A) The Physics of... Unwarranted fuse Operation. For starters (excuse the pun) but a fuse should not "operate for a short to ground" in an ungrounded system. More likely, the fuse operated due to higher than anticipated starting current. The accepted rule of thumb is that starting current is 5 or 6 times, even more, than the motor's normal running current. To be more accurate one can use NEC Table 470-7(b) to calculate inrush kA based on rated Hp. What is not noted is that the calculation yields an rms (symmetrical) value. In reality significant current asymmetry exists between phases. The first-cycle peak can be 10 to 15 times FLA. Furthermore, the subsequent current decrement from peak to rms can take many cycles. Also, starting time could be significantly lengthened if the compressor is not fully-unloaded during run-up. To prove the above, please tell me if the "fuse operation" occurs in the same phase or in different phases. If the latter, substitute a breaker for the fuse. B) The Physics of... "Ungrounded" Distribution Systems". The misinformation is that they are "ungrounded". While is is true that such systems are not deliberately connected to a grounded reference point, they are still capacitively coupled to "ground." All electrical components have a ground capacitance... motors, heaters, cables, power factor correction capacitors, surge arrestors, bus duct, transformers... being the most significant. Perhaps a war story will better illustrate my point. My first job as an Electrical Engineer was in an industry that employed an 'ungrounded' 480V distribution system. Justification of such a system was based on the belief that "an inadvertent ground-fault won't result in an undesirable outage!" (See NEC for the exact verbage). Yes, there were ground detectors in every substation. The detector consisted of three 480 Volts incandescent light bulbs. They were wye connected between phase and ground. One day, I noticed that all bulbs were out, but only two fuses were blown. Measurements showed that the phase with the intact fuse read 0 V to ground, while the other two showed 800 Volts and 1,200 Volts, respectively. And the bulbs were "blown" as well. How, I (you) asked? In addition I learned that every time this happened, 3, 4, and sometimes more, motors would fail within 1 to 2 weeks. Furthermore, failures occurred in seemingly unrelated departments. Detailed investigation revealed the problem was caused by transient-overvoltage. Overvoltage occurred every time a ground fault developed in the system. The overvoltage would then stress the insulation resistance of the sound phases leading to premature failures of other seemingly unrelated equipment. Overvoltages resulted from what we elecrical engineers call resonance. Resonance occurs when capacitive impedance negates the normal inductive impedance. Without going into detail (unless asked for) the resultant voltage equation ends up with a numerator divided by zero. Ergo, infinity!!! The end result was that the cost to repair damaged equipment more than offset the benefit of the "ungrounded" system. Thus I believe that when the fuse blue during starting, system balance was upset, causing the transient overvoltages. The result was loss of the MOV's, then.... To prove this, measure the voltages to ground, with the system in normal operation. Phase to ground deviation is the key. My rule of thumb, if more than 5-10% you can expect more problems! The best fix... ground the system. If additional info is required, just ask! Regards, Phil Corso, PE (Boca Raton, FL)