Alternative Fuel Vehicles Evacuation: Routing, Refueling Infrastructure Planning, and Behavioral Insights
Advisor: Eleftheria Kontou
Abstract
The growing adoption of alternative fuel vehicles (AFVs), including battery electric (EVs), fuel cells, and plug-in hybrid vehicles, presents new challenges for evacuation planning. These emerging vehicle types have unique technological and infrastructure specifications, such as limited driving ranges, sparse refueling and charging networks, and longer refueling/charging times than fossil-fueled vehicles. As AFV adoption increases, particularly in hazard-prone regions, it is crucial to integrate their vehicle specifications and refueling/charging needs into evacuation planning. However, since the introduction of AFVs in the US market, there have been no adjustments to evacuation plans. This dissertation contributes to reducing AFV drivers’ vulnerability during imminent hazards that warrant evacuations and supports emergency response planning by developing four novel evacuation modeling tools, along with decision-making and policy recommendations, for AFV evacuation planning.
The first research endeavor focuses on evacuation route planning for AFVs. Current evacuation route modeling overlooks vehicle range limitations and refueling/charging needs, assuming every evacuee can always reach their destination, which might not apply to AFVs. This dissertation introduces a novel evacuation route planning model for AFVs: the |K|-evacuation tree route problem with hop constraints, which accounts for range limitations and refueling/charging needs across different vehicle fuel types. Numerical experiments show that vehicle fuel type specifications, such as driving range and charging station distribution, are important in determining each evacuation route plan. Additionally, an evacuation route could be unique to a vehicle fuel type and infeasible to others.
The second research objective focuses on the strategic placement of supplementary and emergency alternative refueling/charging stations along designated evacuation routes. Despite the expansion of AFV refueling/charging infrastructure, most of their stations are located in urban centers, and few are along intercity corridors and the outskirts, which are critical for evacuation. This dissertation develops a refueling/charging station location-routing model to deploy supplementary AFV stations to support evacuation demand and minimize evacuation travel times. Numerical experiments show that, when most vehicles in the evacuation network are shorter range EVs, fewer chargers are deployed since evacuation routes are optimized to minimize detours for the majority of vehicles. However, when EV evacuees' shares are low, more chargers are deployed to support their evacuation routing and major detours.
The third research endeavor develops a multi-criteria vulnerability assessment framework for EV drivers and their charging infrastructure during evacuations. The assessment framework considers the impact of power outages and road closures, in accordance with the hazard’s intensity, accounting for the share of gasoline and EV users of the road network, driving range, and refueling/charging dependencies. The study focuses on short-notice flooding evacuations in Chicago, IL, and preemptive hurricane evacuations in Southeast Florida. The findings show that most EV drivers can safely evacuate with or without charging during mild and moderate hazards, even with the expected decrease in charging station accessibility and network disruptions. During rare and severe hazards, those with short-range EVs rely on charging station accessibility to complete the evacuation.
Finally, this dissertation models EV drivers’ evacuation route and charging behavior, using a stated preference survey. The survey consists of 12 route choice experiments of 100-mile and 300-mile mandatory evacuation orders. The experiments examine the behavioral impact of vehicle and charging characteristics, including travel time, queuing and charging duration, battery state of charge, and correlation between route choices. The evacuation behavior is characterized using multinomial and nested logit models. The models show that, in short evacuations, EV evacuees are more likely to evacuate without charging. In long evacuations, some EV evacuees are willing to make frequent stops to charge and maintain their battery state of charge above desirable levels, due to range anxiety.