Search

Urban congestion is getting worse, especially at intersections, where conflicting flows share the same traffic signal cycle. However, traffic demand-responsive control can improve urban traffic operations by directing vehicles to alternative routes with spare capacity. Intersection demand is characterized by left, through and right turning movement (TM) flows from each intersection approach. Information on TMs must be current and consistently reliable for traffic control strategies to be effective. Non-recurring traffic congestion can quickly and unpredictably develop, such as during the Christmas shopping season, causing backups and warranting congestion mitigation. The latest version of the Turning Movement Estimation in Real-Time (TMERT) model, TMERT3, can consistently estimate 5-minute TM flows from 15% of network flow detections. This paper validates TMERT3 by showing it robustly estimates TMs despite widespread detector failure when applied to non-recurrently congested extensive networks. TMERT3 is tested on traffic on the 28-intersection network in Orem/Provo, Utah, USA, during a Christmas shopping season. TMERT3 TM estimates showed to preserve accuracy (Root Mean Square Error- RMSE) even when 20% of the initial detectors fail. So, TMERT3 presented practical applicability in efficiently guiding strategies that mitigate urban traffic congestion. [ABSTRACT FROM AUTHOR]