Network Equilibrium with Activity-Based Microsimulation Models

Travel demand models and network simulation models are distinct sets of procedures that are combined and interact within the framework of regional transportation modeling systems. Conventional four-step models have numerous limitations compared with more advanced activity-based microsimulation models, primarily with respect to internal consistency and detailed behavioral realism. However, two of the remaining advantages of four-step models are an established theory and an effective set of practical rules for achieving global network equilibrium so that travel time and cost simulated in the networks exactly correspond to the demand (trip tables) generated by the model. Nonetheless, this issue remains less explored and somewhat obscure for activity-based models. These models have a more complicated analytical structure compared with four-step models, which makes it difficult to derive equilibrium conditions in a rigorous theoretical manner. In addition, implementation of an activity-based model requires microsimulation of individual outcomes in the form of “crisp” discrete choices that is very different from the summation of fractional probabilities implemented in conventional models. This paper documents the results of testing various equilibrium strategies implemented with the New York City activity-based microsimulation regional travel demand model used by the New York Metropolitan Transportation Council. The purpose of the paper is twofold. First, it is intended to outline some fundamental research directions and extensions of the network equilibrium theory to cover activity-based microsimulation models in a more rigorous way. Second, it describes realistic levels of convergence that can be achieved with activity-based microsimulation models in practice and establishes practical rules and protocols for using these types of models for different projects and policies.