Stochastic Modeling of Shared-Lane Left-Turn Process and Its Effect on Saturation Flow

Abstract

A shared-lane left-turn model was constructed to investigate the ways in which left turners are impeded during the saturation flow period at the beginning of the green when queues are present. The model takes into account the subject and opposing approaches and the length of the two queues. It is event based, viewing each headway as a discrete discharge opportunity. The number of “extra” headways required because of the left-turning activity (and related missed or blocked opportunities) is defined as the “excess impedance.” The shape of the excess impedance distribution is studied, as is its mean. Above all, the considerable randomness of the excess impedance was noted, as were its implications. For instance, this randomness would imply that the 15-min saturation flow rate has so much variability that it is primarily a stochastic process, not a fixed number. Further, the randomness due to the left-turn effects generally exceeds that due to the fact that discharge headways are themselves random. The net effect of the randomness is that data requirements for true calibration of models and for selecting one model over another may be too large to be feasible in many, if not most, cases.