US-131 in Kalamazoo, Michigan

Work Zone and System Description

In the summer of 2004, the Michigan Department of Transportation (MDOT) deployed ITS technology in a work zone on US-131 in Kalamazoo. MDOT deployed a Dynamic Lane Merge (DLM) System in the northbound direction of US-131. The system did not cover the southbound direction. MDOT had used this type of system on multiple occasions in the past to mitigate impacts caused by work zone capacity reductions. The study team collected data during a two-week period in September 2004 to evaluate the benefits of the system. The system deployment area is shown in Figure 4. The construction area covered approximately 11 miles, immediately north of the beginning of the lane closure merge point shown in the figure.

Figure 4. Location of Dynamic Lane Merge System

MDOT designed the DLM System to require traffic to merge early to smooth traffic flow and reduce aggressive driving at the merge point. MDOT installed five trailers 1500 feet apart prior to the merge point, each equipped with lighted "Left Lane Do Not Pass When Flashing" signs. The trailer closest to the merge was always flashing. MDOT equipped each trailer with a Remote Traffic Microwave Sensor (RTMS) unit with the exception of the trailer furthest from the work zone. MDOT procured the system using a subcontract with the vendor through the prime construction contract.

The closed loop system operated based on traffic occupancy. For example, when sensor #1 detected the threshold occupancy, sensor #1 sent a message to sensor #2 alerting it to activate the flashing lights. The occupancy thresholds for sensors 1, 2, 3, and 4 were 5 percent, 7 percent, 9 percent, and 11 percent, respectively. Each sensor had a five-minute minimum activation period.

Focus of the Evaluation

The main objectives of the system deployment were to:

• Reduce aggressive driving at the merge point where two lanes were reduced to one in each direction.
• Smooth traffic flow through the merge area.
• Potentially reduce delay resulting from aggressive passing at the merge area.

The system deployment schedule mirrored the construction schedule, leaving little opportunity for establishment of baseline conditions (the "without ITS" scenario). Since construction was occurring in both directions and demand levels were similar in both directions, the study team planned to use the southbound approach to the merge area as the "without ITS" condition and the northbound approach as the "with ITS" condition. While this approach was not exactly the same as a true 'before and after' study, it would have provided a reasonable comparison of a traditional merge setup with a dynamic merge setup. From the past volume data, the northbound and southbound afternoon peak hours showed the most similar demand levels, so the study team focused data collection efforts on each afternoon period.

The study team collected data for a two-week period from September 20 until October 1, 2004. During this period, study team members performed travel time runs, using Global Positioning System (GPS)-equipped vehicles, in each direction from 3:30 pm until 6 pm each weekday. The study team also collected traffic count data along two mainline areas and seven ramp areas to determine demand at different locations within the merge areas. The study team also collected data on forced merges (last minute merges near the lane drop), dangerous merges (quickly merging after passing a vehicle), and lane straddling during the travel time runs. A modem was installed on one trailer and MDOT archived sensor data for August, September, and October for use in the evaluation.

The study team focused on two main hypotheses for this system, including:

• The use of ITS in work zones will enhance the safety performance of the highway.
• The use of ITS in work zones will reduce traveler delay.

Findings

While the study team originally intended to use the southbound side (without ITS) as the comparison site, congestion was not an issue on the southbound side, which meant that it would not be a good comparison. Therefore, the study team compared the time periods when the system was flashing (i.e., activated) with the time periods when the system was not flashing (i.e., not activated - trailers upstream of trailer 1 were not flashing), all for the northbound direction of travel, instead of comparing northbound to southbound.

The results of the comparisons show that there was a significant reduction in the number of forced merges when the system was activated and the flashers were "on", potentially reducing the risk of collisions near the merge taper. Queues were present on the days the flashers were "on", indicating that vehicles likely merged in advance to avoid forced merges approaching the taper. The study team observed three times the number of dangerous merges and seven times the number of forced merges during periods where the flashers were off compared with periods where the flashers were on.

Based on the effect of "do not pass" flashers in the work zone, the study team concludes that the DLM system enhanced safety by reducing the number of forced merges and dangerous merges at the lane drop. Thus the hypothesis, "The use of ITS enhanced the safety performance of the highway during construction," is valid. The study team was unable to use crash and citation statistics to further prove the hypothesis. It should be noted that there was occasional police presence, and the study team noted it in the data. However, we were unable to directly link the police presence to the effects on traffic.

The study team compared travel time data for two days when "do not pass" lights were flashing with data from the six days when the lights were not flashing. To ensure that the travel time comparisons would be meaningful, the study team checked for differences in volume counts between the two days. The study team found no statistically significant difference in volume count between the two days, and thus, the travel time comparison should be meaningful.

Key Finding: The DLM system improved safety with a reduction in the number of forced merges by a factor of seven, and a reduction in the number of dangerous merges by a factor of three at the lane drop, and by smoothing traffic flow through the work zone taper.

The study team found that the travel time when lights were flashing was significantly higher than when lights were not flashing. This increase in travel time is expected when drivers obey the "Do not pass" sign, as vehicles potentially reduce speed and merge at locations further away from the taper and form an organized queue of vehicles with smoother flow through the bottleneck. By having vehicles line up earlier in the open lane, the distance drivers traveled on the open lane increased, thus increasing the density and reducing the speed for that lane. For lower volume situations, lining up vehicles in the open lane, even though slightly increasing the travel time, is often a better option (if queue storage is available) than allowing traffic flow breakdown due to forced merges or dangerous maneuvers. From these results, the study team deems the second hypothesis "The use of ITS in work zones will reduce traveler delay" inconclusive. The DLM system, such as was used in Michigan, may help reduce traveler delay when traffic is high enough to warrant additional activation time.

Tips and Lessons Learned

Educating the police enforcement community about the system when it is first deployed is extremely beneficial. MDOT held meetings with the local police agencies to ensure adequate understanding of the system, thereby maximizing the benefits of enforcement. Police presence is an important aspect of this type of deployment, as motorists may be less likely to violate the no passing zone when police are present.

Implementing agencies should use the media to the fullest extent possible to help educate the public on the deployment. In Michigan, the DOT often invites the media to the job site to learn about the system, why it is being used, and how it works. MDOT stated that media representatives were a very important stakeholder in the process of deploying the lane merge system in Kalamazoo.

Previous | Next