Quantifying Incident-Induced Travel Delays on Freeways Using Traffic Sensor Data: Phase II

摘要

Traffic incidents cause approximately 50 percent of freeway congestion in metropolitan areas, resulting in extra travel time and fuel cost. Quantifying incident-induced delay (IID) will help people better understand the real costs of incidents, maximize the benefit-to-cost-ratio of investments in incident remedy actions, and facilitate the development of active traffic management and integrated corridor management strategies. Currently, a number of algorithms are available for IID quantification. However, these algorithms were developed with certain theoretical assumptions that are difficult to meet in real-world applications. Furthermore, they have only been applied to simulated cases and have not been sufficiently verified with ground-truth data. To quantify IID over a regional freeway network using existing traffic sensor measurements, a new approach for IID estimation was developed in this study. This new approach combines a modified deterministic queuing diagram with short-term traffic flow forecasting techniques to overcome the limitation of the zero vehicle-length assumption in the traditional deterministic queuing theory. A remarkable advantage with this new approach over most other methods is that it uses only volume data from traffic detectors to compute IID and hence is easy to apply. Verification with the video-extracted ground truth IID data found that the IID estimation errors with the new approach were within 6 percent for the two incident cases studied. This implies that the new approach is capable of producing fairly accurate freeway IID estimates using volumes measured by existing traffic sensors. This approach has been implemented on a regional map-based platform to enable quick, convenient, and reliable freeway IID estimates in the Puget Sound region.