Area-Wide Ramp Metering for Targeted Incidents: The Additive Increase, Multiplicative Decrease Method

Abstract

Ramp metering has been broadly accepted and deployed as an effective countermeasure against both recurrent and nonrecurrent congestion on freeways. However, several current ramp metering algorithms tend to optimize only freeway travels using local detectors’ inputs and overlook the negative impacts on local streets. This may generate two problems: (1) the optimal local settings may not result in a system-wide optimum in terms of traffic operations; and (2) the increased congestion on local streets due to on-ramp overflow may counteract the gains in freeway operations. To solve these problems for nonrecurrent cases, we propose an area-wide ramp metering (AIMD) system to coordinate the previously isolated ramp meters for system-wide optimization. This novel strategy for active ramp metering is inspired by the principles of a computer network congestion control algorithm. In this strategy, certain types of congestion at a targeted freeway location can be significantly reduced by limiting vehicle flows on each ramp to a fraction of ramp demand and then additively increasing rates to avoid ramp queue spillover onto city streets. This approach can be actively used to proactively curb the growth of traffic congestion and therefore shorten travel delays. Six simulated scenarios using this strategy are shown to significantly reduce vehicle delays in various zones when compared to two other strategies: the Fuzzy Logic strategy employed by the Washington State DOT, and the performance of no ramp meter control. Based on the Tukey statistical testing method, the overall improvement of AIMD is shown to be significant. This preliminary research shows promise for future integration with other active traffic management systems as a means to effectively reduce traffic congestion.