Wednesday, August 21, 2013

Adaptive Control vs. Time of Day Signal Timing Plans

Interesting article that is worthy of a blog post by itself: http://www.westernite.org/awards/bestpaper/2013-Weir-Shykowski.pdf

There is some debate over the benefits of Adaptive Control on the ITE listserv, here's my response:
Adaptive control is something that I thought was an effective strategy for dealing with changing traffic conditions. As a practitioner that has experience with the SCATS system, completed evaluation for the (now defunct?) OPAC initiative, and read the evaluations of the ACS Lite implementations completed by contractors (Sabra Wang) for the FHWA, it seems that they have their place, but the costs of the software and the modified detection are a lot for an agency to bear. There is also the cost of learning the software and having the maintenance and engineering staff know how to operate two different systems. As an agency with 1,080 signals, the City of Portland has 13 that are adaptive. The additional benefit of the 1%, that the City spent a significant amount of money to "improve" may have been better spent selectively throughout the system on technology that would have put information in the hands of the engineers that operate the remaining 99% of the signals. One could argue, (I could make this argument myself), that the limitation of resources across the board should not be the key limiting factor that we use in the cases, and there are communities like Bellevue, WA and  Oakland County, MI (among others) that have installed enough of an adaptive system that their costs per intersection have come down significantly. My point is this, if an agency has a critical mass of intersections to apply an adaptive system than the engineers and maintenance staff have the ability to learn the system well enough to be effective in its operation. I will still contend there is a lot of functionality in our existing systems that we aren't using and effective gap setting, improved measurement of performance, and recognition and response of problems would serve us well. Here are my takeaways, I am curious if folks agree.

There are several situations where it is worth considering:
1. Surrounding event centers where the traffic ingress/egress are so variable that there isn't a great way to handle this (Portland uses some responsive control that works reasonably well at the NBA arena in Portland).
2. Applied where there is a significant number of intersections where the focus of several staff will handle the challenge of learning the new system.
3. If an agency is looking to mitigate congestion (as oppose to widening an existing street), adaptive control would likely increase capacity by being responsive to traffic increases in the shoulders of the peak by raising the cycle length earlier than a time of day plan. This is my opinion,  studying these effects has not been done carefully (the Portland system included).

Significant limitations of the systems that need to be considered:
1. Limited ability to prioritize transit within the systems. The first answer from our vendor related to transit priority in the adaptive system was non-responsive.
2. Oversaturated operations. Does the system work better when traffic is queued up between intersections? Have you engineered the adaptive system to recognize when congestion occurs?



No comments: