Let the expert handle it

Picture this: It's the peak of the afternoon rush hour in the city. The mid-July heat is radiating from the freeway pavement and driver frustration is escalating as conditions near gridlock. Just as the collective angst seems to have reached its peak, a maintenance truck begins to close down a lane of the freeway to take care of some routine striping maintenance. The drivers let out a collective sigh as they witness the scene before them: How could this be happening to me? How was the decision made to close a lane right in the middle of the afternoon rush?

While the above scenario represents some creative license, it presents just enough truth to hurt. Transportation decision makers are often faced with the challenge of implementing just enough lane closure time to accomplish the necessary maintenance or construction work while minimizing the negative effects to the traveling public. The vehicle capacity provided by a single lane in a congested urban area can represent the difference between free-flowing traffic and excessive delays and lengthy queues. This reality places a premium on the timing of temporary lane closures.     

With this in view, the Colorado Department of Transportation (CDOT) recently undertook an ambitious endeavor: develop a methodology for scheduling lane closures along state highway facilities within metro Denver to maximize the hours available for necessary maintenance while minimizing the inevitable delays to the driver.

Schedules for each state highway facility would be developed, then summarized in the form of a database that could be quickly consulted when lane closure decisions must be made. The system should be broad in scope, needing to cover approximately 920 miles of freeway and arterial roadways and to reflect traffic operations for over 16,000 different lane-closure scenarios possible in the Denver area.  

The database is in charge

The answer to CDOT's challenge came in the form of the Lane Closure Expert System. An expert system uses human knowledge to automate the solving of problems that typically require the repeated application of human intelligence. These expert systems represent the expert knowledge as data or rules within the computer. These rules and data can be called upon when needed to solve problems.

In the case of the lane closure system, the phrase "expert system" simply refers to a large database of information. This database divides every highway in the CDOT's Denver metro region into sections. A lane-closure schedule is provided for each section, specifying the number of weekday and weekend hours during which the closure of a single lane will be allowed in either direction along a state highway facility.

Providing a consistent and authoritative source for decision making, the system strikes an appropriate balance between delays to the traveling public in the work zone and the cost of construction and maintenance when making lane closure decisions. Each of the 24 hours in a typical day was examined to determine whether a lane closure would be appropriate at that time. 

Hours of operation

The question of how to schedule lane closures is certainly not new to the agenda of public agencies and research entities. Several state departments of transportation (DOT) around the U.S. have developed guidelines supporting the scheduling of lane closures.

The Ohio DOT determines freeway lane closure schedules for individual projects by performing an operational analysis of anticipated impacts of the lane closure on vehicle queueing upstream of the closure. Lane closures are disallowed during time periods when the resulting queue lengths exceed established tolerances. The Florida DOT has developed a policy for both freeways and arterials which establishes the capacity of a roadway segment with a lane closure in place. Lane closures are prohibited whenever traffic flows along the segment exceed the available capacity.

Several lane-closure decision support tools have been developed to assist decision makers in establishing lane-closure decisions. The Texas Transportation Institute (TTI) has developed a computer program called QUEWZ-92 to calculate the queues and user costs associated with work-zone lane closures. The Federal Highway Administration has developed a similar tool, called QuickZone, to estimate traffic delays caused by work zones and the subsequent cost in time to drivers. 

Prior to the development of the Lane Closure Expert System, lane-closure decisions in CDOT's Denver region were made primarily on the basis of field observations, previous experience and engineering judgment. The application of tools similar to those developed and used by other DOTs also was selectively used as a basis for these decisions. Project-specific decisions were required to determine an appropriate lane-closure schedule for each maintenance or construction project. Often these decisions were based more on judgment calls than on actual data and did not account for the unique traffic characteristics of each state highway facility.      

Things have changed with the implementation of the Lane Closure Expert System. The comprehensive system eliminates the need for a tedious analysis process every time a lane closure is necessary. By dividing the region's highways into sections, the unique characteristics of each are accounted for in section-specific schedules. By applying a uniform methodology in developing the schedules, the conditions created by lane closures throughout the region are more consistent and motorists have a sense of what level of delay to expect when a lane closure is encountered. The expert system bases lane closure schedules on actual data, accounting for the spatial and temporal variations in traffic patterns that typically occur throughout a large urban area.  

Closing instructions

An analytical procedure was developed to evaluate traffic characteristics throughout the network and develop appropriate lane closure schedules. The methodology used for freeway facilities compares traffic flows to capacity with a lane closed. Closures are not allowed when and where flows exceed capacity. In the case of the arterial analysis, closures were not allowed when and where the average delay per vehicle caused by the closure exceeded 10 minutes. The arterial delay calculations were performed using a spreadsheet implementation of arrival and departure curves.

Traffic data 

The accuracy of the lane closure schedules is only as good as the traffic data used in their development. Accordingly, a significant amount of traffic data was gathered in the development of the system. Daily traffic flows divided into hourly time periods were necessary to evaluate all hours as options for closure. This information was gathered from a variety of sources, including a comprehensive highway database maintained by CDOT and traffic count information from the Denver Regional Council of Governments. Where necessary, hourly distributions from the available sections were applied to those nearby sections lacking data.

Freeway results

Results of the analysis indicate that single-lane mainline freeway closures for construction and maintenance in the Denver metropolitan area would be allowed for an average of approximately 16 hours of the 24 during a typical weekday. For the weekend scenario, it was found that an average of approximately 18 of 24 hours would be eligible for lane closure. Night closures would be allowed on every freeway segment, while daylight closure time windows would be available on a less frequent basis.  

Results of arterial analyses

The results of the analyses produced lane-closure schedules more lenient than the freeway condition. On average, approximately 20 of 24 weekday hours and 21 of 24 Saturday or Sunday hours are available for single-lane closures along arterials. Allowing for an arterial delay tolerance of 10 minutes increases the closure options over mainline freeway options. Night-only closures apply only to a very limited selection of highly congested arterials. Daylight closure opportunities are evident across the area.  

The results of the arterial lane closure schedule calculations are displayed in a manner similar to the freeway calculations.

Nobody is fighting this system

Upon implementation of the expert system it was anticipated that the schedules would be questioned and critiqued by those using it. Therefore, it was important that a thorough procedure be followed to ensure its credibility (or reasonableness), accuracy, sufficiency and applicability to the region. The schedules were subjected to a review process involving CDOT personnel familiar with regional traffic characteristics. Thanks in part to a thorough quality assurance process prior to implementation, CDOT personnel have found the closure schedules easy to implement and useful as authoritative guidance for project planning. It is anticipated that the system will continue to undergo adjustments as field personnel implement the schedules. In addition, the schedules will be updated approximately every five years to account for inevitable changes in traffic patterns. 

It's still the peak of the afternoon rush hour in the city. The mid-July heat is radiating from the freeway pavement, and driver frustration is escalating as conditions near gridlock. Just as the collective angst seems to have reached its peak, a maintenance truck enters the state highway, and you sense the impending doom of a lane shutdown. Instead, the opposite happens. The crew removes the cones ahead of you and re-opens a lane to provide some needed relief to the harried drivers.

You must be in Denver.