As the resources for expanding street and highway capacity in urban areas have become increasingly scarce, interest has risen in accurately predicting the peak-hour capacity requirements for future years. Much of the travel demand forecasting performed around the country for highway planning has been performed on a twenty-four-hour basis, and peak-hour capacity needs have been estimated by applying a regional factor for the specific facility type or by using the current ratio of peak-hour to twenty-four-hour volume for a specific highway segment under consideration. This method is most often static and does not reflect the reduction in peaking that generally occurs as facilities become congested during the peak hour and trip makers adjust their travel time to avoid the peak. This paper reports the results of research on the peak-spreading phenomenon using traffic data from highway corridors in Arizona, Texas, and California. The data from each corridor covered a period of five to twenty years during which the congestion level in the peak period changed significantly. The research demonstrated that a clear and consistent pattern of peak-spreading emerged for highway facilities as congestion occurred during the three-hour morning and evening peak periods. The relationships derived from the research on peak-spreading have allowed the authors to develop a submodel for the UTPS UROAD assigned package. The new submodel will predict, for each link in the highway network, a peak-hour volume. That peak-hour volume reflects the level of congestion that would result from the predicted three-hour, peak-period volume for the forecast year and the level of capacity planned for each link; and it reflects the effect of peak-spreading that results from the predicted congestion on the facility. More accurate prediction of the peak-hour volumes is also expected to result in better prediction of peak-hour speeds for forecast years. This in turn should result in more accurate forecasting of travel time savings and air quality improvements from highway improvement projects. This paper appears in Transportation Research Record No. 1203, Demand Forecasting and Trip Generation-Route Choice Dynamics.
Abstract