Abstract:
Traffic flow on freeways exists in two states - uncongested (volumes below capacity) and congested. Delay functions used in assignments in regional modeling [such as the classic Bureau of Public Roads (BPR) function] calculate link travel times as monotonic increasing functions of the volume-to-capacity ratio. This works well for uncongested links, but the functions produce the slowest speeds only from volumes well in excess of capacity. In real life, link capacity under congested conditions is significantly below uncongested capacity. This paper describes an assignment algorithm which uses a state variable based on lane occupancy to identify congested links and select from a family of modified BPR functions which reduce capacity and constrain speeds to the lower branch of the volume-delay curve. The new functions, relating speed, volume, and lane occupancy, were derived from three years of 5-minute data gathered from the embedded loops of the Surveillance, Control, & Driver Information System operated on regional freeways by the Washington State Department of Transportation.
