Numerous traffic simulation tools are available to assist in the efficient allocation of resources while operating and managing traffic networks. Their model fidelity ranges from detailed (microscopic) to aggregate (mesoscopic) and intermediate (anisotropic mesoscopic), each approach differing in the level of detail of their component demand and supply models. While each tool may have been demonstrated on specific datasets, these data are generally too varied to allow meaningful comparisons of use to practitioners. Objective evaluations of such tools are often lacking, leading to ambiguity about their modeling assumptions, accuracy, feature sufficiency, running time and scalability. In this paper, we apply three popular traffic simulation tools on a common network and demand data. TransModeler (microscopic), DYNASMART-P (mesoscopic) and DynusT (anisotropic mesoscopic) are tested on a real, urban region in Eureka, CA. Their performance over a range of factors are studied, documented and analyzed from both theoretical and empirical perspectives.
展开▼