Hotspot Identification for Freeways Considering Difference in Single and Multi-vehicle Crashes

ABSTRACT:Previous studies found that the spatial distributions of Single-vehicle (SV) and Multi-vehicle (MV) crashes are quite different, but there has not been much research on hotspots identification considering differences in SV and MV crashes. This study identified hotspots of SV, MV and total crashes separately, using road design data, traffic operational data and crash data collected from a 45-km freeway segment in Shanghai. Full Bayes Poisson Lognormal regression models were developed for SV, MV and total crashes and the potential for safety improvement (PSI) was used to rank hotspots. Model estimation results showed that the significant influencing factors vary in different crash types. Hotspots identification results demonstrated that hotspots of SV crashes are quite different from MV crashes. For example, only three of the top ten hotspots were shared by both SV and MV crashes. Additionally, hotspots of total crashes have a higher consistency with MV crashes than with SV crashes, indicating that a majority of SV crash hotspots may be ignored if total crashes are used to identify hotspots. These conclusions prove the necessity to differentiate SV and MV crashes for hotspot identification and conducting road safety management.

Xuesong Wang*, Mingjie Feng, Qi Shi, Xiaokun Wang, Yan Li. Hotspot Identification for Freeways Considering Difference in Single and Multi-vehicle Crashes. Transportation Research Board 97th Annual Meeting, Washington D.C., USA, 2018. 1.7-11.