By Jay Prigmore II, Ph.D., P.E. Last reviewed · Prepared and reviewed by a licensed Professional Engineer
Investigation of generator and large motor failures — from stator winding faults and out-of-phase synchronization to standby sets that did not start when the utility did not hold.
By Jay Prigmore II, Ph.D., P.E. Last reviewed · Prepared and reviewed by a licensed Professional Engineer
Generator disputes almost always reduce to one of two questions, and they call for different evidence. Either the machine caused the loss — a winding failure that started a fire, a catastrophic mechanical failure — or the machine failed to prevent one, which is the standby generator that did not start, did not accept load, or dropped it partway through an outage.
The first is a materials and electrical failure investigation. The second is an analysis of controls, maintenance history, fuel and starting systems, and transfer scheme behavior — and it usually turns on records rather than on the machine itself.
When a standby set does not perform during an outage, the resulting claim is often larger than the machine. PEFG investigates why the set failed to start or failed to carry load, including:
Sizing disputes are common here. A set that was adequate at commissioning is frequently undersized by the time it is called on, because load was added over years without anyone re-running the calculation.
Generator evidence is unusually perishable. Controllers and engine ECUs hold fault and event logs with limited memory depth; a few subsequent start attempts can overwrite the record of the failure. Windings get cleaned, tested, or rewound before anyone documents the as-found condition. Oil and coolant samples are discarded.
If a machine has just failed, the highest-value early steps are to pull controller and relay event data, photograph the as-found condition before disassembly, and preserve any failed components rather than sending them out for repair.
Analysis is grounded in the applicable industry standards, which commonly include IEEE C50 for machine ratings and testing, NEMA MG-1 for motor and generator performance, NFPA 110 for emergency and standby power systems, and the IEEE C37 series for the protection intended to detect these faults.
A 2.5 MW standby set is paralleled to the utility during a scheduled transfer. Something closes the breaker at the wrong angle. The machine survives the event but is found weeks later with a cracked shaft and damaged coupling, and the dispute is whether the closure caused it or merely revealed a pre-existing defect.
Out-of-phase closure imposes a torque transient on the shaft that can exceed several times rated torque, and the severity is a strong function of closing angle. That angle is recoverable: the synchronizing relay or the generator controller usually logs the closure, and the resulting current and power swing appears in the metering record. Reconstructing the electrical transient gives an estimate of the applied torque, and that estimate is compared against the machine’s C50.13 withstand and the manufacturer’s own shaft torsional analysis.
The physical evidence has to agree. A shaft failure from a single overload transient looks different from a high-cycle fatigue failure that had been propagating for years — beach marks, initiation site, and final fracture zone tell that story. Where the electrical reconstruction says the transient was survivable and the fractography says fatigue, the answer is the fractography, and the causation opinion changes accordingly.
Where the dispute involves a machine starting a large load rather than a machine failing outright, the motor starting calculator at PowerEngCalc shows how far the terminal voltage collapses on inrush and whether the resulting sag is consistent with the equipment behavior being complained of — a question that recurs in standby generation disputes and in power quality attribution.
Questions that come up before a retention decision, answered by the engineer who would do the work.
Stator winding insulation failure, rotor and field winding faults, excitation and voltage regulator malfunction, bearing failure and the shaft currents that sometimes cause it, out-of-phase synchronization, and protection that failed to operate or operated when it should not have. Standby sets add a separate category that has nothing to do with the machine itself: failure to start or failure to carry load, where the cause is usually fuel, batteries, controls, or a transfer scheme rather than the generator.
With the maintenance record, not the machine. NFPA 110 prescribes exercise and load testing intervals, and a set that was not being tested to that schedule frequently has an untested failure waiting in it. The usual causes are starting battery condition, fuel quality and fuel system contamination, control and transfer scheme configuration, and coolant or block heater failures — most of which the exercise regime exists specifically to catch. Whether the owner met the maintenance obligation is often the actual dispute, and the test log is where it is decided.
Usually yes, because the failure initiation site survives as a recognizable feature even in a badly damaged winding. Examination looks at whether the failure began at a slot exit, in the end winding, at a connection, or between turns of the same coil; whether the insulation shows thermal degradation consistent with long-term overheating or the localized damage of a rapid dielectric breakdown; and whether contamination, partial discharge damage, or mechanical abrasion is present. Insulation test history and any prior partial discharge data are what separate an end-of-life failure from a sudden one.
By comparing the as-found relay settings against the machine's own withstand limits and against the protection guide the settings should have been derived from. The recurring questions are whether the differential zone was correctly defined and the CTs correctly matched, whether loss-of-field, reverse power, and negative-sequence elements were enabled and set to the machine's capability, and whether the relay event record shows the fault the relay actually saw. A relay that operated correctly on a fault outside its zone is a coordination problem, not a relay problem, and the distinction usually decides who is responsible.
Controller and relay event records, which sit on rolling buffers, and the machine's as-found condition, which is disturbed the moment a repair contractor begins troubleshooting. Rotors get pulled, windings get cut out, and bearings get replaced as part of getting the plant back up — all of it reasonable operationally, all of it destructive to the evidence. A short written preservation instruction issued in the first days, covering event records, removed components, and photographs before disassembly, preserves nearly everything that matters at almost no cost.
The rotor, the exciter, and the control record usually disagree about what happened. Getting all three read before repair matters.