Abstract
The objective of this paper is to illustrate the sensitivity of evacuation travel time estimates in relation to assumptions about evacuee route choice behavior. The user equilibrium (UE) assignment procedure in which evacuees are assumed to choose the shortest (or fastest) path from their origins to destinations is considered as the base case. Few studies have documented the actual route choice behavior of evacuees during a hurricane. These studies were used to develop realistic route choice behavior assumptions: the types of route choices and proportion of evacuees for those types. The impact of route choice behavior on evacuation performance, especially travel times, was then explored with a large-scale regional simulation model of the Hampton Roads, Virginia, region. The model was developed in an earlier study with real-world network data and actual evacuee behavioral data. The analysis indicates that using UE to generate average travel time estimates for evacuation planning significantly underestimates the actual evacuation travel times. The extent of underestimation of evacuation travel times depends on the total evacuation demand (a function of storm intensity) and the percentage of evacuees willing to use en route information to seek alternate routes when facing congestion. For the three en route percentages reported in the literature (30%, 50%, and 70%), the UE travel times were 58%, 42%, and 33% lower than actual travel times in a Category 1 evacuation; 94%, 71%, and 58% lower than in a Category 2 evacuation; and 90%, 69%, and 54% lower than in a Category 3 evacuation.