Learning to Mitigate Post-Outage Load Surges: A Data-Driven Framework for Electrifying and Decarbonizing Grids

Abstract

Electrification and decarbonization are transforming power system demand and recovery dynamics, yet their implications for post-outage load surges remain poorly understood. Here we analyze a metropolitan-scale heterogeneous dataset for Indianapolis comprising 30,046 feeder-level outages between 2020 and 2024, linked to smart meters and submetering, to quantify the causal impact of electric vehicles (EVs), heat pumps (HPs) and distributed energy resources (DERs) on restoration surges. Statistical analysis and causal forest inference demonstrate that rising penetrations of all three assets significantly increase surge ratios, with effects strongly modulated by restoration timing, outage duration and weather conditions. We develop a component-aware multi-task Transformer estimator that disaggregates EV, HP and DER contributions, and apply it to project historical outages under counterfactual 2035 adoption pathways. In a policy-aligned pathway, evening restorations emerge as the binding reliability constraint, with exceedance probabilities of 0.057 when 30\% of system load is restored within the first 15 minutes. Mitigation measures, probabilistic EV restarts, short thermostat offsets and accelerated DER reconnection, reduce exceedance to 0.019 and eliminate it entirely when 20\% or less of system load is restored. These results demonstrate that transition-era surges are asset-driven and causally linked to electrification and decarbonization, but can be effectively managed through integrated operational strategies.

Figures (7)

• Figure 1: Statistical analysis of post-outage surge ratios under EV, HP, and DER penetrations. (a) EV surge characterization: left, hexbin density of outage events with median (green) and 95th percentile (blue dashed) surge ratios versus EV penetration; middle, boxplots of four penetration bands (5–10%, 10–20%, 20–35%, 35–50%); right, exceedance probability curve showing the fraction of events above the threshold (black dashed) with estimated probability (red). (b) HP surge characterization: left, hexbin plots with median and 95th percentile; middle, boxplots for four bands (5–25%, 25–45%, 45–65%, 65–85%); right, exceedance probability curve. (c) DER surge characterization: left, hexbin plots with median and 95th percentile; middle, boxplots for four bands (5–10%, 10–15%, 15–25%, 25–35%); right, exceedance probability curve.
• Figure 2: Exceedance probabilities of surge ratios across penetration bands. Violin plots show the distribution of exceedance probabilities, yellow dots indicate individual bootstrap estimates, and the horizontal line marks their mean. Each row corresponds to an asset: (a) EVs, (b) HPs, and (c) DERs. Unconditional baseline panels include all events. Subset panels restrict comparisons to events with comparable contextual conditions: EVs (Nighttime & Long Outages: the timing of restoration $\leq 6$ or $\geq 18$, duration $\geq 4$h and Daytime & Long Outages: the timing of restoration 8–16h, duration $\geq 4$h); HP (Cold & Long Outages: temperature $\leq$ 50°F, duration $\geq$ 4h and Hot & Long Outages: temperature $\geq$ 80°F, duration $\geq$ 4h); DERs (Daytime: GHI $\geq$ 200 W/m² and Matched Irradiance: pairs drawn within GHI bins [200,400], [400,700], [700,1000], and [1000,1400] W/m²).
• Figure 3: Distribution of estimated ATEs for a 10 percentage-point increase in asset penetration. Each panel shows causal forest estimates of the effect of a +10 percentage-point increase in EV, HP, or DER penetration on the surge ratio. The gray violin depicts the distribution of local ATE across outage events, and black dots indicate individual estimates. The colored bar marks the mean of these local effects, while the vertical error bar shows the empirical 95% interval of the local effects.
• Figure 4: Parametric analyses of EV-induced surge ratios under varying outage durations and restoration times. (a) EV surge ratios as functions of outage duration (1–4 h) across six representative restoration times (0:00, 4:00, 8:00, 12:00, 16:00, and 20:00). (b) EV surge ratios as functions of restoration time (0–23 h) for six representative outage durations (1–6 h).
• Figure 5: Parametric analyses of HP-induced surge ratios under varying outage durations, ambient temperatures, and seasonal conditions. (a) HP surge ratios as functions of outage duration (1–6 h) across four representative temperature regimes (cold, cool, mild, hot). (b) HP surge ratios as functions of ambient temperature (–20 °C to 40 °C) for four representative outage durations (0.5–4 h). (c) HP surge ratios as functions of penetration across four seasons.