Summarizing the examined literature, one key outcome clearly emerges: the adoption of BEVs has the potential to reduce CO₂ emissions in the transport sector; however, this outcome is not guaranteed. In several cases, BEV may even lead to an increase in CO₂ emissions. It depends on several critical factors, which are detailed below. Therefore, the key question is not whether BEVs could be useful, but how their deployment should be managed to avoid the side effects of their adoption. Some of the main issues identified in the literature and side effects are summarized below:

The paper aims to address some of the gaps outlined in the previous section by proposing a more comprehensive methodology to minimize the indirect CO₂ emissions associated with BEV charging. The environmental impact of battery charging is assessed using the well-known Well-to-Wheel (WTW) approach, which splits total CO₂ emissions into two distinct phases: Well-to-Tank (WTT) and Tank-to-Wheel (TTW) [21]. In the case of BEVs, the TTW phase results in zero CO₂ emissions because no combustion occurs during vehicle operation; in contrast, the WTT is the critical step for assessing the environmental performance of BEVs. In general, WTT CO₂ emissions account for the processes required for the production and distribution of the fuel. In the case of BEVs, the literature mainly focuses on the CO₂ emissions from electricity generation at the power plant level [9], [14], [16]. Building on the issues discussed in point (a), the hourly carbon emission factor for BEV charging is calculated, considering real data from the Italian power grid's generation mix. This approach enables the capture of temporal variability in grid carbon intensity and assesses how charging flexibility can be utilized to reduce indirect CO₂ emissions.

With reference to point (b), the paper proposes a new perspective on smart charging strategies. Generally, V1G and V2G refer to a flexible and smart charging in which the user or the owner of the charging infrastructure adjusts the charging time and power according to grid-oriented signals. In particular, V2G enables BEVs to act as a distributed Battery Energy Storage System (BESS), supporting services such as black start capacity and spinning reserve [24]. In contrast, this work proposes an environmentally driven smart charging strategy, referred to as V1G_CO2. Under this approach, the charging process is scheduled within a predefined time window by prioritizing hours characterized by the lowest grid carbon emission factor, rather than by grid operational constraints or services. It is worth noting that the proposed emission-based charging strategy is conceptually different from price-based approaches, as it explicitly targets minimising indirect CO₂ emissions rather than economic costs. However, the two strategies may be regarded as complementary, since hours characterised by lower electricity prices often coincide with lower-emission hours, as renewable-based generation typically exhibits low or near-zero marginal costs.

Finally, addressing point (c), this study explicitly accounts for the dependence of BEV energy consumption on ambient temperature. Starting from the energy consumption data made available by the vehicle manufacturer, a regression model is developed to describe the energy consumption rate of the selected BEV as a function of outside temperature [25]. This allows the charging demand to be estimated more accurately over time, improving the consistency between vehicle energy needs and the assessment of charging-related emissions.

The paper aims to develop an environmentally driven optimization strategy for BEV charging, focusing on domestic charging after working hours. The primary objective is to minimize indirect CO₂ emissions by scheduling the charging process during periods when the power grid's carbon emission factors are at their lowest, while ensuring vehicle readiness at the beginning of the following day. The methodology combines real hourly emissions factors for the Italian power grid with temperature-dependent vehicle energy consumption.

The remainder of the paper is structured as follows: Section 2.1 describes the user profile, the charging time window; in Section 2.2 a brief overview of the Italian BEV market in Italy is presented, along with a description of the BEV model used, including the regression used to estimate its energy consumption as a function of the outside temperature; finally the Section 2.3 and 2.4 provide an overview of the carbon emission factor for the Italian power grid and the optimization algorithm adopted to schedule the charging process.

The assessment focuses on domestic users who mainly charge their BEV after working hours. According to the outcomes reported by RSE (Ricerca Sistema Energetico), the probability of returning home peaks around 7:00 PM, while it becomes very low around 5:00 AM, [18]. Based on these observations, the time window for vehicle charging is set between 8:00 PM and 5:00 AM of the following day.

A further key aspect of the analysis is the geographical location in which the user drives the BEV. As reported in Section 1.1, the energy consumption factor of EVs is strongly influenced by the outdoor air temperature. The case study focuses on the city of Benevento, situated in the region of Campania in southern Italy. Benevento lies at an altitude of 135 m above sea level, falls within climatic zone C, and has a typical Mediterranean climate. Since the temporal charging windows chosen, the use of BEV is supposed to take place from 6:00 AM to 7:00 PM, therefore, for the influence of temperature, the mean of hourly temperature value, for the year 2023, retrieved by PVGIS [26], belonging to the temporal range from 6:00 AM to 7:00 PM, is computed. The decision to compute the average temperature rather than using hourly temperature data is due to the lack of information on the BEV's actual driving schedule. The information retrieved by [18] allows only for the identification of a plausible daily usage time frame rather than an exact hourly driving pattern; therefore, an average temperature over the assumed driving period is adopted.

Figure 2 presents a graphical representation of daily BEV usage and the domestic charging framework. Driving hours (06:00 AM – 7:00 PM, day d) are used to compute the mean outdoor air temperature, which is later used to estimate the vehicles’ energy consumption. The charging process, indicated by green blocks, is allowed only within the domestic charging window (8:00 PM–5:00 AM, day d to day d+1).

Figure 2.

Schematic representation of the daily BEV usage and domestic charging framework

The distance travelled by the user is another key factor, as it directly affects the electricity required to cover a given route. To define the daily driving distances, a study conducted by the National Union of Foreign Vehicle Representatives (UNRAE, Unione Nazionale Rappresentati Autoveicoli Esteri) was considered [27]. According to this assessment, passenger vehicles in Italy travel an average of 10,712 km/year, corresponding to approximately 30 km per day. Building on this outcome, two driving distances are considered: 30 km, representing a typical daily usage, and 120 km, accounting for scenarios involving longer trips.

As the analysis focuses on residential charging, electricity supply is assumed to occur through either a standard domestic socket (Schuko) or a wallbox. The aim of this paragraph is not to provide a comprehensive overview of charging technologies, but rather to clarify the power levels considered in the simulations. In Italy, one of the most common contractual power levels for residential users ranges between 1.5 kW and 3 kW; therefore, the domestic charge via a Schuko socket typically occurs at 2.3 kW. Conversely, wallbox solutions commonly operate at power levels of 3.7 kW and 7.4 kW [28]. Based on a market analysis, the 7.4 kW option is selected for the case study, [28]–[34]. Table 1 summarizes the domestic charging configurations considered in this study.

Currently, different types of EVs are available on the automotive market. Based on the degree of electrification, they are commonly gathered into four main categories. However, the section's aim is not to review the well-known characteristics, advantages, or drawbacks of each category of EVs, but rather to clarify the focus on BEVs from a charging perspective. These four categories are grouped into two broader classes according to the battery charging mechanism:

• The first class includes vehicles characterized by indirect battery charging. Mild Hybrid Electric Vehicles (MHEVs) and Full Hybrid Electric Vehicles (HEVs) do not allow direct charging through an external charging infrastructure. In MHEVs, battery charging relies solely on regenerative braking, whereas in the second category, it is supported by both regenerative braking and the internal combustion engine.

• The second class involves vehicles enabling direct battery charging as PHEV and BEV. These vehicles can recharge their battery packs directly from the power grid through a charging point [21].

Building on these considerations, since the objective of this work is to optimize the battery charging schedule of vehicles to ensure a given driving range while minimizing the associated indirect CO₂ emissions, the analysis focuses exclusively on vehicles belonging to the direct battery charging category, specifically, the scope of the analysis is restricted to BEVs.

Once the vehicle category is defined, the next step involves assessing the national vehicle fleet to identify a representative passenger car for the case study. In 2024, the Italian vehicle fleet reached approximately 41.3 million units, with an average age of around 13 years. The fleet is still largely dominated by vehicles powered by traditional fuels: gasoline cars account for 42% of the total, followed by diesel vehicles at 40.9%. Regarding electric mobility, HEVs account for the largest share (6.8%), while PHEVs and BEVs together make up 0.7% of the total fleet [35]. The charging process analysis is carried out considering a representative compact BEV as a case study [36]. The selected vehicle belongs to the small city car segment and reflects one of the most widely adopted BEV configurations in the Italian passenger car market from 2021 to 2024, making it a representative choice for the domestic user segment [35]. The considered BEV is available with an electric motor rated at 45 or 65 hp and is equipped with a 26.8 kWh battery pack, which ensures a range of 225 km on a mixed cycle, including urban and extra-urban use, and 305 km on urban-only use. Table 2 reports the main technical specifications, which are useful for contextualizing its dimensions and characteristics [37], [38].

The energy consumption rate is a key parameter for BEVs, as it directly impacts the vehicle’s range and, consequently, the distance that can be traveled. Usually, this data is reported by the manufacturer as results of the Worldwide Harmonised Light-Duty Vehicles Test Procedure (WLTP) cycle [39]. However, as discussed in the previous section, the actual energy consumption of electric vehicles depends on several factors. These factors can be grouped into three main categories: (i) driver related aspects, such as driving style and the use of auxiliary; (ii) external conditions related to the design of the city, the road, and traffic conditions; and (iii) weather conditions, in particular ambient air temperature [23], [40], [41]. Without real-world driving data or vehicle powertrain models, it is challenging to consider all these effects simultaneously. Therefore, this study focuses on the impact of air temperature on the energy consumption factor of BEVs. The WLTP test is conducted at a reference temperature of 23 °C [42]. Several studies generally report that BEVs operate most efficiently within a temperature range of approximately 20 °C to 25 °C [22]. Deviations from this range, both at high and low temperatures, lead to increased energy consumption. This effect is mainly due to variations in the efficiency of power electronics components, changes in aerodynamic drag, rolling resistance, and the additional energy demand of thermal management and auxiliary systems. Although this has been widely demonstrated in the literature, low temperatures, in particular, have a greater impact on all these aspects, especially on the auxiliary system for thermal passenger comfort. Unlike internal combustion engine vehicles, BEVs cannot recover waste heat from the powertrain to heat the cabin. Instead, cabin heating relies on resistive heaters or heat pumps, which significantly increase electricity demand [23], [42], [43]. Although heat pumps improve energy efficiency compared to resistive heating, their performance still degrades as the air temperature drops [23], [42], [43]. Building upon these outcomes, the second step involves regression analysis to define a function that describes the relationships between outside air temperature and energy consumption. Obviously, a universal function applicable to all BEV models has not been developed, as this dependency is specific to each vehicle. The case study focuses on a specific vehicle model, and the required input data were retrieved from the range simulator available on the car manufacturer’s website [25]. This tool enables the estimation of the selected BEV range based on the type of route, temperature, and vehicle settings. In this study, simulations were performed under urban driving conditions, with air conditioning activated and ECO modes deactivated. For each temperature value within the considered range, the estimated driving range was divided by the battery capacity to obtain the corresponding energy consumption factor. Table 3 reports the resulting dataset. Consistent with previous findings in the literature, temperature dependency exhibits different trends. Therefore, two distinct regression functions were carried out: one for temperatures below 20 °C and the other for those above 20 °C. In addition, continuity at the reference temperature was enforced by constraining both regressions to pass through the point (20 °C,12.8 kWh/100km).

Figure 3 shows the energy consumption factor of the selected BEV retrieved from the manufacturer's simulator; the dashed line represents the reference value of 14.6 kWh/100 km declared by the manufacturer. This highlights that for ambient temperatures above approximately 10 °C, the estimated energy consumption remains below the selected BEV reference value. Conversely, at lower temperatures, the consumption increases significantly and exceeds the manufacturer’s declared value. For example, at ‒5 °C, the energy consumption factor reaches a value of 18.5 kWh/100 km, which is approximately 25% higher than the selected BEV reference value.

Figure 3.

Energy consumption data retrieved from the manufacturer's simulator

As mentioned previously, the CO₂ emission factor of the Italian electricity system used in this work is derived from national electricity production data. Since the case study focuses on 2023, it is necessary to provide an overview of the national electricity generation mix for that year. In this context, the statistical report published by Terna, the Italian Transmission System Operator (TSO), represents a reliable source, as it provides a detailed breakdown of electricity production by primary energy source [44]. An analysis of the 2023 electricity mix reveals a decrease in electricity generation from thermoelectric power plants compared to 2022, which nonetheless still accounted for approximately 58% of total national production. Conversely, electricity generation from RESs increased by 16% compared to 2022, reaching a share of 37.1% of gross domestic electricity consumption, net of pumping contributions. A more detailed analysis of production from RESs in 2023 shows a strong recovery in hydroelectric generation, which increased by 42.7% compared to 2022 levels, reaching approximately 40 TWh, thus confirming hydropower as the main RES in the national energy mix. Wind power generation also recorded a significant increase, reaching 23.6 TWh, while photovoltaic production set a new record, at 30.7 TWh. In contrast, electricity generation from bioenergy declined by 9.1%, resulting in a decrease of 16 TWh. Moreover, for the fourth consecutive year, electricity production from geothermal sources continued to decline, settling at 5.7 TWh. Figure 4 shows the percentage contribution of the different energy sources to national electricity production in Italy in 2023.

Figure 4.

Percentage contribution of different energy sources to Italian electricity production in Italy in 2023 ‒ TERNA

By cross-referencing these aggregated national data with the hourly electricity production profiles of all power plants, obtained from the Transparency Platform of ENTSO-E (European Network of Transmission System Operators for Electricity), the national electricity mix was reconstructed at an hourly resolution [45]. Subsequently, ISPRA (Istituto Superiore per la Protezione e la Ricerca Ambientale, The Italian Institute for Environmental Protection and Research) provided technology-specific CO₂ emission factors for electricity production in 2023, differentiated by primary source [46]. Based on these data, total CO₂ emissions associated with electricity generation were calculated for each hour of the year. Knowing the amount of CO₂ emitted as a result of electricity production from thermoelectric plants, three scenarios were defined. For the first emission factor, α TE F, only thermoelectric plants powered by fossil fuels were considered. The second factor, α TE F +B , includes thermoelectric plants fueled by both fossil fuels and biofuels. Finally, the third emissions factor, α TE F +RES, accounts for total electricity production, including both fossil-based and renewable sources. By definition, α TE F +RES represents CO₂ emissions per unit of electricity withdrawn from the grid and therefore reflects, on an hourly basis, the actual national electricity production mix. More details of the methodology adopted are reported in [19]. In the case study, the additional electrical load associated with the BEV charging was assumed to be sufficiently small so as not to significantly alter the national electricity production mix. Therefore, the CO₂ emissions factor α TE F +RES was adopted for the assessment of indirect CO₂ emissions.

Figure 5 shows the values of the emission factor α TE F +RES, calculated at daily and hourly resolution for the whole of 2023. The y-axis reports the emission factor values, the x-axis represents the days of the year, and the colour scale, ranging from blue to red, represents the hour of the day. An analysis of the trend reveals higher average values in the winter months and lower average values during the summer months. This seasonal behavior can be mainly attributed to the increased contribution of renewable sources in the summer months, particularly photovoltaic generation, which benefits from higher solar irradiance. Figure 5 also shows that, for the same season and day, the highest emission factor values are consistently associated with the early morning and nighttime hours, whereas the lowest values occur during the midday hours. Specifically, maximum values of α TE F +RES are recorded in the time slot from approximately 0:00 AM to 4:00 AM (blue dots) and after 8:00 PM (red dots), while the minimum values are recorded between approximately 9:00 AM and 4:00 PM. This intraday variability is again attributable to the higher penetration of renewable generation, especially photovoltaic production, during daylight hours.

For the purposes of this work, the hourly values of the emission factor were converted to a 15-minute time resolution in order to ensure consistency with the temporal resolution adopted for the charging schedule. Each hourly emission factor was assumed to be constant over the corresponding four 15-minute intervals.

Figure 5.

Trend of CO2 emissions factor α_(TE_(F+RES) ) for 2023

In case of large-scale EV adoption, the use of marginal CO₂ emissions factors would be more appropriate, since a substantial increase in electricity demand does not translate into a proportional change in the output of all generating units [47]. Indeed, only a limited subset of power plants can ensure the required operational flexibility and rapid response to load variations, and these units typically determine the marginal CO₂ emissions of the power grid rather than the overall production mix.

Marginal emissions factors can be calculated either ex ante by solving an optimal dispatch problem within an energy model of the electricity system, or ex post by analysing electricity market data or production data. In the latter case, marginal emissions can be inferred through the merit order principle, which links marginal emissions to marginal technology, or through regression-based approaches that quantify the relationship between variations in electricity production and associated CO₂ emissions [17]. While these approaches are well established in the literature, their application would shift the focus of the present study from an emission-aware charging strategy at the individual user level to a system-level assessment of electricity generation dispatch. Since the additional load associated with a single BEV is assumed not to significantly affect the national generation mix, the analysis uses hourly-average grid emission factors. A comprehensive treatment of marginal emission factors and their computation is therefore beyond the scope of this work.

Finally, the hourly unit electricity prices (PUN) for the year 2023 were retrieved from the platform made available by the Italian electricity market operator (GME) [48]. These data were cross-referenced with the percentage share of renewable energy sources in the national electricity production mix, previously calculated. The resulting relationship is illustrated in Figure 6. The y-axis reports the hourly electricity price, while the x-axis represents the days of the year. The colour scale, ranging from blue to red, indicates the share of renewable generation in the production mix. It can be observed that periods characterised by higher renewable penetration (red markers) generally correspond to lower electricity price levels. This evidence supports the argument that emission-based and price-based charging strategies are not necessarily conflicting approaches, but may instead exhibit complementary dynamics under typical market conditions.

Figure 6.

Relationship between hourly electricity prices and renewable generation share in the Italian electricity mix for 2023

This section describes the core of the algorithm implemented for defining an environmentally driven smart BEV charging strategy, V1G_CO2. The optimization framework aims to determine a daily charging schedule, sched d r,p , for each distance r and rated power charging P, that minimizes indirect CO₂ emissions associated with electricity consumption, while ensuring the energy required to meet a predefined driving range. The procedure began with the evaluation of the average daily air temperatures for the city of Benevento over the time interval from 6:00 AM to 7:00 PM. The average temperature, denoted as T_d, is computed for each day d of the year. Based on T_d, the second step of the algorithm consists of calculating the daily energy needed, E d r, to cover a given range, r, for each day of 2023, as shown in the following equation:

where e_c(T_d) represents the energy consumption factor of BEV obtained from the regressions assessment, described in Section 2.2.1, as a function of mean outdoor air temperature.

Once the daily energy demand is defined, the charging window is defined with a temporal resolution of fifteen minutes, between 8:00 PM on day d and 5:00 AM on day d+1. For each time slot within this interval, the corresponding emission factor α TE F +RES is identified. The time slots are then sorted in ascending order according to their associated emission factor values, so that the first elements of the ordered set correspond to the periods with the lowest carbon emissions. The allocation of the energy to be recharged (E d r) is performed to be equal to the energy consumed during the day. For each fifteen-minute time slot in the sorted list, the algorithm assigns an amount of energy equal to the minimum between the maximum energy that can be drawn from the grid, depending on the selected charging power, and the residual energy still to be recharged. After each allocation step, the residual energy demand is updated, and the procedure continues until either the required energy is fully recharged or the charging window is exhausted. This iterative process yields a daily recharging schedule that prioritizes low-emission time slots and is tailored to each combination of driving distance and charging power.

Furthermore, it is important to emphasize that the optimization algorithm is explicitly designed to ensure that the BEV is fully charged and ready for use at the predefined departure time, even when the charging process is temporally shifted within the available charging window. The only circumstance under which the vehicle would not reach full charge by 5:00 AM is not related to the charging strategy itself, but rather to physical constraints of the charging process. Specifically, if the total energy required exceeds the energy that can be delivered within the available charging window, given the available charging power, full recharging cannot be achieved regardless of the scheduling strategy adopted.

Before the charging process starts and after each charging step, the battery's State Of Charge (SOC) is evaluated to verify whether the available charging window is sufficient to meet the energy demand. According to recent literature, maintaining the SOC within the range of 20% - 80% is recommended to preserve battery health and limit degradation phenomena, as a series of deep discharges can compromise the battery's functionality [49], [50]. In line with these findings, an initial SOC value of 80% is assumed in the case under study. With this aim, the residual energy stored in the battery at the end of the day driving, Q residual,d r is calculated as shown in the following equation:

where: Q_max is the battery capacity of the selected BEV, as reported in Table 2; SOC_in is the initial charge level, 80%; and E d r is the energy required to travel a distance r on the day d. The SOC at the end of the driving phase is then calculated as stated by the equation:

During the charging phase, the residual energy stored in the battery is dynamically updated by adding the energy allocated at each time step. Although SOC can be used to modulate charging power in advanced charging strategies in order to preserve the battery's useful life, in this work, the charging process is assumed to occur at constant power, without the implementation of SOC-dependent charging strategies [49].

It is also worth noting that potential issues related to battery degradation are not addressed in this study. Battery degradation phenomena are widely documented in the literature and are commonly classified into calendar ageing and cycle ageing, the latter being directly influenced by charging conditions. In particular, battery ageing is known to depend on several factors, including the charging and discharging C-rate, the SOC level, and the ambient temperature at which charging occurs [51]. Although degradation effects are unavoidable in practice, their impact is expected to be limited under the operating conditions considered in this work. Indeed, several studies report that prolonged charging at high C-rates can accelerate battery ageing, an effect that is further exacerbated when high C-rates are combined with high SOC levels [51]. In the present study, the maximum SOC after charging is capped at 80%, a commonly recommended threshold to preserve battery health. Moreover, the charging process is characterised by low C-rates. For the two charging power levels considered (2.3 kW and 7.4 kW), the corresponding C-rates are approximately 0.09C and 0.27C, respectively (computed as the ratio of charging power to the battery's nominal capacity), i.e., well below 1C. Previous studies indicate that slow AC charging at low C-rates is generally associated with slower degradation processes [51] and higher energy efficiency compared to high-power charging [52].

If battery degradation were explicitly included in the analysis, it would mainly manifest as a gradual reduction in the usable battery capacity over time. From a modelling perspective, this would translate into higher energy demand for the same driving distance and outdoor temperature, resulting in lower SOC at the end of the driving phase. However, the structure of the proposed charging optimisation algorithm would remain unchanged, as it is formulated to allocate the required charging energy within the available charging window, independently of the underlying causes of capacity reduction.

The developed algorithm enables the calculation of CO₂ emissions associated with BEV charging for each combination of day d, travelled distance r, and charging power P. The daily CO₂ emission, C O 2,d r,p are calculated as shown in the following equation:

where h_start and h_end denote the beginning and the end of the charging window, respectively; sched d r,p (i) represents the energy charged during the i-th 15-minute time slot; and α TE F+ RES d (i) is the corresponding carbon emission factor of the Italian power grid at the same time step.

As shown in Table 3, the temperature data used as input for the regression analysis span a wide range, from ‒5 °C to 35 °C. Therefore, before proceeding with the regression analysis, the average daily temperatures were calculated, as described in Section 2.1. Figure 7 shows the resulting temperature profile for the city of Benevento over the reference year. The graph supports the consistency of the adopted regression approach, as it clearly shows that the air temperature range considered in the regression, highlighted by the lower and upper bounds with dashed red lines, encompasses all the average daily temperatures calculated for Benevento.

Based on the analysis of average daily temperatures, it can be inferred that the EV operates predominantly under non-optimal climatic conditions. During the winter, the average temperature frequently falls below 10 °C, with minimum values reaching approximately 1.8 °C. In contrast, during summer, it often exceeds 30 °C, with peak values close to 35 °C. As discussed in Section 2.2.1, both low and high temperatures negatively impact the energy efficiency of BEVs, underscoring the importance of explicitly accounting for temperature effects in estimating energy consumption.

Figure 8 shows the regression functions developed in the MATLAB environment, while Table 4 reports the coefficients of the regression functions. These two functions describe the relationship between the energy consumption factor of the selected EV and the outside temperature, representing a key input to the entire optimization framework. The analysis of the figure reveals two distinct behaviors, depending on whether the temperature is lower or higher than the reference value of 20 °C. This change in trend confirms the need to adopt two separate regression functions to accurately represent the temperature-dependent variation of the consumption factor.

Figure 7.

Average daily temperatures for Benevento

Figure 8.

Regression energy consumption factor outcomes as function of ambient temperature

The results of the analysis on the energy performance of BEVs are presented below, with the aim of highlighting the influence of temperature on the energy consumption factor of BEVs. In this context, the consumption factor of the selected BEV was calculated for each day of 2023 by applying the two previously developed regression functions and the average daily temperatures for Benevento during vehicle operating hours.

The results are summarized in Figure 9, which illustrates the trend in the energy consumption factor throughout 2023. A colour scale is adopted to represent the corresponding average daily temperature, ranging from dark blue (low temperatures) to dark red (high temperatures). The figure clearly shows higher energy consumption factors during the winter months due to the strong impact of low temperatures; conversely, lower values are observed during the summer and mid-season periods.

The lowest temperatures occur at the beginning of February, as indicated by the dark blue markers, for which the corresponding energy consumption factor values are the highest. The maximum value reaches 16.8 kWh/100 km. During the summer months, although temperatures reach around 30 °C ‒ 35 °C, the energy consumption factor remains lower and consistently below the reference threshold value of 14.6 kWh/100 km. The energy consumption factor reaches the minimum value of 12.8 kWh/100 km at an average air temperature of 20 °C. Overall, the results indicated that, when considering temperature effects alone, among the various factors affecting the car's performance, the vehicle performs less efficiently than the manufacturer's stated figures when the temperature drops below 10°C.

Table 5 reports the minimum, maximum, and average values for both temperature and energy consumption factor. Despite the presence of extreme values, the annual average energy consumption factor is 14 kWh/100 km, which remains lower than the reference threshold value.

The analysis of the results related to the electrical energy required by the vehicle to travel a certain distance further confirms the strong dependence of BEV energy performance on temperature. The results show that the energy demand on the electrochemical storage system during the summer months is significantly lower than that observed in winter. In fact, the maximum value, equal to 5.03 kWh, occurs on February 6^th, while the minimum, 3.83 kWh, occurs on May 22^nd. The corresponding annual average value is 4.19 kWh.

Figure 9.

Energy consumption trends for BEV over the reference year

A similar trend is observed for the distance of 120 km. In this case, the minimum value, equal to 15.32 kWh, and the maximum value, 20.13 kWh, occur on the same day identified for the 30 km scenario, while the annual average energy demand amounts to 16.76 kWh. This behavior is primarily attributable to the impact of temperature on BEV energy consumption, which becomes particularly pronounced under cold climatic conditions.

Consequently, low-temperature conditions lead to a disproportionate increase in energy demand compared to high-temperature conditions. This behavior is not specific to the selected BEV but is consistent with findings widely reported in the literature for BEV.

The monthly analysis further corroborates the daily results, showing higher energy demand during the winter months compared to the summer period, as shown in Figure 10.

Figure 10.

Monthly energy demand

In addition, the algorithm developed in MATLAB also calculates the SOC of the selected BEV at the end of each daily trip throughout 2023. Figure 11 shows the daily SOC values as a function of the travelled distance. A red dashed line indicates the 20% SOC threshold, below which it is advisable not to fall in order to preserve the vehicle's battery life and efficiency. The figure shows that, even in the most critical months (winter), the SOC remains above the threshold for short daily distances (30 km). Conversely, for longer distances (120 km), range-related issues begin to arise. Furthermore, for short distances, the influence of the ambient temperature appears to be limited: the SOC never falls below 60% and generally remains close to 65% throughout the years. In the case of longer distances, however, the effects of temperature are more noticeable, resulting in significant SOC variability. For instance, in February, the SOC range spans from a minimum value of 4.9%, recorded on February 5^th, to a maximum value of 17.6%, observed on February 25^th.

These results highlight potentially critical operating scenarios for BEVs, particularly during winter. From a battery preservation perspective, adopting intermediate charging events instead of a single end-of-day recharge could be a viable mitigation measure to avoid prolonged periods at low SOC levels. This consideration does not stem from the environmentally driven charging strategy itself, but rather from the harsh climatic conditions that exacerbate energy consumption and lead to deeper discharge cycles during vehicle operation.

Figure 11.

SOC at the end of the day for each day of the reference year

Initially, the average charging times were calculated for the two charging powers considered, along with the corresponding travelled distances. The results, expressed in hours, are reported in Table 6. Overall, the average charging duration ranges from approximately 35 minutes to more than 7 hours, depending on the combination of distance travelled and charging power. For the purposes of this study, and to facilitate the interpretation of the optimization results, charging events are grouped into two categories: short charging sessions, with durations shorter than 2 hours (typically associated with short distances or high charging power), and long charging sessions, with durations exceeding 5 hours (associated with long distances or low charging power). This distinction is particularly relevant for the subsequent analysis of charging optimization strategies and their interaction with time-varying grid emission factors.

Charging schedule optimization.

Continuing the analysis of the results, this subsection presents the charging schedules obtained by applying the V1G_CO2 optimization algorithm under different operating conditions.

Figure 12 shows the histogram of charging start times, highlighting their temporal distribution with respect to the electricity grid emission factor considered. A clear peak is observed at 8:00 PM, indicating a high frequency of charging initiations at this time. The results, therefore, suggest that, from an environmental perspective, it is preferable to start charging at 8:00 PM and avoid prolonging the charging period until midnight. Analyzing these results in relation to the uncontrolled charging (baseline scenario) reveals that it has a less environmental impact only in cases of short charging durations. In such situations, charging is mainly concentrated between 8:00 PM and 9:00 PM, when emission factors are at their minimum.

Figure 12.

Frequency relative to charging start hours

The following results illustrate the charging schedules obtained according to the V1G_CO2 logic for a representative day, 15 January, in the city of Benevento. This specific day was chosen because it presents conditions that clearly illustrate the result of the optimization process implemented. On this day, the average daily temperature is 8.4 °C, corresponding to an energy consumption factor of 15.3 kWh/100 km. Consequently, the energy demand to travel 30 km and 120 km is equal to 4.6 kWh and 18.3 kWh, respectively.

Figure 13 shows the results of the optimization algorithm for a charging power of 2.3 kW and a distance traveled of 30 km. The primary y-axis displays the distribution of energy across the various 15-minute time slots, while the secondary y-axis represents the corresponding quarter-hourly trend of the grid emission factor. The first step taken by the algorithm is to identify the time slots characterized by the lowest emission factor. In this case, the interval between 1:00 AM and 2:00 AM exhibits the minimum values, with approximately 92 gCO₂ emitted per quarter of an hour.

Figure 13.

Charging schedule for rate charging power 2.3 kW and traveled distance of 30 km ‒ 15 January

The algorithm, therefore, allocates the rechargeable energy to these four time slots. Since the energy required is not fully satisfied at the end of this interval, the algorithm subsequently selects the second most favourable time window, between 8:00 PM and 9:00 PM, during which the emission factor is approximately 93 gCO₂ per quarter of an hour. The algorithm allocates the maximum rechargeable energy, 0.58 kWh, in the first three slots, while the remaining energy (0.55 kWh) is allocated in the last slot.

It should be noted that, since the quarter-hourly emission factors of the electricity grid are derived from hourly values, the charging occurs in consecutive 15-minute intervals. With the update of the methodology for calculating the emission factors of the electricity grid, which is now based on a 15-minute interval rather than an hourly basis, charging schedules could become more fragmented and potentially even more optimized.

For comparison under uncontrolled charging conditions, charging would have occurred continuously between 8:00 PM and 10:00 PM, thus including time intervals associated with relatively high emission factor values.

Figure 14 presents the results of the optimization algorithm for a charging power of 2.3 kW and a travelled distance of 120 km, a case representative of long charging durations. This scenario is particularly informative, as it highlights the limitations of the optimization algorithm when the required charging time exceeds a significant portion of the available charging window. In this case, more than 7 hours of charging are required out of the 10 hours available. As a result, although the algorithm prioritizes low-emission time slots, it can only avoid allocating energy during the two most emission-intensive intervals: from 10:00 PM to 11:00 PM and from 3:00 AM to 4:00 AM, characterized by approximately 97 gCO₂ and 98 gCO₂ per quarter of an hour, respectively.

Figure 14.

Charging schedule and SOC evolution for rate charging power 2.3 kW and traveled distance of 120 km - 15 January

Figure 15 reports the case of a short charging session with a rated charging power of 7.4 kW and a travelled distance of 30 km. Since the total charging duration is approximately 36 minutes, the algorithm allocates the rechargeable energy entirely in the lowest-emission time slot. This scenario clearly demonstrates that in the case of short charging sessions, the V1G_CO2 algorithm is particularly effective in minimizing the indirect CO₂ emissions associated with BEV charging.

Figure 15.

Charging schedule and SOC evolution for rate charging power 7.4 kW and traveled distance of 30 km ‒ 15 January

Indirect CO₂ emission associated with BEV charging.

To evaluate the effectiveness of the proposed optimization algorithm, the daily and annual indirect CO₂ emissions associated with the BEV charging process were calculated for each scenario. Figure 16 reports the daily CO₂ emissions obtained for a rated charging power of 7.4 kW and a travelled distance of 120 km. The results clearly show higher emission levels during the winter months compared to the summer period. The maximum daily emission, equal to 2036 gCO₂, was recorded on January 23^rd, whereas the minimum value of 412 gCO₂ occurred on November 4^th. The corresponding annual average value is 997 gCO₂ per day. This seasonal trend can be explained by the combined effect of two main factors: the energy consumption factor of EVs is higher during winter months due to lower ambient temperatures, and the electricity grid emission factors are generally higher in winter, reflecting a reduced contribution from RESs.

Figure 16.

Daily indirect CO2 emissions ‒ travelled distances 120 km, rated power charging 7.4 kW

For the sake of brevity, the corresponding daily emission profiles for the other scenarios are not reported, as they exhibit analogous trends. A comprehensive overview of the annual indirect CO₂ emissions is provided in Table 7, which summarizes the total annual emissions as a function of travelled distance and charging power.

Table 7.

Annual indirect CO2 emissions [kgCO2/year]

Rated charging power	Distances
	30 km	120 km
2.3 kW	112	478
7.4 kW	111	452

A further key indicator for assessing the environmental performance of EVs is the average annual values of specific CO₂ emissions, expressed per unit distance travelled. Table 8 reports the values obtained in the different scenarios. For short travelled distances, corresponding to short charging sessions, the influence of charging power is negligible: the specific emissions range from 10.1 gCO₂/km to 10.3 gCO₂/km for charging powers of 7.4 kW and 2.3 kW, respectively. This result is explained by the structure of the optimization algorithm, which concentrates charging in time intervals characterized by the lowest emission factors. In contrast, for long travelled distances, and therefore long charging durations, the charging power plays a more significant role. In this case, the maximum specific emissions value of 10.9 gCO₂/km, observed at low charging power, decreases to 10.3 gCO₂/km when a higher charging power is adopted. These results further confirm that the effectiveness of the V1G_CO2 strategy increases as the flexibility in allocating charging energy across low-emission time slots improves.

Table 8.

Average annual values of specific CO2 emissions [gCO2/km]

Rated charging power	Distances
	30 km	120 km
2.3 kW	10.3	10.9
7.4 kW	10.1	10.3

To assess the potential environmental benefits of the proposed V1G_CO2, the CO₂ emissions associated with the optimized BEV charging strategy are compared with those obtained under uncontrolled charging. Table 9 reports the annual CO₂ emissions for all the analyzed scenarios. The comparison clearly shows that, in all scenarios, uncontrolled charging leads to an increase in CO₂ emissions. Annual emissions range from 117 kgCO₂ in the case of short charging sessions to 489 kgCO₂ for long charging sessions. In the case of short recharges, corresponding to a travelled distance of 30 km, the adoption of the V1G_CO2 strategy leads to an annual emission reduction of approximately 5% – 6%; while in the case of long recharges, as discussed previously, the effectiveness of the optimization decreases, with emission reductions limited to around 2%. This behavior is consistent with the reduced flexibility in allocating charging energy when a large fraction of the available charging window is required.

To correctly interpret the relatively modest reductions observed even in the most favourable scenarios, it is important to note that the effectiveness of the optimization algorithm depends not only on the duration of the charging process but also on the temporal position of the minimum grid emission. On days when the minimum coincides with the beginning of the charging window (8:00 PM), the optimization yields limited additional benefits, as uncontrolled charging naturally occurs during low-emission time slots. Figure 12, with reference to the α TE F+RES parameter, shows that, over the year, the minimum emission factor occurs at 8:00 PM on 194 days; therefore, in more than half of the year, the minimum coincides with user-driven charging. Nevertheless, in the remaining cases, when the minimum emissions factor occurs later in the night, the optimization algorithm effectively shifts charging away from periods with higher emissions, leading to measurable CO₂ reductions.

These results highlight that the V1G_CO2 strategy is particularly effective when sufficient temporal flexibility exists and when low-emission electricity is available outside the natural user-driven charging start time. Consequently, the proposed approach demonstrates its greatest potential in scenarios characterized by higher variability in the electricity mix and longer charging windows.

Finally, it is important to note that these results refer to charging a single BEV. However, if a fully electrified Italian passenger vehicle fleet is assumed, the proposed strategy would lead to an annual reduction in 2023 of approximately 4766 ktCO₂/year, while preventing a potential increase of about 7% in total passenger transport sector emissions that would otherwise occur in the absence of charging optimization.