On macroscopic freeway merge behavior: Estimation of merge ratios using asymmetric lane flow distribution

Data from several freeway merges reveal that, contrary to some previous findings, merge ratio can vary within a site with respect to the merge outflow and that the existing merge ratio estimates based on lane counts are not able to predict this within-site variation. Furthermore, the merge ratios estimated based on two well-known merging principles, “fair-share” and “zipper,” are found to be inaccurate for merges where merging streams compete directly due to a lane drop. In light of these findings, we estimate merge ratios using lane flow distribution (LFD) to better predict between and within site variations of merge ratio. In addition, we propose a merging principle specific for merges with a single lane-drop. The model was developed to better represent observed non-uniform redistribution of merging flow not captured by the current merge ratio estimation methods and merging principles. Empirical observations show that the proposed methods are able to improve merge ratio estimates, reproduce within-site variations of merge ratio, and represent more accurately non-uniform redistribution of merging flow dependent on the merge geometry.