A Case for Online Traffic Simulation: Systematic Procedure to Calibrate Car-Following Models Using Vehicle Data

Abstract

Online microsimulation modeling is a congestion mitigation strategy that is gaining traction in the transportation sector. The key limitations to the deployment of this strategy are the availability of data for model calibration and the existence of vetted and transparent methodologies to calibrate models using trajectory-level data. The promise of connected vehicle (CV) data is a solution to the former problem. In this paper, a sample from the second Strategic Highway Research Program's Naturalistic Driving Study (NDS) is used as a surrogate for CV data. A systematic procedure for car-following model (CFM) calibration using trajectory-level data is presented. A case study was developed to illustrate the importance of calibrating microsimulation models using current traffic data. Four CFMs are calibrated and validated using high and low speed trips. Results indicate that behavior varies with speed and better calibration results are achieved when segmenting the data. Finally, validation efforts indicate that calibrated model parameters outperform literature parameters; this underscores the importance of calibrating models with appropriate data.