Calibration and Validation of CORSIM for Work Zones

Accurate estimation of delays at work zones is critical for the following applications: a) scheduling of activities for the work zone; b) determination of road-user costs to determine the incentive/disincentive (I/D) amount for I/D contracts; and c) determination of road-user costs for life-cycle cost analysis of pavements. Approaches to determining delays at work zones have included deterministic queuing, shock wave and stochastic queuing. Microscopic simulation programs provide a greater accuracy in providing estimates of delays because they simulate the behavior of individual vehicles. Calibration of a simulation model is a key step before it can be used for various applications. Two factors, the car-following sensitivity factor and rubbernecking factor were used in calibrating CORSIM for work-zone conditions. CORSIM’s incident specification with a corresponding lane blockage was used to simulate a standards freeway two-to-one lane closure scenario. Results of the calibration experiment indicate that a rubbernecking factor of 50 percent and the default car-following sensitivity factor can replicate the vehicle and driver behavior at work zones. Field validation was performed with data collected on North Carolina freeways. Five sets of hourly volumes, with traffic volumes ranging from light to very high, were used in the validation process. Welch’s confidence intervals were built to determine if any differences existed between the observed average hourly speed and the average hourly speed given by CORSIM. No significant differences were noted between the observed and predicted average speeds at the end of taper, which suggests that CORSIM can be used effectively to modal zones.”