We propose a M/G/1 queueing model with a structure change according to the network state, i.e., non-congestion and congestion, to reflect each stochastic phenomena of traffic. A structural change is explained by either/both an endogenous change of service time and/or exogenous traffic controls such as a reversible lane policy during congestion. Based on the suggested structure-changing queueing model, we develop congestion measures such as congestion flow and cost to evaluate the network efficiency, and perform scenario analysis of various transportation plans aimed at mitigating congestion. Using Korean traffic data, we show that our model capturing a structural change in the traffic queue performs better than the existing queueing models. Also, we find that a reversible lane policy is more efficient than the expansion of regular lanes, possibly due to the intermittent or occasional nature of traffic jams that occurs only during the peak hours.

Topics: Traffic management | congestion pricing | congestion management


Latest Papers