Corridor-Level Approach for Estimating the Contribution of Incidents to Delay

Understanding the relationship between incidents and delay is an important step in developing targeted measures to reduce non-recurrent congestion. Research over the last decade has focused on two key incident characteristics that dictate how much delay an incident will cause: (1) the incident clearance time; and (2) the capacity reduction of the incident. Research relating congestion with incidents has primarily leveraged incident data sets that report on these key variables; however, the reality is that data sets that report this information are uncommon and often take a year or more to produce. Until standards have been developed for incident reporting, work is needed to demonstrate how to best use more commonly available data sets to relate incidents with congestion. The goal of this paper is to present methods for maximizing the use of operational incident data sets for non-recurrent congestion analysis. It uses Computer-Aided-Dispatch data from the California Highway Patrol, which is continuously available in real-time, to estimate the amount of delay caused by different types of incidents. It builds upon previous research by accounting for the fact that different types of incidents have varying congestion impacts depending on the corridor and the time of day. This paper applies the developed methodology to a segment of northbound Interstate-880 in the San Francisco Bay Area, where an Integrated Corridor Management project is in progress. The outputs of the methodology can be used to identify priority time periods and corridors for incident management strategies.