A Comprehensive Analysis of Drivers' Left-turn Lane Preferences at Signalized Intersection

Traffic patterns of the lanes accommodating the same movement at signalized intersection are different with each other in most cases, which could decrease the efficiency of signal control. This problem derives from drivers' preferences to specific lane, which is contrary to common control strategy that usually takes the lane traffics as a whole. Lane-based timing technique can solve this problem, but the first step to apply it is to know external influential factors of the preferences, which should be taken as signal timing variables in optimization framework. This paper takes triple left-turn lanes at signalized intersection as study case. Compositional analysis of variance model and logit model is respectively applied to identify significant static and dynamic influential factors. An important finding is that some factors could have opposite effects on drivers' lane preferences in red and green phases, such as length of longitudinal vehicle movement at intersection and shadowed lane design, while others could impose the same influences on drivers in two phases, such as length of upstream segment, the total number of right-turn and through lanes in the approach. In addition, some factors could only be influential in red phase, which include length of lateral vehicle movement at intersection, location of first left-turn sign at upstream, length of left-turn queue, subject vehicle type and number of heavy vehicles. These factors should be taken into account when design pre-timed, accurate or adaptive lane-based signal timing plan.