Recent statistics show that interstates serve over 25% of the nation’s total Vehicle Mile Traveled (VMT) and over 40% of all truck VMT. Incidents are one of the most frequent and most impactful contributors to unreliable travel times on freeways. Improving travel time reliability on freeways is very important, particularly given the economic impact of freeways to freight logistics as well as the time value for individual travelers.
Dr. Nagui Rouphail from North Carolina State University, participated in STRIDE Project I, a multi-institutional effort focused on freeway travel time reliability. Dr. Rouphail’s portion of the project investigated the effect of incidents on freeway segment capacity and sought to improve the freeway facility reliability method in the Highway Capacity Manual (HCM) and accompanying software, FREEVAL.
Currently, the Highway Capacity Manual provides a lookup table linking the remaining segment capacity fraction during an incident to the total and closed number of lanes on the segment. In reality, segment capacity during an incident will tend to vary over time, with the most severe effects felt early on before any type of incident response is initiated, followed by congestion progressively improving as the appropriate incident management actions are implemented.
The study included a portion of WB I-540 in Raleigh, NC. Using an enhanced Genetic Algorithm developed in this study, time-dependent calibration of capacity adjustments using probe data speeds as the target were carried out. The approach aimed at matching, over time, the HCM6 method speed predictions and those measured from vehicle probes at the time the incident occurred. Matching was done by adjusting the segment capacity under incident conditions until an acceptable match was obtained.
As expected, the trends of the resulting temporal adjustment factors were strongly associated with the progression time of the incident, with low values towards the start of the incident gradually increasing toward the end of the incident. This pattern enabled the researchers to develop a predictive linear model of a standardized capacity adjustment factor (CAF) that varies over time. This calibration factor to the fixed HCM CAF adjusts it to be lower in the early stages of the incident and higher later.
The difference in the fit between the current fixed and time-dependent factor was quite significant, resulting in improving the estimation of incident capacity by about 43%. Another important byproduct is that users will no longer have to guess about the duration of the incident effect, the model will generate the appropriate time factors that would inform the users when those effects will dissipate.
The proposed model will enable traffic engineers and analysts to better plan for incident response through improved knowledge of an incident carrying capacity over time, and inform methods to mitigate its effect on congestion, particularly during the peak hours. The proposed CAF calibration model is readily implementable and is available to all users of the HCM6 methodology and adjoining FREEVAL software.