggregation, division, and statistical operations performed on the set of data thus obtained, one can create various collations, including those described in Section 5.

In order to prepare such a large set of queries, a tool for generating queries of pre-defined parameters needed to be developed and implemented, as described above, in the Visual Studio 2017 programming environment suitable for the C++ language. This programming environment was further used to develop and implement a tool that compared the 33 million replies previously obtained in an aggregated form making it possible to effectively determine correlations and formulate conclusions on the basis of the simulated arrangements of charging stations over the given area, as well as to apply the matrix of travels.

5. Results

The data on travelling between individual zones within the area subject to analysis made it possible to prepare the collations provided in this section of the paper. The test results were compared by taking the following three aspects into consideration:

• Number of travels that cannot be made—this translates into limitation of mobility with regard to electric vehicles (on a specific vehicle battery charge level).
• Travel extension due to the need for vehicle recharging on-route—depending on the locations of the charging stations, they may cause the necessity of significantly departing from the original (optimal) route.
• Energy consumption by electric vehicles—also in this case, specific arrangement of the charging stations can trigger higher energy consumption which makes travelling longer (longer charging time), but first and foremost, causes higher total energy consumption, which can be important for the city.

Table 6. Total values obtained for the transfers on the test routes, assuming no need for on-route recharging.

Number of Simulated Transfers	Total Distance Covered (km)	Total Energy Consumption 1 (kWh)	Total Energy Consumption 2 (kWh)
92,112	2,255,942	262,584.421	258,072.661

¹ Established for the B-segment vehicles, ² established for the C-segment vehicles.

compares the basic values measured for the analysed travels made by means of an electric vehicle which did require on-route recharging (the battery charge level sufficed to cover the entire distance with a 5% energy reserve). The mean travel length was 24.5 km.

Table 7. Number of travels completed under all the scenarios analysed, compared for an area without charging stations and an area with the three alternative layouts of charging stations. Results obtained for the B-segment vehicles.

Battery Charge Level eV 1	Scenario—No Charging Station within the Area	Layout of Charging Stations as per Method I		Layout of Charging Stations as per Method II		Layout of Charging Stations as per Method III
	Number of Transfers Completed without Recharging	Number of Transfers Completed with Recharging	Number of Non-Completed Transfers	Number of Transfers Completed with Recharging	Number of Non-Completed Transfers	Number of Transfers Completed with Recharging	Number of Non-Completed Transfers
5	170	49,245	42,697	29,574	62,368	23,528	68,414
6	3178	81,157	7777	68,355	20,579	59,785	29,149
7	8340	81,091	2681	76,729	7043	74,169	9603
8	15,178	75,457	1477	74,024	2910	74,584	2350
9	23,062	69,050	0	68,458	592	67,887	1163
10	31,552	60,560	0	60,266	294	60,266	294
11	40,282	51,830	0	51,830	0	51,830	0
12	48,518	43,594	0	43,594	0	43,594	0
13	56,330	35,782	0	35,782	0	35,782	0
14	63,330	28,782	0	28,782	0	28,782	0
15	69,610	22,502	0	22,502	0	22,502	0
16	75,074	17,038	0	17,038	0	17,038	0
17	79,492	12,620	0	12,620	0	12,620	0
18	82,960	9152	0	9152	0	9152	0
19	85,640	6472	0	6472	0	6472	0
20	87,734	4378	0	4378	0	4378	0
21	89,310	2802	0	2802	0	2802	0
22	90,426	1686	0	1686	0	1686	0
23	91,154	958	0	958	0	958	0
24	91,592	520	0	520	0	520	0
25	91,836	276	0	276	0	276	0
26	91,982	130	0	130	0	130	0
27	92,042	70	0	70	0	70	0
28	92,076	36	0	36	0	36	0
29	92,100	12	0	12	0	12	0
30	92,108	4	0	4	0	4	0
31	92,112	0	0	0	0	0	0
…	…	…	…	…	…	…	…
50	92,112	0	0	0	0	0	0

¹ eV—electric vehicle.

and

Table 8. Number of travels completed under all the scenarios analysed, compared for an area without charging stations and an area with the three alternative layouts of charging stations. Results obtained for the C-segment vehicles.

Battery Charge Level eV 1	Scenario—No Charging Station within the Area	Layout of Charging Stations as per Method I		Layout of Charging Stations as per Method II		Layout of Charging Stations as per Method III
	Number of Transfers Completed without Recharging	Number of Transfers Completed with Recharging	Number of Non-Completed Transfers	Number of Transfers Completed with Recharging	Number of Non-Completed Transfers	Number of Transfers Completed with Recharging	Number of Non-Completed Transfers
5	34	43,906	48,172	24,819	67,259	20,277	71,801
6	1864	76,133	14,115	59,644	30,604	48,654	41,594
7	5144	81,919	5049	76,012	10,956	68,109	18,859
8	9566	80,168	2378	77,266	5280	75,285	7261
9	14,968	75,666	1478	74,229	2915	74,791	2353
10	20,910	70,607	595	70,320	882	69,747	1455
11	27,298	64,814	0	64,516	298	64,227	587
12	33,958	58,154	0	57,862	292	57,862	292
13	40,672	51,440	0	51,440	0	51,440	0
14	47,056	45,056	0	45,056	0	45,056	0
15	53,064	39,048	0	39,048	0	39,048	0
16	58,838	33,274	0	33,274	0	33,274	0
17	64,128	27,984	0	27,984	0	27,984	0
18	68,968	23,144	0	23,144	0	23,144	0
19	73,376	18,736	0	18,736	0	18,736	0
20	77,116	14,996	0	14,996	0	14,996	0
21	80,278	11,834	0	11,834	0	11,834	0
22	82,894	9218	0	9218	0	9218	0
23	85,060	7052	0	7052	0	7052	0
24	86,786	5326	0	5326	0	5326	0
25	88,246	3866	0	3866	0	3866	0
26	89,386	2726	0	2726	0	2726	0
27	90,250	1862	0	1862	0	1862	0
28	90,886	1226	0	1226	0	1226	0
29	91,330	782	0	782	0	782	0
30	91,618	494	0	494	0	494	0
31	91,818	294	0	294	0	294	0
32	91,936	176	0	176	0	176	0
33	92,022	90	0	90	0	90	0
34	92,050	62	0	62	0	62	0
35	92,076	36	0	36	0	36	0
36	92,096	16	0	16	0	16	0
37	92,102	10	0	10	0	10	0
38	92,110	2	0	2	0	2	0
39	92,112	0	0	0	0	0	0
…	…	…	…	…	…	…	…
50	92,112	0	0	0	0	0	0

¹ eV—electric vehicle.

provide the data of the travels which can be completed assuming that there are no charging stations available and when comparing their diverse layouts. For the B-segment vehicles, the need for on-route recharging was identified starting from a 30% battery charge level. For the C-segment vehicles, this level was 38%. The inability to recharge the vehicle depending on which method of the charging station siting had been used was identified at a much lower battery charge level, i.e., 8–9% for the B-segment vehicles and 10–11% for the C-segment vehicles. The share of the travels not completed and those successfully made with recharging has been illustrated in Figure 8 and Figure 9.

On account of the recharging necessity, some of the travels needed to be extended by the distance required to reach the charging station. The route extension values have been illustrated with a factor that compares the length of the transfer completed with the visit at the charging station and the unaltered route (assuming that the vehicle had departed the point of origin fully charged). Selected combinations of these values have been provided in

Table 9. Comparison of the travel route extension factors on account of the on-route recharging necessity for the three charging station layouts—values obtained for the transfers which required recharging. Results obtained for the B-segment vehicles.

Battery Charge Level eV 1	Route Extension Factor for the Layout of Charging Stations as per Method I			Route Extension Factor for the Layout of Charging Stations as per Method II			Route Extension Factor for the Layout of Charging Stations as per Method III
	Max.	Mean	Std. Dev.	Max.	Mean	Std. Dev.	Max.	Mean	Std. Dev.
5	2.619	1.036	0.073	2.256	1.034	0.072	2.503	1.044	0.091
6	2.800	1.041	0.092	2.866	1.070	0.130	2.937	1.087	0.153
7	2.883	1.029	0.071	2.773	1.055	0.104	2.851	1.081	0.149
8	2.233	1.019	0.044	2.545	1.044	0.088	2.675	1.072	0.141
9	2.068	1.015	0.032	2.226	1.038	0.073	2.286	1.052	0.094
10	1.792	1.013	0.024	1.972	1.032	0.060	1.923	1.041	0.071
11	1.328	1.012	0.022	1.827	1.029	0.053	1.659	1.035	0.057
12	1.313	1.010	0.020	1.704	1.027	0.049	1.598	1.032	0.051
13	1.255	1.010	0.018	1.489	1.025	0.043	1.460	1.031	0.048
14	1.218	1.009	0.016	1.459	1.023	0.041	1.473	1.028	0.044
15	1.186	1.008	0.015	1.425	1.020	0.037	1.361	1.025	0.039
16	1.181	1.007	0.014	1.432	1.019	0.034	1.394	1.023	0.037
17	1.160	1.007	0.013	1.393	1.018	0.031	1.356	1.020	0.031
18	1.161	1.007	0.012	1.412	1.017	0.030	1.356	1.019	0.029
19	1.156	1.007	0.012	1.244	1.018	0.029	1.252	1.017	0.027
20	1.161	1.007	0.012	1.235	1.017	0.029	1.212	1.016	0.026
21	1.131	1.007	0.012	1.191	1.017	0.027	1.193	1.016	0.026
22	1.096	1.007	0.012	1.168	1.018	0.027	1.177	1.018	0.028
23	1.087	1.008	0.012	1.178	1.019	0.029	1.177	1.019	0.029
24	1.087	1.008	0.012	1.164	1.020	0.029	1.165	1.021	0.029
25	1.075	1.009	0.012	1.143	1.018	0.027	1.151	1.019	0.026
26	1.063	1.010	0.013	1.118	1.014	0.024	1.112	1.015	0.021
27	1.042	1.010	0.009	1.062	1.007	0.013	1.071	1.010	0.015
28	1.056	1.009	0.011	1.094	1.014	0.024	1.081	1.012	0.017
29	1.027	1.010	0.009	1.051	1.009	0.015	1.082	1.015	0.023
30	1.017	1.008	0.007	1.001	1.001	0.000	1.030	1.009	0.012

¹ eV—electric vehicle.

and

Table 10. Comparison of the travel route extension factors on account of the on-route recharging necessity for the three charging station layouts—values obtained for the transfers which required recharging. Results obtained for the C-segment vehicles.

Battery Charge Level eV 1	Route Extension Factor for the Layout of Charging Stations as per Method I			Route Extension Factor for the Layout of Charging Stations as per Method II			Route Extension Factor for the Layout of Charging Stations as per Method III
	Max.	Mean	Std. Dev.	Max.	Mean	Std. Dev.	Max.	Mean	Std. Dev.
5	2.588	1.034	0.071	2.615	1.033	0.072	2.531	1.038	0.083
6	2.775	1.044	0.095	2.799	1.066	0.124	2.862	1.068	0.129
7	2.606	1.035	0.079	2.882	1.069	0.126	2.902	1.089	0.151
8	2.423	1.028	0.065	2.579	1.055	0.100	2.955	1.083	0.149
9	2.233	1.021	0.046	2.545	1.047	0.091	2.892	1.075	0.146
10	2.462	1.019	0.044	2.588	1.041	0.079	2.161	1.059	0.107
11	1.792	1.015	0.029	2.169	1.036	0.067	2.034	1.047	0.081
12	1.817	1.014	0.026	1.972	1.033	0.059	1.836	1.040	0.065
13	1.325	1.013	0.024	1.716	1.031	0.053	1.647	1.035	0.056
14	1.301	1.012	0.021	1.708	1.029	0.051	1.647	1.033	0.052
15	1.256	1.011	0.021	1.453	1.027	0.045	1.526	1.032	0.049
16	1.256	1.010	0.019	1.453	1.026	0.042	1.441	1.031	0.046
17	1.220	1.010	0.018	1.477	1.024	0.040	1.432	1.029	0.043
18	1.209	1.009	0.017	1.425	1.022	0.037	1.432	1.027	0.040
19	1.189	1.009	0.016	1.437	1.021	0.035	1.426	1.025	0.038
20	1.183	1.009	0.015	1.419	1.020	0.033	1.410	1.023	0.035
21	1.161	1.009	0.015	1.412	1.019	0.031	1.375	1.021	0.032
22	1.161	1.009	0.015	1.423	1.018	0.030	1.411	1.020	0.030
23	1.164	1.009	0.015	1.347	1.019	0.031	1.342	1.019	0.028
24	1.187	1.009	0.015	1.260	1.019	0.030	1.227	1.018	0.028
25	1.180	1.008	0.015	1.261	1.018	0.029	1.210	1.017	0.026
26	1.152	1.008	0.014	1.247	1.019	0.029	1.217	1.018	0.027
27	1.102	1.009	0.015	1.237	1.020	0.031	1.208	1.019	0.028
28	1.097	1.009	0.015	1.217	1.020	0.031	1.190	1.021	0.029
29	1.087	1.010	0.015	1.219	1.024	0.034	1.175	1.022	0.029
30	1.079	1.011	0.016	1.174	1.023	0.033	1.151	1.024	0.030
31	1.079	1.012	0.017	1.143	1.020	0.030	1.123	1.021	0.027
32	1.083	1.012	0.017	1.173	1.020	0.032	1.151	1.020	0.028
33	1.077	1.013	0.016	1.150	1.012	0.025	1.130	1.015	0.024
34	1.077	1.014	0.017	1.062	1.007	0.013	1.063	1.009	0.014
35	1.067	1.011	0.015	1.100	1.016	0.027	1.130	1.020	0.032
36	1.022	1.008	0.007	1.136	1.017	0.037	1.083	1.015	0.022
37	1.022	1.008	0.007	1.030	1.005	0.009	1.060	1.012	0.018
38	1.017	1.009	0.009	1.001	1.001	0.000	1.031	1.018	0.013

¹ eV—electric vehicle.

and in Figure 10. The maximum value of the mean route extension was recorded for the charging station layout established using method III, but even so the value did not exceed 9%. However, one should note that depending on the vehicle battery charge level at the time of departure from the travel start point, the length of the selected routes could even be greater than twice the optimal route.

The last group of results pertains to energy consumption. It was decided that, in this case, the energy consumption increase would be established as the difference between the energy consumed by an electric vehicle that needed to be recharged on-route and the energy it would have consumed without recharging (assuming that the vehicle had been fully charged when departing from the travel start point). Depending on the layout of the charging stations, this value determined the additional energy needed to reach the charging station.

Table 11. Comparison of the values of energy consumption increase on account of the on-route recharging necessity for the three charging station layouts—values obtained for the transfers which required recharging. Results obtained for the B-segment vehicles.

Battery Charge Level eV 1	Energy Consumption Increase for the Layout of Charging Stations as per Method I (kWh)			Energy Consumption Increase for the Layout of Charging Stations as per Method II (kWh)			Energy Consumption Increase for the Layout of Charging Stations as per Method III (kWh)
	Max.	Mean	Std. Dev.	Max.	Mean	Std. Dev.	Max.	Mean	Std. Dev.
5	9.274	2.693	2.633	9.162	2.609	2.553	9.001	2.504	2.438
6	9.443	2.671	2.588	9.639	2.743	2.601	9.647	2.755	2.587
7	9.124	2.471	2.404	9.469	2.645	2.517	9.901	2.765	2.581
8	8.834	2.262	2.210	9.447	2.465	2.353	9.901	2.659	2.484
9	8.433	2.099	2.053	9.473	2.318	2.213	9.424	2.481	2.338
10	8.410	1.958	1.916	9.447	2.172	2.073	9.438	2.301	2.174
11	7.817	1.831	1.790	7.988	2.047	1.949	7.724	2.155	2.038
12	7.055	1.718	1.679	7.996	1.943	1.845	7.724	2.052	1.935
13	6.842	1.610	1.571	6.848	1.839	1.744	6.974	1.973	1.853
14	6.299	1.518	1.480	6.550	1.745	1.650	6.811	1.886	1.769
15	6.034	1.425	1.390	6.162	1.653	1.565	6.241	1.772	1.662
16	5.989	1.365	1.329	5.696	1.588	1.501	5.687	1.694	1.587
17	5.691	1.318	1.280	5.671	1.536	1.449	5.687	1.619	1.520
18	4.925	1.261	1.223	5.067	1.498	1.410	5.474	1.567	1.469
19	4.746	1.195	1.155	5.117	1.452	1.357	5.253	1.529	1.434
20	4.453	1.128	1.086	4.078	1.402	1.307	4.095	1.490	1.398
21	3.689	1.088	1.045	4.042	1.387	1.288	4.010	1.472	1.375
22	3.442	1.084	1.036	3.609	1.405	1.296	3.656	1.499	1.388
23	3.125	1.094	1.042	3.212	1.413	1.289	3.273	1.518	1.393
24	2.783	1.131	1.071	3.179	1.471	1.335	3.273	1.582	1.438
25	2.723	1.117	1.048	3.078	1.444	1.320	3.122	1.611	1.477
26	2.723	1.220	1.136	2.913	1.533	1.433	2.903	1.668	1.554
27	1.807	1.165	1.078	2.913	1.690	1.636	2.767	1.802	1.725
28	1.804	1.200	1.124	2.913	1.727	1.611	2.767	1.840	1.736
29	1.787	1.123	1.033	2.327	1.558	1.485	2.477	1.779	1.652
30	1.745	1.075	1.003	2.296	1.496	1.489	2.446	1.816	1.735

¹ eV—electric vehicle.

and

Table 12. Comparison of the values of energy consumption increase on account of the on-route recharging necessity for the three charging station layouts—values obtained for the transfers which required recharging. Results obtained for the C-segment vehicles.

Battery Charge Level eV 1	Energy Consumption Increase for the Layout of Charging Stations as per Method I (kWh)			Energy Consumption Increase for the Layout of Charging Stations as per Method II (kWh)			Energy Consumption Increase for the Layout of Charging Stations as per Method III (kWh)
	Max.	Mean	Std. Dev.	Max.	Mean	Std. Dev.	Max.	Mean	Std. Dev.
5	9.115	2.668	2.612	9.005	2.552	2.499	8.846	2.442	2.390
6	9.648	2.653	2.570	9.473	2.662	2.542	9.157	2.633	2.511
7	9.280	2.518	2.444	9.881	2.681	2.537	9.731	2.734	2.548
8	8.922	2.367	2.301	9.372	2.566	2.440	10.129	2.706	2.520
9	8.682	2.214	2.158	9.372	2.424	2.308	9.731	2.617	2.440
10	8.682	2.087	2.034	9.372	2.305	2.195	9.624	2.482	2.330
11	8.266	1.975	1.928	9.372	2.193	2.089	9.262	2.348	2.213
12	8.266	1.871	1.826	9.658	2.091	1.987	9.111	2.216	2.093
13	7.682	1.778	1.733	7.851	1.992	1.891	7.591	2.100	1.982
14	7.135	1.688	1.645	7.858	1.925	1.823	7.591	2.013	1.895
15	6.798	1.605	1.562	7.320	1.822	1.723	7.387	1.960	1.840
16	6.488	1.529	1.487	6.730	1.753	1.654	6.854	1.895	1.774
17	6.191	1.463	1.420	6.438	1.680	1.584	6.345	1.823	1.705
18	5.927	1.393	1.352	6.055	1.620	1.527	6.134	1.748	1.632
19	5.696	1.353	1.311	5.810	1.575	1.483	6.134	1.679	1.566
20	5.671	1.306	1.263	5.573	1.534	1.440	5.716	1.619	1.510
21	5.015	1.256	1.211	5.777	1.489	1.396	5.669	1.565	1.461
22	4.797	1.221	1.176	4.980	1.459	1.365	5.380	1.527	1.423
23	4.707	1.177	1.129	4.980	1.445	1.345	5.162	1.518	1.415
24	4.643	1.124	1.075	4.923	1.405	1.304	4.618	1.464	1.364
25	3.871	1.073	1.024	3.972	1.352	1.249	3.957	1.431	1.333
26	3.626	1.047	0.998	3.691	1.352	1.245	3.846	1.438	1.333
27	3.382	1.041	0.987	3.547	1.353	1.235	3.593	1.458	1.339
28	3.110	1.038	0.981	3.500	1.367	1.244	3.424	1.451	1.320
29	2.778	1.049	0.986	3.330	1.384	1.237	3.217	1.520	1.378
30	2.603	1.080	1.007	3.111	1.441	1.295	3.217	1.562	1.403
31	2.676	1.092	1.004	2.891	1.412	1.278	2.984	1.577	1.430
32	2.676	1.148	1.056	3.189	1.465	1.329	2.984	1.657	1.517
33	2.383	1.174	1.074	3.053	1.591	1.508	2.780	1.729	1.617
34	1.776	1.161	1.054	2.863	1.705	1.652	2.719	1.821	1.745
35	1.773	1.161	1.074	2.863	1.687	1.560	2.876	1.807	1.650
36	1.756	1.124	1.054	2.715	1.643	1.527	2.434	1.717	1.591
37	1.756	1.059	0.985	2.287	1.472	1.428	2.434	1.683	1.580
38	1.715	1.030	0.951	2.256	1.540	1.531	2.404	1.964	1.802

¹ eV—electric vehicle.

summarise these values for the different EV battery charge levels. What the tables clearly highlight is the increase in the mean value of the additional energy consumption as the pre-travel vehicle battery charge level declines (Figure 11).

6. Discussion

The research discussed in the paper has revealed differences in the utility value of charging stations as per the diverse scenarios of their siting in the town. Bearing in mind the need to promote eco-friendly solutions, electromobility, including electric vehicles, is one of the trends which calls for further promotion [29]. One of the reasons why the development of electromobility is hampered appears to be the travelling population’s anxiety about the limitations caused by the necessity of the vehicle recharging on-route, referred to as range anxiety (more broadly discussed in [30,31], inter alia). At the same time, it should be noted that vehicle recharging by means of municipal charging stations usually proceeds much faster than charging at home. Therefore, when assessing the distribution of charging stations over the area of a municipality, it is important to determine the share of travels which cannot be completed. The research performed by the authors implies that the locations chosen as potential charging station sites by application of the three analysed methods are all characterised by high efficiency, since the inability to reach the destination point was recorded only for some vehicles, and only when their initial battery charge level was lower than 8–9% (B-segment vehicles) or 10–11% (C-segment vehicles) at the start of the travel. The conclusion that follows is that users of electric vehicles should maintain a larger energy reserve in order to be sure that they reach the destination. The author of certain papers made an assumption that the minimum state of charge should be kept at 20% [32]. However, depending on how long the travel was to be, the need for onroute recharging was observed starting from 30% (B-segment vehicles) and 38% (C-segment vehicles). Nevertheless, it should be noted that, in this respect, the travels could be successfully completed despite the on-route recharging.

Every need for recharging makes it necessary to depart from the optimal route and additionally head for a charging station. The tool used in the research, namely the travel planner, made it possible to select a specific charging station, considering the given layout of charging stations, for which the total travel time would be the shortest and the travel could be completed in reality (i.e., the vehicle battery charge level upon the arrival at the charging station would not fall below the safe reserve of 5%). However, the analyses performed by the authors showed the degree to which the route should be extended in order to make the travel completion possible. For the vehicle battery charge level at the travel start ranging up to more than 12%, the mean route extension on the charging station layout as per method I did not exceed 1%. When using methods II and III, the mean route extension was higher, reaching 2.7% and exceeding 3%, respectively. However, for some travels, it was determined that the distance of twice the optimal route length was significantly exceeded. Figure 12 and Figure 13 summarise and compare the route extension factors in histograms of the series of data distribution for successive vehicle battery charge levels at the travel start.

When analysed for the B-segment vehicles, the graphs (Figure 12) clearly imply aggregation of the number of route extension factors in the initial histogram formation range, i.e., for a 7–8% vehicle battery charge level at the travel start, regardless of the methods proposed for the siting of charging stations. At the same time, the graph behaviour (Figure 12a, method I) indicates that the route extension factor increases the least under methods II and III, as illustrated in Figure 12b,c, respectively.

For the C-segment vehicles, analogically to the B-segment ones, the analysis of the above graphs implies aggregation of the number of route extension factors in the initial histogram formation range, i.e., for a 7–8% vehicle battery charge level at the travel start, regardless of the methods proposed for the siting of charging stations. At the same time, the graph behaviour (Figure 13a, method I) indicates that the route extension factor increases the least under methods II and III, as illustrated in Figure 13b,c, respectively. Thus, the layout of the charging stations proposed as per method I, regardless of the vehicle segment assumed, is characterised by the best distribution of the stations over the territory of the GZM Metropolis subject to the research, which is confirmed by a lower number of route extension cases.

Different spatial arrangement of the charging stations entails differences in the energy consumption by the electric vehicles performing transfers with necessary recharging. From the perspective of municipalities, this is an important decision making factor affecting the choice of the charging station siting method. The research results have showed that, for the travels completed with on-route vehicle recharging, assuming the basic battery charge level (above 11% of the battery capacity), where method I was used, the mean consumption increase ranged at 1–2 kWh, while for the charging station layout obtained by methods II and III, the additional energy consumption was ca. 1 kWh higher.

This provides local authorities with valid information concerning the total energy consumption. However, since the research covered only selected travels (between 304 transport territories in the area subject to analysis), such collations have not been included in this article, since they do not exhaust the actual demand—the actual number of travels within the analysed area. The evaluation method proposed in this paper basically focuses on the accessibility of charging stations over the entire area. However, it should be noted that the total number of battery electric vehicles currently in operation in Poland does not exceed 7000 (data as of the end of June 2020 [33]), which—in the authors’ option—makes the analysis covering over 92 thousand transfers performed within a relatively small area a convincing sample for the chosen part of the country’s territory (ca. 8.2% of the area of Poland, ca. 5.8% of the total population of Poland).

The methods analysed in the article for in-city siting of electric vehicle charging stations have ultimately made it possible to establish the potential according to which a given area of 100 m × 100 m could be assessed as more or less relevant to installation of charging stations. Therefore, for the sake of the most transparent presentation of the method proposed for the evaluation of the layout of charging stations, it was assumed that, in the areas of the highest potential, identical Type 2 charging stations with a capacity of 7.4 kW would be installed. However, it should be noted that such a mode of evaluation can also be applied to charging stations of different types when using the travel planner. The layout of the charging stations, including specification of their plug types, power capacity, and the current station occupancy, provides the input to the network graph in the travel planning tool. Considering the present unavailability of the given charging station, the travel planner can take into account not only the physical occupancy by an electric vehicle being charged, but also the current grid overloading (having received the relevant information form the electricity distribution network operator).

7. Conclusions and Further Research

The authors of the article have proposed that a travel planner should be used to analyse the alternative scenarios for siting of charging stations. This would provide local authorities with significant decision making support. In the case study discussed in the paper, the optimum layout of charging stations from the perspective of the potential to complete travels was produced by method I. However, it should be noted that such a result does not necessarily mean that this siting method will prove superior in other urban areas. What the research results take into account is the local conditions, and primarily the road network layout, and the evaluation process should definitely be repeated for another area.

Having determined an adequate layout of charging stations by taking the specifics of the given area into consideration, one can continue to mould the development of the municipal electrical power supply and distribution system in such a manner as to deliver energy to the designated locations intended for installation of charging stations.

The method proposed in the paper can also be used to increase the travelling population’s confidence in the use of electric vehicles in the given area, even in situations when the vehicle is not 100% charged. The route calculation process takes the vehicle recharging option into account. Research makes it possible to identify a safe energy reserve for a travel to be completed without recharging as well as for a travel that includes on-route recharging in a manner which ensures that the vehicle is not immobilised on the way, thus minimising the phenomenon referred to as range anxiety.

It is planned that the analyses to be performed under further research should account for the actual transfers performed by the travelling population (field surveys), which does not affect the method itself, but may increase the accuracy of the results it delivers. The series of travels pre-assumed for and examined in the article are combinations of centroids of 304 transport territories, and when combined with the review of the available travelling alternatives, they reflect a wide range of individual transfers. As demonstrated in the article, the current size of the fleet of battery electric vehicles in Poland substantiates the approach adopted by the authors. For a large number of such vehicles operated in a relatively small area, using the matrix of travels based on real-life data makes it possible to take also some other research problems into consideration, like those related to charging station sharing during the day. One of further important problems to be addressed in the future research is the need for scheduling of vehicle charging procedures, while temporary occupancy of each charging station makes it necessary for vehicle users to either queue or to plan a visit to another charging station, more distant from the optimal route. However, in order to account for this problem in a simulation of a series of travels, one must know not only the actual traffic streams, but also their distribution in time (variability over the day).