AN ALGORITHM FOR SUEED - STOCHASTIC USER EQUILIBRIUM WITH ELASTIC DEMAND

Abstract This paper proposes a model for Stochastic User Equilibrium assignment with Elastic Demand (SUEED), which involves the simultaneous satisfaction of two equilibrium conditions. An algorithm is developed which estimates, at each iteration, the optimal step lengths in the flow and demand directions. The algorithm is applied to a number of test networks, ranging from a simple two-link, single O-D pair network to one with over 2,800 links and 4,800 O-D pairs. These tests show that although the new algorithm requires more computational effort at each iteration than the conventional alternative algorithm, the Method of Successive Averages, it converges considerably faster.