Summary and Objectives

On-ramp metering strategies that increase the capacity of a freeway / on-ramp merge will be experimentally tested. We emphasize the objective here is to determine if strategic ramp metering can increase the outflow or capacity of an active merge bottleneck. (An "active" bottleneck is one marked by queueing immediately upstream and by freely flowing traffic conditions downstream.) The objective is not merely to verify that metering can diminish freeway queueing and promote higher freeway flows upstream of bottlenecks. That metering can produce these kinds of higher flows is indisputable. But whether or not metering can generate higher capacity discharging from an active merge bottleneck has not, until now, been properly verified.

With this key distinction in mind, the work will build upon some of our preliminary experiments at a freeway merge. We have recently shown that sufficiently restrictive ramp metering can increase outflow from an active merge bottleneck by 10 percent or more. This is because the metering can diminish disruptive vehicle lane-changing maneuvers that would otherwise arise at the merge. These preliminary findings serve as a basis for the work to be performed under the upcoming PATH contract.

In that capacity reductions are apparently caused by disruptive lane-changing maneuvers, we suspect the relation between on-ramp metering rate and merge capacity resemble s something like what is shown in the accompanying figure; i.e., the relation takes a convex form. If this proves the case, an improved freeway management strategy would be as described below.

The ramp's metering rate periodically switches between a low rate (as exemplified by the rate corresponding to point 1 in the figure) and a high rate (point 2). The combined effect of these alternating metering rates would be a longer-run average comparable to the rush-period rate currently used at the study site's on-ramp (700 vph). Note that by periodically deploying the high rate (point 2), the on-ramp's queues are precluded from growing too long. And very importantly, the merge capacity would no longer be the one generated by a fixed metering rate of 700 vph (qa in the figure). Rather, the longer-run average capacity would be a higher outflow, qA, as described by point "A" lying along the chord connecting points 1 and 2. The increase in capacity would therefore be the vertical distance between points qa and qA and this might be quite large.

Methodology and Work Plan

We will perform on-ramp metering experiments to determine if the relation between metering rate and capacity actually takes a convex shape, similar to what is shown in the figure. To this end, we have a identified a study site; the northbound Freeway 805 at the 47th St/Palm Ave on-ramp in San Diego, California. (This is the site used in our previous studies of metering and merge capacity.) It is ideal for our upcoming experiments in every respect: the merge becomes an active bottleneck during morning rush periods and the site has upstream freeway loop detectors for identifying this bottleneck's activation times; there are multiple freeway over-crossings nearby from which to videotape traffic and thereby record merge operation in great detail; the on-ramp is metered and there is sufficient space for storing the on-ramp queues without impacting upstream signalized intersections.

Each day when the merge bottleneck becomes active (as identified from measurements taken by the freeway loop detectors), the on-ramp's metering rate will be reduced to perhaps 400 vph. This restrictive rate would persist for no more than 10 minutes. After this time, the metering rate will gradually increase (e.g. in increments of 100 vph) until reaching a high rate (of perhaps 900 or 1,000 vph). During each day's experiment, we will collect traffic data using multiple video cameras positioned on the site's nearby over-crossings. We have techniques for processing these data in high resolution (as described in the original proposal) so that we can conclusively identify the relation between metering rate and capacity.
figure 1

Project Completed Final Report: UCB-ITS-PRR-2005-28