Programs & Labs
Spring 2001
Under the virtual microscope
Dr. Eil Kwon
Ramp metering, the practice of regulating the rate at which cars enter an urban freeway by means of traffic signals at entrance ramps, is an increasingly common traffic management technique in metropolitan areas with extensive freeway networks. But despite the broad support that ramp metering in general enjoys among traffic managers and researchers, the benefits and drawbacks of particular metering algorithms continue to be widely debated by academics, politicians, and the press. In the Minneapolis-St. Paul metro area, three research teams studied different aspects of ramp metering during and after an experimental area-wide meter shutoff.
Although many cities have traffic signals on freeway entrance ramps, the algorithms that govern these signals' operation vary widely. Dr. Eil Kwon, director of the Advanced Traffic Systems Program at the ITS Institute, grouped currently operational algorithms into three categories and used computer simulations to compare the performance of these representative metering approaches under identical conditions.
Three types of traffic-responsive control algorithms were considered, each representing a strategy currently in place in a different U.S. metropolitan area.
Denver, Colorado, implements a strategy known as incremental group coordination, which divides the freeway into groups of metered entrance ramps. If any ramp in a group reaches a critical state, the upstream meters respond one by one to restrict traffic flow.
In Seattle, there are no explicit groups of meters; the system regulates each meter individually using a "fuzzy logic" algorithm which approximates human reasoning. Data from detectors surrounding each ramp are processed according to a set of weighted rules, and the the resulting rate implicitly reflects the coordination of multiple ramps.
In Minnesota's algorithm, meters are explicitly grouped into zones, with each zone's downstream boundary considered a traffic bottleneck. All meters in a given zone are adjusted simultaneously based on volume and occupancy data from detectors in the roadway. Also, unlike the other two algorithms, Minnesota's does not take into account the number of vehicles waiting on entrance ramps.
The simulation results led Kwon to conclude that, with proper calibration, Minnesota's brand of explicit zone-based coordination can outperform the other two metering strategies when it comes to maximizing flow on the mainline. However, the Minnesota algorithm yielded greater vehicle-hours of delay on entrance ramps because it does not explicitly reflect ramp queue conditions in rate calculations. Meanwhile, the other metering approaches showed their own strengths in different areas. Incremental group coordination proved relatively simple to implement, and the fuzzy-logic approach showed great strength in its flexibility when adapting to unusual use patterns.
One implication of this study, Kwon says, is that system-wide efficiency must be balanced against ramp-queue delay. A new control algorithm that targets the balanced management of corridor-wide traffic by adaptively coordinating ramp meters with adjacent intersections is currently being developed by a group led by Kwon.