Breakers for ACs and Heat Pumps have a delay built in to allow for overamperage at start up just like fused disconnects for these units use a dual element time delay fuse. This would still allow time for a loose/poor connection or poor quality part to overheat and snowball deteriorate due to degradation of the connection/part from overheating and repetative tries to start.
Speaking with Square D engineer, their theory is that is was not high inductive load (high amperage) that caused the melting, but a high resistitive load (example of a space heater), and that was caused by a defective compressor at the weakest link. They also stated that the odds of 2 different breakers not tripping under load is highly unlikely. If it was truly a high inductive load that the wire would not have melted at the far end.
Loads are made up of inductive, resistive and capacitive loads or a combination of the three. A poor mechanical connection is an example of a resisitive load. The wires burnt off only where a mechanical connection was made. If the wire that was burnt off in several places was run straight to the compressor without any connections at the disconnect or contactor it would not have burnt off randomly at places in the middle of the wire. The primary cause of the the problem would be the compressor not starting up which would cause 4-5 times the normal running amp draw until the winding sensor opened the circuit. This could be due to a problem with the compressor capacitor or a mechanical problem with the compressor bearings or the rotor not being centered in the stator due to factory misalignment or the unit being dropped during transport.
I was just trying to point out that the problem could be as simple as a weak or open capacitor, which you couldn't tell unless you measured the capacitance to see if it was within plus or minus 10% of its rating. But that it might also be a mechanical problem with the compressor that was causing a locked rotor condition. The introduction of the two poor connections in the circuit adding resistance to the circuit would tend to decrease the amperage. Enough to prevent the circuit breaker from tripping? It is hard to tell quantitatively.