I would check your voltage to ground per phase.
Have seen this in systems where the fault isn't enough to trip OCPD but bled off voltage to ground
I don't quite see "bleeding off voltage" to ground so much as I maybe see (and have seen happen before) where a lead faults to ground, burns completely open, the supply side is "isolated" and the side connected to the load is still faulted - and draws abnormal current because the voltage to ground isn't the same as it would have been to the proper supply conductor.
Had this happen somewhat recently on a single phase motor, 240 volts. One of the "thru bolts" that holds the ends on the motor had one lead to the main winding fault to it. Burned clear from the 120 volt input lead, motor side of that burned conductor was still contacting the "thru bolt" and therefore the main winding was operating from the other lead to this fault - only seeing 120 volts instead of 240. Motor still worked, but they kept blowing start capacitors. I finally figured it all out after replacing a few capacitors - I stayed around to watch them use it (they never were using it other times I had been there, and to run it without a truck there to take on the grain it was moving would have left it piled on the ground) and when it got loaded enough it began to surge as the rotor slowed enough to kick in the start capacitors, sped up, kicked capacitors out, then slowed down again - lather, rinse, repeat - eventually took a toll on start capacitors.
"bleeding off voltage" would have to do so through some resistance - which would also cause heating and a hot spot, or it eventually burns itself clear or turns into low enough resistance to get fault level current that overcurent device can respond to.