This study focused on field implementation and testing of a Coordinated Ramp Metering (CRM) algorithm on 11 on-ramps located along a nine-mile-long corridor of California State Route 99 Northbound (SR99 NB) in Sacramento, CA in 2016. The CRM algorithm was fined-tuned using a microscopic simulation model of the study site developed in a commercially available traffic simulator. The algorithm calculated the optimal ramp metering rate for maximizing Vehicle-Miles-Traveled (VMT) and minimizing Vehicle-Hours-Traveled (VHT).
After fine-tuning the CRM algorithm using a microscopic simulation of the test site, the research team worked closely with California DOT (Caltrans) for (a) refining the Concept of Operations (ConOps) for the overall system structure; (b) real-time data acquisition from a commercially available controller in the field and establishing correct data mapping between field detectors and the controller in the CRM algorithm; (c) implementing the CRM algorithms as real-time code running on a Partners for Advanced Transportation Technology (PATH) computer; (d) estimating real-time traffic state parameters; (e) system integration of all software modules and hardware components; (f) conducting three weeks of dry-run tests, two weeks of progressive switching-on, system tuning and preliminary test, and five weeks of extensive testing and data collection; and (g) accomplishing performance analysis with Performance Measurement System (PeMS). VMT, VHT, and the ratio VMT/VHT obtained during the field tests in 2016 (after CRM application) were compared with data from 2015 in the same period (before CRM application).
- The performance evaluation over the five weeks of data showed that VMT was increased by 5.39 percent on average, and VHT was decreased by 1.64 percent on average, for AM peak hours (6:00am – 9:00am) which usually had congested traffic.
- The ratio VMT/VHT (freeway efficiency) was increased by 7.25 percent on average, for the AM peak hours.
- For PM peak hours (3:00pm – 6:00pm), VMT was increased by 2.56 percent on average, and VHT was increased by 3.04 percent on average.
- The ratio VMT/VHT was decreased by 0.44 percent on average during the PM peak hours, which was within the statistical error margin. The result meant that the CRM algorithm could not improve PM traffic. The reason could be that the traffic was not congested most of the time in the PM peak hours, so the CRM algorithm could not improve it. Therefore, it was concluded that the CRM algorithm could be effective for congested traffic.