Motor failure root cause analysis.

§1Scope of work

R-03 covers integral-horsepower induction motors from 1 HP up to approximately 1,500 HP; DC motors of comparable rating; and the VFD or soft-starter controllers driving them where the controller is implicated in the failure. The Lake Erie industrial corridor has a particularly dense population of mid-1980s through mid-1990s motors in service — the 1990 RIM (Rebuild In Motor) program, run by a since-merged Reliance Electric distributorship in Sharon, PA, rewound a large fraction of Erie’s installed industrial base, and the rewinds are now 30 to 35 years old. Many of those motors are still functioning, and a meaningful subset are failing in ways traceable to the original rewind specification rather than to current operating conditions.

The firm does not undertake R-03 work on fractional-horsepower motors (below 1 HP) absent unusual circumstances; the cost of bench analysis at the firm’s rates routinely exceeds the cost of a like-for-like replacement, and most fractional-HP failures are not commercially worth investigating. The firm also does not undertake submersible motor failures, vertical hollowshaft pump motors above 250 HP, or hermetic-refrigeration motors; these require manufacturer-controlled disassembly we cannot replicate.

R-03 frequently cross-references R-04 (harmonic distortion) where the suspect failure mechanism is VFD common-mode bearing currents or supply-side harmonic heating. The firm routinely opens both files when warranted; see the cross-protocol notation on the Reports index.

§2Failure modes catalogued

The firm maintains a standing taxonomy of motor failure modes that the principal investigator works through in fixed order. The order is not arbitrary — it reflects the relative ease of ruling each mode in or out from bench evidence, with the cheapest-to-rule-out at the top:

• M1 · Bearing failure (mechanical). Tested first because it’s the most common (over 50% of integral-HP failures by NEMA’s survey data) and the cheapest to confirm. Bearing race spalling, brinelling, smearing, or fluting visible on disassembly; lubrication contamination on inspection; vibration spectrum recovered if recorded prior to failure.
• M2 · Bearing failure (electrical/EDM). Tested where M1 is supported and the bearings show fluting consistent with electric discharge machining damage; characteristic of VFD-driven motors without shaft-grounding rings or insulated bearings under high common-mode dv/dt.
• M3 · Single-phasing winding burn. Tested where one winding shows uniform thermal damage and the others are clean; supported when supply-side data shows loss of one phase coincident with failure.
• M4 · Voltage unbalance overheating. Tested where all three windings show thermal damage but one shows more; supported by NEMA MG-1 §14.36 calculation against supply-side voltage measurement.
• M5 · Overload thermal damage. Tested where uniform thermal damage to all three windings is consistent with sustained over-current; supported by NEMA MG-1 service factor calculation against measured load.
• M6 · Turn-to-turn winding short. Tested by surge comparison test (e.g. Baker AWA-IV); supported by characteristic asymmetric surge response.
• M7 · Rewind specification error. Tested where the rewind shop’s records are available and the winding turns count, wire size, or insulation class can be compared to the manufacturer’s original specification.
• M8 · Manufacturing defect. Tested last because it requires ruling out all other modes; supported by characteristic defect signatures and by manufacturer recall or field-failure data.

The standing taxonomy is adapted from EPRI’s Motor Failure Workbook (TR-114951, 2000) and from the EASA failure-mode publication AR100. We do not deviate from the taxonomy on individual files.

§3Bench methodology

The subject motor arrives at Bay 4 under chain of custody. The first bench procedure is non-destructive: an insulation-resistance test per IEEE 43-2013 using the Megger MIT525 (5 kV insulation tester) and a polarization-index measurement over ten minutes. The IR and PI values are recorded against the motor’s nameplate insulation class. A surge comparison test follows on the Doble M4100 to identify any turn-to-turn faults that the IR test cannot see.

If the non-destructive testing identifies a clear winding fault, the bench procedure pauses; the carrier is notified, and disassembly proceeds only after the carrier’s consent is on file. Disassembly under chain of custody is photographed at every step (rotor extracted; one end-bell removed; the other end-bell removed; the stator core fully exposed; the rotor cage examined for broken bars). The bearings are removed last and examined under a binocular microscope for the characteristic damage signatures noted under M1 and M2 in the taxonomy.

Where a third-party metallurgical examination is warranted — typically for rewind-specification disputes where copper wire size is the question — the suspect winding sample is sent to a NVLAP-accredited materials lab. We use Intertek’s Mentor, OH facility for these examinations; the lab’s sealed report is appended to the matter file.

§4VFD-induced failures

VFD-induced motor failures — primarily M2 bearing failures from common-mode shaft voltage discharging through the bearing grease film — are a growing share of the firm’s R-03 docket. The mechanism was first formally characterized in the late 1990s (see Erdman et al., 1995, IEEE PCIC) and is now well-understood; the field is now in its third decade of installed VFD motors, and the early failures are being tracked back to motors and drives commissioned without shaft-grounding rings or insulated bearings.

Where the failed motor is VFD-driven, the firm conducts a parallel R-04 power-quality survey of the drive output during normal operation (or, where the drive is offline, simulates expected dv/dt from the drive’s topology and switching frequency). dv/dt on the order of 5,000 V/µs and switching frequencies of 4–8 kHz on standard PWM drives produce shaft voltages of 20–70 V peak in moderately-sized motors absent mitigation — easily enough to break down a 1–2 µm grease film on the bearing races. AEGIS shaft-grounding rings, SKF INSOCOAT insulated bearings, and dV/dt output filters are the standard mitigations; the firm reports on whether any of them was specified at original installation.

§5Deliverable & fee

The R-03 deliverable is a sealed PE report with: nameplate data; chain-of-custody log; non-destructive bench test results (IR, PI, surge comparison); disassembly documentation; the working failure-mode taxonomy with each mode marked supported / ruled out / inconclusive; the finding; and any third-party laboratory reports appended. Where the finding implicates a specific party (manufacturer, rewind shop, drive supplier), the report names the party.

Fee band $9,500 to $36,000; lead time 21 to 42 days; cross-protocol R-04 work billed separately. Engagement terms at the Engage page.

§6References