Comparing simulated road safety performance to observed crash frequency at signalized intersections

Abstract

Microscopic traffic simulation has been developed and applied over the past two decades with the main focus towards the design and operations of transportation systems. Recently, due to advancements in data collection techniques and microscopic algorithms, the potential of microscopic simulation as a tool for safety assessments has been under considerable debate. This type of approach may allow better knowledge regarding the chain of events preceding crash occurrences; therefore, leading to a more comprehensive methodology for safety studies when compared to traditional observational studies. This paper presented a validation effort between observed rear-end collisions and simulated traffic conflicts, as reflected by three Safety Performance Measures (SPM) namely: Time to Collision (TTC), Deceleration Rate to Avoid the Crash (DRAC) and Crash Potential Index (CPI). Three years of accident data (2007-2009) for two-hour peak (7:00AM - 9:00AM) and off-peak (9:00AM - 11:00AM) were compared to the estimated number of traffic conflicts as obtained by a microscopic simulation experiment. The results suggested that simulated SPM did not reflect the apparent decrease in observed rear-end crashes for peak and off-peak periods. This might be due to the inherently rare random characteristic of traffic accidents and the somewhat simplified microscopic environment provided by current algorithms. On the other hand, all tested SPM were capable of capturing differences in the number of accidents among the three different study sites.