Modeling Driver Lane Choice Behavior at Auxiliary Through Lanes (ATLs).

摘要

Auxiliary through lanes (ATLs) have typically been implemented as a cost-effective means of congestion relief at signalized intersections as an alternative to full roadway widening. However, the capacity of an ATL is generally less than that of a continuous through lane (CTL) because drivers tend to only use an ATL to avoid becoming stopped at the intersection at the end of the signal cycle. The research presented here describes the author’s efforts to quantify the rationale behind driver lane choice at signalized intersections with an ATL. Specifically, this research contains two methods to quantify the flow in the ATL—using an aggregate and a disaggregate model. The aggregate model predicts the ATL flow as a function of macroscopic elements such as through-movement demand and signal timing. The disaggregate model describes individual driver lane choice as a function of instantaneous elements the driver experiences such as the queue in either lane. The results of the aggregate model development indicate that ATL flow tends to increase with through-movement demand and congestion, while right turns from a shared ATL tend to detract from ATL use.;Alternatively, the resulting disaggregate model indicates that the utility of the ATL increases as the CTL queue increases but that a driver’s lane choice may also be dependent upon the displayed signal phase (effective red or effective green). After presenting the reasoning and methodology for the development of each type of model, the research shows a comparison of the two models and a summary of how each can be implemented within the context of current engineering process—both in the Highway Capacity Manual and using traffic simulation. The appropriate comparison between the models is achieved via Monte Carlo simulation to assume that the form of the input and output data for each model is identical. The results of the comparison indicate that the models yield statistically indistinguishable results, and these results are also statistically indistinguishable from the field data. The author then introduces a methodology to implement the aggregate model into the Highway Capacity Manual. Finally, the author recommends that the disaggregate models be implemented in microsimulation to either improve or replace the lane change algorithms currently used to control lane utilization in these systems.