As for the photocell, the ones in the Brookfield PATH tunnel seem, as best as I can tell, maybe prime the motor for opening effort assistance when they detect someone approach, as I believe when you walk up to them while closed a light comes on or off on the bar at the top. Then I think there's a standard 5-second hold-open phase, but I believe that it waits until it detects nobody approaching to close, or maybe at least extends the 5-second hold to 15 or 20 seconds before requiring somebody nudge it back open.
And many hold-open systems can be programmable to be disabled during an alarm -- to stay compliant with a fire code. I'm not sure if it's configured this way at Union to revert to complete manual operation during a fire alarm, or simply modify variables (e.g. don't continuously hold open as long, disable motion sensors, run in only touch-activated/assisted-open mode).
Also, one possible pro
and con of hold-open is if there are doors open, lazy people begin walking towards them, avoiding the closed doors! This may create a peak-period bottleneck if opening effort is intense. On the other hand, this is good during the winter. If the opening effort is extremely light, this is not a problem since when there's any scant slowdown in pedestrian flow, people will approach the unused doors if it speeds up their walk.
So when there's lots of pedestrian traffic, regular commuters will bother to open unused doors, if the door-opening effort is already known to be light. So extra open doors appear and appear, and all doors continuously stay open (continually triggered by sensor) right on the dime of the peak period crush.
Many
older doors on TTC require amazingly strongman effort to open just one door. We need to avoid this situation, or everyone is just going to walk through the only open doors once the hold-open activates, even during peak period.
Ideal door opening effort is motor assisted, yet not-speed-governed, still allowing rushed commuters to yank-swing open as usual. You don't want a situation where you try to yank open the door but the motor insists opening the door at its own speed (like many automatic wheelchair door opening systems).
The best systems for motor-assisted door opener is fully programmable (adjustable door-opening resistance, like adjustable-resistance power steering in cars) -- while not self-speed-limiting (letting people swing open faster, just like a regular door). These good systems also can increase closing-assist power during windy days when the door is unable to close by itself.
But the best motor-assisted keep-open systems are great; you just yank at door and it opens lovely and easy just like a reasonably lightweight wooden door (despite being a heavy metal-glass door), without its manual-open swing speed being too severely limited (like when a wheelchair automatic door is currently motoring itself open).
There were also flawed systems deployed elsewhere in the world. For example, you pull, and the door insists on opening slowly. You don't want to create undue new bottlenecks because of a cheap, flawed hold-open system with high-door-opening resistance and/or speed-limited door-opening that you can't yank faster.
Hopefully the opening-effort is fairly light (e.g. motor power-assisted) so that it will quickly cascade to an all-open scenario during a peak-period crush. People avoid high-resistance heavy doors intentionally, preferring to crush through fewer open doors.
That said, door-opening effort needs to be a sweet spot (not supermarket automatic, just lighter opening effort, enough to avoid "press the wheelchair button" laziness) so some doors stay closed during winter, in moderate pedestrian traffic situations, to keep the teamway warm. Door opening resistance adjusted to balancing intentional encouragement of laziness (during lower pedestrian traffic, avoid unnecessary extra numbers of open doors in winter) and encouraging of all doors open (for peak period crush).
