Fair and Safe EV Charging Under Renewable Variability: Multi-objective Scheduling with V2G

We address city-scale fair and safe EV charging under renewable variability with bidirectional V2G. We model the city as two coupled graphs—traffic and distribution grid—and propose a hierarchical scheduler that operates in a receding-horizon loop. The upper OPF layer produces station power envelopes that are feasible with respect to feeder thermal limits, voltages, and transformer aging; the lower mobility–charging layer assigns sessions, regulates charging/V2G within envelopes, and enforces queue-aware QoS and fairness (SoC satisfaction, waiting, detours) using α-fair or lexicographic allocations. Uncertainty in renewables and arrivals is handled via robust budgets, chance constraints, and scenario sampling. The design exposes operator knobs (objective weights, risk levels, robust budgets) and requires only envelope/dual exchange, preserving privacy. In a city digital twin with 24 stations across 5 feeders (PV share ≈ 40%), our method improves safety and user experience versus Greedy/Valley-only/Grid-only baselines: peak loading 0.86 p.u. (vs 0.98/0.93/0.88), voltage violations 0.06 feeder-h/day (vs 0.42/0.21/0.09), and curtailment 3.82 MWh (vs 4.97/4.35). User metrics also improve: avg. wait 11.30 min, CVaR90_{90}90 17.20 min, Gini (wait) 0.22 (from 0.31), and SoC shortfall 2.60 p.p. (from 4.20 p.p.). Carbon is −6.8% versus Grid-only. Disabling V2G raises peak to 0.89 p.u. and curtailment by +0.71 MWh, confirming the value of controlled discharge during renewable-rich congested periods. Each receding-horizon step completes in <30 s on commodity hardware.