Announcing the Final Examination of Md Tarek Hasan for the degree of Doctor of Philosophy
Traditional crash frequency prediction models cannot capture the temporal effects of traffic characteristics due to the high level of data aggregation. Also, this approach is less suitable to address the crash risk for active traffic management strategies that typically operate for short-time intervals. Hence, this research proposes short-term crash prediction models for traffic management strategies such as Variable Speed Limit (VSL)/Variable Advisory Speed (VAS), and Part-time Shoulder Use (PTSU). By using high-resolution traffic detectors and VSL/VAS operational data, short-term Safety Performance Functions (SPFs) are estimated at weekday hourly and peak period aggregation levels. The results indicate that the short-term SPFs capture various crash contributing factors and safety aspects of VSL/VAS more effectively than the traditional highly aggregated Average Annual Daily Traffic (AADT)-based approach. The study also investigates the safety effectiveness of VSL/VAS for different types and severity levels of traffic crashes. The results specify that the VSL/VAS system is effective in reducing rear-end crashes in the Multivariate Poisson Lognormal (MVPLN) crash type model as well as Property Damage Only (PDO) and C (non-incapacitating) crashes in the MVPLN crash severity model. Recommendations include deploying the VSL/VAS system combined with other traffic management strategies, strong enforcement policies, and drivers' compliance to increase the effectiveness of this strategy. Further, this research estimates the Random Parameters Negative Binomial-Lindley (RPNB-L) model for PTSU sections and provides valuable insights on potential crash contributing factors related to PTSU operation, design elements, and high-risk areas. Last, the study proposes a novel integrated crash prediction approach for freeway sections with combined traffic management strategies. By incorporating historical safety conditions from SPFs, the real-time crash prediction performance could be improved as a part of proactive traffic management systems. The findings could assist transportation agencies, policymakers, and practitioners in taking appropriate countermeasures for preventing and reducing crash occurrences by incorporating safety aspects while implementing traffic management strategies on freeways
Committee in Charge:
Mohamed Abdelâty, Chair, Civil, Environmental, and Construction Engineering
Naveen Eluru, Professor in Department of Civil, Environmental and Construction Engineering
Nada Mahmoud, Consultant in Kittelson & Associates, Inc.
Liqiang Ni, Associate Professor in Department of Statistics and Data Science