Traffic Conflict–Based Safety Assessment Framework for Evaluating Treatment Effects at Urban Unsignalized Intersections: A Case Study from Surat, India

Traditional methods of relying on historical crash records present significant challenges for developing countries with flawed crash databases. This study presents a novel framework for assessing the safety of geometric treatments at unsignalized intersections, illustrated through case studies of two recently modified intersections in Surat, Gujarat, India. The present research employs traffic conflict analysis for a before–after (BA) assessment. Utilizing the extreme value theory (EVT) and post-encroachment time (PET) as a measure of traffic conflict, the study evaluates changes in traffic conflict intensities both before and following the implementation of countermeasure treatments. The developed framework analyzes several key metrics, including shifts in crash risk, the number of theoretical crashes, and the probability of critical crossing conflicts (PCCC). Additionally, it incorporates geometric improvements through heat maps that depict traffic conflict locations during the BA periods, allowing for an assessment of changes in their spatial distribution and intensity. The safety benefits of these geometric modifications are quantified by evaluating PCCC at similar traffic volumes using a Beta regression framework, effectively mitigating regression-to-mean bias. Findings indicate substantial reductions in both crash risk and PCCC after the implementation of geometric improvements, with decreases ranging from 33.43% to 57% compared to pretreatment conditions. This study introduces a fresh perspective for evaluating safety enhancements in contexts where historical crash data are lacking or where establishing control sites and comparison groups is challenging.