Electric vehicle fleet penetration helps address inequalities in air quality and improves environmental justice

The study addresses persistent inequities in air pollution exposure in the United States, where communities with higher proportions of people of color and lower incomes disproportionately live near major roads and experience elevated levels of traffic-related air pollutants (TRAPs). Despite overall declines in on-road emissions, exposure disparities remain, particularly for NO2 and PM2.5 near highways. Electric vehicle (EV) adoption is a key strategy to reduce on-road emissions and greenhouse gases, but its distributional air quality benefits in near-road, environmental justice (EJ) communities have not been well quantified. This research tests the hypothesis that increasing EV penetration will yield greater incremental air quality improvements for EJ communities adjacent to major roadways compared with the general population. Focusing on southern California communities along Interstate 710, the study evaluates six 2040 EV penetration scenarios and quantifies changes in NOx/NO2 and primary PM2.5 at census block group resolution, differentiating outcomes by EJ indicators (race/ethnicity, white population percentage, and educational attainment).

Prior work shows U.S. air quality management has not produced equitable outcomes, with people of color and low-income communities disproportionately exposed to pollutants and often residing near major roadways due to historic discriminatory policies. Near-road TRAP concentrations can be two to four times higher than background levels. While NO2 and PM2.5 levels have declined, disparities persist; for example, NO2 exposure in 2010 remained higher for non-White communities despite larger declines from 2000 to 2010. Numerous studies in the U.S., Europe, and Asia indicate that EV penetration reduces regional PM2.5, ozone (O3), and NOx concentrations, although O3 increases have been modeled in some VOC-limited areas due to reduced NOx titration. Skipper et al. found EV benefits scale approximately linearly, with reduced O3 disbenefits by 2028 compared to 2016. Schnell et al. reported O3 generally decreases with EV adoption regardless of electricity source, with exceptions in specific conditions; PM2.5 changes are more dependent on power generation mix. However, most prior analyses are regional in scope and do not resolve near-road settings where primary vehicular pollutants drive exposure disparities. Policy context includes U.S. federal goals for 40–50% EV sales by 2030 and California’s ACC, ACT, and ACF regulations to accelerate LDV and MHDV electrification. This study fills a gap by quantifying EJ-relevant, near-road concentration changes from EV penetration.

Study area and scenarios: The analysis focuses on southern California communities adjacent to Interstate 710, a major goods movement corridor serving the Ports of Los Angeles and Long Beach. Six calendar year (CY) 2040 EV penetration scenarios were evaluated: a Reference case (no additional EV policies) and five alternative futures, including policy cases and idealized bounding cases. Policy cases: Scenario 1 (High emissions reduction) combined CARB policies and cost parity assumptions, increasing LDV EV share to 19% and MHDV EV share to 30.1% (fleetwide 17.5%). Scenario 2 (Medium emissions reduction) assumed gasoline price increases ($0.07/gal/year) affecting LDVs minimally (10.7% EV) and ACT for MHDVs (27.8% EV). Scenario 3 (MHDV-focused) assumed ACT and ACF for MHDVs with no LDV changes (fleetwide 10.8%). Idealized bounding cases: Scenario 4 implemented ACC II for LDVs (100% ZEV/PHEV sales by 2035; modeled as ZEV-only for analyses) yielding 69.7% LDV EV share, and accelerated MHDV electrification based on Raju et al., achieving 56.4% MHDV EV share by 2040. Scenario 5 assumed 100% EV sales beginning 2023, reaching 89% LDV, 76.7% MHDV, and 85% fleetwide EV share by 2040. EV population modeling: A Python tool estimated future EV populations using model-year-specific market shares and EMFAC baseline populations. LDV EV shares in Scenarios 1–2 used ORNL MA3T model outputs; MHDV shares used CARB ACT/ACF phase-in schedules. For idealized cases, accelerated phase-ins were assumed. Emissions modeling: EMFAC2021 provided link-level emissions factors and activity for the South Coast region. PM2.5 included exhaust, tire wear, and brake wear; resuspended road dust was excluded and assumed unaffected by EV penetration. EMFAC assumes EV brake wear is 50% of ICE vehicles due to regenerative braking. Air quality modeling: Total concentrations at census block group centroids were the sum of regional background and on-road contributions. Regional background was estimated from 2000–2020 monitored data (EPA AQS), projected to 2040 (e.g., ~4%/yr NOx decline), and spatially interpolated via inverse distance weighting, excluding near-road sites. On-road contributions were modeled with EPA’s R-LINE line-source dispersion model (with NOx chemistry), driven by spatially resolved meteorology (e.g., RTMA 2.5-km inputs processed via AERMET/AERMOD tools). Approximately 113,000 roadway segments were grouped into 5×5 km grids to reduce computational cost, accounting for contributions from adjacent grids and buffers. EJ analysis: CalEnviroScreen parameters (race/ethnicity, white population percentage, education level, composite EJ score) were linked to census block groups. Concentrations and reductions were summarized by EJ groupings, including analysis of maximum disparities (difference between groups with highest and lowest mean concentrations). Statistical tests (ANOVA, post-hoc Tukey-Kramer) evaluated differences between groups. Health-related metric: The attributable fraction (AF) for avoided mortality was calculated using a log-linear concentration-response function: AF = 1 − exp(−βΔX), using RR of 1.04 per 5.32 ppb NO2 and 1.03 per 5 μg m−3 PM2.5 to derive β. ΔX was the concentration change between Reference and scenario for each group.

- Emissions: By 2040, fleetwide NOx emissions reductions were 27% (Scenario 1), 14% (Scenario 2), 26% (Scenario 3), 44% (Scenario 4), and 62% (Scenario 5). PM2.5 emissions reductions were 8%, 4%, 4%, 29%, and 40%, respectively. MHDVs dominated NOx emissions; thus, NOx reductions largely tracked MHDV electrification. LDVs dominated PM2.5 (non-exhaust), limiting reductions in policy cases with modest LDV EV gains. - Concentrations (domain-wide means at census block group centroids): Reference case on-road and background components were 0.51 ± 1.12 ppb and 8.08 ± 9.89 ppb (NO2), and 1.88 ± 3.61 μg m−3 and 12.49 ± 7.93 μg m−3 (PM2.5). Total concentration reductions versus Reference ranged from −0.40 to −1.75 ppb (−4.7% to −20.4%) for NO2 and −0.08 to −0.98 μg m−3 (−0.6% to −6.8%) for PM2.5 across Scenarios 2–5. Larger reductions occurred near major roadways; in Scenario 5, maximum near-road reductions exceeded 3 ppb (NO2) and 2 μg m−3 (PM2.5). - EJ disparities in Reference: Communities with higher proportions of non-White residents and lower educational attainment had higher NO2 and PM2.5 concentrations. Example: communities with more Latino members had 12% higher NO2 than those with more White members (9.2 vs. 8.2 ppb); communities with more Asian members had 8% higher PM2.5 than those with more White members (15.2 vs. 14.2 μg m−3). - Differential benefits with EV penetration: Under aggressive electrification (Scenario 5; ~85% fleet EV), mean NO2 reductions were larger in communities with more people of color than in those with more White residents: 1.9 vs. 1.6 ppb. PM2.5 reductions showed the same pattern: 1.1 vs. 0.94 μg m−3. Differences between racial groups in reductions were statistically significant for all scenarios (ANOVA p < 0.05; Tukey-Kramer: reductions in communities with more White members significantly smaller than other groups). - Disparity reductions: Maximum disparity (difference between highest- and lowest-exposed EJ groups) decreased with higher EV penetration. For NO2, disparity decreased 30% in Scenario 5 (from 0.95 to 0.67 ppb) and 22% in Scenario 4. For PM2.5, disparity decreased 14% in Scenario 5 (from 1.06 to 0.91 μg m−3) and 10% in Scenario 4. - Health-related implications: Using the attributable fraction metric, avoided mortality benefits (relative) were 19% (White vs. Latino) and 16% (White vs. Asian) higher for NO2 and PM2.5 in communities with more people of color than in those with more White members under Scenario 5. - Policy salience: Truck-focused policies (ACT/ACF) drove large NOx reductions and disparity declines; LDV-focused policies are critical for PM2.5 via reduced brake wear due to regenerative braking, though overall PM2.5 benefits are constrained by persistent non-exhaust and background contributions.

The findings confirm the hypothesis that accelerating EV penetration reduces air pollutant exposures for all communities while conferring greater incremental benefits to EJ communities near major roads. By reducing primary on-road emissions, particularly NOx from MHDVs, electrification lowers near-road NO2 hotspots responsible for disproportionate exposures. The evidence that disparity declines scale with EV penetration underscores the importance of aggressive policies. Truck-focused electrification (ACT/ACF) is pivotal for NOx and NO2 disparity reductions because MHDVs dominate NOx emissions, especially along freight corridors like I-710. For PM2.5, benefits are more modest due to the dominance of non-exhaust and background sources by 2040, though LDV electrification still reduces brake wear emissions via regenerative braking. Early action amplifies benefits by accelerating turnover of high-emitting older vehicles. Given that a substantial share of the U.S. population lives near major roads, and near-road pollution gradients are observed globally, these results have broad relevance domestically and internationally for environmental justice policy and planning.

Using a chain of fleet, emissions, and dispersion modeling at census block group resolution, the study demonstrates that EV fleet penetration by 2040 substantially reduces NOx/NO2 and yields modest PM2.5 reductions, with disproportionately larger benefits for EJ communities near major roads. Under an aggressive scenario (~85% EV fleet share), NO2 and PM2.5 reductions are greater in communities with more people of color than in communities with more White residents, and exposure disparities decline by 30% (NO2) and 14% (PM2.5). A policy-aligned scenario reflecting ACC II and accelerated MHDV electrification also produces meaningful disparity reductions (22% for NO2; 10% for PM2.5). The results support policies that rapidly electrify both LDVs and especially MHDVs, emphasize early implementation, and incorporate EJ considerations into transportation and air quality planning. Future research should refine background projections, quantify evolving non-exhaust emissions, improve exposure modeling, and assess interactions with power sector decarbonization to fully characterize air quality and health equity outcomes.

- Scope of pollutants: The study focused on primary NOx/NO2 and PM2.5 and did not simulate photochemical formation of ozone or secondary PM2.5; potential secondary pollutant disbenefits/benefits were not assessed. - Power generation: Changes in emissions from electricity generation due to EV charging were not included, based on California’s planned phase-out of fossil-based generation by 2045; regional impacts from power plants were outside the near-road scope. - PM2.5 components: Road dust resuspension was excluded and assumed unchanged with EV penetration; EV and ICE vehicle weight differences and their impact on non-exhaust emissions were simplified, with an assumption of reduced EV brake wear (50%) but no change in road dust. - Background projections: Regional background concentrations were projected to 2040 using historical trends and may not capture future regulatory changes or technology shifts affecting background levels. - Modeling uncertainties: R-LINE dispersion and EMFAC emissions carry inherent uncertainties; grid simplifications and interpolation may introduce spatial errors. Regional on-road contributions were not subtracted from background to avoid nonlinearity issues, potentially adding small uncertainties. - Health assessment: The health analysis used an attributable fraction metric with literature-derived concentration-response functions and did not constitute a full health impact assessment accounting for demographics, baseline rates, or morbidity outcomes. - Generalizability: Results are specific to the South Coast/I-710 context; while mechanisms are general, absolute magnitudes may differ in other regions with different fleets, meteorology, or backgrounds.