The incident duration and traffic recovery time are the two critical periods for any incident management process. Since incident durations are available in the post-evaluation analyses, the reliability of estimating incident-induced delay significantly depends on the accuracy of recovery time estimation. The dynamics of the traffic system under the combined impact ofrecurrent and non-recurrent congestions are complex and thus make it difficult to quantify the incident impact from recurrent congestion. This paper extends the difference-in-travel-time method to estimate the traffic recovery time using both the incident and travel time data. The proposed method utilizes percentile statistics to establish the background conditions which represent travelers' anticipation under incident-free conditions and then employ the difference-in-the-travel-time concept and the information from the incident database to estimate the traffic recovery time. The variability of traffic recovery time estimates was also obtained using the proposed method. The case study conducted using the proposed method revealed that the method is fairly straightforward and robust in that the "ground truth" representation of incident impacts can be captured without any modelassumptions and calibrations. In addition, the method can be used to support the calculation of various measures as part of freeway and incident management performance monitoring programs.
Abstract