I always tell them, you’re not mad at the button, you’re mad at the signal timing.
When a traffic signal is running under a coordinated plan, sometimes funny things can happen if splits don't cover any Ped phases fully (W+FDW+Y+AR), at least for an exclusive pedestrian phase and also if the force-off (Veh and PedApply) calculations are off. It's also dependent on what controller is on the field, as their software and methods can vary.
Broadway and Evans St in South Portland ME has a Trafficware Commander ATC controller and initially when the signal was retimed this June, Ped Phase 9 wouldn't get serviced at all when a call was placed. The new cycle lengths were half-cycled at 56s AM, 50s Md and 60s PM. The controller was still running the programmed split time (6-8s), however the ped phase (4 W) and clearance (30 FDW) would not be timed at all, resulting in the Ped phase getting skipped. Eventually after three cycles, the controller would detect that it hasn't serviced at all, resulting in the coordination plan failing and going into Free timing until the next programmed coordination pattern. While not the exact same thing as described below in this post... we did have instances of peds crossing against DW and lots of red light runners due to the controller acting wonky.
Until a pedestrian pushes the button, then just walks across on his own against the light after waiting for a few seconds and realizing there is no conflicting traffic anywhere near. Then by the time the walk signal activates and the arrow turns red, a handful of cars have arrived and now have to wait for no reason to make their left turn, which is not only annoying but also increases idling emissions.
This is a problem with HAWKs too, by the way. People push the button, get sick of waiting for the light to change, and jaywalk on their own as soon as a gap in traffic shows up. By the time the light does change, any pedestrians have long since crossed and left the scene.
Eventually it was discovered the following things caused this:
- The split for Phase 9 was too low (Must cover Min Green + Y + AR).
- No "dummy" phase was assigned to each side of the barrier in the coordinated ring (There was only a Phase 9 on Ring 1, nothing below on Ring 2).
- Stop-in-Walk (below) wasn't working correctly because of this - it only worked if Phase 5 (lagging prot left) was active and gapped out.
- Easy Calcs (default yield points, or window where phases can be serviced) weren't being calculated correctly (the yield calculation was resulting in a negative number).
The split for Phase 9 was bumped to 8s (W + AR), a "dummy" phase 13 was added to Ring 2 of the Ring/Barrier and the coordination strategy had to be changed from End to Begin of Green. In addition, since the Ped permissive window wasn't being calculated correctly, the Easy Calcs had to be manually calculated for the Veh/PedApply (permissive window) for all of the phases. Since these changes, there haven't been any issues with skipped peds causing long waits and causing the controller to fail the coord plan to go into Free.
Stop-In-Walk is a very important feature that allows the split time of a phase less than the minimum pedestrian requirements (sum of the walk + ped clearance + yellow + all-red clearance).
Stop-In-Walk causes the local cycle counter to “stop” during coordination if a force-off is applied to the phase and it is still timing walk or pedestrian clearance.
TL;DR: Broadway and Evans St in South Portland ME, under new Coordination plans, was skipping the exclusive Ped phase and failing Coord plans due to the split being too low and the controller incorrectly calculating when phases can be serviced, resulting in long ped wait times. Remedies were done to correct this issue and the intersection works as intended now with no more long ped waits or Coord Faults.
