Developing and Validating a Model of Left-Turn Crashes to Support Safer Design and Operations

Abstract

This report documents work done to advance the state of art in crash simulation. This includes: (1) A field study to collect data on drivers’ left-turn gap acceptance and turning times, and development of statistical models that can be incorporated into a crash simulation model; (2) The use of Markov Chain Monte Carlo computational tools to quantify uncertainty in planar impact reconstruction of two-vehicle crashes; (3) A method for combing the results from planar impact reconstruction with event data recorder pre-crash data to estimate descriptive features of actual left-turn crashes. This is applied to several left-turn crashes from the National Highway Traffic Safety Administration’s NASS/CDS database; (4) A left-turn crash simulation model incorporating the above results. Initial model checking is performed using estimates from the reconstructed NASS/CDS cases as well as results from a previous study on left-turn crash risk. Also described is a method for simulating crash modification effects without having to first simulate crashes as rare outcomes in very large numbers of gap acceptances.