Abstract:
Reducing the impact of incidents on the freeway network, which represents a key goal for most regional traffic management systems, requires an effective traffic incident management system. Early and accurate recognition of incident impact the network represents one of the key challenges associated with incident management. Microscopic simulation represents an effective strategy for forecasting incident impact because it provides opportunities to study diverse conditions that may occur irregularly both spatially and temporally in the field. Often, the traffic simulation parameters appear be calibrated for uncongested conditions. This research explores macroscopic network performance when applying microscopic simulation models calibrated under both congested and uncongested conditions; the research considers a default uncongested parameter set and compares it with a parameter set calibrated using macroscopic data during congested conditions and a parameter set calibrated using microscopic data during congested conditions. Using a testbed along I-210 in Pasadena, California, the study varies incident duration, traffic volume, and lane closures to investigate the change in results across all three models. The study also investigates six actual incidents and compares the simulation results with the observed field conditions collected from the California Department of Transportation Performance Measurement System database. During the experimental cases, the macroscopic congested parameter set consistently estimates longer network travel times and slower speeds. The comparison with the field conditions confirms that the macroscopic congested parameter set better reflects field performance especially during more congested conditions. This research indicates the need to use congested conditions for calibration when investigating incident impact.
