Perturbation strategies for optimal traffic reassignment

Abstract

Suboptimal static traffic reassignment strategies are developed following a traffic perturbation due to an incident or change in flow in a large freeway corridor network. The basic technique is perturbation analysis using a quadratic approximation to the cost of traversing the roadway links. A dynamic programming approach, using the subnetwork number as the stage variable, leads to an optimization problem which is a variant of the linear-quadratic optimal control problem. Explicit relationships between decay of perturbations, controllability, and lateral access in the network are obtained. The perturbation strategies are useful because they can be used 1) to isolate the subset of the network that needs to be reoptimized by nonlinear programming, or 2) as a real-time quasistatic traffic reassignment algorithm; in either case computational requirements are reduced.