So you mention vertical bus bars, is this part of an A-B MCC then? And so is this an A-B factory built Y-Delta starter? If you are measuring current at only 260A and it trips in 13 minutes, then something squirrely is going on. Could be a bad breaker, it happens. Could also be the ground fault or load spike as Pierre suggested; a faulty unload valve on a compressor can do that to you. It could also be a sort circuit fault that takes a while to show up; i.e. something heats up and makes contact with something else, either in the starter, the motor or even the conduit. I think from the sounds of this it warrants a data logger as Len suggested.
Yes this is an AB MCC, however the Y-Delta starter was supplied by the compressor manufacturer inside their compressor. I'll have to go look exactly what make it is.
If you are referring to a ground fault then I guess you are saying that it is one that only happens when something heats up. I dont believe this breaker can detect ground fault current, so the ground current would have to to be of a magnitude that would either exceed the thermal sensing or the instaneous setting. Right?
I will put a data logger on this circuit tomorrow to see if anything strange is happening. Should I put the data logger to measure all three pases on the line side of the breaker to see the true current that is going through the breaker?
If I look at current on the load side of the breaker which set of cables should I monitor? I am still confused how one set will show 58% of line current and the other set will show 42% as Jraef mentioned? I would think that both sets would show 58% of the line current. What would happen if I put the CT around both sets of cables at the tap point on the breaker for each phase? Would I see zero current similar to a zero sequence CT?
If i'm using a Fluke 1735 should this be quick enough to catch a possbile fault or transient if sampling time is set to 30s?