The paper describes a methodology and its application to measure total, recurrent, and non-recurrent (incident related) delay on urban freeways. The methodology uses data from loop detectors and calculates the average and the probability distribution of delays. Application of the methodology to two real-life freeway corridors—one in Los Angeles and the other in the Bay Area—indicates that reliable measurement of congestion should also provide measures of uncertainty in congestion. In the two applications, incident-related delay is found to be between 13 to 30 percent of the total congestion delay during peak periods. The methodology also quantifies the congestion impacts on travel time and travel time variability.
