Model Validation and Reasonableness Checking Manual

6.0 Time-Of-Day/Directional Split Factors

Peak period information serves many uses in transportation planning. Recent model improvement efforts have focused on improving the forecasting of peak period speeds which are used for air quality analysis and also for determining the competitiveness of transit over the automobile. Peak period volumes, both for highway and transit, are used to determine the necessary capacity of facilities and the resulting level-of-service.

The historical use of time-of-day factors (TODF) has been through post-processing of assignment results. Peak hour factors are applied to daily traffic volumes after assignment of a daily vehicle trip table. For example, peak hour volumes are often assumed to range between 8 and 12 percent of daily traffic volumes. These factors can vary by area type or facility type.

In light of recent emphasis on detailed analysis of congestion levels and peak period spreading, best-practice manuals have advocated models which use a pre-assignment approach. Three possible approaches include the following:

Factors applied before trip distribution
Factors applied before mode choice
Factors applied before traffic assignment

In each of these approaches, separate peak period and off-peak period trip tables are created before assignment. Daily traffic volumes are produced by summing the results of the time-of-day assignments for each link in the network. The pre-assignment method recognizes that the traffic volume on a link is composed of trips with different purposes, each having its own peaking characteristics. For example, work trips have well-defined peaks during the morning and afternoon. Shopping trips are more pronounced in the afternoon and also on weekends.

To improve the application of the peak factors, they can be stratified by mode of travel. Distinguishing factors by mode is important since auto and transit trips exhibit very different temporal distributions. Transit trips tend to have a more concentrated morning peak with evening trips dropping off substantially compared with auto trips.

Peaking characteristics also vary by geographic location, depending on the function of the corridor (radial vs. circumferential) and the presence of special generators (such as hospitals, universities). Much of the variation is accounted for by the stratification of trip purposes in trip generation and trip distribution.

In a highway assignment, peak period trip tables representing more than one hour are normally assigned while link capacities are specified in vehicles per hour. As a result, factors specifying the percentage of trips that take place within the peak hour of the time period being assigned are used to relate the hourly capacities to multiple-hour trip tables. The peak hour percentage of daily traffic varies according to the area type and functional class of a roadway link. For example, on urban freeways the peak hour might account for only 6 to 8 percent of the daily traffic because the road is congested all day long. A suburban collector might have as much as 12 to 14 percent of the daily volume during the peak hour.

Based on the detailed trip characteristics and impedances associated with equilibrium traffic assignment, it is illogical to perform twenty-four hour traffic assignments assuming a ten percent peak-hour factor as has been done in the past. Congestion on roadways occurs at specific times-of-day. In addition, the traffic mix at the different times-of-day is different. Morning peak period traffic is composed mainly of relatively long-distance work trips, whereas the mid-day period is composed mainly of shorter home-based non-work and non-home-based trips. The afternoon peak period is composed of both the longer work trips and the shorter non-work trips. Thus, in order to be consistent with the detailed theory of equilibrium traffic assignment, trips must be assigned by time-of-day.

Based on the information included in each trip record, the direction of the trip can also be determined as being a trip from home to a non-home location (i.e. a production zone to attraction zone trip) or a trip from a non-home location to the home of the trip maker (i.e. an attraction zone to production zone trip). The trip data should be summarized by the trip purposes.

Figure 6-1 presents a diurnal distribution derived from the National Personal Transportation Survey (from NCHRP 365).

Figure 6-1
Diurnal Distribution of Trips (NPTS)

Validation Tests
The following reasonableness checks for time-of-day factors should be performed:

Compare TOD factors used to create time-specific trip tables with secondary sources such as the NPTS and CTPP. In particular, review the following:
- Percent of trips by time-of-day by purpose
- Percent of trips by time-of-day by mode (total, in autos, and in transit)
- Percent of trips by time-of-day by direction (home to non-home, non-home to home)
Review and adjust peak hour factors used in assignment to relate volumes to hourly lane capacities.

Sensitivity analysis can be used to test the affects of changes in the peak hour factor, i.e. peak spreading, on assigned traffic volumes and speeds.

Initially the peak hour factors should be based on the hour with the highest continuous volume for each of the three time periods. However, during the validation process these factors may be adjusted if it becomes apparent that these factors produce volumes which are too high for one hour. This would indicate that there is a great deal of peak period spreading, in which congested conditions are spread over a longer time period than one hour.

Peak hour factors provide an indication of the peak hour volume within the peak period. For the morning and afternoon peaks, trips may be spread evenly throughout the peak period. In this case, the capacity factor is simply the inverse of the length of the period. For example, the two hour a.m. peak period has a peak hour factor of 0.5. The three and one-half hour p.m. peak period has a peak hour factor of 0.286.

For the off-peak period, the congestion is usually not as severe and a different method is used. Since speeds should reflect the off-peak conditions in which most of the trips take place, the off-peak hour factor typically represents the middle of the day (as opposed to the middle of the night).