In the past, many traffic forecasting efforts have focused on estimating 24-hr volumes, leaving the determination of design-hour volumes up to highway designers. These days, planners are becoming moreinvolved in estimating peak-hour volumes, but the available techniques for calculating peak-hour traffic are somewhat limited. In particular, it is often assumed that the percentage of daily traffic occurring in the peak hour will not change in the future. In addition, planners sometimes forget that, for a given link, the peak-hour volume cannot exceed the link's capacity, regardless of the increase in daily volume. It is hypothesized that, as 24-hr traffic volumes continue to increase, peak-hour volumes will not increase at the same rate. In fact, future roadway capacity limitations (as well as other factors) will force drivers to modify their trip departure times, mostlikely so as to travel in the shoulders of the peak. Other researchers have also hypothesized this, but a literature review disclosed no practical methods for forecasting the future flattening or shifting of the peak hour on a link-specific basis in response to increasedcongestion. A methodology is presented for projecting such a changein temporal patterns, on the basis of research conducted in the i-80 corridor in northern new jersey. This technique uses a modified poisson distribution to describe the spread of 4-hr volumes across each 15-min period. Although the resulting model structure is not free of flaws, it represents a reasonable attempt to estimate future changes in peaking and will hopefully stimulate further research into this important subject. This paper appears in transportation research record no. 1305, Finance, planning, programming, economic analysis, and land development 1991.
Abstract