Very interesting story, that is definitely not a common occurrence. The power cycling behavior is almost certainly the doings of a "recloser", which is essentially a high voltage circuit breaker that resets automatically a set amount of times before finally quitting and keeping the power off. I have never experience a brownout, but reliably once a year the high voltage expulsion fuse right outside my house opens with a deafening bang and then we have to wait to get it refused. Don't you just love overhead distribution?

I can remember at the place I used to work (large building so 3-phase power) there were a few separate times where there was a partial outage.... one phase went out, one was on at around half-voltage and one was on at full voltage.
That ofcorse caused issues. Refrigerators & freezers don't like running half-voltage!
And 3-phase motors like those in HVAC systems especially don't like such a setup..
Fluorescent lights in the hallway must have been on the half-voltage phase, because turned it into a wild blinky light-show. .lol.
Then there were various that were either on or off.
Stuff like LED bulbs just ran dim.

One of the times it happened I heard the cause was a truck hitting a power pole.
Another I think was road construction.

What surprises me on this story is how long the situation remained: 5 minutes of clearly damaged distribution system, doing who konws what and no protection shut it off automatically? That is the craziest part on it form me.

I consider the grid here as pretty outdated, but this has never happened here, even going as far as into the 80's.
To be clear idiots ramming poles and breaking some wires or causing flashovers do happen. But such crazy situation laststing for more than a second or two, before the power gets completely cut off by the protections.

Over here, distribution transofrmers (from 22kV phase-phase to 230/400V) are 3 phase, Delta primary, Y secondary. If there are similar transformers used in your grid then here is what might have happened :

Initially there was a short, either between phases or from a phase to ground. (Probably not to the LV wiring, as that would cause extensive damage in all connected properties and possibly some house fires)

There was either a remaining metal contact short, or remaining conductors in close proximity to each other, which would arc again when voltage is applied

After few reclosing attempts either the shorting part burned off, or the arc happened to not strike again anymore. In this case the substation does not see a fault anymore, so it has no reason to shut down again. (@Medved)

However, one of the phase wires continuing towards your transformer was open circuit, so you got 2 phases and 1 open ended wire, connected to the 3rd phase of all transformers in the remaining part of the branch. For simplicity let's assume there is just your transformer there. Assume the broken HV phase is HV3

The transformer can be imagined as 3 individual transformers wired as follows :

T1 : Primary HV1-HV2 : Secondary L1-N
T2 : Primary HV2-HV3 : Secondary L2-N
T3 : Primary HV3-HV1 : Secondary L3-N

T1 is getting correct input voltage, and puts out the correct voltage on L1 (120V or 230V)
The wire HV3 is broken upstream, so it is only connected to T2 and T3. So T2 and T3 are effectively connected in series between HV1 and HV2

If the load on T2 and T3 is identical, the voltage will divide equally, each will get 1/2 of the voltage it is supposed to get on the input, and put out 1/2 of the normal voltage on the output

In practice the load is probably not identical, and will fluctate - especially as some loads either cut out from the undervoltage condition, or draw excessive current (such as motors unable to start) which may burn them open circuit or trip overload protections or breakers. If say L2 is loaded more, it will pull down the voltage of L2, and rise the voltage of L3. The sum of the two will be around 120V or 230V



In the US many transformers are single phase and wired Y on the primary i think ?

In this case, the transformers will be :
T1 : Primary HV1-HV0 : Secondary L1-N
T2 : Primary HV2-HV0 : Secondary L2-N
T3 : Primary HV3-HV0 : Secondary L3-N

(And they dont have to be all located in the same place)

As long as this is the circuit, HV3 and L3 will have no power at all, so just outage and not brownout. However :

 - Three phase loads - whether a 3 phase transformer as above, or 3 phase LV load connected in Delta to the output of 3 individual transformers, will backfeed into HV3, and this way provide power to other transformers that are connected to HV3

 - In SWER systems, the highly assymetrical load of HV1 and HV2 going to HV0, may cause voltage drop in the ground between HV0 and Earth near the location where there is high load. Now consider two single phase transformers connected between HV3-HV0 : One right next to the location with the high load, and one far away, but both in the bad area of the network. There is a voltage difference between HV0 in one and HV0 in the other location. The two transformers are connected between those locations in series (through the open HV3 wire), so they do get some voltage, and will output a matching voltage on the secondary (probably very low voltages in this case)

Here the 22kV has imbalance sensors (something similar to RCD in what faults it is looking for).
I would guess disconnecting one phase gives quite significant imbalance (the capacitive currents from wires towards ground are not that small), triggering the protection as well.
Plus I would guess there are more protection elements in the system, "looking" for mainly such common faults.

What I'm pretty sure phase wires being broken or open circuit is quite common fault (not that far behind short/leakage to ground), yet the result is pretty always shut down of the affected branches within seconds.
Yes, there are automatic reclosers autromatically attempting to recover the service mainly for situations when the fault signature signals was some kind of glitch (e.g. nearby lightning or even induced magnetic storm events, I guess), but even that means attempts for no longer than a second or so as well in case the fault persist.
Don't ask me for details, the only thing I know about it is how it behaves at the consumer end...

Here a single 22kV line from the subsation can run many 10's of mostly 400kVA transformers. (Well, lots of them are single transformers powering small stand-alone loads like a single farm or traffic lights etc, so the actual load is many times less than N*400kVA). I think a lost phase in some final branch, powering just 2..3 transformers won't even stand out in the measurement

Also, lets say you can discover the existence of such fault. Is it viable to cut the entire line from the substation due to this fault ? I.e. causing full downtime to 50 customers because 3 lost a phase ?

Very interesting story, that is definitely not a common occurrence. The power cycling behavior is almost certainly the doings of a "recloser", which is essentially a high voltage circuit breaker that resets automatically a set amount of times before finally quitting and keeping the power off. I have never experience a brownout, but reliably once a year the high voltage expulsion fuse right outside my house opens with a deafening bang and then we have to wait to get it refused. Don't you just love overhead distribution?

Very interesting fact about the reclosers; I had never even heard of such a thing until now.  I'm not the most familiar with high-voltage distribution.  But, that makes a lot of sense and would certainly explain why the power was "cycling" on and off!

Also, lets say you can discover the existence of such fault. Is it viable to cut the entire line from the substation due to this fault ? I.e. causing full downtime to 50 customers because 3 lost a phase ?

Of course, better to cut off power for 50 custoers for few hours (till the thing gets fixed) then irreversibly blow up systems of 3 customers.
Normally the operator is liable for damages from wrong power, but allows certain percentage of cut power.
Plus it is not just about the 3 customers. An open fault could be just an indication of a fault that may cause huge damage along the line, like forest fire or so, i.e. the broken wire did not falled to the bushes yet, but it is a matter of few minutes till it will do so. And then even fraction of a second discharge can ignite the forest in a remote area.
So no, if there is anything wrong detected, the system needs to be shut down until it could be positively identified it is safe to keep it on. And in most cases if you are that far, you can as well fix the thing...

And if some line is prone to such faults for some reason and the situation is so such faults can not turn into something dangerous, power companies use local detectors and automatic load disconnects on the customer site, so it cuts out power only to those where the power is bad. But the "can not turn into something dangerous" means it is really rare...

There are many ways how equipment at the customer can lose a phase - Some of them result of problems in the grid and some of faults at the customer's own premises. To name a few :

 - Loose connections in distribution boards, boxes, sockets, plugs, and in the terminal box of the motor itself
 - EOL worn out contactors
 - Bad quality breakers and switches losing phases out of the blue
 - Building dust getting into switchgear and preventing a phase from closing
 - Blown fuse in feeder (most of the time result of overloading by the customer, or short circuits in the customer premises)

A simple motor overload breaker for each 3 phase motor (which costs less than the motor itself) would prevent 99% of the consequences of lost phases which i seen in commercial equipment. A dedicated voltage monitoring unit would cover the remaining 1%

If the customer has no protections on his 3 phase equipment, he is an idiot. (Either for not installing protection, or for using the services of cowboy-grade techs to install equipment)



The 50 additional customers may include :

 - Water supply (loss of pressure in pipes may lead to chemical and biological hazards)
 - Pipes heated against freezing
 - People sitting at home with medical apparatus (it should have backup, but they are explicitly mentioned as a risk group in many evaluations of power outages)
 - People who heat their home with electricity (again, may involve eldery and ill people, possibly in badly insulated houses that can chill quickly)
 - Traffic lights (the traffic laws do say how to drive if traffic lights dont work, but an outage does increase the risks)
 - Public transportation
 - Lifts (people getting stuck)
 - Lighting in private and commercial spaces - Even if everyone's emergency lighting is intact and working, the time to repair the fault may exceed the battery runtime of emergency lighting installations
 - Lighting in outdoor spaces - There is rarely any EM lighting there, but for some areas such lighting is important for safety or crime prevention
 - Agricultural and industrial processes....

So i won't be so sure that it is better to cut off 50 customers, as it may indeed lead to risks there too



As for setting forest fire, i see your point and have thought about it many times in another context : The investment the power company does into protecting areas under power lines

Over here, the standard is that an overhead 22kV power line going above public road (maybe above a certain road size) must be safe against falling down due to a single fault in an insulator. So lines in the part that goes above a road are fitted with double insulators (either string or pin type)

Yet i have never seen double insulators used in lines stretched just through forest. Including in a few select places where a part of the line is stretched over an oasis between 2 distant poles, and the wires in this section obviously weight much more than in any other section

So it seems that our power company is not too concerned by this



Nowadays the meter at the customer's place ca n report in real time a phase loss, and out power company are indeed installing additional metering points throughout the LV netorks. However, this was not the case as little as 20 years ago - With everything post subsation being purely electromechanical

If the fault happens at the "customer side of the meter", it is his responsibility, his loss, so his money.
But if the fault happens on the distributor side, any damage caused by that is the distributors liability, so the distributor has to pay. Without that distributors would ignore such faults and the timely fix, when they will still be able to collect money from the 50 and leave the repair cost to the one whos equipment god damaged. They would not have any incentive to even maintain it properly so such faults won't happen that often in the first place.

And even without this liability, equipment damage of a single customer of the 3 would be way more expensive than the temporary loss of power service of the 50.

So no, the "who cares something gets damaged, better to keep the 50 running and keep collecting" is vry greedy and selfish attitude from the power distribution company, and it is really not acceptable in a civilized world.

Of course there could be exceptional situations where you intentionally want to recover at least a single phase when you can not fix it completely in a reasonable time, but that is then communicated to the customers so they can switch off the potentionally sensitive equipment, and also before turning it on in that mode, someone will inspect and secure the system at least to a state when the fault can not cause any other harm.