Left-Turn Spillback Probability Estimation in a Connected Vehicle Environment

At intersections in which the left-turn bay does not have sufficient length or the left-turn volume is relatively high, left-turn vehicles may spill back and block the adjacent through traffic. This paper aims to develop quantitative measures of the left-turn spillback, and by using the results on spillback probability, develop a suitable signal control strategy. We first develop an improved queue length estimation method for vehicles in the left-turn bay based on Convert and Cetin's general queue length estimation method with connected vehicles, after which we propose a probabilistic model to measure the left-turn spillback probability at an intersection in a connected environment. The model accuracy is validated with results from microscopic traffic simulation. The effect of bay length is also studied. In the end, a signal control demonstration is presented to show the efficiency of the proposed method in signal control.