1Ramp Meters on Trial: Evidence from the Twin CitiesRamp Meters Shut-offDavid Levinson1 Lei Zhang2, Shantanu Das2, Atif Sheikh2Department of Civil EngineeringUniversity of Minnesota500 Pillsbury Drive SEMinneapolis, MN [email protected]. Assistant Professor, 2. Research AssistantDRAFT Nov. 16, 2001ABSTRACTRamp meters in the Twin Cities have been the subject of a recent test of theireffectiveness, involving turning them off for 8 weeks. This paper analyzes the resultswith and without ramp metering for several representative freeways during the afternoonpeak period, depending on data availability. Seven performance measures: accessibility,mobility, equity, productivity, consumer surplus, travel time variation and travel demandresponses are compared. It is found that ramp meters are particularly helpful for longtrips relative to short trips. On Route 169, trips more than 3 exits in length benefit, whilethose 3 exits or less are hurt by ramp meters. Ramp metering, while generally beneficialto freeway segments, may not improve trip travel times (including ramp delays). Traveltime variation is reduced as another benefit from ramp meters. Non-work trips and worktrips do respond to ramp meters, but via different ways. The results are mixed, suggestinga more refined ramp control algorithm which explicitly considers ramp delay is in order.Key Words: Ramp Meters, Evaluation, Equity, Mobility, Accessibility, Productivity,Consumer's Surplus, Travel Time Variation, Travel Demand1. INTRODUCTIONIn what may be the single most comprehensive experiment in the history ofsurface transportation, ramp meters in the Twin Cities of Minneapolis and St. PaulMinnesota have been the subject of a recent test of their effectiveness. By turning themeters off for 8 weeks in October, November, and December 2001, it has been possibleto determine how well the metering system meets stated objectives, and to provide newinsight into the appropriateness of those objectives. This paper presents the detailedresults of an analysis of observed data of several representative freeways, US 169Northbound and Southbound, I-94 eastbound, I494 Outer-loop/Inner-loop and TH62westbound during the afternoon peak period, and considers a variety of measures ofeffectiveness resulting with and without ramp meters to control freeway traffic in theTwin Cities area. Earlier research (1) (2) identified a number of alternative performance2measures. These include accessibility, mobility, equity, productivity, consumers’surplus, travel time variation, and changes in travel demand.The first ramp meter was installed in the Twin Cities in 1970 on southbound I-35E north of downtown Saint Paul. Now after 30 years of evolution, meters are standardon many freeways. There are currently 443 meters regulating ramps throughout themetropolitan area. Figure 1 displays the number of ramp meters that were put intooperation each year. Since the first installation of ramp meters, which operated as anisolated system there has been sustained improvement in the system. Now most of theramp meters are controlled centrally in real-time. Also, initially there were single laneramp meters, but to better utilize the system so that it does not affect the arterial orconnecting roads, the usual practice now is to have two-lane ramp meters, whichincreases the storage capacity of ramps.Ramp meters in the Twin Cities were intended “to optimize flow in metro areafreeway corridors by making efficient use of available transportation facilities” (Mn/DOT1996 (3)). The Minnesota Department of Transportation made this goal operationalthrough a control strategy that divided freeways into zones terminating at bottlenecks.The number of vehicles in each zone at a given time was constrained by the capacity ofthe bottleneck. Ramp metering was used to limit those vehicles. The metering zoneequation can be expressed as:SBXFMUA++=+++(1)s.t. incident override and occupancy overrideWhere:A upstream mainline volume (measured variable);U sum of unmetered entrance ramp volumes (measured variable);M sum of metered local access ramp volumes (controlled variable);F sum of metered freeway to freeway access ramp volumes (controlled variable);X sum of exit ramp volumes (measured variables);B downstream bottleneck volume at capacity (constant);S space available within the zone (volume based on a measured variable).For more details on the Minnesota algorithm, readers may refer to reference (3)(4)(5).However, controversy arose when a State Senator from rural Minnesotachallenged the conventional wisdom of the state Department of Transportation. The longdelays at some ramps (at times, though not generally, in excess of 20 minutes) to ensurethat the freeway remained free-flowing drew the ire of some commuters, who believedthe system was at best inefficiently managed. The state legislature passed a bill in Spring2000 requiring a ramp meter shut off experiment. This paper is an analysis of the datacollected with and without metering. All data used in this paper has passed the Mn/DOTTraffic Management Center continuity test for detector readings, which is an algorithmbased on flow conservation to check the accuracy of detector data.3The next section details the method used to measure travel times on ramps andfreeway segments from the data available. The following section outlines the variousperformance measures and shows how they are computed. Then results are presented forrepresentative freeways for each of the performance measures. Recommendations andconclusions are delivered at the end of this paper.2. MEASURING TRAVEL TIMESThis section summarizes the calculation methodology required to measure traveltimes (and speeds) on entrance ramps, freeway segments, and O-D pairs on a highwaywith and without ramp meters.The data collected for entrance ramps come in two types of paired data:• Departure rate (Qk) arrival rate (qk) pair in each time interval (k) obtained fromvolume detectors.• Departure rate (Qk) queue length (nk) pair for each time interval obtained fromvolume detectors and periodic visual observation of queue length by remote cameras.The second type data can be transformed to the first type by equation 2i (see end notes):1−−+=kkkknnQq (2)Where:qk the arrival rate in time interval k (vehicles/hour);Qk the departure rate in time interval k (vehicles/hour);nkthe queue length in time interval k (number of vehicles).Throughout the studied peak periods,
View Full Document
Unlocking...