The photovoltaic potential for electric vehicle charging along highways: A Dutch case study

The large‐scale deployment of photovoltaics (PVs) along highways has the potential for the generation of clean electricity without competing for land use or burdening the power grid since energy for electric vehicles (EVs) can be generated locally on wastelands along highways near service stations. An analysis was carried out to evaluate the feasibility of integrating vertical bifacial solar modules into noise barriers. The approach involved integrating geospatial data with PV potential data using geographic information systems (GIS) technology. The results show a potential of around 200 GWh/year if all current noise barriers along highways in the Netherlands are considered suitable for PV module integration. Three case studies have been analysed regarding specific service stations for specific road orientations. It is shown that solar energy can charge more than 300 vehicles per day by combining bifacial PV noise barriers and standard mono‐facial PV modules on publicly available land along the highway in all three case studies, which is sufficient to meet 80% of the expected EV charging demand along highways in 2030.


| INTRODUCTION
The decarbonisation of the power sector requires the large-scale deployment of photovoltaics (PVs), which may lead to competition with other land uses. 1 On the other hand, the decarbonisation of the transport sector, and in particular road travel, is expected to be achieved mostly using battery-powered electric vehicles (EVs), increasing high power consumption points in service stations. 2PV deployment along highways addresses the increasing demand for clean energy for EV charging without competing for land use or burdening the power grid since energy for EVs is generated locally, near services stations.
The deployment of PV along and on roads has been studied using a variety of approaches, 3,4 such as utilising spaces that are not affected by traffic including the side slopes of highways or open areas at interchanges. 5,6Another approach is using PV noise barriers, which not only reduce noise pollution but also generate clean energy. 7Additionally, PV can be installed in parking lots or rest areas 8 or even in structures that are built over the road lanes. 9,10These structures can not only provide shading and protection from the elements but also generate electricity.The goal of these different approaches is to maximise the use of available space and resources to generate clean energy while also providing other benefits such as noise reduction and shading.Prior studies have shown that PV charging stations for EVs can be technically feasible and financially and environmentally beneficial, and it would therefore be worthwhile to expand this context from single cars to larger systems, which can charge multiple EVs. 11,12e Netherlands has one of the densest road networks in the world with 73 km of highways per 1000 km 2 of land. 13The Dutch state-owned highway network is made up of around 3055 km of road.This often includes the nearby surroundings to provide space for guardrails and safety barriers.While uncultivated land in the Netherlands is scarce, the unused space in the vicinity of highways could be suitable for PV energy production.Several large-scale solar projects have been executed or are expected to be completed along these highways.The first PV sound barrier was located along the A27 highway near Utrecht and was extensively monitored and analysed from 1994 until 1999. 14This type of system has been installed all over Europe; however, its nominal power of 55 kWp seems minuscule compared to most recent projects.For instance, at the A16 highway near Rotterdam, around 20 000 m 2 of solar modules are expected to be installed, by 2025, at several intersections producing a total of 3.34 GWh per year.The energy will be used to power the lights, traffic signs, tunnel ventilation and other road-related equipment along a part of the A16. 15Another project will be the installation of a solar wall 1.6 km long in the vicinity of Oostwold.While also functioning as a noise barrier, the yearly expected yield will be around 1.35 GWh. 16 2013, Jochems studied the potential of PV along Dutch national highways. 17He concluded that most existing projects were not economically feasible without subsidies, at the time.Other factors affecting the potential were regulatory barriers and high installation costs.Meppelink focused on the integration of PV modules within existing noise barriers along the highway. 18He determined the theoretical potential of energy production by integrated mono-facial PV modules in the year 2015 to be around 211 GWh.More recent studies focused on maximising the annual yield of bifacial PV noise barriers. 19is paper explores the solar potential of highways in the Netherlands.Data and spatial analysis tools are described in Section 2.
In Section 3, the total potential of PV noise barriers is estimated at the national level.Then, in Section 4, three case studies are analysed in detail to explore the potential of using solar energy generation to power EV charging in service stations along the highways.Conclusions are drawn in Section 5.

| DATA AND METHODS
The potential energy generation of PV modules along Dutch highways was analysed by combining spatial analysis, using ArcGIS Desktop 10.8.1 by Esri, with solar radiation modelling, using PVGIS.
As a reference, the Topo Basiskaart by Esri has been used as a base layer.This reference map layer includes the indication of road segments and objects in the vicinity of this infrastructure.In addition, satellite and street-level images from Googlemaps and Streetview are used to validate the exact location and size of these objects.All imported vector and raster data sets, such as the base layer reference map, have a specific geographical coordinate system (GCS).Each layer should be projected into the correct coordinate system based on the working area.The projected coordinate system for the Netherlands onshore used is the EPSG:28992 Amersfoort/RD New.
Rijkswaterstaat (RWS) is responsible for almost all highways in the Netherlands.This executive agency of the government has a wellmaintained and monthly updated version of all roads in the data set 'Nationaal Wegenbestand' (NWB) 20 It contains several shape files with simple or detailed versions of road segments.Based on the attributes, the highways in possession of RWS are determined and exported as an additional layer.Additional road information regarding surrounding objects can also be found within NWB.It contains the exact location and length of the current installed noise-reducing measures.Moreover, the road intensity is determined using the RWS data set 'INWEVA'. 21To determine the potential PV module location, the state-owned land in the vicinity of the national highways should be known.It is considered that the placement of PV modules is only suitable on land owned by the state.This information is not publicly available, except for the province of Zuid-Holland.
Moreover, information about the exact location of the charging stations, service stations and parking lots along the highway is obtained using the data register of RWS. 22

| TOTAL POTENTIAL PV NOISE BARRIERS ON A NATIONAL SCALE
The integration of PV modules within current infrastructural objects such as noise barriers, parallel to highways, would be an efficient way of using available land.Vertical bifacial PV modules can capture solar radiation on both sides making them a suitable and efficient application for noise barrier integration. 7 estimation of the total potential of bifacial PV integration in noise barriers can be obtained using GIS modelling.The current existing noise-barrier data set is imported as a line-type geometry layer into ArcGIS.In the GIS database, each highway is assigned a unique ID.In the subsequent analysis, each highway is divided into several segments with constant orientation (azimuth).These segments are then grouped into 8 intervals with an amplitude of 22.5 , ranging from 0 (North) to 180 (South).The length of each segment is calculated and summed within each group to determine the total length of highway segments in that group.Segments with opposite directions can be grouped due to similar characteristics; for example, a noise barrier from North to South has both orientations of 0 and 180 .The total length of 1075 km of noise barriers in the Netherlands for each of the orientation intervals is presented in Figure 1 (dashed line).Despite a slightly higher frequency of east-west and north-south road segments, one may observe that there is a rather uniform distribution of road orientations, unlike in other geographies such as in Italy where the north-south orientation is predominant. 23 determine the theoretical PV potential, it is assumed that all the current existing noise-reducing measures are suitable to be replaced by new ones integrating vertical bifacial PV modules.The hourly irradiance can be determined for both the front and rear sides of the bifacial PV modules for each orientation using PVGIS.Then, the PV output power P can be estimated using Equation (1), where G is the irradiance [W/m 2 ], A the total area of the PV module [m 2 ], η is the PV conversion efficiency and PR the performance ratio, used to evaluate the difference between optimal Standard Test Conditions (STC) and the actual output including environmental loss factors such as temperature, shading and soiling (dirt, snow) as well as other electrical losses.
It is assumed that the noise barriers are equipped with bifacial solar modules with an average height of 1 m.The efficiency of the south-facing side of the module is set to be 20% with a bifaciality factor of 75% (e.g. the rear side of the module, always facing the least favourable orientation, has an efficiency of 15%).Finally, the performance ratio is set to be 75%.
The height of the modules is multiplied by the length in each direction to obtain the total PV module area.The total potential of bifacial PV modules on existing noise barriers follows by adding the yield in all directions for a full year.
The total potential energy generation integrated within or on top of the existing noise barriers is estimated at 196.37 GWh/year.The total energy production for each orientation interval of noise barriers can also be seen in Figure 1 (solid line).One can observe that the use of the bifacial layout leads to low sensitivity to the road orientation, with a slightly higher performance for the north-south road orientation (+10% above the yield for the east-west orientation).
The actual solar potential could however deviate from these results depending on several factors.The estimated potential assumes a PV module height of 1 m, which can be significantly increased in some locations.Furthermore, PVGIS does not consider reflected irradiance, thus underestimating PV generation for the vertical layout considered.On the other hand, the performance ratio was assumed to Noise barrier total length (dashed line) and yearly PV generation (solid line), per orientation bin.
F I G U R E 2 Geographical overview of the three case studies.
be the typical 75%, which may be considered optimistic for long linear PV systems, with larger cabling losses.
The energy generated by solar panels on noise barriers along Dutch highways is sufficient to charge approximately 15 000 EVs per day, which is equivalent to the charging needs of 10% of the current Dutch car fleet if all vehicles were converted to electric.This is based on an average charging requirement of 36 kWh per EV and the assumption of weekly charges (c.f.Section 4.3 below).Hence, it becomes reasonable to explore the potential of using PV noise barriers to solar power service stations for EV charging along highways.This is the purpose of the next section.

| PV-POWERED SERVICE STATIONS
In comparison with the noise barrier integrated PV module potential, the PV potential to charge EVs at service stations along the highway will be determined on a more local scale.
The PV potential in the vicinity of service stations is determined based on three different case studies.Each case study focuses on an existing service station along an important commuting road.The selection of these service stations is based on several criteria.First, the main orientation of the corresponding national highway differs in each case study.Moreover, the service station is located between two large cities on a road with relatively high traffic intensity.Finally, the land on and around the service station is under the management and ownership of RWS.Three service stations in the province of Zuid-Holland have been selected.Each is located along roads with different average orientations and high traffic intensity, along the A4, A15 and A12.

| PV deployment criteria
Before the PV potential can be determined, the concept of 'vicinity' should be clarified.Le et al. investigated the sizing of large-scale linear PV systems. 24They determined that the costs of long linear power plants (5 km) already consist of 20% cable cost.Since the cable cost grows with the square of the number of strings, thus installed capacity, longer solar power plants would be too expensive.Moreover, the installation costs and cable power losses will increase with increasing distance from the service station.Therefore, a 5-km maximum radius around a service station is chosen as a potential PV module area.
The determination of PV potential around a service station is divided into two categories.First is the integration of vertical bifacial PV modules within or on top of existing noise barriers.This group also includes the potential new placement of vertical bifacial modules along the roadside.For example, the replacement of normal fences or adding PV partition walls.
The second group consists of standard mono-facial modules installed at optimal inclination for the available space.The locations for these groups are determined based on RWS-owned land adjacent to the corresponding highway.An area is specified as suitable when it does not contain water bodies, cultivation, vegetation or other infrastructural objects.To minimise the effect of local shading, only locations with a reasonable distance from local vegetation have been assigned as potential locations.The total area of bifacial and standard PV systems can then be obtained.The area, azimuth and slope of both categories are extracted from ArcGIS and used as input data to PVGIS to determine solar irradiation on the different surfaces.
The energy production of the three case studies is determined for each season of the year using PVGIS.The exact location of the service station is specified as well as the type of PV technology, installed peak PV power and system losses.The azimuth of the vertical bifacial modules depends on the orientation of the noise barriers, while the slope and azimuth of the mono-facial modules are optimised for a freestanding fixed mounting.

| Case studies
For each case, the potential placement of PV modules is mapped in Figure 2. The locations of PV potential on suitable existing noise barriers are marked as orange lines, while new potential barriers are marked as green lines.In these locations, it would be possible to instal bifacial PV modules.The assigned horizontal yellow areas comply with the requirements for the placement of mono-facial optimal angle PV panels as detailed above.
Table 1 shows additional information and values corresponding to the case studies.Despite their differences, it is interesting to notice T A B L E 1 PV potential installed capacity for three case studies.that the potential installed capacity is similar for all case studies, of the order of 4-5 MWp, with a relevant contribution from PV noise barriers.

| Energy yield
The GIS database provides the base data to determine the daily yield.
The average daily energy generation (GWh/day) is then obtained using PVGIS for January, April, July and October (Figure 3) representing the typical seasonal days.The seasonal generation profile and the specific generation (kWh/day/kWp) are rather similar for all cases despite the different shares and orientations of the vertical bifacial installed power.
At the hourly scale, the different orientations of the vertical bifacial PV noise barriers lead to significantly different behaviours (Figure 4).For case study 1, we can observe generation peaks corresponding to the bifacial systems (mostly oriented towards east and west) and the standard PV system.For case study 2, with the PV noise barriers facing south (their rear side facing north), generation peaks at midday.The highway in case study 3 has a mostly northeast-southwest orientation with a generation profile (not shown in the figure) peaking in the morning period.

| EV charging potential
To estimate the potential of solar generation to EV charging, one needs to consider the number of vehicles and their energy needs.
F I G U R E 4 Hourly PV generation profile for a summer day (top row) and a winter day (bottom row) for case study 1, in Peulwijk-A4 (left), and case 2, in Hardinxveld-A15 (right): bifacial front and rear generation (dashed and dotted blue lines, respectively), mono-facial tilted (green line) and total PV generation (orange line).
F I G U R E 3 Daily PV generation for four typical seasonal days in January, April, July and October, for the three case studies, where 1 stands for Peulwijk (A4), 2 for Hardinxveld (A15) and 3 for De Andel (A12).
Considering the forecasted growth of the EV industry in the upcoming years, this EV charging potential is analysed for the year 2030, when the general characteristics and capacity of EVs are expected to differ from the current conditions.[27] Considering the results presented in the previous section, one may determine the potential number of charging sessions throughout the year for the three case studies.The results are shown in Figure 5.
One may observe that the number of charges per day in winter ranges from 100 to 130, while in the summer period, about 400-500 EVs could be charged every day.The dotted line shows the average daily charging sessions in a year for all three cases, respectively.
Considering the occupancy rate and charging time (c.f.Table 2), the number of charging sessions per day determines the number of fast charging points FCP that could be powered by solar energy (also shown in Figure 5, secondary vertical axis).
Data about the energy demand for EVs (or other vehicles) for specific service stations are not publicly available, but Verbeek and Cuelenaere 25 have presented the required number of fast charging points to be installed in all service stations along Dutch highways by 2030, featuring an average of 14 FCP per service station.The results in Figure 5 show that, with the proposed layout for the three case studies, solar energy could provide all electricity demand for EV charging between February/March and October.During the winter months, PV could provide more than half of the electricity demand for EV fast charging.Overall, the solar fraction of EV charging is higher than 80% for all case studies.
The service stations of the case studies are located along highways in a densely populated province with high traffic intensity.Due to the lack of publicly available data, no direct comparison could be made between the potential energy production and EV energy demand for a specific service station.However, on a national scale, the results show the generation of PV energy can be sufficient for EV charging in most cases, especially when considering that the available area will be higher and traffic intensity lower in most of the other Dutch provinces.
It should also be pointed out that considering daily, instead of hourly, charging sessions ignores the intraday mismatch between charging events and solar power generation.This mismatch is probably more relevant on weekdays, when charging is expected to take place at the rush hour early in the morning or later in the afternoon, than on weekends when the traffic intensity profile during the day is more like the PV generation. 28The supply-demand mismatch can be addressed by energy exchange with the grid, importing electricity at peak charging times and exporting to the grid excess PV power at midday, which would lead to the need for reinforcement of the grid, higher costs and CO 2 emissions (before the power system becomes fully decarbonised).Supply and demand mismatch can also be addressed by local electricity storage. 29Local storage could be provided by slowly charging EV batteries, in a battery-swapping model 30 or dedicated stationary second-life EV batteries.Ratio charged along the highway 10% It is worth mentioning that, perhaps on a horizon beyond 2030 considered for this analysis, EV charging technology may evolve towards catenary 32 or wireless charging 33,34 along the highway, instead of fast charging at service stations.These approaches would open the opportunity for PV generation along the full length of the road, not just in the vicinity of service stations.
The study was carried out in the Netherlands, which has a high density of heavily trafficked highways and less-than-ideal solar conditions.It thus can be expected that in other regions of the world, in particular those that are located closer to the equator, the solar potential near service stations would be sufficient to charge EVs travelling along the highway.

| CONCLUSIONS
Using a geographical information system and a proper analytical methodology of geospatial data for the Netherlands, it is shown that upgrading all existing noise barriers with 1-m-high PV modules would lead to the generation of about 200 GWh/year of clean energy.This generation would occur mostly in urban or semi-urban areas with high electricity demand and would thus be expected to be consumed locally hence reducing greenhouse gas emissions without straining the power grid.Furthermore, this would be achieved without any further land use or relevant environmental impacts since the infrastructure is already in place.
It is particularly interesting to explore the use of this locally generated electricity to charge EVs at service stations along the highway.
For this purpose, we have explored in detail three case studies, focusing on service stations along high-traffic highways (which are more challenging because demand is higher) with different orientations, to explore different generation profiles.
For each case study, potential PV capacity combines vertical PV bifacial modules (on existing and new noise barriers, when suitable) and standard mono-facial tilted modules on publicly available land along the highway in a 5-km radius around the service station.The total installed capacity in the different case studies is of the order of 5 MW, about half of which as vertical modules.For these setups, it was shown that PV could charge an average of 300 vehicles per day thus addressing more than 80% of the annual EV charging needs expected for 2030 on Dutch highways.
These results are expected to be favourably extrapolated for other sunnier locations and/or with lower traffic intensity thus highlighting the potential for large-scale deployment of PV along highways.