Freeway On-ramp Bottleneck Activation, Capacity and the Fundamental Relationship

This study examines traffic behavior in the vicinity of a freeway bottleneck, revisiting commonly held assumptions and uncovering systematic biases that likely have distorted empirical studies of bottleneck formation, capacity drop, and the fundamental relationship (FR). This simulation-based study examines an on-ramp bottleneck using Newell's lower order car following model with a driver relaxation factor added for the vehicles that enter or are immediately behind an entering vehicle (termed “affected vehicles”). The affected vehicles will tolerate a truncated headway for a little while after an entrance but slowly relax back to their preferred speed-spacing relationship. All other vehicles remain on their preferred speed-spacing relationship throughout.Simulating conventional detector measurements, we show that flow is supersaturated in any sample containing an affected vehicle with a truncated headway, i.e., the flow is higher than the underlying FR would predict. This systematic bias is not readily apparent in the detector measurements and during the initial queue formation the supersaturated states can exceed the bottleneck capacity. As the affected drivers relax, the high flows become unsustainable so a queue initially forms downstream of the on-ramp (consistent with earlier empirical results) only later receding upstream past the on-ramp. This initial phase of activation often lasts several minutes. Without any evidence of queuing upstream of the ramp, the conventional point bottleneck model would erroneously indicate that the bottleneck is inactive. Thus, an empirical study or traffic responsive ramp meter could easily mistake the supersaturated flows to be capacity flow, when in fact these supersaturated flows simply represent system loading during the earliest portion of bottleneck activation. Instead of flow dropping “from capacity”, we see flow drop “to capacity” from supersaturation. We also discuss how the supersaturated states distort empirically observed FR. We speculate that these subtle mechanisms are very common and have confounded the results of many past empirical studies.