EVACUATION NETWORK OPTIMIZATION WITH LANE REVERSAL AND CROSSING ELIMINATION Evacuation planning with lane reversal and crossing elimination

environment throughout the evacuation network. Although both have proven reasonably effective as capacity-increasing measures (for accelerating the evacuation process), their combination can enhance performance further. Solving this design problem with lane reversal and crossing elimination, however, poses a great challenge, because of the complex mutual connectivity requirement imposed by the two lane-based network design strategies and the large number of solution spaces generated by the lane-based network setup, in which each intersection must be explicitly modeled as a subnetwork. One way to reduce the problem’s complexity is to relax the crossing elimination constraints and redefine the relaxed problem on the standard node–arc network. Such simplification can be mathematically characterized by Lagrangian relaxation. To compensate for relaxation of the crossing elimination constraints, one needs to evaluate the number of traffic crossing points at each intersection and substitute the evaluation result back into the solution algorithm to guide its search itinerary. This evaluation requirement results in an intersection origin–destination (O-D) flow optimization problem. As will be shown later, under this specific modeling context, the O-D flow optimization problem’s objective is to find an intersection O-D flow pattern that minimizes the number of traffic crossing points without altering the overall (arc-based) network flow pattern. It is clear that this problem differs from more traditional work in intersection O-D flow estimation, which generally seeks to replicate a most likely O-D flow pattern (1–6). The remainder of this paper is organized as follows: the evacuation network design problem with lane reversal and crossing elimination is introduced. The intersection subnetwork O-D flow optimization problem is then formulated, and conditions for the existence and validity of the problem are discussed. The major contribution of this work is the development of a simplex-based solution method for the proposed O-D flow optimization problem. The fourth section elaborates on the method, and the fifth section provides an illustration. The final section of the paper concludes with some modeling extensions.