METHODOLOGY

The coordinated ramp metering algorithm takes available on-ramp storage into account and dynamically reduces the cycle length of the feeding intersection signal control in order to avoid on-ramp queue spillback and mitigate unnecessary delay in the conflicting directions. The proposed algorithm was tested in the morning peak during a four-month period in 2019 along a four-mile long four-lane corridor on an interstate highway in San Jose, California. Total travel time (TTT), total delay (TD), and vehicle-miles traveled (VMT) were used to determine the effectiveness of the control strategies developed in this study to improve the day-to-day operation on the freeway corridor and the relevant arterial intersections. The on-ramp occupancy based on queue detectors at each on-ramp were used to determine of the presence or severity of on-ramp queue spillback with the coordination of freeway ramp metering and arterial traffic signals. Four different scenarios were tested during the four-month test period:

• Scenario One was the baseline that used the local responsive ramp metering without the coordination of ramp metering and arterial traffic signals.
• Scenario Two was the local responsive ramp metering with coordination between ramp metering and arterial traffic signals.
• Scenario Three was the Coordinated Ramp Metering (CRM) Algorithm developed in this study but implemented without coordination between ramp metering and arterial traffic signals.
• Scenario Four was the CRM algorithm implemented with coordination between ramp metering and arterial traffic signals.

FINDINGS

• Comparing Scenarios one and two, it was found that the proposed CRM algorithm reduced the freeway TD by up to 11.49 percent.
• Comparing Scenarios three and four, it was determined that a 5.92 percent reduction in TD was achieved.
• In terms of TTT, a 5.79 percent reduction was observed based on a comparison between Scenarios one and two.
• Comparing Scenarios three and four revealed a 2.75 percent reduction in TTT.

Based on the observed data from on-ramp queue detectors, the CRM algorithm was also able to mitigate the arterial spillback. In Scenario 4, none of the on-ramp queue detectors showed higher than 40 percent occupancy (the threshold for queue spillback and the ramp metering controller to relax ramp metering rates to alleviate queue spillback) while the queue detector occupancies were higher than 40 percent in Scenario 3 without the coordination of freeway ramp metering and arterial traffic signals.

The study recommended that further tests needed to be conducted during different times of the year with more fluctuations in traffic demand. Also, larger and more complex freeway corridors could be used for implementation to assess the longer-term feasibility and robustness of the proposed coordination and control strategies.