Calibration of Micro-Simulation Model Parameters for Heterogeneous Traffic using Mode-Specific Performance Measure

Abstract

Calibration is an essential prerequisite to scenario evaluations using traffic micro-simulation models (TMMs). In the context of mixed-traffic operations, where different fast and slow moving vehicular modes form a heterogeneous environment, a well-calibrated model needs to give adequate importance to each mode to realistically replicate the complex interactions in the traffic stream. This paper presents a methodology for calibrating TMMs for such mixed-traffic conditions. A combination of vehicle mode-specific travel time distributions is adopted as the performance measure for the calibration. To aid practitioners, each step of the methodology is demonstrated using a VISSIM simulator considering a signalized corridor in the Kolkata metro city, India. The work includes genetic algorithm (GA)-based optimization for obtaining mode-specific parameter sets. The Kolmogorov-Smirnov test is carried out to compare the travel time distributions of different modes. The calibrated model is also validated considering several signalized approaches along the calibrated study corridor. The results show that the methodology is successful in developing a model for non-lane based mixed-traffic operations with vehicle mode-specific optimized parameter sets.