Electric Motor Failure Statistics: Causes, Failure Rates, and Survey Data

Ask five vendors what causes motor failures and you will get five pie charts with five different sets of percentages, all presented as settled fact and rarely with a citation attached. The numbers are real. They just come from different studies of different motor populations conducted across two decades, and the differences between them are more useful than any single chart. This page traces each of the major datasets back to its primary source, explains what population each one measured, and covers the failure-rate and maintenance findings that the pie-chart reproductions usually leave out.

This page covers motors specifically. For failure data on transformers, switchgear, and other electrical assets, see our electrical equipment failure rates page. For the cause breakdown inside the bearing slice itself, see our bearing failure statistics page.

Motor failure causes at a glance

Some reproductions in the secondary literature shift these splits by a point or two. The EPRI and IEEE figures above match the comparison reproduced in EPRI’s own maintenance guidance. The ranked order is identical in every version.

The four datasets behind the motor failure pie charts

Nearly every motor failure statistic in circulation traces back to one of four sources. Knowing which one a given chart came from tells you what kind of motors it describes.

The EPRI utility study (1982)

The Electric Power Research Institute commissioned General Electric to survey motor reliability in electric utilities. The results were published as EPRI report EL-2678 in 1982, with follow-up analysis by Albrecht and colleagues in IEEE Transactions on Energy Conversion in 1986 and 1987. The study covered roughly 4,800 motors of 100 horsepower and larger and recorded 1,227 failures on 872 motors, meaning hundreds of the failures were repeat failures on machines that had already failed at least once.

The component breakdown: bearings 41 percent, stator windings 36 percent, rotor 9 percent, everything else 14 percent. The original report puts the overall failure rate at roughly 3.4 to 3.5 failures per 100 motor-years; figures closer to 3.1 percent circulate from later updates and reproductions, so quote the number with its edition attached.

Two findings from the primary reports rarely survive into the reproductions. First, the failures were heavily concentrated. In the EPRI data, 50 percent of failed motors occurred on 17 percent of generating units, and 90 percent occurred on 54 percent of generating units. When EPRI grouped units by failure rate, the high group represented 17 percent of units and had a weighted failure rate of about 9.5 percent per motor-year; the low group represented 30 percent of units and had about 0.9 percent. Motor reliability in this dataset was concentrated in specific units rather than spread evenly across the fleet. Second, within the winding failures, insulation-to-ground faults dominated (18.5 percent of all recorded failure modes) while turn-to-turn insulation shorts were far less common (3.7 percent). The frequently repeated claim that most winding failures begin as turn faults does not appear in the EPRI reports.

The IEEE-IAS industrial survey (1985)

The Motor Reliability Working Group of the IEEE Industry Applications Society surveyed large motors in industrial and commercial installations, covering machines above 200 horsepower at voltages up to 13.8 kV, including induction, synchronous, wound rotor, and DC machines. Results were published in three parts in IEEE Transactions on Industry Applications (Parts I and II in 1985, Part III in 1987). The dataset represents 5,085 motor unit-years of service history with 360 recorded failures.

The component split: bearings 44 percent, winding-related failures 26 percent, rotor-related 8 percent, all other causes 22 percent. The survey put the average failure rate at 0.0708 failures per unit per year, or roughly 7 failures per 100 motor-years, about double the utility rate EPRI found.

The 1985 survey also broke results out by maintenance quality and frequency, motor size, and speed, which is where its most actionable findings live. More on those below.

The Thorsen and Dalva petrochemical survey (1995)

Olav Thorsen and Magnus Dalva surveyed squirrel cage induction motors on offshore platforms and in petrochemical plants, gas terminals, and refineries, much of it in the North Sea environment. Results were published in IEEE Transactions on Industry Applications in 1995. Bearings accounted for approximately 51 percent of failures, with stator windings and external causes each contributing roughly 15 to 16 percent. The survey also found that motors in the 101 to 500 kW range failed at a notably higher rate than other size classes, and that machines on a maintenance cycle of 12 months or shorter showed a significantly lower failure rate.

The Bonnett and Yung compilation (2008)

The chart that most vendor literature reproduces today, with bearings at 51 percent, comes from “Increased Efficiency Versus Increased Reliability” by Austin Bonnett and Chuck Yung, published in IEEE Industry Applications Magazine in 2008. Both authors are closely associated with EASA, the Electrical Apparatus Service Association: Bonnett served as an EASA technology consultant and Yung as an EASA senior technical support specialist. The breakdown is reproduced in EASA/AEMT and EASA-affiliated literature, which is why it often circulates under the label of EASA data.

Their compiled figures: bearings 51 percent, stator windings 16 percent, external causes 16 percent, rotor bar 5 percent, shaft and coupling 2 percent, unknown 10 percent. The numbers align closely with Thorsen and Dalva’s petrochemical findings. OEM literature reproduces the same breakdown: ABB’s motor maintenance guide, for one, shows the identical 51/16/16/5/2 split while attributing it to IEA data, an example of how these figures circulate with the provenance scrambled.

Why bearings lead the surveys

The bearing share rises from 41 percent in the oldest study to 51 percent in the more recent ones, and the failure analysis literature offered an explanation early on. In his 1993 analysis of anti-friction bearing failures in AC induction motors, Austin Bonnett observed that bearing failures had accounted for an increased percentage of motor failures, due in part to improved stator and rotor construction that can leave the bearings as the weak link. Insulation materials and winding treatment processes improved substantially over that period while rolling element bearing design stayed comparatively stable.

The bearing percentage is also where survey methodology matters most. Some plants replace bearings proactively when vibration data crosses a threshold; others run to functional failure. Whether those condition-based replacements count as “failures” varies by survey, and that definitional choice alone can move the reported bearing share by several points. When comparing your own plant data against these benchmarks, confirm your failure definition matches before concluding anything.

A bearing failure is also where the gap between failed component and root cause is widest. The bearing is what seized, but the cause is usually lubrication practice, contamination, misalignment, belt tension, or shaft currents. EASA’s own root cause methodology makes exactly this point: the component that failed identifies where to start the investigation, and stopping there guarantees a repeat failure. The EPRI data quantified how often that happens, with hundreds of its recorded failures being second or third failures on the same machine. We break down the causes inside the bearing slice on our bearing failure statistics page.

One modern factor deserves its own mention because it is absent from the classic surveys at anything like today’s prevalence: bearing damage from shaft currents on motors fed by variable frequency drives. PWM inverters can induce voltages that discharge through the bearing, producing electrical discharge machining damage (fluting) that vibration analysis eventually picks up as a bearing defect. In the failure record it gets logged as one more bearing failure, which means VFD-driven fleets can push the bearing share above what any of the legacy surveys reported.

What the winding data actually says

Stator winding failures run second in all four surveys, at anywhere from 16 to 36 percent depending on the population. The spread itself is informative: winding failures claimed a much larger share in the EPRI utility population of large, older machines than in the later industrial and petrochemical surveys, consistent with the improvement in insulation systems over the intervening years.

The EPRI failure-mode detail is worth restating because it contradicts a common claim. Insulation-to-ground faults made up the single largest identified winding failure mode at 18.5 percent of recorded failure modes, roughly five times more frequent than turn insulation shorts at 3.7 percent. Vendor content frequently asserts that winding failures begin as turn-to-turn faults that cascade to ground faults, and uses that claim to sell early-detection technology. Whatever the merits of the technology, the claim is unsupported by the EPRI failure records.

The classic root cause analysis of winding and rotor failures is the pair of papers by Bonnett and Soukup in IEEE Transactions on Industry Applications (rotor failures in 1988, stator and rotor failures in 1992). Their framework attributes winding failures to thermal, electrical, mechanical, and environmental stresses, with overheating and contamination as the recurring themes.

Failure rates: how often motors actually fail

The headline rates from the surveys:

Maintenance frequency and protection strategy are among the strongest correlates reported in the survey data, alongside motor size and speed:

Maintenance frequency. Thorsen and Dalva’s finding is clean: a maintenance cycle of 12 months or shorter correlated with a significantly lower failure rate. The IEEE 1985 maintenance data needs more careful reading. Plants rated as having excellent maintenance practices on cycles under 12 months logged a numerically higher failure rate (0.1115 FPU) than plants with fair maintenance on cycles beyond 24 months (0.0719 FPU), because frequent inspections caught developing faults that would otherwise have surfaced later as in-service failures. The tell is in the downtime data from the same table. Failures at the excellent, sub-12-month plants cost a median of 8 hours of downtime each, while failures at the fair, beyond-24-month plants cost a median of 165 hours each. Frequent maintenance converts long unplanned outages into short planned ones, and a raw failure count misses that entirely.

Protection. In the 1995 data, embedded thermal protection was associated with less than a third of the failure rate of motors in the survey’s minimum electrical protection category.

Size and speed. In the IEEE 1985 data, motors from 5,001 to 10,000 horsepower had a reported failure rate roughly three times higher than motors from 500 to 5,000 horsepower (0.2169 versus 0.0730 FPU), with a small sample in the 5,001 to 10,000 hp class (46.1 unit-years and 10 failures) worth keeping in mind. Slower machines failed more often than faster ones (0.1004 FPU below 720 RPM versus 0.0519 FPU from 1,801 to 3,600 RPM).

The rewind question

Because bearings and windings dominate the failure record, most failed motors are candidates for repair rather than scrap, and the standard objection to rewinding is efficiency loss. That objection has been tested twice under controlled conditions.

The EASA/AEMT rewind study (2003) had motors performance-tested before and after rewinding at an independent laboratory using the IEEE 112 Method B loss segregation approach. Its finding: rewinds performed to documented good practices can maintain efficiency within the measurement accuracy of the test procedure itself (about plus or minus 0.2 percent), and occasionally improve it. The same study showed how bad practice produces the losses the objection is based on, with uncontrolled burnout temperatures and rough coil stripping damaging the core.

The follow-up study (2019, published 2021) repeated the test on ten premium efficiency and IE3 motors from 30 to 75 kW, rewound to ANSI/EASA AR100 and the EASA/AEMT Good Practice Guide. Efficiency changes ranged from minus 0.5 to plus 0.3 percent with an overall average of minus 0.1 percent, reaffirming the 2003 result for modern high-efficiency designs. The practical takeaway for the repair-or-replace decision: a good-practice rewind carries no inherent efficiency penalty, so the decision rests on repair cost versus replacement cost, lead time, and whether an older motor should be upgraded to a higher efficiency class anyway.

Why motor failure data carries so much weight

Motors are the single largest end use of electricity on the planet. The IEA’s 2011 analysis by Waide and Brunner put electric motor-driven systems at more than 40 percent of global electricity consumption (the paper’s estimate range was 43 to 46 percent), and the IEA 4E Electric Motor Systems platform currently cites 53 percent when pumps, fans, and compressors in buildings are included. Over a motor’s service life, electricity typically accounts for around 90 percent of total life-cycle cost, with the purchase price closer to 1 percent.

That is the context that makes a percentage point of failure rate or efficiency worth arguing about, and it is why motor OEMs, repair shops, drive vendors, and monitoring companies all reproduce these statistics. The numbers are load-bearing in a lot of sales collateral. They deserve to be quoted with their sources and their limitations attached.

How to read these numbers

The surveys are old. The newest primary failure survey in wide circulation is from 1995, and the Bonnett and Yung compilation from 2008 draws on that era of data. Bonnett’s 2010 root cause paper for the Petroleum and Chemical Industry Conference found the earlier results still held, and a newer public survey with comparable scope has yet to clearly supersede the 1982 to 1995 datasets. Motor designs, insulation systems, and drive technology have all changed since, most visibly in the VFD-related bearing damage that the old surveys predate at scale.

Population determines percentages. A utility fleet of large machines, an industrial plant, and a North Sea platform produced three different pie charts from the same physics. Your plant is a fourth population. The ranked order (bearings, then windings, then everything else) transfers well. The exact percentages do not.

The component is where the investigation starts. Every one of these surveys counts failed components. None of them tells you why your bearing failed. The EPRI repeat-failure data is the strongest argument in the whole record for doing the root cause work: replace the component without finding the cause and the failure comes back.

Related reading: our pump reliability and MTBF statistics page covers the asset class where many of these motors spend their working lives, and our maintenance and reliability glossary defines the failure terminology used throughout.

Frequently asked questions

What is the most common cause of electric motor failure?

Bearing failure, in each of the major public surveys. The EPRI utility study put bearings at 41 percent of failures, the IEEE-IAS industrial survey at 44 percent, and the Thorsen and Dalva petrochemical survey and the Bonnett and Yung compilation at 51 percent. The bearing is the failed component in these counts; the underlying cause is most often lubrication, contamination, misalignment, or, on drive-fed motors, shaft currents.

What percentage of motor failures are caused by winding problems?

Between 16 and 36 percent depending on the survey population. The EPRI utility study found stator windings involved in 36 percent of failures among large, older machines, while the IEEE-IAS industrial survey found 26 percent and the petrochemical-era data found around 16 percent. Within the EPRI winding data, insulation-to-ground faults were roughly five times more common than turn-to-turn insulation shorts.

How often do electric motors fail?

The major surveys report between roughly 3 and 7 failures per 100 motor-years. The original EPRI report found about 3.4 to 3.5 percent per motor per year in utilities (later reproductions often cite closer to 3.1), and the IEEE-IAS survey found about 7.1 percent per year in industrial and commercial installations. Maintenance frequency, thermal protection, motor size, and speed all correlated strongly with the rate: embedded thermal protection was associated with less than a third of the failure rate of motors in the survey’s minimum electrical protection category.

Do rewound motors lose efficiency?

Rewinds performed to documented good practices can maintain efficiency within test measurement accuracy. The EASA/AEMT rewind studies (2003, repeated in 2019 on premium efficiency and IE3 motors) found post-rewind efficiency changes ranging from minus 0.5 to plus 0.3 percent, with an overall average change of minus 0.1 percent in the 2019 study. Efficiency losses from rewinding trace to poor practice, particularly uncontrolled burnout temperatures and core damage during stripping.

Why do bearing failure percentages differ between studies?

Three reasons: different motor populations (utility, industrial, petrochemical), different eras (winding reliability improved over the survey period, a shift Bonnett identified in the early 1990s as raising the bearing share of remaining failures), and different failure definitions (some surveys count condition-based bearing replacements as failures, others count only in-service breakdowns, a definitional choice that can move the bearing share by several points).

Are the classic motor failure studies still valid?

The ranked order of failure causes has held up: bearings first, windings second, rotor a distant third, across the major surveys from 1982 to 2008, and a 2010 root cause review found the earlier findings still applied. The exact percentages should be treated as historical benchmarks rather than current measurements, since a newer public survey with comparable scope has yet to supersede the legacy datasets, and modern factors like VFD-induced bearing currents are absent from them.

Sources

• EPRI EL-2678, “Improved Motors for Utility Applications,” Industry Assessment Study, Vol. 2, Final Report, October 1982 (study performed by General Electric)
• Albrecht, P.F., et al., “Assessment of the Reliability of Motors in Utility Applications – Updated,” IEEE Transactions on Energy Conversion, Vol. EC-1, No. 1, March 1986
• Albrecht, P.F., et al., “Assessment of the Reliability of Motors in Utility Applications,” IEEE Transactions on Energy Conversion, Vol. EC-2, No. 3, September 1987
• IEEE Motor Reliability Working Group, “Report of Large Motor Reliability Survey of Industrial and Commercial Installations,” Parts I and II, IEEE Transactions on Industry Applications, Vol. IA-21, No. 4, 1985; Part III, Vol. IA-23, No. 1, 1987
• Thorsen, O.V., and Dalva, M., “A Survey of Faults on Induction Motors in Offshore Oil Industry, Petrochemical Industry, Gas Terminals, and Oil Refineries,” IEEE Transactions on Industry Applications, Vol. 31, No. 5, 1995
• Bonnett, A.H., and Yung, C., “Increased Efficiency Versus Increased Reliability,” IEEE Industry Applications Magazine, Vol. 14, No. 1, 2008
• Bonnett, A.H., and Soukup, G.C., “Analysis of Rotor Failures in Squirrel-Cage Induction Motors,” IEEE Transactions on Industry Applications, Vol. 24, No. 6, 1988
• Bonnett, A.H., and Soukup, G.C., “Cause and Analysis of Stator and Rotor Failures in Three-Phase Squirrel-Cage Induction Motors,” IEEE Transactions on Industry Applications, Vol. 28, No. 4, 1992
• Bonnett, A.H., “Cause and Analysis of Anti-Friction Bearing Failures in A.C. Induction Motors,” IEEE Industry Applications Society Newsletter, September/October 1993
• Bonnett, A.H., “Root Cause Failure Analysis for AC Induction Motors in the Petroleum and Chemical Industry,” Proceedings, 57th Annual IEEE Petroleum and Chemical Industry Conference, 2010
• IEEE Std 493 (Gold Book), “Recommended Practice for the Design of Reliable Industrial and Commercial Power Systems”
• EASA/AEMT, “The Effect of Repair/Rewinding on Motor Efficiency,” 2003
• EASA/AEMT, “The Effect of Repair/Rewinding on Premium Efficiency/IE3 Motors,” 2019 study, published 2021
• ABB, “Motors don’t just fail… do they? A guide to preventing failure” (referenced as an example of OEM reproduction of the Bonnett and Yung breakdown)
• Waide, P., and Brunner, C.U., “Energy-Efficiency Policy Opportunities for Electric Motor-Driven Systems,” IEA Energy Papers No. 2011/07, OECD Publishing, 2011
• IEA 4E Electric Motor Systems Platform (EMSA), current motor-systems electricity share