Summary

The quantitative and qualitative separation of recurrent and non-recurrent congestion, and the contribution of different causes to freeway congestion, will enable the design of better correctional strategies. Typically, recurrent congestion is defined as "the regular, everyday rush-hour delays that occur when the design capacity of a freeway is exceeded and low vehicular speeds result" or, more technically, a "condition lasting for 15 minutes or longer where travel demand exceeds freeway design capacity [and] vehicular speeds [decline] to 35 mph or less during peak commute periods on a typical incident-free weekday." Non-recurrent congestion is the extra delay caused by causes like weather, incidents, lane closures/work zones, and special events, among others.

Little is known of the causes, magnitude and impact of non-recurrent congestion in the state. Depending on the agency and usage, different proportion of total congestion is attributed to non-recurrent congestion, ranging from 13% to 50% (1999 HICOMP Report [1], Lomax et al. [2], Skabardonis [3], and the Dowling Report below) but more comprehensive, quantitative and empirical work is needed in order to properly design, implement, and evaluate congestion mitigation strategies.

Among several studies addressing these issues, the recent study by Dowling Associates, Berkeley Transportation Systems and Systems Metric Group [4] presents a comprehensive survey of the issues and proposes a methodology for estimating non-recurrent and recurrent congestion. We refer to this work as the Dowling Report. The proposed research starts with the Dowling Report study and proposes improvements.

Goal

The goal of the proposed study is to develop an empirical model of causes and impact of non-recurrent congestion, similar in nature to that proposed in the Dowling Report, which can be applied to statewide facilities. We also aim to calibrate the model and apply it to different facilities in the state using loop detector data from PeMS [5]. Specific products of the proposed research will be:

1. An empirical model of causes and delay impact of non-recurrent congestion.
2. An application of the model to a long-term statewide data set from PeMS; the application will determine the magnitudes of non-recurrent congestion to the magnitude of recurrent congestion.
3. A study of the safety impact of non-recurrent congestion.
4. A procedure that Caltrans can use to put in place a continuing Non-recurrent Congestion Monitoring Program, based on PeMS. The procedure will be a refinement of the PeMS' HICOMP application, which estimates total congestion.