A Data-Driven Optimization Approach for Community Resilience Enhancement During Disasters Leveraging Electric Vehicles

Abstract

Natural disasters pose a significant risk to critical infrastructure, leaving essential facilities such as hospitals, and fire stations vulnerable to power outages. Traditional backup power solutions, such as diesel generators, often lack the capacity for prolonged outages, while repair crews face substantial delays in restoring the power grid. This study presents an innovative optimization approach using electric vehicles (EVs) to restore critical isolated loads, ensuring that vital services remain operational during disasters. This problem’s complexity surpasses traditional vehicle routing problems due to several unique features: (1) it combines elements of the traveling salesman problem, timely demand, load shifting, and split delivery; (2) it requires energy and transportation layers with state-of-charge (SOC) tracking; (3) EVs may need to make multiple back-and-forth trips between shelters and charging stations; and (4) route interdependency of vehicles. To address these complexities, we developed an efficient mixed-integer-linear programming model and employed a branch and price solution approach that efficiently manages the routing and scheduling of EVs to meet the demand of critical isolated loads. Realistic case studies, based on data from Florida hospitals, demonstrate the effectiveness of our model and solution approach, solving the problem accurately within a one-hour time limit. Specifically, our proposed branch and price algorithm can solve the problem up to 120 times faster than Gurobi. This study highlights the potential of EV-powered backup systems to enhance community resilience, offering a sustainable, adaptive alternative to traditional generators in supporting critical infrastructure during extended power outages.