Observed Dynamic Traffic Features on Freeway Section with Merges and Diverges

Abstract

Traffic data measured on a four-lane freeway and its on-ramps (with metered flows) and off-ramps were studied for 4 days during the peak hour. Certain features were reproducible from day to day. A bottleneck arose on all 4 days because of traffic merging and diverging at a freeway location. During these times queuing arose in a freeway section with an auxiliary lane some distance downstream of the merge. Corresponding flow reductions were also measured at a downstream off-ramp. Queuing was apparently caused by drivers who had just entered the shoulder lane from the on-ramp and had slowed down to merge. The slowing of vehicles spread to both lanes and propagated upstream. The average discharge flows that accompanied the onset of queuing were 4% lower than flows measured just before the queue's formation. Upon bottleneck activation, flow reductions occurred sequentially in time and space marking the passage of the backward-moving shock. Mean shock velocities ranged from 20 to 24 mph (32 to 40 km/h). The analysis used transformed curves of cumulative vehicle counts and cumulative occupancy to obtain the measurement resolution necessary for studying important traffic features. The site is located in Minneapolis, Minnesota, and data were collected before a ramp meter shutdown experiment conducted in 2000.