Trajectory-Integrated Accessibility Analysis of Public Electric Vehicle Charging Stations

TL;DR

This work introduces TI-acs, a novel trajectory-informed metric that measures public EV charging accessibility by integrating individual mobility with the spatial distribution of charging ports. TI-acs distinguishes accessibility near homes, workplaces, and other activity sites, and aligns with time-of-use grid windows, enabling analysis of potential grid-friendly charging. Applied to the San Francisco Bay Area using TimeGeo mobility data and AFDC charger data across 2012–2024, the study reveals substantial improvements in access but persistent spatial and racial disparities, with Hispanic-dominated communities experiencing substantially lower TI-acs than White-dominated ones. The framework provides actionable insights for equitable infrastructure deployment and can be extended to evaluate accessibility to other urban services, balancing electrification goals with social equity and grid considerations.

Abstract

Electric vehicle (EV) charging infrastructure is crucial for advancing EV adoption, managing charging loads, and ensuring equitable transportation electrification. However, there remains a notable gap in comprehensive accessibility metrics that integrate the mobility of the users. This study introduces a novel accessibility metric, termed Trajectory-Integrated Public EVCS Accessibility (TI-acs), and uses it to assess public electric vehicle charging station (EVCS) accessibility for approximately 6 million residents in the San Francisco Bay Area based on detailed individual trajectory data in one week. Unlike conventional home-based metrics, TI-acs incorporates the accessibility of EVCS along individuals' travel trajectories, bringing insights on more public charging contexts, including public charging near workplaces and charging during grid off-peak periods. As of June 2024, given the current public EVCS network, Bay Area residents have, on average, 7.5 hours and 5.2 hours of access per day during which their stay locations are within 1 km (i.e. 10-12 min walking) of a public L2 and DCFC charging port, respectively. Over the past decade, TI-acs has steadily increased from the rapid expansion of the EV market and charging infrastructure. However, spatial disparities remain significant, as reflected in Gini indices of 0.38 (L2) and 0.44 (DCFC) across census tracts. Additionally, our analysis reveals racial disparities in TI-acs, driven not only by variations in charging infrastructure near residential areas but also by differences in their mobility patterns.

Figures (8)

• Figure 1: Graphic overview of the work.a Our research (top) integrates two data layers: one (bottom) is the geographical distribution of public charging resources, the other (middle) is week-long individual trajectories. b An illustrative example of TI-acs calculation. c Workflow of processing data, calculating TI-acs, and analyzing aggregated TI-acs. Yellow boxes are data types, with gray text below them annotating specific data sources.
• Figure 2: Average TI-acs (within 1 km) across Bay area census tracts.a,b plot the census-tract-level averaged TI-acs to L2 and DCFC public chargers as of June 2024, respectively. Each census tract is colored by the average trajectory-integrated accessible hours (TI-acs) of individuals whose home location is within the area. Means, Medians and 1/4, 3/4-quantiles of census-tract-level statistics are annotated. c, d compare the growth of total installed charging ports (blue dashed lines) and the improvement of TI-acs (in terms of accessible hours within 1 km range) from 2012 to 2024. Refer to the left $y$-axis for values of TI-acs, and right $y$-axis for values of total installed ports. A 1-km distance typically means 10 $\sim$ 15 minutes walk.
• Figure 3: TI-acs breakdown.a,b plot the breakdown of TI-acs (accessible hours) into different stay location segments: home, work, other by different years from 2012 to 2024, for L2 and DCFC respectively. Height of bars represent the means of census-tract-averaged metric, and the darker error bars annotate quartiles (25%, 50%, 75%) of the metric (with $0$ from the bottom of corresponding bars). Subplots (a1), (b1) are density of accessible hours in different segments across all census tracts in year 2024. c, d plot the annual change of Gini index of TI-acs, as well as its breakdown into different location segments from 2012 to 2024, for L2 and DCFC respectively. The higher Gini index is, the more uneven charger accessibility is for people reside in different census tracts. e, f plot the breakdown of TI-acs into different time segments, defined by power grid TOU periods, by different years from 2012 to 2024, for L2 and DCFC respectively, with subplots (e1), (f1) visualizing the density in year 2024. The meanings of different plot elements are the same as they are in a, b. The total duration of different TOU periods in a day are: super off-peak: 5 hr, off-peak: 14 hr, peak: 5 hr, as indicated by the light yellow shades in (e1), (f1).
• Figure 4: Racial disparities on TI-acsa, b plot the cumulative density function (CDF) of census-tract-level-averaged TI-acs (accessible hours) in year 2024 grouped by the dominate race (ethnic identity) of the census tracts, for public L2 and DCFC, respectively. Horizontal bars and markers at the top mark the quartiles (25%, 50%, 75%) of TI-acs for different groups, sharing the same $x$-axis with the main plots. Subplots (a1), (b1) plot the CDF of accessible hours in different stay location segments in year 2024. c, d visualize the coefficients $\beta_r$ from the regression: $y_i = \sum_r \beta_r \delta_{ir} + \beta_I \log I_i + \beta_0 + \epsilon_i$. For census tract $i$, $y_i$ is its accessibility metric, $\delta_{ir} = 1$ if it is in group $r$ (white, black, Asian, Hispanic), otherwise $0$, and $I_i$ is its median household income. $\beta_r$ (unit: hours) means the difference in daily accessible hours between people residing in the census tracts with a given dominate race and people in census tracts with no dominate race. We ran the regression for each year in 2012 to 2024 separately. Markers are the ordinary least square (OLS) estimate of $\beta_r$'s, and the error bars visualize the confident interval (CI) of the corresponding estimate. A solid marker indicates that the coefficient is significantly ($p<.05$, double side) greater/smaller than 0. For all results plotted, we only include census tracts which more than half of its households live in 2-unit (or more) structures.
• Figure 5: TI-acs [hours] breakdown by different distance thresholdsSubplot titles specify the accessible radius. $x$-axis marks EVCS snapshots in different years. top row: L2, bottom row: DCFC. left half: breakdown by stay location types. right half: breakdown by time in the day. Please refer to Fig.\ref{['fig:breakdown']} for detailed legends.