The structure of user equilibria: Dynamic coevolutionary simulations vs. cyclically expanded networks
A variety of approaches exist that model traffic time-dependently. While all approaches have their advantages and disadvantages but have to find a balance between modeling traffic as realistic as possible and being still manageable in combinational terms. While transport simulations are efficient in evaluating user equilibria in large scale scenarios, their potential to be used for optimization is limited. On the other hand, analytical formulations like models based on cyclically time-expanded networks can be used to optimize traffic flow, but are not suitable for large scale scenarios. By optimizing the network structure in a mathematical model and evaluating its effect in a more realistic transport simulation, two models can benefit from each other. Detailed knowledge about model properties and differences in traffic flow behavior help to understand results and potential difficulties of such a model combination. In this paper, properties of two such models are compared regarding traffic flow modeling. It is shown that the set of user equilibria in both models and, therefore, the resulting route distributions can be structurally different.
Published in: Procedia Computer Science, 10.1016/j.procs.2017.05.371, Elsevier