Navigation – Plan du site

AccueilPublicationsCollection générale2021Possible models for the roll-out ...

2021
162

Possible models for the roll-out of a public charging infrastructure for electric vehicles in the Brussels-Capital Region

Quels modèles pour le déploiement d’une infrastructure de recharge publique pour véhicules électriques en Région de Bruxelles-Capitale ?
Welke modellen voor de uitrol van publieke laadinfrastructuur voor elektrische voertuigen in het Brussels Hoofdstedelijk Gewest?
Quentin De Clerck et Lieselot Vanhaverbeke
Traduction de Philippe Bruel
Cet article est une traduction de :
Welke modellen voor de uitrol van publieke laadinfrastructuur voor elektrische voertuigen in het Brussels Hoofdstedelijk Gewest?  [nl]
Autre(s) traduction(s) de cet article :
Quels modèles pour le déploiement d’une infrastructure de recharge publique pour véhicules électriques en Région de Bruxelles-Capitale ? [fr]

Résumés

L’impact du transport sur le changement climatique ne cesse de s’intensifier, faisant des véhicules durables une étape importante pour réduire les émissions. Dans ce contexte, un grand réseau de bornes de recharge permettrait de stimuler l’adoption de la voiture électrique, mais ce réseau demeure trop limité à Bruxelles. Dans cet article, nous présentons deux méthodes complémentaires susceptibles de soutenir le déploiement d’une infrastructure de recharge. L’analyse pour la recharge opportuniste publique consiste en une approche basée sur un point d’intérêt (point of interest, POI) qui démontre que l’accessibilité de l’infrastructure de recharge tend à s’améliorer avec l’installation de bornes à proximité d’importants pôles de mobilité bruxellois. Pour la recharge résidentielle publique, des emplacements sont identifiés sur la base de modèles de localisation tenant compte des prévisions de la demande en véhicules électriques (VE). Le déploiement qui en résulte s’impose en premier lieu dans les quartiers où l’adoption escomptée de VE est forte, et s’étend ensuite de manière équilibrée sur l’ensemble du territoire de la Région. La combinaison des deux méthodes offre une solution tant aux habitants (modèles de localisation) qu’aux visiteurs et aux navetteurs (approche POI).

Haut de page

Notes de la rédaction

To see the figures in a better resolution, open the article online and click on “Original” below them.

Texte intégral

List of Abbreviations

1BAU: Business as usual

2BEV: Battery Electric Vehicle

3CBC: Choice-based conjoint

4CO2: Carbon Dioxide

5DMCLP: Dynamic Maximal Covering Location Problem

6EU: European Union

7EV: Electric Vehicle

8E2SFCA: Enhanced two-step floating catchment area method

9MCLP: Maximal Covering Location Problem

10NOX: Nitrogen Oxides

11PHEV: Plug-in hybrid electric vehicle

12PM: Fine dust

13POI: Point of interest

Introduction

14Passenger transportation is a major contributor to climate change because of its use of fossil fuel engines. In Europe, passenger cars are responsible for 12 % of all CO2 emissions [European Commission, 2018], but in addition to CO2, other greenhouse gases are emitted that can cause even more damage in densely populated cities. These more local emissions are, for example, nitrogen oxides (NOX) or fine dust (PM) [Van Mierlo, et al., 2017]. In addition to being harmful to the environment, these emissions are also harmful to citizens' health. In the Brussels-Capital Region, these emissions are mainly emitted by road transportation. In 2012, road transportation was responsible for 67 % of NOX emissions, 47 % of PM2.5 emissions and 27 % of CO2 emissions [Willocx, 2015]. A switch from fossil fuel engines to alternative lower emissions vehicle technologies could reduce these values. One of the alternatives to conventional fuel vehicles are the electric vehicles (EVs). Since these vehicles do not emit any local emissions, they could offer a solution to improve Brussels-Capital Region's air quality. The low emission zone introduced in the Region on 1 January 2018 demonstrates the Region's will to tackle this problem. This zone should encourage the inhabitants of Brussels to make the vehicle fleet greener by purchasing alternative vehicle technologies. Today, however, the use of electric vehicles is still limited. One of the possible causes could be the availability of the charging infrastructure. Initially, the roll-out of charging stations is slow, as charging station operators without users run the risk of a loss-making business case and consumers do not want to purchase an EV without being sure of finding a charging station to recharge their vehicle. The latter is even more appropriate for the Brussels-Capital Region, where 22 % of families have a private parking space to recharge at home [Ermans, 2019]. In this situation, a large network of charging stations should stimulate the switch to electromobility [Mersky et al., 2016]. The lack of an extensive charging infrastructure in the Brussels-Capital Region is therefore an impeding factor to the use of EVs. In this regard, the Brussels-Capital Region published a vision on charging infrastructure in June 2020, which foresees 11 040 charging stations by 2035 [Brussels Environment, 2020]. This vision also emphasizes the Region's ambition to organize the planning of the roll-out on the basis of cartographies. Meanwhile, since 2018, the Region has been rolling out a basic network of public charging infrastructure, of which 162 charging stations have been installed and 25 are pending government approval [Charge.brussels, 2021]. This vision and the need for charging infrastructure in the Brussels-Capital Region are the starting point of this paper.

15The roll-out of the charging infrastructure also fits within the context of the recent mobility strategy developed by the Brussels-Capital Region – Good Move – which aims, among other things, to improve air quality through adjustments in Brussels mobility [Brussels-Capital Region, 2019]. Although this plan deals more broadly with the various facets of mobility, such as improvements in public transportation, the promotion of the sharing economy, etc., the car is the most popular means of transportation in the Region and this aspect of mobility also needs to be improved (46 % of journeys are made by car in the Brussels-Capital Region [Derauw et al., 2019]). To achieve this, the Region aims to reduce the number of cars and make the Brussels car fleet greener. Since EVs are better for air quality in an urban context [Hooftman et al., 2018], a shift in vehicle technologies from fossil fuels to electricity is a step towards better air quality in the Region.

16In this paper, we propose two complementary analyses to identify locations for installing public charging infrastructure for passenger cars in the Brussels-Capital Region. Both analyses outline an evolution of the charging infrastructure in the Brussels-Capital Region between 2015 and 2020. These analyses were carried out on behalf of the Brussels-Capital Region authorities between 2014 and 2016 in the context of two policy preparatory studies.

17The first analysis describes the charging infrastructure's accessibility in the Brussels-Capital Region and proposes a method to extend the public charging infrastructure for occasional charging, ie recharging the vehicle that is parked while the user is carrying out an activity (such as shopping, visiting a tourist attraction, eating or drinking...). So this analysis focuses on both residents and visitors and can also be used for planning charging infrastructure for electric shared cars. The second analysis focuses on residential charging, for example for EVs that are parked along the public road at night. For this type of charging, a location model is used that proposes optimal locations based on the inhabitants' socio-demographic characteristics. Since socio-demographic characteristics of the inhabitants are used as a basis, this analysis mainly focuses on the use of the charging stations for private vehicles of Brussels residents. Both analyses can be performed separately for their specific purpose, but the results can be considered complementary. Since this paper focuses on normal charging (maximum 22 kW), electric taxis are outside the scope. Electric taxis generally use fast charging stations to remain operational longer and faster and benefit less from normal charging infrastructure. Also, the quantification of the impact of the increasing number of EVs on the electricity grid is not studied in this paper. However, the location of low-voltage cabins is taken into account in order to limit the costs related to the installation of the charging stations.

18The next section first outlines the Brussels situation in terms of the number of electric vehicles and compares these figures with predictions made in 2015 for the number of EVs in 2020. Section 2 describes the two methods used in this paper to determine the infrastructure's locations for public occasional charging and residential charging. The results of both methods are discussed in section 3. Finally, this paper concludes with some recommendations.

1. Electric vehicles in the Brussels-Capital Region

19At the beginning of 2015, 7 402 851 vehicles were registered in Belgium, of which 637 809 in the Brussels-Capital Region [FEBIAC, 2017]. The number of newly registered vehicles in the Brussels-Capital Region was 78 470 in 2014 and 82 889 in 2015. Of these new registrations, 434 in 2014 and 542 in 2015 were electric vehicles (both battery electric vehicles (BEV) and plug-in hybrid electric vehicles (PHEV)) [Denys et al., 2020]. Numbers from 2016 to 2019 show that the number of purchased EVs continues to increase. In 2016, 2017, 2018 and 2019, respectively 1 016, 1 666, 1 823 and 2 357 new EVs were registered in the Brussels-Capital Region [Denys et al., 2020].

20Based on the results of a previously performed choice-based conjoint (CBC) study for Flanders [Lebeau et al., 2012], the market share of BEVs and PHEVs was predicted for the newly registered vehicles between 2016 and 2020 based on consumer preferences regarding vehicle properties ( such as purchase price, charging speed or driving range) and the expected evolution of the properties. When we take into account the real sales of 2014 and 2015, we obtain the cumulative number of BEVs and PHEVs for the next five years in the Brussels-Capital Region (assuming that a similar trend is present in the Region as in Flanders). This forecast was executed in 2016 and includes 3 scenarios: business as usual (BAU), subsidy policy, and subsidy-and-infrastructure policy. In the BAU scenario, sales are forecast on the assumption that no policies are in place to drive EV uptake. This results in a total number of 14 000 vehicles in 2020. In the subsidy policy scenario, it is assumed that a financial incentive is given (an eco-bonus) to stimulate market development. In 2016 the bonus would be € 6 000 and would decrease over time, namely € 3 000 in 2017, € 2 000 in 2018 and € 1 500 in 2019. The effect of this incentive measure, compared to the BAU scenario, is a faster uptake of EVs and therefore a higher cumulated number of EVs in the Brussels-Capital Region, namely more than 16 000. The subsidy and infrastructure policy scenario builds on the subsidy policy scenario with the additional assumption that the focus will be on charging infrastructure, with the number of charging stations increasing by 100 each year. The combination of these measures was also analysed with the results of the CBC study. It would initially lead to the same effects as the subsidy scenario, but in the longer term, from 2019 on, the total number of EVs would rise to more than 18 000. This delayed effect can be explained by the in time progressive number of charging stations in combination with the in time degressive financial incentive. The results of this scenario-based demand forecast are depicted in Figure 1.

Figure 1. Forecast number of EVs in Brussels-Capital Region 2016-2020

Figure 1. Forecast number of EVs in Brussels-Capital Region 2016-2020

Sources: MOBI for the projections and visualization and [Denys et al., 2020] for the historical data)

21If we compare the predicted numbers with the BAU scenario, we notice that the forecast is more positive than the actual numbers. The historical numbers also indicate an increase in the number of registered vehicles, but the number of annual registered vehicles was on average 500 units lower than predicted. Taking into account people who exchanged their EV for a new EV or another type of vehicle or means of transportation, the total number of EVs registered in the Brussels-Capital Region at the end of 2019 amounts to 6 672 EVs [Denys et al., 2020]. The actual total number of EVs in 2019 lies between the 2017 and 2018 forecasts of the BAU scenario and is 35 % smaller than predicted. This comparison with the actual numbers demonstrates the difficulty of making accurate predictions for a market that is in the early stages of its development. Several factors play a role: the greater uncertainty surrounding the evolution of vehicle properties is a major challenge in this regard. Changes in vehicle properties, such as purchase price, running costs, driving range, density of the charging infrastructure, or even charging speed, can all have an impact on the final forecast and must be accurately estimated for the coming years. It is not an easy job and requires a good knowledge of the technology and the trends in the field. Another possible factor that could have influenced the forecast is that the measured preferences for Flemings would not be fully transferable to the inhabitants of Brussels. These preferential differences can be observed in the EVs' market share evolution between 2014-2019 for Flanders and Brussels (see Table 1) [Denys et al., 2020]. It can therefore be concluded that one should be careful with the use of CBC data for making predictions, both in terms of technological evolution and in the way to apply the most correct preference.

Table 1. Evolution of the market shares of electric vehicles in Flanders and the Brussels-Capital Region [Denys et al., 2020]

2014

2015

2016

2017

2018

2019

Brussels-Capital Region

0,55 %

0,66 %

1,31 %

2,18 %

2,42 %

3,01 %

Flanders

0,50 %

1,01 %

2,38 %

3,55 %

3,09 %

3,99 %

  • 1 These detailed figures were personally provided by Mr Kwanten of FPS Mobility. The figures for mid- (...)

22When comparing the spatial distribution of the adoption with the predictions per municipality for 2020, shown in Figure 2, one notices that the trends are broadly correct1. The municipalities of Brussels, Ixelles, Woluwe-Saint-Lambert, Woluwe-Saint-Pierre and Uccle show the highest adoption rates as predicted. On the other hand, there are municipalities that show a lower adoption rate than predicted, such as Evere, Auderghem, Molenbeek-Saint-Jean and Watermael-Boitsfort. The latter municipality in particular has a slower adoption than predicted, as this municipality was considered a possible frontrunner. Saint-Josse-ten-Noode, on the other hand, shows a much stronger adoption than predicted. The adoption in this municipality is mainly due to the increasing number of many electric commercial vehicles. Except for the municipalities of Watermael-Boitsfort and Saint-Josse-ten-Noode, the spatial distribution of adoption follows the predicted pattern, but as mentioned above, to a lesser extent than predicted in terms of numbers. The spatial distribution of demand in the location model can still be regarded as quite adequate, with those exceptions. This leads us to three conclusions regarding the use of EV projections to forecast demand: demand is overestimated due to the unpredictable nature of the EV market (which is in an initial phase), vehicle preferences need to be better tailored to the specific Brussels context and the spatial distribution of demand was reasonably well predicted. Since the spatial distribution of the prediction is more important than accurately predicting the number of vehicles, the use of this method is not a problem as long as it is taken into account that the numbers are an overestimation.

Figure 2. Number of electric vehicles in the Brussels Capital Region end 2018 and mid 2019 compared to the forecast made in 2015 for 2020

Figure 2. Number of electric vehicles in the Brussels Capital Region end 2018 and mid 2019 compared to the forecast made in 2015 for 2020

Source: FPS Mobility for the historical data and MOBI for the projection and visualization

2. Methodology

23These sections describe the methodology for public occasional charging, public residential charging and the location model.

2.1. Public occasional charging

24The “enhanced two-step floating catchment area method” (E2SFCA) is used to identify the accessibility of charging locations for public occasional charging [Luo and Qi, 2009; Marion et al., 2015]. In the scientific literature, the term accessibility is defined as “the relative ease with which a facility or activity can be reached from a given location” [Luo and Wang, 2003]. In other words, accessibility expresses the relationship between supply and demand for the use of a particular type of facility.

25In this analysis, accessibility will measure the ratio between the number of charging points (supply) and the number of address points (demand). This method takes into account the distance between demand and supply by measuring the amount of demand within a defined area around the charging station. The size of the area is determined by a threshold value that is equal to a walking distance of 9 minutes from the charging station. The demand is therefore equal to the number of address points that are located in the area of a charging station. An address point is a reference to one or more house numbers within a building. The address point can be used for multiple purposes, eg residences, shops, offices, schools, recreation, etc. For each address point, a score is calculated that indicates to what extent it is served. Since an address point can have different functions, the accessibility of address points is not only calculated for Brussels residents, but also for visitors or commuters planning an activity (shopping, visiting family or friends, work...) at an address point. The motives or profiles of the potential users are therefore not taken into consideration, only their quantity by means of the number of address points. This emphasizes the applicability of this method for public occasional charging and the different segments of the target audience.

26The first step in calculating the scores resides in first calculating an accessibility score for each charging station based on the amount of address points located in the area. A score is calculated for each address point by summing the accessibility scores of the charging points in their range (usually 2 charging points per charging station). This score is adjusted by the walking distance from the address point to the charging station and the number of other addresses in the range of the charging points. This means that the further an address is located from a charging station or the more other addresses there are in the area of the charging station, the less accessible this charging station will be for this address. The result of this method is a score per address point in the range of one or more charging stations. This score therefore indicates how accessible the charging infrastructure is for the address points in the Brussels-Capital Region.

27Three datasets are necessary for the implementation of the E2SFCA method, namely, a dataset of charging stations, a network of the streets and the addresses of the buildings in the Brussels-Capital Region. In this study, the administrative database of UrbIS [CIBG, 2021] was used for the street and address details. Since there was no official overview of the charging infrastructure in the Brussels-Capital Region in 2015, a list of the existing charging infrastructure was drawn up by consulting 5 different charging infrastructure databases (ASBE, Chargemap, Oplaadpunten.nl, Oplaadpunten.org and Openchargemap). This initial list was supplemented with data obtained from some charging station operators and was eventually validated by conducting a field study of all listed charging stations. In total, 144 charging points for 220 004 address points were identified in the Brussels-Capital Region.

2.2. Public residential charging: splitting up the EV demand by building blocks

28The previously performed CBC analysis [Lebeau et al., 2012], discussed in section 1, was used to forecast EV adoption in the Brussels-Capital Region. Given the inherent spatial nature of the location analysis, the predicted EV demand should be more finely assigned to spatial entities. Ideally, all data is available at the same spatial level, but in practice this is not the case and therefore a step-by-step approach is applied. As a result, the number of registered EVs is uniformly distributed, starting from metropolitan Region level up to the building blocks that form the smallest spatial entity in our analysis. We calculate an EV suitability score for each district to account for spatial socio-demographic differences. The remainder of this calculation is based on the results of the BAU scenario.

29In the first allocation phase, the total number of predicted EVs is distributed proportionally among the 19 municipalities according to the number of newly registered electric vehicles by private individuals in each municipality in the year 2015. Since one of the objectives of this study is to optimally locate the public charging infrastructure for the benefit of residents, only electric vehicles sold to private individuals are taken into account (see appendix for more details).

30The distribution of this number is presented on the map in Figure 3. Note that in 2 municipalities, Berchem-Sainte-Agathe and Saint-Gilles, no private EVs were registered; the highest number was registered in Uccle. In order to simulate the uptake of EVs in the first mentioned municipalities, a small correction factor was applied so that a valid prognosis for the future is obtained.

Figure 3. Number of registered EVs by private individuals per municipality in 2015

Figure 3. Number of registered EVs by private individuals per municipality in 2015

Source: FPS Mobility for the historical data and MOBI for the visualisation

31In the next phase, the demand for EVs (both PHEVs and BEVs) is distributed over the different neighbourhoods in a municipality. This step is crucial to take into account the spatial variation with regard to socio-demographic characteristics of the inhabitants. The Brussels Neighbourhood Monitoring database was used as a source for this information.

32Note that the Brussels-Capital Region is divided into 118 residential districts (in which 99,7 % of the population lives), 6 industrial zones and stations, 18 green areas and 3 cemeteries. In our analysis we exclude the 3 cemeteries and there is no information about building blocks in the neighbourhoods Parc des Étangs and Parc Astrid (in Anderlecht), Schaerbeek Station and Parc du Cinquantenaire (in Brussels).

33Note also that the boundaries of the Monitor neighbourhoods do not necessarily coincide with the boundaries of the municipalities, therefore the demand of a neighbourhood is calculated based proportionally on the area it comprises within two (or sporadically three or four) municipalities.

34In order to take into account the socio-demographic profiles of the inhabitants, the variables income, living area per inhabitant and household size are used to break down the demand at the municipality level to the demand at neighbourhood level. An indicator – EV suitability score – was calculated for each district. This score is multiplicatively proportional to the average income and living area per inhabitant in a neighbourhood and inversely proportional to the average household size. In the scientific literature Plötz et al. [2014]; Axsen et al. [2016]; Hardman et al. [2016] high income and multi-person households are often cited as characteristic of households that own an EV. Income is also mentioned as planning criteria for the roll-out planned in the vision of the Brussels-Capital Region [Brussels Environment, 2020]. The living area per inhabitant is not explicitly mentioned in the literature. The study by Plötz et al. [2014] does determine that smaller or medium-sized cities are considered more perfect for EV adoption than large cities. The reasons are that vehicle ownership is higher there and that there are also more households owning a private parking space. These characteristics can be found in the municipalities with a larger living area per inhabitant such as Uccle, Auderghem, Woluwe-Saint-Lambert, Woluwe-Saint-Pierre and Watermael-Boitsfort [BISA, 2018]. The average predicted demand per neighbourhood ranges from 14,68 EVs in 2016 to 69,45 in 2020; the median from 9,18 to 43,43 (see Figure 4).

35In the final step towards creating question points for the location model, the demand per district is allocated uniformly to the number of building blocks in a district. A building block is delimited by streets or by municipal boundaries. 4 992 building blocks are taken into account in the analysis. Average forecast demand per block ranges from 0,6 EVs in 2016 to 2,81 EVs in 2020; the median from 0,26 to 1,23. We do not depict the building blocks individually, but they can be identified by the centroids in Figure 5, ranging from yellow (low predicted demand) to brown (high predicted demand). As mentioned in Section 1, the demand per block is spread in a similar way to the actual adoption of EVs in the Region for mid-2019 (with the exceptions of Saint-Josse-ten-Noode and Watermael-Boitsfort). This validates for the other municipalities the spatial distribution of the demand and the locations that are gradually identified in the dynamic location model.

Figure 4. Prognosis EV uptake per neighbourhood

Figure 4. Prognosis EV uptake per neighbourhood

Source: MOBI

Figure 5. Forecast EV uptake per building block

Figure 5. Forecast EV uptake per building block

Source: MOBI

2.3. Location models for public residential charging

36The aim of this analysis is to provide – from the perspective of the public authorities – a maximal service to the inhabitants of the Brussels-Capital Region by installing or having installed a given number of charging stations. To model this mathematically, the Maximal Covering Location Problem (MCLP) is applied [Church and ReVelle, 1974].

37We consider di > 0 to be the above predicted demand for public charging infrastructure in building block i I. Public charging stations are opened at candidate locations j J. A given budget B is available for opening new charging stations that will serve the EV owners . The Ni set indicates which candidate locations are within distance d of block i.

38The MCLP model is fully formulated below:

39where xi indicates whether a building block is served by one or more public charging stations and yj indicates which candidate locations were chosen. The objective function is to maximize the degree of service for all building blocks given the constraint on the sets coverage (3) and the constraint on the budget (4).

40The MCLP model allows to use the demand predicted above at building block level. In the first scenario, the candidate locations for a charging station are the 1 292 existing cabins of the grid operator that are already equipped with a 400 V connection and a low voltage panel and for which no substantial costs have to be incurred for the power supply of a charging station. In the second scenario, the candidate locations for a charging station are also the grid operator's existing cabins, but in this scenario investments are needed to make the cabin compliant. These are 2 345 cabins. In both scenarios, the electricity grid is therefore taken into account, which is also mentioned as planning criteria in the vision of the Brussels-Capital Region [Brussels Environment, 2020]. A radius d is determined that indicates how far a resident has to walk to the charging station (via the road network); the basic principle is that the residents of a block in the radius of 350 m can meet with their charging needs. The model maximizes the charging stations service by placing the given number of charging stations at the greatest expected demand.

41The starting point is the roll-out by 2020 of 500 charging stations, each equipped with 2 charging points, i.e. 1 000 charging points. This corresponds to the rule of thumb of 1 charging point per 10 EVs when the forecast of 14 000 vehicles is used and it is assumed that semi-public infrastructure can also serve the demand. To simulate the rollout over several years, the Dynamic Maximal Covering Problem (DMCLP) is applied [Gunawardane, 1982]. Since the solution must be optimal in the long term, the solution of the MCLP in the year 2020 will be used as a starting point. An overview is then provided for each year t T (2016-2019) of the charging stations that will be opened.

3. Results

42In this section, the results of the applications of the models for the public occasional charging (section 3.1) and residential charging (section 3.2) are discussed separately.

3.1. Public Occasional Charging

43Figure 6 shows the result of the E2FSCA method for the public charging infrastructure in 2015. The squares on the Region's map represent the charging points. The yellow, orange and red zones represent the address points that are located at a walking distance of 9 minutes from a charging station. The charging infrastructure's accessibility is illustrated by the address points' score based on the demand (number of other addresses in the area) and the number of charging stations that are accessible from the address points. This accessibility score can therefore be interpreted as the number of charging points that are accessible per address point, taking into account the distances between the address points and the charging points. A score of 1 is equal to an address point that would have a charging point on its doorstep without any competition from other address points. In other words, the darker the red colour of the address points, the more accessible the charging infrastructure in their vicinity. The following factors (alone or combined) promote the charging infrastructure's accessibility for an address point, namely, a short distance to charging infrastructure from the address point, more supply of charging infrastructure from the address point or a smaller number of other address points in the area around the charging point. In total, 36 % of the addresses were served by the infrastructure in 2015, i.e. slightly more than one third of the Brussels addresses.

Figure 6. The charging infrastructure's accessibility in 2015

Figure 6. The charging infrastructure's accessibility in 2015

Source: MOBI

44The highest accessibility scores are obtained in neighbourhoods that are well served by the charging infrastructure. According to the metric, the charging infrastructure is very accessible in the western part of the Brussels Region. This can be explained by a low degree of competition, as there are very few buildings for the number of charging points. An example of a charging station located in this area is the charging station at IKEA. This charging station is therefore typically a charging station that will be used for occasional charging. It is a charging station located in an environment with few other buildings and in a commercial zone. In contrast to the charging stations in the centre of the Capital Region, which are also located at an economic core and are characterized by a reasonable degree of accessibility due to the large number of charging stations in the same neighbourhood. Lower accessibility scores can be found at charging stations where there is little charging infrastructure and many buildings in their vicinity. Such cases lead to a higher degree of competition for using the charging points in those neighbourhoods.

45In order to stimulate occasional charging, we could install new public charging infrastructure at the important poles (POI) for public transportation according to the Iris-2 plan [Brussels-Capital Region, 2011] and public car parks in the Brussels-Capital Region. These important poles are specific locations in the Brussels Region where public transportation and multimodality need to be improved. A large number of these poles are discussed in detail in the Ontwerp van Gewestelijk Plan voor Duurzaam Ontwikkeling (Draft of the Regional Plan for Sustainable Development) [Brussels-Capital Region, 2013]. These locations include economic centres (commercial centres, centres of activity such as offices or industry...), universities, hospitals, major train stations, etc... This POI-based strategy identifies locations that are potential destinations for EV users (in addition to their work and their place of residence) [Wagner et al., 2013]. These locations then allow the EV users to charge during their activities at these POIs. Placing charging stations at POIs would change the coverage of the charging infrastructure as can be seen in Figure 7. 4 charging points are allocated per location, this corresponds to two charging stations with two charging points each. In total, there are 620 charging points serving about 65 % of the address points in the Region, almost two-thirds of the addresses in Brussels. Note also that the scores in this figure are higher than in Figure 6. This means that the accessibility of the infrastructure has improved in general.

Figure 7. Charging infrastructure's accessibility after a POI approach

Figure 7. Charging infrastructure's accessibility after a POI approach

Source: MOBI

3.2. Public Residential Charging

46In the first scenario, it is assumed that the charging stations to be installed must be placed in the immediate vicinity of a low-voltage cabin.

47The result of the optimization is shown in Figure 8: the blue dots indicate the optimal locations for charging stations in 2020.

Figure 8. Optimal locations for charging stations in 2020 based on low-voltage cabins

Figure 8. Optimal locations for charging stations in 2020 based on low-voltage cabins

Source: MOBI

48Based on this configuration of the public charging stations' network, approximately 11 800 of the estimated 14 000 EV owners in 2020 will have access to a charging station within 350 m walking distance. This must of course be immediately nuanced, as only two cars can connect per charging station at the same time, and the service level will therefore be much lower in practice than the figures above suggest.

49The roll-out of the infrastructure (as a result of the dynamic model) over the years 2016-2019 is shown in Figure 9. It is clear that first there will be an extensive roll-out near the places where the estimated EV adoption is high (south-eastern municipalities ). Later on, a more balanced distribution of the charging infrastructure over the Region will follow. That is to say, the charging infrastructure will be first and mainly made available in neighbourhoods with affluent residents. Although these neighbourhoods more often have houses with a private parking space where families can charge their vehicle, there are also other families who have the financial means to buy an EV but do not have a private parking space. When a public charging point becomes available in a neighbourhood, households without a private parking space may be easier inclined to switch to electric driving. It is important to note that even if charging infrastructure is initially provided in more affluent neighbourhoods, the benefits of an increasing number of EVs in the Region are available to all thanks to better air quality and a reduction in noise pollution. This roll-out strategy also has the beneficial effect that the business case is more interesting for the charging station suppliers in the start-up years than a geographically uniform roll-out without taking the predicted adoption into account. A favourable business case is important to ensure that these investments are made, because the vision of the Brussels-Capital Region assumes that the charging station suppliers themselves will have to invest without being subsidized. Possibly, a “vehicle first and station later” strategy could supplement the need for charging infrastructure in neighbourhoods where there would be a lack of coverage and the demand for infrastructure would come from residents. This strategy is currently used by PitPoint and the Brussels-Capital Region and is described in the Regional vision [Brussels Environment, 2020]. In a later phase, the DMCLP model ensures that the charging infrastructure is further rolled out in neighbourhoods with less wealthy residents. Finally, the POI approach proposed for occasional charging provides complementarity to this method. This approach does not take into account the income of the households in the vicinity of the POI and could be implemented simultaneously to combat social inequality.

Figure 9. Optimal locations for the roll-out of charging stations in 2016-2019 based on low-voltage cabins

Figure 9. Optimal locations for the roll-out of charging stations in 2016-2019 based on low-voltage cabins

Source: MOBI

50Since the number of charging stations determines the quality of the solution, i.e. the level of service, a sensitivity analysis on this parameter was also performed. If the number of charging stations that are rolled out per year would be limited to 50 (resulting in 250 new charging stations in 2020), the service level logically turns out to be significantly lower than with the reference scenario of 100 charging stations per year. Figure 10 shows how the service level evolves over the years using the DMCLP with parameters 50, 75, 100 and 125 new charging stations, with the last group of bars representing the total number of predicted EVs per year that was used as input for the model (and would therefore represent a 100 % service level).

Figure 10. Sensitivity analysis with regard to charging stations parameter

Figure 10. Sensitivity analysis with regard to charging stations parameter

Source: MOBI

51Figure 11 shows what the laying out of the 500 charging stations looks like when all cabins are considered as possible locations. Given the more even distribution over the territory of the cabins that are not yet compliant, the distribution of the charging stations is also less concentrated, although the general pattern is similar to the optimal configuration of Figure 8.

Figure 11. Optimal locations for charging stations in 2020 based on all cabins

Figure 11. Optimal locations for charging stations in 2020 based on all cabins

Source: MOBI

52Since there are more possibilities with regard to possible locations for the charging stations, the service level with this laying out is also somewhat higher than in scenario 1. 13 281 of the 14 028 estimated owners of an EV in 2020 have access to a charging station within a walking distance of 350 m. Again, the nuance must be made that this only guarantees accessibility, and not availability, given the charging points/EVs ratio is less than 1 in 14.

Conclusion

53This paper proposes two different analyses identifying locations for the roll-out of public charging infrastructure for passenger cars. The first analysis aims at implementing a charging infrastructure network for public occasional charging based on important mobility poles and public car parks. The second analysis locates charging stations on the basis of socio-demographic information from the residents, in order to maximize the service level. Fast charging infrastructure falls outside the scope of this paper.

54Wagner et al. [2013] indicates that POIs are a good location to install charging stations for occasional charging. Placing charging infrastructure at important mobility poles and public car parks improves accessibility by one third of the coverage of the number of Brussels addresses compared to the situation in 2015. The infrastructure's accessibility is also improved as can be seen from the higher accessibility scores. This confirms that a good starting point to further roll out charging infrastructure are the locations of important POIs in the Brussels-Capital Region. Since the demand for a possible location is measured on the basis of the number of address points within a walking distance of 9 minutes, and not on the basis of usage motives or profiles, the analysis applies to residents of Brussels as well as to visitors or commuters.

55The second analysis examines where new charging locations can be implemented for residential charging. The location model uses the predicted demand down to the spatial level of building blocks as input. The demand is equal to the predicted number of electric vehicles per block between 2015 and 2020. As a control for the use of the data in the analysis, the demand forecast was tested against the latest numbers of electric vehicles in the Brussels-Capital Region. The result of this comparison indicates that the predictions were too optimistic. On the other hand, the spatial distribution of the demand largely corresponds to the spatial distribution of the forecast. Using spatial optimization models MCLP and DMCLP, the service level of new charging stations to be localized was maximized, using the existing 400 V network. Two scenarios were developed: on the one hand, on the basis of the existing low-voltage cabins, and on the other, on the basis of the cabins that are not yet fully compliant. Both approaches provide insight into the proposed location of the new charging stations. Hereby, the greater availability of non-compliant cabins compared to low voltage cabins will result in a greater service level for EV owners in 2020, although conforming these cabins to provide a connection to a charging station entails an additional cost.

56In summary, the combination of both methods focuses on neighbourhoods where the EV adoption of residents is expected to be higher, but also covers most areas of the Region through a complementary POI approach. This complementarity ensures that the infrastructure is provided at relevant locations for visitors, commuters and residents alike.

57A number of recommendations can be formulated in connection with the data and assumptions used in the methods. In the first place, in 2015, there was no official overview of all existing charging infrastructure in the Brussels-Capital Region available yet. In the meantime, the situation has partly changed, as the Region has awarded a concession to PitPoint (Total) for the expansion of the charging infrastructure network since 2018. An overview of the public charging infrastructure is now available on the charge.brussels website [Charge.brussels, 2021]. However, this is also not the complete overview of the charging infrastructure in the Brussels-Capital Region. This dataset contains all public charging stations that PitPoint has installed, but not the semi-public charging stations (charging stations that are not on public roads but are still available to users) that we listed in 2015. A dataset that contains both types of charging stations would provide a more complete picture of the current situation in the Region. In second place, with regard to predicting numbers of EVs, one can conclude that the resulting numbers should be handled with care. The numbers can easily be overestimated, especially if the market is still in its early development, such as the EV market in 2015.

58A possible avenue for further research is to integrate the results of the accessibility analysis into the location model. In other words, the accessibility score resulting from the analysis could be added to the model to take into account not only the distance between the points but also the number of surrounding addresses representing the demand. Another option could be to consider the POI as fixed charging stations in the location model. This could constitute an additional scenario if the Brussels-Capital Region would like to carry out a complementary POI-oriented roll-out. Since the planning criteria mentioned for the cartography of the Regional vision are similar to the parameters in the proposed location model, a similar model can be used in the future for the roll-out of the 11 040 charging stations from the Regional vision. However, for the amount of charging stations that the Region wishes to install by 2035, this will only be possible if some adjustments are made in the parameters such as e.g. finding other candidate locations (e.g. streets or parking spaces), defining a smaller radius to cover the demand and using a demand that is distributed at a finer level. A final avenue of research is to work with CBC data that is specifically tailored to Brussels residents. This data, coupled with the most recent EV numbers, would make it easier to predict the preferences and adoption of EVs by the inhabitants of Brussels. Hereby, the announced transition to zero-emission mobility can also be taken into account, both at the EU level [European Commission, 2021] and at the Brussels level [Brussels Environment, 2021], with the aim of achieving a zero-emission fleet for passenger cars by 2035.

This research was partly financed within the framework of projects for Brussels Environment, Brussels Mobility and BRUGEL.

Haut de page

Bibliographie

AXSEN, J., GOLDBERG, S. and BAILEY, J. 2016. How might potential future plugin electric vehicle buyers differ from current “pioneer” owners? In: Transportation Research Part D: Transport and Environment. 06/07/2016, Vol. 47, pp. 357–370.

BISA, 2018. Brussels Institute for Statistics and Analysis. In: bisa.brussels. [Online] 2018. [Accessed on 22/11/2018]. Available at the address: http://ibsa.brussels/?set_language=fr.

BRIC, 2021. UrbIS data. In: cibg.brussels. [Online] 2021. [Accessed on 26/03/2021]. Available at the address: https://cirb.brussels/fr/nos-solutions/urbis-solutions/urbis-data?set_language=fr.

BRUSSELS-CAPITAL REGION, 2011. Iris 2 mobiliteitsplan. Brussels.

BRUSSELS-CAPITAL REGION, 2013. Ontwerp van gewestelijk plan voor duurzaam ontwikkeling. Brussels.

BRUSSELS-CAPITAL REGION, 2019. Ontwerp van Gewestelijk Mobiliteitsplan: Strategisch en operationeel plan. Brussels.

BRUSSELS ENVIRONMENT, 2020. Visie op de uitrol van een oplaadinfrastructuur voor elektrische voertuigen. Brussels.

BRUSSELS ENVIRONMENT, 2021. Strategie ‘Low Emission Mobility’. In: Brussels Environment. [Online] 2021. [Accessed on 3/09/2021]. Available at: https://leefmilieu.brussels/themas/mobiliteit/strategie-low-emission-mobility

CHARGE.BRUSSELS, 2021. Network status. In : charge.brussels. [Online] 2021. [Accessed on 26/03/2021]. Available at the address: https://www.charge.brussels/netwerkstatus/.

CHURCH, R. and REVELLE, C., 1974. The maximum covering location problem. Papers of the Regional Science Association. In: Papers of the Regional Science Association. 12/1974, vol. 32, no. 1, pp. 101–118.

DENYS, T., BECKX, C., and VANHUTSEL, M., 2020. Analysis of the Belgian Car New Registrations (2008-2019). In: ecoscore.be. [Online] 2020. [Accessed on 28/12/2020]. Available at the address: https://ecoscore.be/en/fiches/new

DERAUW, S., GELAES, S. and PAUWELS, C., 2019. Enquête Monitor over de mobiliteit van de Belgen.

ERMANS, T., 2019. Brusselse huishoudens en de auto. Brussels: Perspective Brussels.

EUROPEAN COMMISSION, 2021. CO₂ emission performance standards for cars and vans. In: European Commission. [Online] 2021. [Accessed on 02/11/2021]. Available at the address: https://ec.europa.eu/clima/eu-action/transport-emissions/road-transport-reducing-co2-emissions-vehicles/co2-emission_en

EUROPEAN COMMISSION, 2021. CO2 emission performance standards for cars and vans. In: European Commission. [Online] 2019. [Accessed on 03/09/2021]. Available at the address: https://ec.europa.eu/clima/eu-action/european-green-deal/delivering-european-green-deal/co2-emission-performance-standards_en

FEBIAC, 2017. Data Digest 2017. In: FEBIAC. [Online] 2017. [Accessed on 26/02/2018]. Available at the address: http://www.febiac.be/public/statistics.aspx?FID=23&lang=NL.

GUNAWARDANE, G., 1982. Dynamic versions of set covering type public facility location problems. In: European Journal of Operational Research. 06/1982, vol. 10, no. 2, pp. 190–195.

HARDMAN, S., SHIU, E., and STEINBERGER-WILCKENS, R., 2016. Comparing high-end and low-end early adopters of battery electric vehicles. In: Transportation Research Part A: Policy and Practice. 12/04/2016, vol. 88, pp. 40–57.

HOOFTMAN, N., MESSAGIE, M., JOINT, F., SEGARD, J. B., and COOSEMANS, T., 2018. In-life range modularity for electric vehicles: The environmental impact of a range-extender trailer system. In: Applied Sciences, vol. 8, no. 7 pp. 1016.

LEBEAU, K., VAN MIERLO, J., LEBEAU, P., MAIRESSE, O., MACHARIS, C., 2012. The market potential for plug-in hybrid and battery electric vehicles in Flanders : A choice-based conjoint analysis. In: Transportation Research Part D: Transportation and Environment. 06/09/2012, vol. 17, no. 8, pp. 592–597.

LUO, W. and WANG, F., 2003. Measures of spatial accessibility to health care in a GIS environment: synthesis and a case study in the Chicago Region. In: Environment and Planning B: Planning and Design. 1/12/2003, vol. 30, no. 6, pp. 865-884.

LUO, W. and QI, Y., 2009. An enhanced two-step floating catchment area (E2SFCA) method for measuring spatial accessibility to primary care physicians. In: Health & Place. 01/12/2009, vol. 15, no. 4, pp. 1100-1107.

MARION, G., VANHAVERBEKE, L., MESSAGIE, M., MACHARIS, C. and VAN MIERLO, J., 2015. Spatial accessibility study of slow charging infrastructure for electric vehicles: case study in a belgian mid-sized city. In: XIII NECTAR International Conference. NECTAR International Conference. Ann Arbor, MI. 14-15/06/2015.

MERSKY, A. C., SPREI, F., SAMARAS, C., and QIAN, Z. S., 2016. Effectiveness of incentives on electric vehicle adoption in Norway. In: Transportation Research Part D: Transportation and Environment. 31/03/2016, vol. 46, pp. 56–68.

PLÖTZ, P., SCHNEIDER, U., GLOBISCH, J. and DÜTSCHKE, E., 2014. Who will buy electric vehicles? identifying early adopters in germany. In: Transportation Research Part A: Policy and Practice. 30/07/2014, vol. 67, pp. 96–109.

VAN MIERLO, J., MESSAGIE, M. and RANGARAJU, S., 2017. Comparative environmental assessment of alternative fueled vehicles using a life cycle assessment. In: Transportation research processes. 08/06/2017, Vol. 25, pp. 3435–3445.

WAGNER, S., GÖTZINGER, M., and NEUMANN, D., 2013. Optimal location of charging stations in smart cities: A points of interest based approach. In: Thirty Fourth International Conference on Information Systems. International Conference on Information Systems. Milan. 15-18/12/2013.

WILLOCX, B., 2015. Improvement of air quality through mobility management in Brussels. Brussels: Brussels Environment.

Haut de page

Annexe

Table 2. Number of registered EVs per municipality in 2015 and 2018

Haut de page

Notes

1 These detailed figures were personally provided by Mr Kwanten of FPS Mobility. The figures for mid-2019 are an underestimate of the 2019 numbers of EVs per municipality.

Haut de page

Table des illustrations

Titre Figure 1. Forecast number of EVs in Brussels-Capital Region 2016-2020
Crédits Sources: MOBI for the projections and visualization and [Denys et al., 2020] for the historical data)
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-1.png
Fichier image/png, 114k
Titre Figure 2. Number of electric vehicles in the Brussels Capital Region end 2018 and mid 2019 compared to the forecast made in 2015 for 2020
Crédits Source: FPS Mobility for the historical data and MOBI for the projection and visualization
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-2.png
Fichier image/png, 208k
Titre Figure 3. Number of registered EVs by private individuals per municipality in 2015
Crédits Source: FPS Mobility for the historical data and MOBI for the visualisation
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-3.png
Fichier image/png, 120k
Titre Figure 4. Prognosis EV uptake per neighbourhood
Crédits Source: MOBI
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-4.png
Fichier image/png, 223k
Titre Figure 5. Forecast EV uptake per building block
Crédits Source: MOBI
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-5.png
Fichier image/png, 349k
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-6.png
Fichier image/png, 3,5k
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-7.png
Fichier image/png, 14k
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-8.png
Fichier image/png, 18k
Titre Figure 6. The charging infrastructure's accessibility in 2015
Crédits Source: MOBI
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-9.png
Fichier image/png, 193k
Titre Figure 7. Charging infrastructure's accessibility after a POI approach
Crédits Source: MOBI
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-10.png
Fichier image/png, 179k
Titre Figure 8. Optimal locations for charging stations in 2020 based on low-voltage cabins
Crédits Source: MOBI
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-11.png
Fichier image/png, 418k
Titre Figure 9. Optimal locations for the roll-out of charging stations in 2016-2019 based on low-voltage cabins
Crédits Source: MOBI
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-12.png
Fichier image/png, 352k
Titre Figure 10. Sensitivity analysis with regard to charging stations parameter
Crédits Source: MOBI
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-13.png
Fichier image/png, 55k
Titre Figure 11. Optimal locations for charging stations in 2020 based on all cabins
Crédits Source: MOBI
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-14.png
Fichier image/png, 415k
URL http://journals.openedition.org/brussels/docannexe/image/5809/img-15.png
Fichier image/png, 106k
Haut de page

Pour citer cet article

Référence électronique

Quentin De Clerck et Lieselot Vanhaverbeke, « Possible models for the roll-out of a public charging infrastructure for electric vehicles in the Brussels-Capital Region  »Brussels Studies [En ligne], Collection générale, n° 162, mis en ligne le 07 novembre 2021, consulté le 27 septembre 2022. URL : http://journals.openedition.org/brussels/5809 ; DOI : https://doi.org/10.4000/brussels.5809

Haut de page

Auteurs

Quentin De Clerck

Quentin De Clerck graduated in 2014 as a Master of Science in Applied Sciences and Engineering: Computer Science at the Vrije Universiteit Brussel. Since 2014 he’s a researcher under the supervision of Lieselot Vanhaverbeke, for the MOBI research centre. His fields of expertise are electric vehicles, location analysis for charging infrastructure and Total Cost of Ownership.
quentin.de.clerck[at]vub.be

Lieselot Vanhaverbeke

Prof. dr. Lieselot Vanhaverbeke is Assistant Professor at the Vrije Universiteit Brussel, at the department of Business Technology and Operations (BUTO) and the research centre Mobility, Logistics and Automotive Technology (MOBI). She teaches Operations Research and Research Methods. Her research on location analysis, consumer mobility and economic aspects of electric vehicles is published in academic ISI journals. She is member of the Belgian Operational Research Society (ORBEL) and The Institute for Operations Research and the Management Sciences (INFORMS).
lieselot.vanhaverbeke[at]vub.be

Haut de page

Droits d’auteur

CC BY 4.0

Creative Commons - Attribution 4.0 International - CC BY 4.0

https://creativecommons.org/licenses/by/4.0/

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search