Comparing costs and climate impacts of various electric vehicle charging systems across the United States

An integrated techno-economic analysis (TEA) and life cycle assessment (LCA) was developed to comprehensively compare the widescale deployment of DCFC, BSS, and DWPT charging infrastructure in the contiguous U.S. In this analysis, the three charging systems were deployed independently to facilitate comparison. The analysis focused on four vehicle categories: car, LDT, MDV, and HDV. For each of the vehicle categories, the TCO and cradle-to-grave GHG-intensity were evaluated for each of the EV charging systems and compared to a representative HEV and ICEV with a functional unit of one VKT. The analysis considered the implementation of charging infrastructure in 2030, aligning with the expected mass market adoption of EVs^4,20. Infrastructure deployment was exclusively modeled in 2030 to ensure a consistent comparison across charging systems, followed by their 20-year operational period from 2031 to 2050^12,37.

Charging system energy usage

The usage of public charging was evaluated for DCFC, BSS, and DWPT. The DWPT system was assumed to maintain the vehicle’s state of charge while driving on the electrified roadway, whereas the DCFC and BSS systems were modeled to provide energy only during daytime trips. The vehicle energy efficiencies per VKT were 0.19 kWh for cars, 0.30 kWh for LDTs, 0.68 kWh for MDVs, 1.34 kWh for HDVs, and 1.35 kWh for buses³⁸. The usage of these systems was categorized into stationary charging usage (DCFC and BSS) and DWPT roadway usage. Cars, LDTs, MDVs, and HDVs were assumed to utilize the stationary charging systems and DWPT roads. Buses were included in the usage of the DWPT roadway; however, their TCO and GHG-intensity were not explicitly modeled in the analysis due to their minimal VKT³⁹.

For each vehicle category, the usage of stationary charging systems was modeled separately. Cars and LDTs were assumed to have the same usage since they are both considered light-duty vehicles. Among light-duty vehicles, only battery electric vehicles were assumed to utilize the infrastructure, as plug-in hybrid electric vehicles are typically incompatible with the high-power rates of DCFC and the standardized battery sizes required for BSS. Hence, it was estimated that 82% of electric cars and LDTs would use the infrastructure⁴⁰. Observational data indicated that public charging usage for light-duty vehicles is around 6%, resulting in a 5% usage of public charging for electric cars and LDTs⁴¹.

However, due to the lack of available data for electric MDVs and HDVs, the usage of public charging was simulated for multiple vehicle operating ranges and corresponding battery sizes⁴². MDVs and HDVs were assumed to undergo overnight charging and start each day with a fully charged battery²⁹. The battery sizes were divided into 90-kWh battery modules⁴³, and the number of modules onboard the EV was determined to minimize battery expenses while ensuring that the vehicle required no more than one public charging event per day during the operator’s required driving break⁴⁴. The simulated driving break was modeled such that the vehicle’s state of charge would be above 20% before the charging event to maintain battery health and below 80% at the end of the charging event to optimize charging time⁴⁵. The vehicle was assumed to only charge the minimum amount to complete its trip. The battery size, portion of VKT in the vehicle category, and portion of public charging usage are presented in Table 1 for each operating range. The average portion of energy supplied from public charging, weighted by VKT in each operating range, was found to be 0.6% for MDVs and 14% for HDVs.

The electrified roadway was assumed to provide continuous power, maintain the vehicle’s state of charge, and be used by cars, LDTs, MDVs, HDVs, and buses. The minimum portion (R) of each roadway segment (i) that is needed to be electrified for each vehicle category was calculated in Eq. (2) based on the segment’s speed limit (S), receiving pad power rating (P) of 50-kW, number of receiving pads (N) on the vehicle (v), vehicle energy efficiencies per VKT (EE), charging efficiency (CE) of 85%⁴⁶, and amount failed (F). The amount failed was calculated in Eq. (1) based on the VKT by each vehicle type over the life of the roadway segment, number of receiving pads on the vehicle, failure rate (FR) of 2.87 pads per million hours⁴⁷, number of roadway pads (RP) per segment (200 per kilometer (km))¹², and speed limit.

The electrified portion of the roadway was modeled based on the vehicle category that needed the highest portion electrified. The number of receiving pads for each vehicle category depended on their charging requirements and wheelbase allowances, with cars having one pad, LDTs having two pads, MDVs having four pads, and HDVs and buses having five pads¹².

Time of day usage

Time-of-day resolution was added to the EV energy demand using arrival and departure time data for cars, LDTs, MDVs, and HDVs. The charging schedule for DWPT and BSS was aligned with the vehicles’ in-route periods since these systems have minimal charging times (Supplementary Fig. 6). In contrast, the DCFC schedule corresponded to the vehicles’ dwell periods due to the slower charging rate of DCFC (Supplementary Fig. 7).

The arrival and departure time data for cars and LDTs were extracted from the 2017 National Household Travel Survey, which provides trip data for various vehicles⁴⁸. Specifically, the schedule for cars was derived from 280k automobile trips, while the LDT schedule was based on 289k van, sport utility vehicle, pickup truck, other truck, and recreational vehicle trips. The energy consumed during each trip was assumed to represent the amount of energy replenished through charging. For in-route charging, the trip energy was evenly distributed throughout the trip, calculated based on the trip distance and vehicle energy efficiency. For charging during car and LDT dwell periods, the energy was replenished up to the amount consumed during the trip.

The charging schedules for MDVs and HDVs were determined using the National Renewable Energy Laboratory’s Fleet DNA database, which contains operating data for commercial fleet vehicles⁴⁹. The MDV schedule was developed from 1471 trips made by class 3 to 7 delivery trucks and vans, while the HDV schedule was based on 969 trips made by class 8 tractors. The trip data was categorized into each vehicle operating range and weighted by VKT from Table 1; trip data for over 322 VKT (200 vehicle miles travelled) was used for all operating ranges above 322 km. It was assumed that the VKT for each trip was evenly distributed between the arrival and departure times, resulting in a distribution of the in-route charging profile throughout the day. The weighted average charging schedules for MDVs and HDVs are illustrated in Supplementary Figs. 6 and 7.

Deployment of infrastructure

Deployment scenarios were developed for DCFC, BSS, and DWPT to assess the charging cost and GHG-intensity of individual charging locations across the U.S. The estimated usage of each charging system was scaled using yearly traffic data and EV adoption projections. Vehicle traffic data from the Freight Analysis Framework Version 4 (FAF4) provided VKT estimates for 2012 and 2045 on 663k individual roadways for freight trucks (MDV and HDV) and all vehicles¹⁹. These estimates were interpolated and extrapolated linearly up to 2050.

To break down the FAF4 data by vehicle category, multiple datasets from the FHWA were utilized, incorporating the 2019 FHWA VKT data and their projected increase in 2049²⁴. The FHWA VKT data included breakdowns for cars, LDTs, single-unit trucks (MDVs), combination trucks (HDVs), motorcycles, and buses^39,50. The FHWA VKT data were used to calculate the portion of vehicles in the FAF4 data that fell into the categories of cars, LDTs, MDVs, and HDVs based on the percentage of FHWA VKT from each vehicle category on state roadways, including interstates (FAF4 interstates), other arterials (FAF4 freeways, principal arterials, and minor arterials), and other road types (FAF4 major collector and minor collector)³⁹. It is worth noting that the FAF4 data did not include VKT on local roads. As a result, the FAF4 VKT data for stationary charging systems (DCFC and BSS) were scaled to match the total VKT of the FHWA data for each vehicle category in 2019 and 2049. In contrast, the FAF4 data for DWPT were not scaled to match the FHWA VKT total, as the VKT on individual roads was directly used in the analysis.

The VKT data, categorized by vehicle type on each roadway segment, were combined with EV adoption forecasts (Supplementary Figs. 1 and 2), a charging efficiency of 85% for each system^38,46, and vehicle energy efficiencies to estimate the charging demand from EVs. To account for uncertainty, three EV adoption scenarios were considered: optimistic, baseline, and conservative. The optimistic adoption curves for MDVs and HDVs were derived from Konstantinou and Gkritza (2023)⁴, while the conservative and optimistic scenarios for cars, LDTs, and buses, as well as the conservative and baseline scenarios for MDVs and HDVs, were obtained from Mai et al.²⁰. The baseline scenario for cars, LDTs, and buses represented the average of the conservative and optimistic EV adoption rates.

In summary, the energy demand for public EV charging was computed on major roadways in the contiguous U.S. from 2031 to 2050. The energy demand included hourly and yearly resolution for cars, LDTs, MDVs, HDVs, and buses.

The yearly energy demand on the roadways from each EV adoption scenario was used to allocate EV charging to suitable charging sites. A total of 122k potential charging site locations were identified, including 85k gas stations⁵¹, 30k public surface parking lots⁵¹, and 7k existing DCFC sites³. The suitability of the sites for high-power charging stations was evaluated based on their proximity to grid interconnections and minimum EV charging utilization. Since load growth from widescale EV adoption was expected to require new substations²³, the location of grid interconnections was modeled as transmission lines with voltages under 200-kV rather than existing substations⁵². Site locations within 6-km of grid interconnections⁵³, based on the 95th percentile of existing DCFC sites, were deemed to be within the maximum proximity to grid interconnections. Further, sites within 1.5-km were not restricted on their maximum power due to the allowances of line extension policy⁵⁴. Sites with power limitations were restricted to a maximum power of 2.5 megawatts⁵⁵.

DCFC sites without power limitations were restricted to a maximum daily energy dispensed of 30% of the time⁵⁶ for 32 stations with space constraints, as observed. The DCFC stations were set to use either 150-kW or 350-kW chargers as observed from major charging networks⁵⁷. In contrast, BSS energy dispensation was constrained by a 3-min swap time¹⁰, limiting the maximum number of swaps during peak demand hours to prevent queuing. Each BSS site was designed to have two sizes of swapping stations: a small size for cars and LDTs, and a large size for MDVs and HDVs.

The maximum capacity (m1) of each site (j) was used in a gravity model (Eq. (4)) to allocate the yearly demand for EV charging on each roadway segment to the nearest 30 sites⁵⁸. The allocation of EV charging (F) to each site was also based on the amount of charging demand on the roadways (m2), the distance between the charging site and roadway (d), and a scalar (g). The scalar g was computed in Eq. (3) to ensure that the sum of F was equal to m2 for each roadway segment.

The allocation of EV charging to each site was then corrected to ensure that the maximum capacity of the site was not exceeded, and sites with very low usage were removed to avoid poor economics. The minimum allowed energy allocated to a DCFC site was set such that one 150-kW charger would dispense energy at least 5% of the time based on today’s conditions⁵⁷, which represented the minimum usage threshold for the highest demand year. Similarly, each BSS site was designed to have a total energy demand of at least 4 swaps per day for both sizes of swapping stations during the highest demand year. Equations (3–4) were then used to allocate the energy for the remaining viable locations.

The required amount of charging equipment at each site was then determined based on its expected usage. For DCFC sites, the number of chargers needed was calculated separately for the first (2031–2040) and second (2041–2050) 10-year life of the equipment³⁷. DCFC sites with low expected usage were equipped with 150-kW chargers, whereas those with high expected usage were equipped with 350-kW chargers. Specifically, 150-kW chargers were only deployed at sites where the highest usage was below the maximum capacity of a single 350-kW charger during the initial 10-year period, and also below the maximum combined spatial capacity of 32 150-kW chargers over the full 20-year period. Alternatively, for BSS sites, the number of batteries and support equipment needed was determined annually, with a minimum usage of 4 batteries per site. Supplementary Fig. 3 shows the coverage of DCFC and BSS infrastructure within 80 km (50 miles¹) for each EV adoption scenario.

DWPT infrastructure was deployed on major roadways in the contiguous U.S. to maintain every vehicle’s state of charge. The energy dispensed from the DWPT roadway was set to match the vehicle energy consumption on each roadway. If over half of vehicle traffic saturates the DWPT lane, a second lane is assumed to be electrified in each direction, reducing the utilization by half for the analyzed lane.

Techno-economic analysis

The TEA conducted an evaluation of the charging cost and TCO for EVs using DCFC, BSS, and DWPT charging systems. In this study, only public charging costs from these systems were considered for the TCO comparison, although the actual TCO would include a mix of charging costs from home, workplace, fleet, public, and other locations. To capture the full range of values, optimistic, baseline, and conservative scenarios were developed for capital costs, electricity prices (Supplementary Fig. 4), and EV adoption (Supplementary Figs. 1 and 2), resulting in 27 charging cost and TCO scenarios for EVs. In addition, optimistic, baseline, and conservative scenarios were developed for traditional fuel prices (Supplementary Fig. 5) to evaluate refueling costs for ICEVs and HEVs.

The charging cost for each DCFC site, BSS site, and DWPT roadway segment was evaluated individually using a discounted cash flow rate of return (DCFROR). The DCFROR considered capital costs, operational costs, electricity costs, and utilization. The DCFROR assumed a 5% internal rate of return, capital debt financing of 50% with 6% interest and 10-year loan term, state and average local sales tax (Supplementary Table 1)⁵⁹, corporate federal (21%), and state income tax (Supplementary Table 1)⁶⁰, and a modified accelerated cost recovery system depreciation schedule. The cash flow spanned 21 years, including a 1-year build period and a 20-year operating life (2031 to 2050). The charging cost was calculated such that a net present value of zero was achieved. All costs were converted to 2022 USD using consumer price indexes^61,62 and producer price indexes^63,64,65,66. A summary of the costs for each charging system and vehicle is presented in Table 2.

The capital costs for DCFC, BSS, and DWPT were evaluated individually and listed in Table 2. For every system, it was assumed that utilities would cover the cost of substations and line-extensions up to a certain distance⁵⁴, with expenses being recouped through electricity sales. As noted in Nelder and Rogers (2019), however, certain utilities might have imposed line-extension fees⁵⁷. Further, this study assumed that the land of each site was owned already and did not depreciate. Therefore, capital costs were limited to the installation and procurement of all necessary components of each charging system.

The capital costs for DCFC were calculated by scaling the costs (Table 2) with the number of chargers at the site. These costs included a procurement component that was scaled linearly and an installation component that decreased on a per charger basis as the number increased. The procurement cost was incurred in 2030 and 2040, corresponding to the number of chargers deployed during each respective 10-year period. The installation costs, on the other hand, were incurred upfront in 2030 to future-proof the charging system for both sets of charger lifespans³⁷.

In contrast, the capital costs for BSS (Table 2) included fixed costs for the small (19 square meters) and large (46 square meters) sizes of BSSs, which covered the automated storage and retrieval system required to swap batteries and the building housing the system. The number of cabinets, comprising containers, thermal management systems, and fire suppression systems, were determined based on the number of batteries needed to meet the annual demand. Furthermore, the number of chargers for each BSS was calculated based on the maximum charging load from 2031 to 2040 and from 2041 to 2050, utilizing 7.7-kW chargers for the small BSS and 50-kW chargers for the large BSS. Similar to DCFC, the procurement costs for the chargers were incurred in 2030 and 2040 for the first and second set of chargers required, respectively. The installation, automated storage and retrieval system, and building costs were incurred in 2030, while the battery and cabinet costs were incurred in the respective years when they were added to the BSS.

For DWPT, the capital cost (Table 2) was scaled according to the electrified distance of each roadway. Separate estimates were utilized for urban and rural roads, taking into account the substantial difference in civil costs between the two⁶⁷. The electronics cost was assumed to be the same for both urban and rural roads. For the optimistic scenarios, the low estimate from Limb et al. (2019) was used for rural roads, while the high estimate was used for urban roads¹¹. As for the baseline and conservative scenarios, the electronics cost of 1.6 million USD per km (adjusted to 2022 USD⁶³), as reported by Haddad et al. (2022), was employed²³. The conservative civil cost for urban roads was derived from the 1st-of-a-kind case in Trinko et al. (2022), with a lower contingency cost of 10% compared to the original 30%¹². The baseline urban civil cost was also adapted from Trinko et al. (2022), incorporating a combination of the 1st-of-kind and nth-of-a-kind cases, which is further detailed in Supplementary Table 2¹². Regarding the civil costs for rural roads, the nth-of-kind case from Trinko et al. (2022) was utilized for the baseline scenario, while the estimate from Haddad et al. (2022) was adapted (without substation) for the conservative scenario²³.

The DCFC sites were modeled to have data contracts, network contracts, and maintenance costs annually (Table 2). BSS sites were modeled to only have maintenance costs (Table 2) on the chargers. The replacement of BSS batteries was assumed to be the responsibility of the vehicle owner. The operational costs for DWPT consisted of replacing failed roadside inverters (Table 2) with a mean time to failure of 101 years⁶⁸. Since the failed DWPT roadway pads were modeled to have excess capacity in the design (Eq. (2)), they were not replaced. The maintenance costs of the roadway were assumed to be out of scope since they are typically paid for by taxes, which are not part of the DCFC and BSS analysis.

Commercial electricity schedules from the U.S. Utility Rate Database⁶⁹ were collected⁷⁰ for the largest utility company in each state to determine electricity costs for each DCFC site, BSS site, and DWPT roadway segment on an annual basis. The electricity schedules were categorized into demand charges (USD/kW-month), electricity rates (USD/kWh), and fixed charges (USD/month), with applicable demand charges and electricity rates determined by the time-of-day charging profiles (Supplementary Figs. 6 and 7). The fixed charges were assessed to each BSS and DCFC site as well as to each DWPT roadway segment per 16 lane-km of electrified road. The most affordable schedule was selected based on the service location and power range for each load, with BSS charging profiles optimized to minimize electricity costs.

The BSS charging profile was optimized to ensure that each battery could be fully charged prior to the swap, with a one-hour buffer period. The charging time (t) required to charge the battery of each vehicle (v) was calculated using Eq. (5), which accounts for the charger rating (p) (7.7 kW for small BSS and 50 kW for large BSS), average power rate (a) (95%), charging efficiency (e) of 85%³⁸, starting state of charge (soc_s) (20%)⁴⁵, final state of charge (soc_f) (80%)⁴⁵, and battery size (b) (Table 2).

The minimum number of batteries needed was determined by considering the required charging time and the swap schedule of batteries within the BSS (Supplementary Fig. 6). These constraints were integrated into the charging load optimization algorithm, which aimed to minimize electricity costs while ensuring that each battery was charged within the designated window and that the daily charging volume met the demand from EVs.

Once the electricity costs were calculated for each charging system, the electricity costs were adjusted using the 2022 and 2031 to 2050 price projections for generation (electricity rate) and distribution (demand charge) from the Annual Energy Outlook (2023)⁷¹. Three scenarios were considered to account for future changes in electricity prices (Supplementary Fig. 4): optimistic, baseline, and conservative.

The TCO analysis utilized the charging cost results for DCFC, BSS, and DWPT to estimate the charging cost for EVs per VKT. The TCO of each modeled EV was compared to that of a HEV and an ICEV. The fuel prices for HEVs and ICEVs were broken out by state and adjusted to 2031 through 2050 values (Supplementary Fig. 5) for optimistic, baseline, and conservative scenarios⁷¹.

The TCO was computed over the first 10 years of the vehicle’s life and included the charging or fueling cost (Supplementary Table 3), depreciation (Table 2), maintenance (Table 2), license and registration (Supplementary Table 4), and insurance (Supplementary Table 5) expenses. The analysis of ICEVs and HEVs considered gasoline fuel for cars and LDTs, and diesel fuel for MDVs and HDVs. Cars and LDTs were assumed to have a discount factor of 1.2% for yearly expenses, while MDVs and HDVs had a discount factor of 3%²¹.

The purchase price (Table 2) for each vehicle category was determined based on an average vehicle. For EVs, the purchase price included the cost of the EV without the battery and the marked-up cost of the EV battery (Table 2). The price of an EV without the battery was obtained from Burnham et al. (2021) for a 2025 model year midsize sedan (car), pickup truck (LDT), class 6 pickup/delivery truck (MDV), and sleeper tractor (HDV)²¹. The battery size for electric LDTs was adjusted to match the range of an electric car by considering vehicle efficiencies⁷². Reduced battery sizes for DWPT EVs were determined for a 56 km operating range with a maximum depth of discharge of 80%.

The vehicle depreciation cost was calculated annually based on the purchase price. Cars and LDTs lost 29% of their original value in the first year and 11% of their remaining value in each consecutive year²¹. For MDVs and HDVs, 9% of their remaining value was lost every year²¹. Insurance costs were assessed based on the remaining value of the vehicle each year and the location of the charging system. License and registration costs were fixed annually and varied by state. Maintenance costs were based on the U.S. average fixed rate per VKT plus any battery replacement costs. Battery life was assumed to be 1000 full cycles for full-size batteries⁷³, while reduced battery sizes charged via DWPT were assumed to have the same life in years due to optimal operating characteristics, such as a smaller depth of discharge and a state of charge that can be maintained around 50%^74,75,76. Based on these assumptions, only electric HDVs needed battery replacements in the first 10-year period due to their high annual VKT⁷⁷. The sum of the vehicle costs was discounted along with their yearly utilization to obtain the TCO on a per VKT basis.

The impact of switching from ICEVs to EVs charged with DCFC, BSS, or DWPT systems (cs) to the overall cost of on-road transportation was calculated using Eq. (6) for the percentage change (∆TCO%) and Eq. (7) for the change in USD (∆TCO$). These equations used the TCO of an ICEV (TCO_ICEV) and an EV (TCO_EV) for each vehicle category (v), as well as the VKT of EVs (VKT_EV) and the VKT of all vehicle powertrains (VKT_All).

Life cycle assessment

An attributional LCA was conducted to compare the GHG-intensity of EVs charged with DCFC, BSS, and DWPT, as well as ICEVs and HEVs. Specifically, the impact assessment used an economic allocation method and the 100-year global warming potential from the Intergovernmental Panel on Climate Change’s (IPCC) 6th impact assessment report⁷⁸. The study used a cradle-to-grave system boundary with a functional unit of one VKT. For EVs, the emissions were divided into charging emissions, embodied charging infrastructure emissions, and embodied vehicle emissions for different vehicle categories: cars, LDTs, MDVs, and HDVs. Charging and infrastructure emissions were allocated on a per unit of energy dispensed basis (kWh). The study used Ecoinvent 3.8 and openLCA 3.10 to collect the life cycle inventory data for charging infrastructure and charging emissions⁷⁹. The Greenhouse gases, Regulated Emissions, and Energy use in Technologies model (GREET) 2022 was used to determine embodied vehicle emissions, as well as the HEV and ICEV feedstock, fuel, and vehicle operation emissions³⁸. The feedstock and fuel emissions were combined as the equivalent infrastructure emissions for HEVs and ICEVs.

The DCFC infrastructure emissions were calculated based on the charger pedestal, power cabinet, implementation, and construction (Supplementary Table 6)¹⁶. The pedestal inventory data were taken from Ecoinvent 3.8 and scaled to a weight of 250-kg for 150-kW and 350-kW chargers^79,80. Emissions from the power cabinet were broken out by material (Supplementary Table 7) using primary data for a weight of 1340-kg per 150-kW charger and 2680-kg per 350-kW charger⁸⁰. Implementation emissions for DCFC were adapted from Lucas et al.⁸¹.

The BSS infrastructure emissions included the charger pedestal, battery, battery cabinet, automated storage and retrieval system, building, and construction (Supplementary Table 8). The emissions of each BSS site were scaled based on the amount of equipment used.

The DWPT infrastructure emissions were based on the electronics, pavement, and construction (Supplementary Table 9). DWPT infrastructure components were based on Marmiroli et al. (2019), however, the emissions factors were adjusted to use concrete rather than asphalt and use the IPCC 6th impact assessment¹⁷. The emissions for DWPT were scaled based on the electrified distance.

The emissions from EV charging were calculated using the forecasted hourly electricity mix in 134 Cambium (2022) zones from 2031 through 2050⁸². Three electricity mix scenarios were examined: optimistic, based on the Cambium (2022) mid-case with 100% decarbonization by 2035; baseline, based on the Cambium (2022) mid-case; and conservative, based on the Cambium (2022) high renewable energy cost. The charging emissions were calculated using the average electricity consumption mix from each zone, rather than the marginal mix⁸³.

The consumption mix was determined by tracking the net energy imports and exports from each zone within the Western, Eastern, and Texas U.S. interconnections. Furthermore, the electricity mix used for charging energy storage resources was accounted for and assigned to the mix at the time of EV charging. Emissions factors for various grid resources in North American Reliability Corporation regions were obtained from Ecoinvent 3.8, encompassing both operating and embodied emissions (Supplementary Table 10)⁷⁹. Notably, carbon capture associated with the electricity grid was not attributed to EV charging as it was beyond the system boundary.

The charging emissions (ChgGHG) were computed on a per unit of energy basis (kWh) using Eq. (8). The calculation was based on each grid resource’s (r) emissions factor (EF) and fraction of the consumption mix (M) at the time-of-day (h), year (y), and location (z) of charging.

The charging emissions were then scaled (Eq. (9)) by the hourly and yearly charging load (load) from each DCFC site, BSS site, and DWPT road segment to get the average charging emissions per unit of energy (kWh) over the life of each system.

Embodied vehicle emissions for each vehicle type were calculated using a representative 2025 vehicle with conventional materials from GREET (2022)³⁸. Specifically, the vehicle modeled for each vehicle category from GREET (2022) were a passenger car, pickup truck (LDT), class 6 pickup-and-delivery truck (MDV), and class 8 sleeper-cab truck (HDV).

The vehicle emissions were divided into components; assembly, disposal, and recycling (ADR); batteries; and fluids. EV battery sizes (Table 2) were input into GREET (2022) for the corresponding charging system and EV-adoption scenario³⁸. The batteries for both EVs and HEVs were assumed to be manufactured in China and use a lithium-ion chemistry. One replacement of the hybrid electric HDV battery was assumed to occur over the vehicle’s lifetime. Further, electric MDVs and HDVs were calculated to average 1.7 and 2.9 battery replacements in their lifetime, while cars and LDTs had none. The modeled vehicle emissions are summarized in Table 3.