High-Resolution Monitored Data Analysis of EV Public Charging Stations for Modelled Grid Impact Validation

Abstract

As electric vehicle adoption grows, understanding the impact of electric vehicle charging on electricity grids becomes increasingly important. Accurate grid impact modelling requires high-quality charging infrastructure data. This study examined the electric vehicle recharging infrastructure and usage patterns in a region of the Italian Alps over a three-year period from 2021 to 2023. The primary objectives were to analyze the growth and distribution of electric vehicle charging stations, assess energy consumption, and evaluate charging behaviours across various recharging points. The research involved collecting empirical data from 411,800 recharging sessions and simulated data using the emobpy tool to model energy consumption and charging behavior. Key findings reveal a substantial increase in the number of recharging points, from 673 in 2021 to 970 in 2023, with the total energy delivered increasing from 938 MWh in 2021 to 4133 MWh in 2023. The data showed distinct temporal trends: AC points were primarily used during the day, while DC points saw higher usage during morning and late afternoon peaks, aligning with travelling times. The study’s validation of simulation results against empirical data emphasized the importance of high-quality input for accurate grid impact assessments. These findings suggest the necessity for strategic placement of recharging infrastructure and provide practical insights for policymakers, urban planners, and utility companies to support sustainable electric vehicle integration.

1. Introduction

The rapid adoption of electric vehicles (EVs) is a key strategy for reducing greenhouse gas emissions and dependence on fossil fuels in the transportation sector. EVs have the potential to significantly decrease the environmental impact of transportation, which currently accounts for a substantial portion of global carbon emissions [1]. However, the growing number of EVs presents challenges for electricity grids, as charging infrastructure expands and demand for vehicle charging increases [2]. The additional load from EV charging can strain electrical distribution networks, potentially leading to issues such as voltage deviations, transformer overloading, and power quality problems [3].

To ensure reliable operation and plan necessary grid reinforcements, it is critical to accurately predict the impact of EV charging on electricity networks [4]. Grid impact modelling plays a crucial role in understanding the effects of EV charging on the power system, allowing utilities and policymakers to make informed decisions about infrastructure investments, charging strategies, and regulatory frameworks [5]. These models simulate the behavior of EV drivers, the distribution of charging points, and the resulting load on the grid, helping to identify potential bottlenecks and optimize charging patterns to minimize negative impacts [6].

However, the accuracy of grid impact models depends heavily on the quality and granularity of the input data [7]. Many existing models rely on assumptions or aggregated data about EV charging behavior, which may not capture the real-world variability and complexity of charging patterns [8]. This can lead to inaccurate predictions and suboptimal planning decisions [9]. To improve the reliability of grid impact models, it is essential to validate them against real-world data from actual EV charging infrastructure [10].

Analyzing monitored data from EV charging stations can provide valuable insights into actual charging behavior, energy consumption, and infrastructure utilization [11]. This information can be used to refine model assumptions, calibrate simulation parameters, and verify the accuracy of model predictions [12]. By comparing model results with empirical data, researchers can identify discrepancies and improve the representativeness of their models [13]. This iterative process of model validation and refinement is crucial for developing robust and reliable tools for assessing the impact of EV charging on electricity grids [14].

The charging behavior of EVs has already been the subject of several studies and has been investigated using various data sources and methodological approaches.

The analysis of the electrical energy required from the grid by the charging points has been studied in the literature through two main approaches: the first through the analysis of real data from the charging points (through private or public data), and the second through a simulative approach, in which different tools are used to calculate the energy demand based on specific input data.

Regarding the first approach, the usage behavior of electric vehicle drivers at public charging stations (CSs) was investigated for the first time based on a comprehensive data set by Hoed et al. [15], who investigated the Amsterdam metropolitan region from April 2012 to April 2013, analyzing the number of charging events and the spatial distribution of the energy used, thus being at an early stage of electric mobility and only at the urban level. These results were also integrated by Wolbertus et al. as a comparison within a benchmark analysis [16] and for an analysis considering the introduction of specific parking spaces for recharging vehicles, available only at certain times of the day [17]. These studies, contextualized at a time of early development of electric mobility, focused on the relationship between spatial and temporal variables and the charging behavior of vehicles but only considered a limited number of decision-making variables, and only at the urban level.

Other studies have therefore placed the focus on quantifying the factors that have an influence in determining user connection periods at public charging stations. Almaghrebi et al. [18] analyzed more than 17,000 charging events to evaluate the impact of the charging location (i.e., school, workplace, etc.). In a similar study, they considered instead the variation of the charging starting time and duration, based on the period of the day [19]. Although these studies provide a detailed analysis and a high degree of disaggregation of results, they refer to a small number of charging points and have a spatial frame at the urban level, not including spatial variability from a broader point of view.

Following instead the technological development of vehicle charging technology, other studies have analyzed the differences between fast and slow charging points. Xiao et al. [20] proposed a techno-economic analysis of the impact of the charging strategy on the Total Cost of Ownership (TCO) of a fleet of electric buses in Shanghai. The results included the hourly electricity demand for charging, which is also analyzed in this paper, but for vehicles with strict daily schedules, which did not allow consideration of the charging behavior due to the geographical position of the charging stations. Friese et al. [21] instead shifted the focus to private cars, monitoring the data of public charging points, but only at the urban level, only partially providing an analysis of geographical and spatial influences on the use of charging points. Morrissey et al. [22], expanded the case study, monitoring more than 700 charging points across Ireland to study the different use of slow and fast charging points. Although the data provided represents a strong step forward in defining the charging habits of electric vehicles, the analysis falls short in providing data on daily charging profiles, which are crucial in defining the increase in power demand due to the electrification of the transport sector. A different approach is proposed by Wang et al. [23], in which real data of daily taxi movements in the Beijing area were considered to identify the optimal locations in which to build charging points. Although this study presents an innovative approach in the use of real data, it refers only to taxis and does not consider private cars, whose users may follow different rules for the choice of charging points. Furthermore, although economic considerations are present, analyses of potential charging profiles and energy considerations are missing.

Due to the difficult availability of open data, another possible approach is to use simulations to estimate future charging infrastructure requirements and potential usage. Pasha et al. [24] provide a comprehensive review of the state of EV scheduling, identifying key challenges such as energy management, battery degradation, and grid stability. They emphasize the need for adaptive algorithms that account for real-time fluctuations in energy demand and the importance of considering the impact of EVs on grid infrastructure. The paper also highlights future research opportunities, including the development of predictive models and decentralized scheduling algorithms to enhance EV network resilience. Among the studies that assessed electric vehicle scheduling, including power grid considerations, Kamankesh et al. [25] proposed a system that included both charging demand and renewable energy sources to evaluate the differences between controlled and uncontrolled charging, but this was a micro-grid case study, thus lacking the consideration of the influence of different geographical position on the production and demand profiles. Zhang et al. [26] proposed a comparable charging station model but employed a Markov Chain process to represent EV arrivals. However, their approach does not account for user behavior variables, such as charging point selection or charging duration, which can significantly impact the charging process.

Another approach followed in the scientific literature is to include survey or similar data to define the charging infrastructure usage. Using survey data and basic assumptions regarding parking behavior and fleet composition, Anderson et al. [27] employed an agent-based simulation to estimate the number of charging points (CPs) required in Germany. The results indicate significant impacts from on-street charging, home charging, and the proportion of plug-in hybrid electric vehicles (PHEVs) in the fleet on the number of necessary CPs. However, no information is provided on the impact of charging points on the network, considering the increase in electricity demand overall or the hourly distribution of charges. A more comprehensive analysis is provided by a similar study [28], which also used an agent-based simulation, incorporating charging behavior and location choice for charging as explicit inputs based on realistic assumptions. The needed number of charging points and their daily use was calculated for an urban case study.

Other studies used instead a different methodology to evaluate the charging infrastructure usage. Prasetyo et al. [29] analyzed the operation of a charging point in combination with a photovoltaic system in Indonesia using the HOMER framework. However, the case study relates to a single recharging point in an urban context, as the geographical variability in both photovoltaic production and electric vehicle recharging demand cannot be considered. In a similar study [30], the case study is expanded to include these aspects. In this case, however, a limitation is to have considered an idealized charging station where there is a perfect distribution of vehicle charging, and all the charging points are always used at any time of the day. In another study, Sohnen et al. [31] considered the operation of various charging points, within an analysis of California’s energy system. In this case, the vehicle charging profiles are not based on real data but on dumb or perfectly smart charging assumptions, where the vehicle charging is arbitrarily set afterwards to fill the valleys created by the already existing load. Although this approach may be considered sufficient for the purposes of the study, it does not allow consideration of the different energy demands of charging, depending on the geographical location of the points considered, such as the difference between tourist and non-tourist areas.

The present study aims to provide a comprehensive analysis of electric vehicle charging patterns and infrastructure utilization in the Trentino–South Tyrol region of Italy, thus including urban and non-urban areas, by examining high-resolution temporal and spatial data from public charging stations over a three-year period (2021–2023).

The innovative aspect of this article lies in the extensive scope of the regional case study, which, unlike other studies in the literature, encompasses a diverse range of spatial and geographical areas and includes various types of charging points, both AC and DC, with different nominal capacities. This breadth allows for the examination of multiple factors influencing the utilization of charging points. The study provides a detailed and varied representation of the charging behavior, including not only common data such as the duration and start time of charging but also crucial data to assess the impact on the network, such as the hourly power demand from each type of charging point.

Furthermore, the measured data are compared with data obtained through simulations using the emobpy tool (version 0.6.2, https://emobpy.readthedocs.io/en/latest/, accessed on 28 August 2024). Unlike other approaches in the literature, emobpy includes parameters such as daily vehicle movements, which influence the likelihood of finding an available charging point and the usable nominal capacity. This allows for a more precise definition of the power profiles required from the network for vehicle charging.

Hence, the objectives of this study are as follows:

• Characterize the growth and distribution of EV charging infrastructure in the region.
• Analyze energy consumption patterns and charging behaviours across different types of charging stations.
• Identify temporal and spatial trends in charging station usage.
• Compare the monitored data with results from grid impact simulation models to validate and refine predictive approaches.

The remainder of this article is structured as follows: Section 2 describes the study area, provides an overview of the recharging infrastructure, and details the data collection and analysis methods. Section 3 presents the results, including the distribution and expansion of charging points, energy consumption and charging session statistics, and detailed profiles of charging behaviours. Section 4 discusses the implications of these findings for grid impact modelling and infrastructure planning. Section 5 concludes the study by summarizing key insights and suggesting directions for future research.

2. Materials and Methods

This section provides a comprehensive overview of the study’s design and methodology as follows: Section 2.1 describes the study area, focusing on the geographical and demographic characteristics of the Trentino-South Tyrol region, which influences local transportation preferences. Then, Section 2.2 provides detailed information on the recharging infrastructure, including the number and types of recharging points, and the expansion of the network over time Next, Section 2.3 outlines the data collection process, describing both the empirical data gathered from real-world observations and the simulated data generated using the emobpy tool. The combination of these data sources allows for a robust analysis of energy consumption patterns and charging behaviors across different types of charging stations.

2.1. Study Area

Trentino–South Tyrol is an autonomous region located in the northern part of Italy, and it is composed of two provinces: the province of South Tyrol in the north, and the province of Trentino in the south. This region features a mountainous landscape occupied by around 1.07 million inhabitants [32,33]. In terms of local transport, the private vehicle is by far the preferred means to reach the workplace (62% in South Tyrol and 80% in Trentino) or place of studies (13% in South Tyrol and 29% in Trentino) [34]. This area is also characterized by an elevated touristic presence of about 15.2 millions of arrivals between 2021 and 2022 [35,36]. The expansion of the network for public transportation has always been a challenge in this region due to its geographical position. Therefore, to reach remote destinations within the area, visitors prefer private vehicles; in the province of Bolzano alone, around 80% of tourists arrive by car [37,38].

Nevertheless, the whole area presents several favorable conditions for the diffusion of electric vehicles. In 2021, 83.3% of the electricity share produced in the region was produced from hydropower [39]. Moreover, the regional government is committed to reduce CO₂ emissions from the transport sector, which responsible for most CO₂ emissions in the region. For instance, in 2011 the Autonomous province of South Tyrol established an interdisciplinary strategy called “Climate Plan South Tyrol”. The plan has just been revised, with the aim of achieving climate neutrality by 2040 and CO₂ emission reductions of 55% (from 2019 level) by 2030 [40]. The strategy includes actions such as the implementation of an optimal recharging infrastructure, the application of economic incentives for the transition to zero-emission vehicles (ZEVs), allowing new registrations and limiting the circulation inside urban areas only to ZEVs, and the implementation of solutions to promote sustainable mobility, among others.

2.2. Recharging Infrastructure

In Italy, there were 36,772 public recharging points available in 2022 [41]. Compared to the previous year, the number of recharging points increased by 41%, which represents a significant milestone in terms of expansion of the recharging infrastructure on a national scale. Of these recharging points, 3% were present in Trentino–South Tyrol (1361 recharging points available).

On the other hand, even if the growth of such infrastructure is not running alongside the number of registered ZEVs (EV and Fuel-Cell Vehicles), it is possible to witness that the number of circulating ZEVs is also increasing every year, especially in the region of Trentino–South Tyrol, where there is a high concentration of new vehicle registrations for leasing purposes (Figure 1) [41,42].

Neogy (Bolzano, Italy) is one of the main Charging Point Operators (CPOs) in Italy; it emerged in 2019 as a joint venture between the renewable energy utilities Alperia and Gruppo Dolomiti Energia, aiming at the promotion of electric mobility [43]. Neogy manages the largest recharging infrastructure network in Trentino–South Tyrol; such infrastructure includes not only a large percentage of recharging points with maximum power outputs higher than 22 kW and the first Hypercharger points installed in Italy, but also the offering of different services to promote sustainable mobility both in the private and in the touristic sector [43].

2.3. Data Collection

2.3.1. Operational Data Collection

The data was collected through the monitoring process of the recharging events from the points of the CPO present in Trentino–South Tyrol. Such monitoring data was collected from an automatic report that the system of the CPO generates and automatically sends every month. These files include the following information for each recharging event: (i) the ID of the recharging point, (ii) the start and end time of the event, and (iii) the consumption of the Battery Electric Vehicle (BEV) [kWh]. Data was further completed by scraping the interactive map of the CPO to retrieve the list of available charging points. By matching the ID of every charging point from the file with the downloaded data for recharging stations, it was possible to obtain: (iv) the coordinates of the recharging station, (v) the brand and model of the recharging station, (vi) the maximum power output of the recharging points, and (vii) the type of connector(s) available for the recharging point. Based on the data reported previously, this study presents several performance features of the recharging infrastructure from the CPO present in the territory, indicated in

Table 1. Overview of the performance features for the analyzed recharging points, based on data collected from January 2021 to December 2023.

Category	Performance Feature	Assumptions
Distribution and expansion of recharging points	(a) Recharging stations by brand and model
	(b) Expansion of recharging points
	(c) Location and number of recharging points
Energy consumption and recharging sessions	(d) Monthly consumption and number of sessions
	(e) Monthly consumption (yearly comparison)
	(f) Consumption by municipality
	(g) Avg. consumption per recharging point by municipality
Profiles for energy consumption and session duration	(h) Consumption and duration distributions	Considered only values up to 250 kWh and up to 1800 min; data for DC ultra-fast (level 2) (p ≥ 350 kW) recharging points not included due to incomplete dataset.
	(i) Duration vs. energy consumption
	(j) Normalized profile of occupancy
	(k) Avg. consumption and duration by hour of session start

, including a list of assumptions considered for the analyses. The analyses contain data collected from January 2021 to December 2023.

2.3.2. Simulation Data

The simulation data regarding the hourly power demand required from the grid are instead obtained using the emobpy open-source tool [44]. Such tool employs a sampling approach to generate profiles BEVs by integrating customizable assumptions, vehicle physical properties, and empirical mobility data from German institutions. To create passenger car charging profiles, the tool requires the percentage distribution of the population among non-commuters, part-time commuters, and full-time commuters. After establishing Origin–Destination vehicle patterns, the tool needs the characteristics of the vehicle fleet to determine battery consumption for each trip.

To evaluate the energy consumption for each time step t when the vehicle is in the “driving” state, the tool calculates a power balance across the battery as outlined in Equation (1). The positive power outputs considered are the auxiliary power ($P_t^{aux}$), the electrical power for heating/cooling device ($P_t^{device}$) and the power provided to the motor ($P_t^{m,in}$), while the negative power input is provided by the generator in case of regenerative braking ($P_t^{G,out}$):

$$P_t^{all}=P_t^{m,in}-P_t^{G,out}+P_t^{aux}+P_t^{device}$$

(1)

As shown in Equation (2), if the sum of all the powers $P_t^{all}$ is positive, the battery supplies energy to the vehicle and the discharging efficiency ${\eta }_{discharge,BEV}$ is used. If $P_t^{all}$ is negative, then the battery is charged via regenerative braking. In that case, the battery load $P_t^{battery}$ is negative, hence the charging efficiency ${\eta }_{charge,BEV}$ is utilized.

$$P_t^{battery}=\left\{\begin{array}{c}{\displaystyle \frac{P_t^{all}}{{\eta }_{discharge,BEV}}},if P_t^{all}>0\\ P_t^{all}\times {\eta }_{charge,BEV}, otherwise\end{array}\right.$$

(2)

The total energy consumption per trip $E^{BEV,total}$ is defined in Equation (3), where battery load is aggregated through the duration of the trip at every hourly time step $t$.

$$E^{BEV,total}=\sum _{t=1h}^T{P}_t^{battery}$$

(3)

Once vehicle consumption is assessed, it is crucial to provide probability distributions for charging stations, outlining the likelihood of finding an available charging station with a specific power rating at each location. Using this information, emobpy determines whether a BEV is connected to the grid at any given timestep and at what power capacity. Finally, to assess the overall grid electricity demand, an “immediate” charging strategy is considered (using the nomenclature of the emobpy developers [44]), where BEVs charge at the maximum power rating as soon as a charging station becomes available.

Specifically, for the region of South Tyrol, data on the most sold cars and population distribution based on real data were utilized to calculate the total electricity demand on the grid. This calculation aggregates data for the entire study area, allowing for distinctions based on points of interest where charging points are located, such as workplaces, shopping areas, and leisure sites. However, the exact geographical locations of the charging points could not be specified.

This methodology, previously employed by another study in the literature [45], enables the calculation of the energy required from the electricity grid for charging passenger vehicles.

The results obtained were then filtered, considering only the recharging in public stations (excluding, for example, charging at home) and maintaining the distinction between fast charging and charging points at reduced power, thus obtaining sufficiently valid data to be compared with the actual aggregated data obtained by measurement.

3. Results

This section presents the findings of the study, beginning with Section 3.1, which examines the growth in EV charging infrastructure, identifying key trends over the analyzed period. Section 3.2 then analyzes electricity consumption patterns, focusing on differences between AC and DC charging points and geographical variations. Section 3.3 delves into the characteristics of individual charging sessions, including duration, energy consumed, and the temporal distribution of charging events. Section 3.4 compares the measured charging data with simulated profiles generated using the emobpy tool, evaluating the accuracy of the simulation tool. The results build upon each other, progressing from infrastructure growth to consumption patterns, and culminating in the validation of the simulation tool against real-world data.

3.1. Distribution and Expansion of Recharging Points

Figure 2 and Figure 3 show the location, the brand and model, and the maximum power output of the public recharging stations installed by Neogy from 2021 to 2023 along the region of Trentino–South Tyrol. Currently, there are more than 488 recharging stations from at least six different brands and 20 different models. Such recharging stations provide 970 recharging points in Alternating Current (AC) and/or Direct Current (DC) configurations, with maximum power output values that range from 3.7 kW to 400 kW. All the recharging points present in this analysis are classified according to the Regulation for the deployment of Alternative Fuels Infrastructure (AFIR) [46].

Figure 4 shows the total number of public recharging points up to December 2023 (left side) and the expansion of such points from 2021 to 2023 (right side), while Figure 5 shows the current location of the recharging points, divided by category of maximum power output. During this period, the availability of recharging points increased by around 44% (from 673 recharging points in 2021 to 970 in 2023), with yearly expansion rates of 24% from 2021 to 2022 and 16% from 2022 to 2023, respectively. Most of the available recharging points are from the AC category (760 recharging points), and the predominant part are those classified as AC medium, triple phase (7.4 kW ≤ p ≤ 22 kW). The province of Trento holds 52% of the AC recharging points in the region. As mentioned before, in 2022, there were 1361 public recharging points available in the study area. From these, 835 points (around 61%) belong to the CPO from this study.

The availability of recharging points from the DC category also increased during this period. Up to December 2023, there were 210 such recharging points, 84% of them classified as DC fast (50 kW ≤ p < 150 kW). In this period, the availability of DC recharging points increased by around 89% (from 105 recharging points in 2021 to 199 in 2023). The province of Bolzano holds more than 80% of the DC recharging points in the region. Moreover, the first recharging station with five recharging points classified as DC ultra-fast (level 2) (p ≥ 350 kW) was installed in November 2023 in this province, providing maximum power outputs of 400 kW [47].

In terms of types of connectors available, the AC recharging points are equipped with the IEC 62196 Type 2 standard, while the DC recharging points are based on either Combined Charging System (CCS) or CHAdeMO connector types. The former connector is present in all the categories, while the latter is only present in recharging points classified as DC fast (50 kW ≤ p < 150 kW). Only a small percentage of recharging points have the possibility of using both CCS and CHAdeMO, which are only present in DC fast recharging points.

3.2. Energy Consumption and Recharging Sessions

Figure 6 shows the number of recharging sessions and the energy consumption from the analyzed recharging points from 2021 to 2023. The figure also shows the sessions and the energy consumption on a yearly and monthly basis, respectively. Overall, there were more than 411,800 recharging sessions with an energy consumption of 7936 MWh destined to recharge BEVs. As the number of recharging points expanded in the region, the recharging sessions also increased. Indeed, the number of recharging sessions increased by around 127% from 2021 to 2022, and by around 151% from 2022 to 2023. Considering that the hydropower plants in South Tyrol produced on average around 6000 GWh in the years 2010–2021 [48], the energy consumption from the recharging points corresponds to 0.04% of the hydro production.

As expected, most of the recharging sessions come from AC recharging points, since most of the available recharging points belong to this category. Nonetheless, the recharging sessions from DC recharging points also increased over the analyzed period as the availability of such points is also higher, getting closer to the number of sessions from AC recharging points.

In contrast to the number of sessions, the energy consumption from AC and DC recharging points is almost the same. The energy consumption in 2022 was 157% higher than in 2021, while the energy consumption in 2023 was around 170% higher than the previous year. However, in 2023, the DC recharging points surpassed the energy consumption from AC ones. By comparing the expansion of installed AC and DC recharging points in terms of energy consumption, installing 210 DC recharging points leads to a similar impact on the electric grid as the installation of 760 AC recharging points.

Looking at the consumption over different seasons (view Figure 7), there was a higher energy demand during summer. Moreover, the consumption decreased during the cold seasons, excepting December, in which the consumption was closer or higher than in August. The lowest consumption values were reached in February and March, while the highest were reached in August and December. The reason for this seasonality is probably the tourism presence in Trentino–South Tyrol, which is highest in August and around Christmas.

Figure 8 shows the overall energy consumption from the recharging points installed in each municipality of the analyzed region (left), and the energy consumption from the municipalities with the highest consumption, plus the number of available recharging points (right). The province of South Tyrol achieved 72% of the consumption, which is linked to the higher availability of DC recharging points in this territory.

Nonetheless, most of the energy consumption comes from large municipalities in both provinces, as they usually have more recharging points available for residents and tourists than the small ones. However, in some municipalities, the energy consumption is higher, despite the low availability of recharging points. This is the case in municipalities such as Vipiteno and Riva del Garda, which are touristic locations in South Tyrol and Trentino, respectively. Hence, by calculating the average energy consumption per recharging point in each municipality (view Figure 8—Bottom), it results in higher consumption from municipalities with touristic locations, but also from municipalities with roads that are between such locations.

3.3. Profiles for Energy Consumption and Session Duration

Figure 9 and Figure 10 show the distribution of energy consumption (in kWh) and duration values (in minutes) for each recharging session from AC and DC recharging points. The figures present the Kernel Density Estimation (KDE) of all the values (top section), plus the boxplot of consumption values (bottom section), containing the summary of the distribution for each classification of maximum power output (minimum and maximum, quartiles, median, average and outliers). In addition, Figure 11 shows the comparison between duration and energy consumption for each registered recharging session from AC and DC recharging points, including their respective median and average values; such average and median values are also reported in

Table 2. Average and median values for the consumption [kWh] and duration [min] for the analyzed recharging points, classified by type of maximum power output.

Maximum Power Output	Consumption [kWh]		Duration [min]
	Average	Median	Average	Median
AC slow, single-phase (p < 7.4 kW)	9.8	6.6	163	106
AC medium, triple-phase (7.4 kW ≤ p ≤ 22 kW)	15.6	11.1	165	116
AC fast, triple-phase (22 kW < p ≤ 45 kW)	13.2	9.4	124	88
DC fast (50 kW ≤ p < 150 kW)	26.9	23.1	36	31
DC ultra-fast (level 1) (150 kW ≤ p < 350 kW)	33.3	31.5	38	34

. For data cleaning purposes, we considered only values up to 250 kWh and up to 1800 min, and X axes were truncated for a clearer visualization. Also, data for DC ultra-fast (level 2) (p ≥ 350 kW) recharging points was not included, since the dataset is not as complete as the other categories.

As expected from the maximum power output, the energy consumption from AC recharging points is lower than the consumption from DC ones, while the duration is higher. Results show that the consumption values from AC recharging points are more concentrated under 15 kWh and under 300 min, while the consumption values from DC recharging points are more distributed. The skewness of the distributions is also confirmed by the larger difference between average and median values from the AC recharging points, if compared to the same values from DC recharging points. In many cases, AC recharging points are used also for parking purposes, resulting in long sessions with low or high energy consumption. However, more information would be required to better understand the distributions since both consumption and duration values are affected by many factors, such as the initial state of charge from the EV, the reached power output due to simultaneous recharges, the difference of battery sizes that influences both the duration and consumption, and even the different preferences of users, in terms of charging up to 80% for battery longevity versus a full recharge of the vehicle for longer energy availability, among others.

Results from AC recharging points show that both average and median values from AC fast, triple phase (22 kW < p ≤ 45 kW) recharging points are lower than those classified as AC medium, triple phase (7.4 kW ≤ p ≤ 22 kW). This could happen since the AC fast recharging points are present in the same recharging stations in which also DC recharging points are present, so the maximum power output rate cannot be reached because the other recharging points are occupied. Nonetheless, the duration values from AC fast recharging points are also lower, which is an effect of providing maximum power output values.

Moreover, data about energy consumption and duration from DC recharging points show that users from DC ultra-fast (level 1) (150 kW ≤ p < 350 kW) recharging points usually consume more energy than users from DC fast, triple phase (75 kW ≤ p < 150 kW) recharging points, but they also stay longer. However, by considering median values, it takes 3 more minutes to recharge an additional 7.4 kWh with a higher maximum output power.

Figure 12 shows the profile of occupancy from the analyzed recharging points. Such profiles were calculated by using the weighted arithmetic average of the occupancy per hour and per day of the week, and then scaling the results with a min-max normalization. Moreover, Figure 13 and Figure 14 show the average energy consumption (in kWh) and the session duration (in minutes) for AC and DC recharging points, allocated by the hour in which the EV started recharging. The lines represent the average values, while the bars indicate the number of registered recharging sessions.

In general, the occupancy is usually lower at the beginning of the week, increasing more when the weekend is closer; the highest demand is on Saturdays. The only exception is for AC slow, single phase (p < 7.4 kW) recharging points, which are busier during working days.

AC recharging points are requested in most of the cases during the first part of the day. In addition, the average energy consumption during working hours is similar, while in early mornings and during the night, it tends to increase; still, the average duration during late hours is higher. Such results could be related to users leaving their EVs connected to the recharging points during these time slots, using the space for parking purposes.

In the case of AC fast, triple phase (22 kW < p ≤ 45 kW) and DC recharging points, results show that they are more requested in the morning and in the late afternoon; this behavior corresponds to users recharging their EVs before and after working hours and before and after longer trips. Then, the average energy consumption values from DC recharging points are similar among different working hours, but the average duration of sessions decreases early during the morning and in the night hours. This could be associated to a higher power output during such time slots, resulting in shorter recharging sessions. As mentioned before, there are many factors that affect the consumption and duration values, so more data would be required to better understand such behaviours.

3.4. Comparison with Data from Simulations

This section compares the values obtained by measurement with the values obtained by simulation using the emobpy tool, the aim being to analyze the similarities and differences obtained about different reference values, to then effectively study the behavior of the simulation tool. Figure 15 shows the total energy consumption required by users in the analyzed infrastructure. The total consumption is then allocated in the hour at the start of the session and finally normalized for each of the maximum output powers available. Figure 16 displays the simulation results obtained using the emobpy tool. The upper section, which illustrates the profile of public charging points with a maximum output power of 22 kW, shows an average daily pattern consistent with actual measured values (Figure 15—Top), thereby validating the methodology used. Notably, there is a distinct peak in charging activity in the late afternoon, when many users leave workplaces and head to other destinations, recharging when charging stations are available.

Conversely, when examining the electricity demand from high-power charging points on the grid, a discrepancy is observed compared to the measured data (Figure 15—Bottom). The profile does not exhibit regular and consistent values throughout the day, with lower values during the night; instead, it shows a pronounced peak in the mid-afternoon. This behavior can be attributed to the assumptions used within the tool, which considers that all possible destinations have both slow and fast charging points, with the latter being less frequently used due to their limited availability. Since no destinations exclusively feature fast charging points, the profile more closely resembles that of slow charging points, maintaining the afternoon peak.

In the next comparison, two critical parameters are analyzed to evaluate the use of charging points: the average duration of charging sessions and the average energy drawn from the grid. The analysis involves a detailed comparison between real-world data obtained through monitoring and the data generated by simulations. This comparison specifically focuses on the different behaviours observed when using slow charging points and fast charging points.

The top part of Figure 17 illustrates the comparison between the simulated data for slow charging points, which have a nominal capacity of less than 45 kW. Compared to the measured data in the top of Figure 11, the simulation data indicate a longer average duration of charging sessions, accompanied by a simultaneous reduction in the average electricity drawn from the grid for each recharge. A possible explanation for this discrepancy could be related to the structure of the simulation code. The code includes a high frequency of home charging and does not impose a maximum state of charge (SOC) limit before vehicles are recharged. Consequently, vehicles often arrive at public charging points with a relatively high battery level, utilizing the available time to recharge but drawing only a limited amount of electricity from the grid.

The same comparison between measured data and simulated data through emobpy can be done by considering instead the fast DC charging points. When comparing the measured data (Figure 11—Bottom) from the fast DC charging points with the simulation results (Figure 17—Bottom), a similar pattern emerges as observed with the AC charging points. There is an increase in the average charging time coupled with a reduction in the energy required for charging, for the reasons previously mentioned. However, an aspect that validates the correct functioning of the simulation tool is that, in this case as well, the comparison shows that the fast DC charging sessions are shorter but demand more energy than the AC charging sessions.

4. Discussion

In this work, actual data obtained by measuring the utilization of the charging infrastructure in the region of Trentino–South Tyrol were compared with data obtained by means of the emobpy simulation tool, with which it was possible to derive hourly charging profiles and the average duration and demand of energy to the grid.

The extensive detail and variety of the results provided by the measurements allows an initial comparison with the results so far found in the literature. Compared to the study of Van der Hoed et al. [15], the wider time span considered allows analysis of the strong increase in the number of recharging points installed in the area, with the number of charging points increasing by 44%, and an even-higher rise in charging sessions, increasing by over 150% annually. These values, higher than those present in a similar study by Almaghrebi et al. [18], indicate a clear sign of the strong diffusion of electric mobility (and its necessary infrastructure) within this Italian region. Compared to the study of Xiao et al. [20], this study also provided a differentiation considering the diversity in the charging infrastructure, showing that while AC charging points were more numerous, DC chargers delivered nearly equal the total energy. This aspect highlights the importance of a diverse charging infrastructure to meet various user needs.

The advantage of having also considered a case study, as broad as the regional one, made it possible to show the influence of geographical location on the amount of charging points present and their utilisation. In their study, Sohnen et al. [31] calculated the consumption of a single charging station and then redistributed the load, depending mainly on parameters such as the population present. This paper shows instead the influence of other parameters that were not considered until now, such as the presence of areas with a lower population, but which represent important tourist areas, able to attract many people and, therefore, potentially also electric cars. Those tourist areas show a high demand that requires a tailored infrastructure deployment.

The influence of tourism and other annual phenomena is also crucial in the analysis of the variation of energy required for charging during the year. The presence of an increase in the energy required for charging in August and December is closely related to the characteristics of the territory considered. Compared to the study of Prasetyo et al. [30], this demonstrates the need to consider a long period for the collection of monitoring data, so that this variance can be understood. In addition, any data obtained by simulation must be correctly redistributed over the year if they start from a shorter simulation period.

From the point of view of charging technology, the presented study focuses on different charging behaviours depending on whether charging is fast or slow, hence on the nominal capacity available, less evaluated by a similar study of Prasetyo et al. [29]. AC charging sessions tended to have lower energy consumption but longer durations, compared to DC sessions. This difference is likely due to the technical capabilities of the chargers and their typical use cases (e.g., AC chargers being used for longer-term parking).

Another aspect presented in this study is the differentiation between the type of plug used for charging, distinguishing between CHAdeMO and CCS, showing how most of the recharges occur with the second type, especially regarding the recharges in DC. This aspect is an important analytical tool for considering the type of vehicles that use these charging systems.

An important novelty of the article is the comparison between measured and simulated data, allowing a more complete analysis of the modelling pattern of the tool. Hourly electricity demand shows similar profiles, with peaks in the late afternoon hours. However, the simulated recharges via emobpy are longer and have lower consumption, due to the structure of the code, whereby vehicles are often recharged with already high SOCs and use all the time available while parked.

The findings of this study have significant theoretical and practical implications. Theoretically, this research contributes to the body of knowledge on EV infrastructure deployment and usage patterns, providing a comprehensive dataset that can be used to refine and validate predictive models of energy consumption and charging behaviors. This data is essential for upcoming studies related to sustainable transportation systems and grid integration of electric vehicles. Practically, the results offer valuable insights for policymakers, urban planners, and utility companies. Policymakers can use the data to inform decisions on where to allocate resources for new charging infrastructure, ensuring it meets the actual demand and supports widespread EV adoption. Urban planners can benefit from understanding the spatial and temporal distribution of charging sessions to design more efficient and user-friendly recharging networks. Utility companies can use the findings to optimize grid management and plan for future demand, ensuring reliable energy supply and minimizing the risk of overloading the grid. Additionally, businesses involved in the EV market, such as charging station operators and automotive manufacturers, can leverage these insights to enhance their services and products, ultimately contributing to the development of a more robust and sustainable EV ecosystem. The results regarding the presence of charging points installed in the territory and their actual use can provide crucial insights for the operator in the charging infrastructure market. If the data can identify areas that may require an investment for the construction of new charging points and the analysis of total electricity consumption, together with the time of use, they also provide important information on the use of charging points currently available, thus providing an important tool on the effectiveness of installations that have already been completed in recent years. For manufacturers and fleet operators, incorporating advanced simulation tools can lead to more accurate predictions of energy consumption, enhancing range accuracy and extending battery lifespan. This results in better vehicle design, more reliable performance, and greater customer satisfaction. Additionally, the analysis of the share of charging connectors used during the monitored timeframe provides crucial information that can guide manufacturers in designing vehicles with charging systems that align with actual usage patterns. Understanding these preferences allows for more targeted development of new EV models, ensuring they meet specific user needs and driving conditions. By leveraging these insights, the EV industry can refine its products and contribute to the creation of more efficient and sustainable electric vehicles.

The present study collects detailed and granular data of public charging stations. Nevertheless, to have a full picture of the impact of e-mobility on the electricity grid, the same data would be needed for charging stations in private homes, offices, hotels, and restaurants. Usually, most charging events occur at these places, leading to a higher electricity demand than the one of public chargers. However, these private charging stations are usually just a few of the many electricity consumption points connected to a consumer’s electricity meter, making data collection and analysis more challenging. Another limitation of the study presented is related to emobpy, the tool used to simulate daily recharging events, evaluating their duration and finally deriving the hourly recharging profile. Since the emobpy tool provides the type of place where the vehicles are located during the day (e.g., school, work, etc.) but not the exact geographic location, it is not possible to combine the obtained charging profiles and the daily trips to find the optimal location for public charging points for users. Therefore, real and measured data can only be compared regarding the specific field of recharging and impact on the grid through the definition of hourly recharging profiles. Despite this limitation, these two types of data can be considered sufficient to have an initial validation of the data obtained through simulation, as well as showing the different characteristics of the reloads obtained, to refine the future use of predictive systems.

Building on the results of this study, several future research directions emerge and could further enrich the field of e-mobility. First, future studies should explore the long-term impact of increased EV adoption on regional and national electricity grids, incorporating diverse geographic and climatic conditions. Second, there is a need to investigate the effectiveness of various incentive programs and policies in promoting the deployment and usage of EV recharging infrastructure. Third, research could focus on the integration of renewable energy sources with EV charging stations to enhance sustainability and reduce grid dependency. Fourth, the development of advanced predictive models using machine learning and AI could improve the accuracy of demand forecasting and grid impact assessments. Finally, examining user behavior and preferences in different urban and rural settings could provide insights for optimizing the design and placement of future recharging stations. As EV adoption continues to grow, such detailed understanding of charging patterns will be crucial for ensuring the sustainable and efficient integration of EVs into our transportation and energy systems.

5. Conclusions

This study provides a comprehensive analysis of the electric vehicle recharging infrastructure in the Trentino–South Tyrol region, focusing on data from a three-year period (2021–2023). During this time, the number of charging points grew from 673 to 970, representing a 1.4-fold increase, while the consumed energy surged from 938 MWh to 4133 MWh, a 4.4-fold increase. This significant discrepancy between the growth rates of charging points and energy consumption highlights a rapidly increasing demand for recharging, leading to higher utilization rates per charging point. Given the steady growth of the overall EV fleet, this trend is expected to continue, particularly as the current penetration of battery electric vehicles in Italy is still in its early stages. As more battery electric vehicles enter the market, the demand for efficient and strategically located charging infrastructure will likely intensify, resulting in higher utilization rates and increased revenue potential for charging point operators.

The data analysis reveals that the most effective charging point locations are small villages with high tourist attraction (e.g., Riva del Garda, Vipiteno, Badia, Dobbiacco, Ortisei), cities with conveniently located charging stations (e.g., Bolzano, Trento, Merano, Bressanone), and sites close to major travel routes (e.g., Vipiteno, Chiusa). These locations consistently demonstrate higher charging rates and utilization, underscoring the importance of strategic placement in the planning of future charging infrastructure.

In terms of utilization patterns, the study found that the median utilization time per charging session for AC stations is approximately three times longer than that for DC chargers (around 100 min versus 30 min), while the median energy delivered per session by DC stations is about three times that of AC stations (27 kWh versus 9 kWh). Despite the differences in charging time and energy delivery between AC and DC categories, the variation within each category remains limited.

A daily usage profile resembling a “mountain” shape was observed for both AC and DC chargers, with a steep rise in usage beginning at 7 AM (AC) and 8 AM (DC), a flat profile through the day with a slight dip around noon, and a clear decline starting at 4 PM. This pattern is consistent across all days of the week, with Saturdays showing particularly strong occupancy rates, especially for DC chargers.

By comparing these findings with simulation results, the study validated the accuracy of predictive models for grid-impact assessments. The results underline the importance of high-quality input data for reliable grid planning and emphasize the need for strategically located charging points to support the ongoing transition to electric mobility. Future research should focus on understanding the long-term grid impacts of increasing EV adoption, exploring the integration of renewable energy with EV charging infrastructure, and refining predictive models to better anticipate future trends. These efforts will be crucial in ensuring the sustainable growth and effective management of EV infrastructure.