Neutral are connected properly though.
yet opening one phase of primary causes secondary overvoltage...
just a thought experiment here-
get a pencil and sketch some phasors to see if the following is right:
Primary side voltages with neutral connected, phase B open
let me assume phase to phase primary voltage is one unit...
Van = 1/√3 unit
Vbn = 1/√3 unit because it's backfed from secondary
Vcn = 1/√3 unit
so secondary phase to phase voltages are balanced, each is 1/√3 unit X turns ratio,
and the 120 degree phase relationship is intact.
Now remove the neutral connection between transformer and its source...
Phases An and Cn no longer are 120 degrees out because
On primary side , voltages across windings A and C are now
in phase not out by 120 degrees, because the neutral is missing so can't pull their midpoint over to center of phasor triangle.
They are just in series .
So each sees half Vac, or 0.5 unit. No problem yet, for 0.5 is less than 1/√3
BUT
On secondary side, phase B is connected between A and C.
So phase B secondary winding sees (Vac/2 +Vac/2) X turns ratio,
or (0.5unit + 0.5 unit ) = 1 unit X turns ratio
and that is
more than 1/√3unit X turns ratio.
...
So phase B secondary winding sees 1 whole unit X turns ratio, not 1/√3 unit X turns ratio.
That's a 15% overvoltage.
That increased phase B secondary winding voltage will backfeed to the primary winding.
And it might saturate the transformer you just disconnected ... Did it hum ?
What damage did you observe? Overvoltage arrestors actuating? Phase B transformer overheating? Phase AC secondary voltage too high?
I'd look for an open in the neutral wire that connects those transformer primaries back to the next transformer or bus upstream.
I guess i'd summarize this so:
Draw two unit vectors emanating from a common point at 120 degrees to each other.
What is distance between their outboard ends?
Now rotate them so they align.
Their ends moved further apart by ratio 2/√3. That's 15% more distance.
So --- will your phase AC voltage, if you disconnect phase B with an open present( but not noticed because things were balanced ) in the neutral wire somewhere back on the line, not increase by 15% ?
If neutral current is returning via earth, you'll get overvoltage less than 15%.
But - this is just a thought experiment from probably a long distance away.
Probably it's something else . It's very late here and i probably made a mistake - catch you in the morning..
old jim