Calibration of Car Following Parameters in VISSIM for Traffic in Kathmandu

Abstract

Traffic simulation is widely used to perform analysis of traffic operations. Traffic in Nepal is non-lane based, heterogeneous and mixed while car-following models are developed for lane based and homogeneous traffic condition. Traffic simulation software needs to be calibrated to represent local conditions. The main objective of this study was to calibrate Car-following parameters in VISSIM. The parameters for calibration were selected based on past studies on VISSIM calibration in heterogeneous traffic condition. Eleven parameters were selected for analysis. Latin hypercube sampling technique was used to create the sample set of parameters for simulation. One-way ANOVA was used to determine the sensitive parameters. Linear equations were developed using sample set prepared by Latin hypercube sampling technique. Multi objective Genetic algorithm tool available in MATLAB was used to perform optimization of linear equations to minimize the difference between field delay and simulated delay. Three car-following parameters of Widemann-74 model were found sensitive which were calibrated using genetic algorithm to obtain optimal values. For further study, it is recommended that more than one intersection can be studied to generate a range of values, and more than one measure of effectiveness can be selected for robust calibration and validation.