NETWORK DESIGN EFFECTS OF DYNAMIC TRAFFIC ASSIGNMENT

This paper examines the effects of dynamic user-equilibrium (DUE) traffic assignment with scheduled trip arrival times on network design outcomes in comparison to outcomes with steady-state travel demands. The objective is to minimize systemwide travel cost by considering alternative link improvements to an existing network (e.g., select among budget-constrained subsets of link-improvement candidates). DUE is a temporal generalization of static user-equilibrium (SUE) assignment with additional constraints to insure temporally continuous trip paths and first-in first-out (FIFO) trip ordering between all origin-destination pairs. Previous research has not investigated the effects of dynamic travel demands and schedule delay (i.e., shifts by trips to earlier or later arrival times) on network design with multiple trip origins and destinations. DUE is formulated as a bilevel program of two subproblems solved successively by an iterative algorithm that consistently converges to solutions that closely satisfy the necessary optimality conditions of this problem. Examples show the impacts of alternative combinations of network changes affecting capacities and/or free-flow travel times (e.g., ramp metering or road widening) to depend on temporal travel demands and schedule delay distributions.