Predicting lane utilization and merge behavior at signalized intersections with auxiliary lanes: A Buffalo, New York, study

Auxiliary through lanes (ATLs) are often used to increase the capacity at signalized intersections. The effectiveness of these lanes is a direct function of the level of their utilization by motorists, which tends to be significantly lower than continuous through lanes. Currently, there is no official model that can be used to predict ATL utilization, although a handful of recent studies have attempted to develop such models. This study, conducted in the Buffalo, New York, area, was designed to: (1) determine whether recently-proposed models for predicting ATL utilization are applicable to Buffalo; (2) develop new models for predicting lane drop and auxiliary lane utilization in the area; and (3) develop models for predicting drivers' merge behavior at intersections with ATLs. To achieve these objectives, geometric, traffic count and merge data were collected from select sites in the Buffalo metropolitan area. The data were then used to assess the transferability of ATL utilization models, and to develop Buffalo-specific models for predicting ATL utilization and average merge distance. The study's results show that models for predicting ATL utilization tend to be site-specific. Given this, several new models are presented to better predict the lane usage patterns observed in Buffalo; the variable most represented across all models is the average lane volume, whereas the short lane length (a variable shown by many other studies to be important in lane utilization) seemed to be much less important. Additionally, several new land use variables were developed in this study to better quantify the effects of land use trip generations on lane use. In a first attempt to describe drivers merging behavior, the study also developed models for predicting the average merge distance, which was found to be a function of the short lane length, the first lane drop warning sign, the downstream speed, and the total through traffic volume. Finally, the study investigated using data on the merging behaviors to help calibrate traffic simulation models to better represent traffic conditions at these types of intersections.