Abstract
This paper presents a semi-dynamic traffic equilibrium assignment with the queue evolution that is hard to be described by the conventional static assignment models. To be specific, in our model, 1) time is divided into finite intervals; 2) each time period is treated as the static equilibrium state while queue evolution on each link in successive time periods is explicitly dealt with; 3) not only link flows but also OD demands for each period are endogenously given (they are consistent with the random utility theory). Furthermore, we prove existence and uniqueness of the equilibria and develop some efficient algorithms based on the variational inequality approach.
