Stationary Models of Unqueued Traffic and Number of Freeway Travel Lanes

Occupancies and flows were jointly sampled from freeway segments in nearly stationary, unqueued traffic. When plots of occupancy per lane versus flow per lane were normalized by n (the number of travel lanes in the freeway segment from which a data set came), the plots took shapes that were piecewise linear in form (except for conditions that were near capacity) and were clearly influenced by n. Drivers adopted a higher speed (for a given occupancy) while traveling on segments of greater n. Yet, the speeds on these wider segments exhibited greater sensitivity: drivers began decelerating at relatively low occupancies. These findings came from a comparison of a data plot from each of five different freeway segments with the plot from its neighboring segment. Because each segment appeared to differ from its neighbor only in its n, the comparisons (approximately) controlled for other influential factors, including geometric design standards, speed limit, and driver population. The five pairwise comparisons, which verify the reproducibility of the effects of n on the data, were performed for freeways in and near Toronto, Ontario, Canada, and California. The findings are compared with the information currently provided in traffic handbooks.