Reliability optimization of electric vehicle fleets under mobility uncertainty

Electric vehicles (EVs) can enhance power-system outage reliability by supplying local demand through vehicle-to-grid operation, but their effective contribution depends on un-certain mobility patterns and battery limits. This paper presents a reliability-oriented framework that combines a time-varying two-state Markov model of EV availability with a dual-objective outage dispatch optimization. Unlike static or deterministic availability assumptions, the proposed model captures stochastic departure and arrival behavior using empirical mobility distributions and therefore provides more realistic estimates of fleet support during outages. The optimization balances two reliability measures: (i) minimizing the energy not supplied over short analysis windows and (ii) maximizing the duration of sustained service. A 50-EV case study on a commercial/institutional feeder, derived from the DOE/NREL Commercial Reference Buildings dataset and scaled upward by 15% to create a stressed operating condition, demonstrates how mobility uncertainty affects available capacity and the trade-off between energy not supplied and continuity of supply. The resulting Pareto frontier highlights the compromise between reducing unserved energy and maintaining service duration. The framework provides a computationally efficient, behaviorally grounded, and scalable tool for evaluating outage reliability using EV fleets.