Ask how often an industrial transformer or circuit breaker fails, and you’ll get a confident number back. Ask where that number comes from, and the room goes quiet.
Many of the failure rates repeated across industrial and commercial power systems trace to one place: the IEEE Gold Book, IEEE Std 493. The figures are real, and they’re useful for the right job. The catch is that the surveys behind them were run decades ago, and almost nobody who quotes them says so.
This page rates the failure-rate benchmarks that hold up, the ones that need a caveat, and the one interpretation that’s flat wrong.
What These Numbers Measure (and Where They Come From)
Electrical failure rates are usually given in failures per unit-year. One unit running for one year is one unit-year. A rate of 0.0041 failures per unit-year means you’d expect about 41 failures across 10,000 transformer-years of operation. It describes how a whole population behaves over time. It says little about any single machine.
The historical source is IEEE Std 493-2007, the Gold Book, which IEEE now classifies as inactive-reserved. IEEE reorganized its Color Books into the 3000 series, and IEEE Std 3006.8-2018 is the active recommended practice for analyzing equipment reliability data. Both are paid standards.
The provenance matters here, so it’s worth stating plainly. The original IEEE survey was completed in 1972 and reported 1,982 equipment failures from 30 companies covering 68 plants across nine industries in the United States and Canada. A separate 1979 survey updated the transformer data, and later work fed the editions that followed. That body of data is what most modern failure-rate citations still rest on.
Manufacturers reproduce the figures too. Eaton’s medium-voltage breaker reliability paper pulls its numbers straight from IEEE Std 493-1980 Table 24, which is a good sign the data is real and a good reminder of how old the source is.
The Reliable Confidence Score
Each row rates our confidence in the claim, not always in a single number. A narrow negative finding can score High. A figure with correct arithmetic and a wrong interpretation scores Low.
| Source or Claim | Figure | Reliable Confidence | What It Really Means |
|---|---|---|---|
| Liquid-filled power transformer failure rate (IEEE 493 surveys) | ~0.0041 (1973-74 survey) to ~0.0062 (1979 survey) failures/unit-year | Mediumhistorical range | Treat it as a planning range, not one universal benchmark. The newer survey ran roughly 51% higher than the older one. |
| Metalclad drawout circuit breaker failure rate (IEEE 493-1980, Table 24, via Eaton) | 0.0036 failures/unit-year | Mediumsmall sample, old data | Drawn from 58 failures across 16,280 unit-years. IEEE reports breakers separately by construction and voltage class, so this is one category, not all breakers. |
| Downtime per breaker failure (IEEE 493 breaker data) | ~83 hours to repair, ~2 hours to replace | Mediumsame vintage caveat | Two ways to recover from the same failure. Replacing the unit rather than repairing it in place is what closes most of the gap. |
| “MTBF of about 280 years” for a circuit breaker | 16,280 / 58 ≈ 281 years | Lowcommon misread | The arithmetic holds. Reading it as a lifespan does not. It expresses a fleet-average event rate across accumulated exposure. |
| A single universal electrical equipment failure rate | none exists | Highnarrow negative | Rates split by equipment type, construction, voltage class, operating environment, and survey vintage. One blended number hides all of that. |
| Provenance of the original 1972 IEEE survey | 1,982 failures, 30 companies, 68 plants | High | The data is real and documented. The open question is its age. |
The Big Takeaway
The honest answer to “how often does electrical equipment fail” is that we have workable numbers for fleets and weak numbers for single units, and the good numbers are getting old.
A rate like 0.0041 per unit-year earns its keep when you multiply it across a population. It falls apart the moment someone tries to turn it into a guarantee about the transformer in front of them.
The famous electrical failure rates are real. What’s missing from most citations is the date stamp.
Why the Numbers Vary or Disagree
Different surveys covered different industries in different decades, so two “IEEE failure rates” for the same equipment can disagree and both be correctly cited. The transformer figure is the clearest case: the 1973-74 survey and the 1979 survey produced different rates for the same equipment class.
Construction and voltage class matter for breakers. IEEE reports separate rates for fixed, molded-case, and metalclad drawout equipment, and breaks several of those out by voltage. A single breaker number blends categories that behave differently.
Sample sizes are thin for some equipment. The metalclad breaker figure comes from 58 reported failures. That’s enough to estimate a rate and not enough to be precise about it.
There’s also a modeling step that’s easy to miss. Reporting failures per unit-year does not, by itself, assume the rate stays constant for an asset’s whole life. That assumption enters when an analyst applies a constant-hazard model and takes the reciprocal of the rate as an MTBF or MTTF.
How to Use These Benchmarks Safely
Use them at the fleet level. Multiply the rate by your population and your time horizon to estimate expected failures, then plan spares and budgets around that.
Take EDT Engineers’ worked example: a solar farm with 47 oil-filled transformers over an assumed 20-year life. EDT used 0.0041 per unit-year and landed on about four expected failures (3.85 before rounding). Run the same math at 0.0062 and you get about 5.83. That spread is the point. Your answer swings with which survey you pick, which is why the range belongs in the estimate.
Pair the failure rate with downtime per failure to size your exposure, then apply your own cost-of-downtime figure rather than a generic one. Our Industrial Downtime Cost Benchmarks page covers why that last number has to be yours.
Treat anything paywalled or single-sourced as a planning estimate. The inputs that move the decision are plant-specific: equipment condition, duty, environment, maintenance history, and the consequences of a failure. ANSI/NETA MTS-2023 specifies the field tests and inspections used to judge whether equipment is fit for continued service, and NFPA 70B-2026 sets out the electrical maintenance program and allows the potential-failure to functional-failure curve to determine maximum maintenance intervals. Our piece on insulation resistance testing and IEEE 43 gets into one of those tests in detail.
Where Teams Go Wrong
The most common mistake is reading an MTBF in years as a service life. A 280-year MTBF does not mean a breaker lasts 280 years. It means that across a large fleet, you’d expect one failure for every 280 unit-years of operation, on average.
The second is confusing failure rate with failure cause. “Bearings account for 50 to 70% of motor failures” is a cause distribution, not a rate, and the two answer different questions. Our Bearing Failure Cause Statistics article handles the cause side.
The third is quoting a single rate with no context. Without the equipment type, the construction, the voltage class, and the operating environment, the number is close to meaningless.
The fourth is assuming the data reflects today’s equipment. Eaton characterizes the historical IEEE figure as a likely worst-case estimate for its modern vacuum breakers. That’s a manufacturer assessment of its own product, not an independent field comparison, but it’s a fair reminder that the survey base predates a lot of current designs.
Methodology: How We Rated These
We rated a claim High when it’s documented and stable, such as the survey provenance, or when it’s a narrowly stated negative, such as “no single universal rate exists.”
We rated a figure Medium when it’s real and traceable but rests on old or small survey samples, or sits behind a paywall. IEEE Std 493 and IEEE Std 3006.8 are paid standards, so we confirmed specific numbers against open material where we could: Eaton’s technical paper for the breaker data, and the published record of the 1973-74 and 1979 IEEE transformer surveys for the transformer range.
We rated a claim Low when the arithmetic is correct but the interpretation isn’t, as with the MTBF-as-lifespan reading.
We did not use vendor blogs as load-bearing sources. The anchors here are IEEE standards and survey papers, plus a manufacturer technical document. A secondary engineering-firm article is cited only for its worked arithmetic example.
The Short Version
The numbers are real. They come from IEEE Std 493, with the analysis of equipment reliability data now addressed in IEEE Std 3006.8, and they work as historical screening inputs for fleet-level planning: expected failures across a population over time.
Three things to keep straight. Don’t read a roughly 280-year MTBF as a lifespan. Don’t confuse a failure rate with a failure cause. And remember the surveys behind these figures predate much of today’s equipment design and operating context, which is why, for any single unit, its condition and duty may tell you more than a generic fleet average.
Sources
- IEEE Std 493-2007 (inactive-reserved), Recommended Practice for the Design of Reliable Industrial and Commercial Power Systems (Gold Book).
- IEEE Std 3006.8-2018 (active), Recommended Practice for Analyzing Reliability Data for Equipment Used in Industrial and Commercial Power Systems.
- J. W. Aquilino, Report of Transformer Reliability Survey, IEEE Transactions on Industry Applications, vol. IA-19, pp. 858-866, 1983 (published 1983, reporting the 1979 survey that updated the 1973-74 transformer data).
- IEEE 3006 Historical Reliability Data (2012), summarizing equipment reliability surveys over 35-plus years.
- Eaton, Medium-voltage circuit breakers: reliability and availability for service, Application Paper AP083006EN, reproducing IEEE Std 493-1980 Table 24.
- ANSI/NETA MTS-2023, Standard for Maintenance Testing Specifications for Electrical Power Equipment and Systems (official NETA page).
- NFPA 70B-2026, Standard for Electrical Equipment Maintenance (official NFPA page).
- EDT Engineers, Transformer Failure: Frequency and Causes (used here for the worked arithmetic example, not as a primary rate source).
