I found this on a funny site but I can't help but wonder how this could happen? Maybe the MMU shorted out? Anyone that has experience in traffic lights, I'd like to get your views. BTW, no idea if this is real or photoshopped but it looked very real on the site.

It could theoretically happen if the controller got screwed up, but I'm no traffic light engineer.

The picture on the site certainly looks real enough.
Real or not I'm certain that I've seen this happen but I figured if it did that the MMU would throw the intersection to flashing mode unless of course the MMU is faulty.

I've seen it happen before. One of the intersections here has the red on the protected left signal heads stuck on. I'm guessing the load switch for that phase shorted closed and the MMU didn't catch it. The red just stays on during the green arrow and amber lights.

You can get a similar picture at any traffic light with some photographic trickery. You need to take a long exposure of a signal as it changes state from red to green. Adjust aperture to suit and/or use a filter to ensure correct exposure.

With LED signals it is enough to snap at the moment the lights switch from red to green: The green turns on immediately, but the red holds on from its filter capacitor for some 100ms or so. If you snap the picture exactly at that time, you get both on on it.

Other aspect may be the real fault: The "stuck ON red" may not be considered a danger, so it is not treated by the conflict monitor.
I may agree with the part that it is not as severe problem as "stuck ON green", but still very potent to create dangerous confusion, so I would treat it the same way as any other conflict: Shut the signal down (and so turn the intersection to the default priorities as defined in laws valid there - here it means the traffic signs take over and when none, the right hand priority rule).

Or it could be a design weakness in the conflict detector vs the fault:
If the signals are controlled by AC solid state switching elements, they may fail in the way they pass only one polarity of current. With triacs the frequent failure is the A1->A2 direction the triac conducts all the time, the A2->A1 current direction still responding to the control.
And if at the same time the conflict monitor is implemented responding only to one polarity (often the positive; typical when using optocouplers with a single diode rectifier driving its LED).
When the detected polarity is the opposite than the triacs tend to fail, that failure remains undetected.
And many LED ballasts still drive the LEDs on that single polarity, so the lamp stays lit, often not even dimmed down as typical for incandescents.
With LV incandescents with their local transformers (each lamp having its own transformer), this fault results into transformer saturation and so blown out fuse (and that is then detected correctly).

Or when using just a capacitive dropper circuit to ballast the optocoupler LED. Then the rectifying triac faul;t yields to just a DC voltage, so there is no current to the optocoupler LED, so the circuit evaluates the state as OFF. If the same happened with incandescents, their resistive nature would mean there will still be 115V (in 230V area)/60V (in 120V area) AC component still driving the optocoupler LED, usually still enough to make the circuit to see it as powered, so detecting the fault.


The thing is, many LED ballasts contain input filter capacitors. And triacs are sensitive to di/dt, so switching even rather small capacitor directly is very damaging for the triacs, leading often to the rectification failure mentioned above.
So the problem may have been easily triggered by just replacing the incandescents for LEDs, without adopting the controller (adding inrush current limiters,...). But even there, the main fault would be (if that is what happened) the "rectifying triac" fault not being detected properly.

At Zevulun/Hatealla junction, there was once a pedestrian light that both green and red light turned on at once during the green phase: https://www.lighting-gallery.net/gallery/displayimage.php?pos=-188307

...here it means the traffic signs take over and when none, the right hand priority rule.

So the car to the right gets priority if it reaches the junction first, but what happens if all the cars arrive at the same time.?? 

With LED signals it is enough to snap at the moment the lights switch from red to green: The green turns on immediately, but the red holds on from its filter capacitor for some 100ms or so. If you snap the picture exactly at that time, you get both on on it.

LED signals don't have filtering capacitor, as they have 100hz flickering when observing with the camera, as the flicker free light isn't important here like in general lighting.

LED signals don't have filtering capacitor,

I wouldn't make this as a general statement about all LED signals, but rather some LED signals.
Pretty sure some here do have filtering/or are running on true DC. There's no flickering, and even on a cheap dashcam they just look normal 'on' (unlike those aweful LED tail-lights and/or the un-filtered LED signals)

So the car to the right gets priority if it reaches the junction first, but what happens if all the cars arrive at the same time.?? 

Here it does not matter who arrives first. If you have conflicting traffic from your right hand side, he has the priority, so you have to wait.

I thought UK used similar rule (??), maybe sides reversed (because of the left hand side driving there?), but otherwise the same concept, when the priority is not distinguished by other means (signage, signals, a cop controlling the traffic by hands,...).

@ Medved - When I've driven in the US and come to one of those 4-way intersections with no lights and apparently equal priority from all directions, it seemed to be whoever had the most nerve went first 

In the UK, we don't tend to have those junctions, most of them will always have some form of marked priority, with stop or give way signs. Even in very rural areas, where you might not have any signs, it's usually obvious which is the major road and that will always have priority.

But the question was, if 4 cars all arrive at the same time, who has priority..? Unlikely, admittedly, but still possible.

@ Medved - When I've driven in the US and come to one of those 4-way intersections with no lights and apparently equal priority from all directions, it seemed to be whoever had the most nerve went first 

In the UK, we don't tend to have those junctions, most of them will always have some form of marked priority, with stop or give way signs. Even in very rural areas, where you might not have any signs, it's usually obvious which is the major road and that will always have priority.

But the question was, if 4 cars all arrive at the same time, who has priority..? Unlikely, admittedly, but still possible.

I was not talking about US... :-D

If 4 cars arrive, then the solution relies on at least one of them not being a moron and guesturing the next one to go ahead...
But all cars have to stop anyway (because all have to give way to some other), so even if it turns into a total confusion, all start approaching, the entropy in the world will make sure one of them becomes the first to pass.
And if all are morons and so collide, they deserve that for being morons. Anyway the collision won't be from any significant speed, so not that much heavy damage... :-D

I found this on a funny site but I can't help but wonder how this could happen? Maybe the MMU shorted out? Anyone that has experience in traffic lights, I'd like to get your views. BTW, no idea if this is real or photoshopped but it looked very real on the site.

The mercury contactor (MC) is "stuck".

Likely, this signal was in "flash" with the reds flashing, and the CM (conflict monitor, most newer MMU cabinets have replaced the MC with an SSR) dropped the MC, but the MC gets "stuck" and keeps powering the loadswitches.
When I see this happen, kicking the side of the cabinet with the MC will allow the MC to open, removing loadswitch power and leaving just the reds or red/yellows flashing.

Percussive maintenance, works every time... 

I wonder if he cusses when applying percussive maintenance

Seriously though I appreciate takemorepills answer. It helps thank you.