Great post in Talking Transportation blog regarding what must be done to reach consumers of transportation news in this age of social media:
The fourth TRB paper I would like to discuss is paper number 14-4109: “Lane Bias Issues in Work Zone Travel Time Measurement and Reporting” by a group of 8 authors from the Georgia Institute of Technology and several other organizations.
For this study they chose to test BlueTooth and License Plate Recognition (ALPR) systems as ways to collect and analyze travel times through work zones. They found many problems with doing this but I’d like to share them as the results provide information that may help you deploying any type of travel time system.
The study was conducted on two projects in the Atlanta area on 4 and 5 lane freeways. System data was compared to manual times. These manual times were arrived at by watching date & time stamped film from each end of the project.
The researchers found that in free flow traffic these sensors provided reasonably accurate travel time information. But when traffic slowed, they found both systems showed slower travel times than were actually the case.
The primary reason for this error is that the sensors are generally placed on the shoulder. Traffic in the slow lane often stays there. Slower traffic like large trucks don’t generally use the fast lanes. So if you collect data on the shoulder it will naturally collect speeds of slower traffic.
Another issue with the ALPR was occlusion. The cameras were mounted on a tripod on the road shoulder. The likelihood of capturing the plate number of a vehicle traveling in the fast lane at both the beginning and end of the job was quite low. The sample will be heavily weighted with slower vehicles as the camera often won’t be able to see the interior lanes.
Another finding was that this bias got worse as traffic slowed. The slower the traffic the worse the travel time was over-stated. The authors felt it is better to err on the side of over stating travel time, than understating it. And I agree. But over-stating it by more than a few minutes on a consistent basis undermines credibility.
It also makes these two systems poor choices for incentive/disincentive programs with the contractor. The over stated travel time would favor the agency and would cost the contractor, both in money and in the opportunity cost equal to the time they could have left the lane(s) closed to continue working.
Finally, this data is not something agencies will want to share with FHWA as part of any work zone performance measure program. In this case, it is the DOT that would be penalized.
In fairness, this study was conducted on 4 and 5 lane freeways. The error would have been significantly less on 2 lane freeways or expressways. And if you are only using these systems to report travel time to motorists, these errors could be acceptable. But if that data will be used for anything more than that, it sounds like another technology might be better.
You must always take the time to find the best sensors for each situation. Each type has its own particular strengths and weaknesses. More than anything else, I think that’s what this study points out. You just can’t take a one size fits all approach when it comes to measuring travel times.
The third TRB paper I would like to discuss is paper number 14-3840: “Work Zone VS. Non-Work Zone: Risk Factors Leading to Rear-End and Sideswipe Collisions” by Claire Silverstein, Justin Schorr and Samer Hamdar of George Washington University.
I co-chair California’s work zone safety committee. One of our members, Pat Fyhrie of U. C. Davis Advanced Highway Maintenance & Construction Technology (AHMCT) research group has been reconstructing work zone crashes to better understand where the majority of crashes occur. She has found that most are rear-end or sideswipe crashes and a majority of those occur alongside the work area within the work zone.
This TRB paper takes this a step further. They have found that rear end and sideswipe crashes occur in work zones more often when “there is no precipitation, during daylight conditions, with limited roadway curvature, and when increasing number of lanes and speed limits.” In other words, the results are counter-intuitive. The “safer” the work zone appears, the less safe they become.
This makes more sense when you look at it from the driver’s perspective. He or she will slow down when it rains, or when the lanes are narrowed or shifted, or when they are traveling near temporary concrete barrier. But when the road straightens out and opens up, it’s pedal to the metal time!
The paper comes to the conclusion that there are many opportunities to use technology to help prevent these crashes. Some of that technology will (eventually) be in-vehicle. Automated enforcement should be considered. But a third area of opportunity is work zone ITS. The authors write, “potential ITS applications should be targeted at creating safer traffic flow conditions by encouraging safer driver maneuvers.”
Queue warning systems, Variable Speed Limit systems, and Dynamic Merge systems come quickly to mind. All three are low cost systems that are easy to deploy and that help to smooth traffic flow and reduce speed variance – the primary cause of both rear-end and sideswipe crashes. V2V systems will one day do this automatically. But that day is still many years away. Work zone ITS can do it today, dependably and cost effectively.