Optimizing Automatic Traffic Recorders Network in Minnesota
Principal Investigator(s):Diwakar Gupta, Former Professor, Mechanical Engineering
Accurate traffic counts are important for budgeting, traffic planning, and roadway design. With thousands of centerline miles of roadways, it is not possible to install continuous counters at all locations of interest (e.g., intersections). Therefore, at the vast majority of locations, MnDOT samples axle counts for short durations, called portable traffic recorder (PTR) sites, and obtains continuous counts from a small number of strategically important locations. The continuous-count data is leveraged to convert short-duration axle counts into average-annual-daily traffic counts. This requires estimation of seasonal adjustment factors (SAFs) and axle correction factors at shortcount locations. This project focused on developing a method for estimating SAFs for PTR sites. The continuous-count data were grouped into a small number of groups based on seasonal traffic-volume patterns. Traffic patterns at PTR sites were hypothesized by polling professional opinions and then verified by performing statistical tests. PTRs with matching seasonal patterns inherited SAFs from the corresponding continuous-count locations.
Researchers developed a survey tool, based on the analytic hierarchy process, to elicit professional judgments. MnDOT staff tested this tool. The statistical testing approach was based on bootstrapping and computer simulation and the tool was tested using simulated data. The results of this analysis show that in the majority of cases, three weekly samples, one in each of the three seasons, will suffice to reliably estimate traffic patterns. Data could be collected over several years to fit MnDOT’s available resources. Sites that require many weeks of data (say, more than five) may be candidates for installation of continuous counters.
- Project number: 2013075
- Start date: 04/2013
- Project status: Completed
- Research area: Transportation Safety and Traffic Flow
- Topics: Data and modeling