Stochastic User Equilibrium Model with a Bounded Perceived Travel Time

Abstract: Stochastic User Equilibrium (SUE) models depict the perception differences in traffic assignment problems. According to the assumption of an unbounded perceived travel time distribution, the conventional SUE problems result in a positive choice probability for all available routes, regardless of their unappealing travel time. This study provides an eUnit-SUE model to relax this assumption. The eUnit model is derived from a bounded probability distribution. This closed-form model aligns with an exponentiated random utility maximization (ERUM) paradigm with the exponentiated uniform distributed random error, where the lower and upper bounds endogeneously determine the route usage. Specifically, a Beckmann-type mathematical programming formulation is presented for the eUnit-SUE problem. The equivalency and uniqueness properties are rigorously proven. Numerical examples reveal that the eUnit bound range between the lower and upper bounds greatly affects the SUE assignment results. A larger bound range increases not only the number of routes in the choice set but also the degree of dispersion in the assignment results due to a larger route-specific perception variance. The misperception is contingent upon the disparity between the shortest and longest travel times and the bounds. As the bound range decreases, the shortest route receives significant flow allocation, and the assignment result approaches the deterministic user equilibrium (DUE) flow pattern.