Interactive Process of Microsimulation and Logistic Regression for Short-Term Work Zone Traffic Diversion

Abstract

The rapidly growing number of work zones on national highways is having significant operational impacts due to the temporary loss of capacity. Work zone impact on safety and mobility creates a strong need to alleviate work zone congestion and protect road users and workers, which requires a sufficient understanding of work zone impact on traffic flow. Previous studies and field observations demonstrated the importance of considering diversion phenomena when performing work zone impact analysis. To overcome the limitations of deterministic queuing approaches applied in most work zone impact analysis tools, an interactive process combining microsimulation and logistic regression was developed to imitate diversion behavior dynamically in the upstream of work zones with a number of entrance and exit ramps. Specifically, the logistic regression model based on the field observations was incorporated into a well-calibrated VISSIM model to simulate traffic flow in work zones with diversion behavior. The integration of the two models was achieved via the development of diversion calculation module using a COM interface provided by VISSIM. The comparison between simulated results and field observations suggested that the diversion calculation module using logistic regression can simulate the queue propagation process due to lane closure in an efficient and effective manner. It was demonstrated that the interactive process can improve work zone impact analysis by using real-time traffic feedback information to emulate the diversion phenomenon upstream of work zones.