Finding Near-Optimal Locations for Variable Message Signs for Real-Time Network Traffic Management

This study addressed the problem of finding the best locations for portable variable message signs to divert traffic to alternative paths when an incident occurs so that impact on the network is minimized. The study proposed and evaluated a solution procedure for finding such locations in the context of real-time network traffic management. In this context, it was essential that the procedure find the solution to the formulated mathematical program in a relatively short time. The procedure relied on a heuristic to guide the search and a simulation-based dynamic traffic assignment program to evaluate the solution. The proposed heuristic combined principles of greedy and drop heuristics. To evaluate the proposed solution procedure, four sets of experiments were conducted on the Fort Worth, Texas, network. The results from the proposed solution procedure are compared with those obtained by other methods - (a) an a priori solution to a stochastic programming formulation, and (b) the optimal solution with an exact (but slow to execute) procedure. It is found that the solutions obtained from the proposed solution procedure consistently outperform the a priori solutions and that they are consistently within 15% of the optimal solutions.