Simulation-based dynamic traffic assignment is not a black box

The dynamic traffic assignment (DTA) problem is to find the state of a transportation system where demand (travel behavior) and supply (network performance) are mutually consistent. The difficulty of solving this problem with mathematical rigor led to the development of microsimulation-based solution procedures. Microsimulations bring along vastly increased descriptive capabilities, but their equally increased complexity often motivates both practitioners and researchers to treat them as black box systems. This presentation demonstrates how the structure of the DTA problem can be exploited when dealing with microsimulations. The findings are made concrete in the context of a DTA calibration problem.