Abstract
This study developed a validation procedure for a dynamic traffic assignment–based mesoscopic simulator and identified its relevant data requirements. Model inconsistencies at the network level can be detected through route-based validation with probe vehicle travel time data. Simultaneously, link-based validation uses point detector–based volume and speed data. A procedure for filtering the validation data set to make it consistent and comparable with the simulation results is outlined in the paper. Two assignment methods were tested with a detailed validation data set through a case study. The no-diversion assignment approach yielded higher travel time than the field-observed average time. In contrast, simulated travel time with the optimal diversion method implied a more rational behavior by drivers than was observed in the field, along with more diversion of trips on arterial routes in the vicinity of a work zone. Finally, a link-based performance comparison revealed worse queue spillback effects on the upstream links of the work zone than in reality.
Get full access to this article
View all access options for this article.