Go back to Carlos Daganzo's Publications PageThis paper presents a qualitative description of queuing patterns under an idealized scenario analogous to the evolution of traffic congestion during the morning rush hour in a long corridor leading to a single destination. The simplicity of our scenario allows the results to be verified independently by hand.
Initially the corridor is assumed to consist of a single freeway. Traffic is generated at the freeway's many on-ramps during a short period of time and then is assumed to subside. Capacity limitations create queues on the ramps and the freeway, whose evolution is then described. A special case with just a few parameters is analyzed in detail.
The solution obtained under the assumptions of the hydrodynamic theory of traffic flow (which explicitly recognizes vehicle storage limitations on the freeway) is shown to be drastically different from the solution obtained using "point queue" models, which ignore these limitations. Because the latter models are currently a favored approach in the dynamic traffic assignment literature, our results illustrate the need for re-evaluating the conditions under which current theories may be applicable.
The note also discusses briefly the effect that a slower parallel arterial would have on the system's traffic. It is found that a route choice mechanism where drivers do not anticipate the system's evolution leads to unreasonable traffic patterns; i.e. patterns that would not be expected in reality. The anticipation phenomenon, thus, must be incorporated into any realistic model of dynamic network flows. Unfortunately this increases the difficulty of developing detailed control strategies.
Finally, the paper shows that ramp metering would not reduce
the total number of vehicle-hour spent in the system for our
scenario. This result, however, is not general; ramp metering
can be of benefit when all the traffic doesn't flow to the same
destination.
Go back to Carlos Daganzo's Publications Page