Most traffic engineers in the United States do level of service calculations per the Transportation Research Board's Highway Capacity Manual. It has been the default manual on how to determine if roads or intersections are operating acceptably or not since the 1950's (both under existing conditions and future forecasted conditions).
There are several factors in the level of service analyses that attempt to hone in on different characteristics that might affect how a road or intersection operates. A big one is the Peak Hour Factor (PHF), which is a measure of how concentrated traffic is during the busiest portion of the peak hour.
The PHF at a given intersection is the sum of the traffic entering the intersection over the busiest 60 minutes divided by four times the entering volume of the busiest 15 minutes within that hour.
A PHF of 1.0 means traffic levels are evenly spread out over the whole hour where a lower number of 0.80 or less means traffic spikes for a short period (usually near a factory, church, school or some other use with a surge in traffic).
Highway Capacity Manual recommends a
default of 0.92 for intersections that have more than 1,000 vehicles entering
in the peak hour and 0.90 for less busy intersections.
We decided to enter the counts we've done since 2006 into a spreadsheet to see if the 0.92 and 0.90 defaults make sense (thanks Lindsay). Here's the spreadsheet -
Download Peak Hour Factor MN Analysis.
It contains a little data from Wisconsin and Iowa, but mostly has a.m. and p.m. peak hour PHFs. Below is a pivot table of the average PHFs broken down by Urban (Minneapolis & St. Paul), Suburban (part of the Minneapolis/St. Paul metro), and Out State (locations not adjacent to the Minneapolis/St. Paul metro). Also below is a chart of the average PHF, by classification and by year.
This data confirms the 0.92 default is reasonable for Minnesota. I am a little surprised though by the Urban and Suburban PHFs – I thought they would be a little higher. Even though we have 3,000+ PHFs in the spreadsheet, I don't think there's a large enough sample size to start honing in on annual variations. To play off last week's post, there's probably more noise in that chart than signal. And please don't put to much weight in the Iowa and Wisconsin data – they are very small samples.
Feel free to use the spreadsheet and these results. If there's any interest, we could setup some type of open source google docs spreadsheet. I think it would be really interesting to have a large enough dataset to develop historic PHFs for specific communities.
Thanks for compiling this data. Did you consider checking the PHF against total intersection volume? I’ve noticed that all else being equal, the larger the intersection (in terms of volume), the higher the PHF.