---
title: "Power-Quality Problems: Manufacturing Costs Beyond Blackouts"
slug: power-quality-problems-manufacturing-costs
url: /topics/guides/power-quality-problems-manufacturing-costs/index.md
publishedAt: "2026-11-05T21:20:00"
updatedAt: "2026-11-05T21:20:00"
excerpt: "By the editors at Reliable. Last updated November 5, 2026. The short version: Momentary interruptions of five minutes or less accounted for about two-thirds of the cost of U.S. power interruptions, $52 billion of roughly $80 billion a year, in Lawrence Berkeley National Laboratory’s 2004 estimate (2002 dollars). Sustained outages, the blackouts everyone counts, accounted […]"
authors:
  - Reliable Media
---

# Power-Quality Problems: Manufacturing Costs Beyond Blackouts

By [Reliable Media](/authors/reliable-media/index.md)

*By the editors at Reliable. Last updated November 5, 2026.*

## What "power quality" covers, and where the five-minute line sits

[IEEE Std 1159-2019](https://standards.ieee.org/ieee/1159/6124/), the recommended practice for monitoring electric power quality, sorts disturbances by magnitude and duration. A voltage sag is a drop in rms voltage to between 10% and 90% of nominal lasting from half a cycle to one minute. An interruption is a drop below 10%. Interruptions are momentary at half a cycle to three seconds, temporary at three seconds to one minute, and sustained beyond one minute.

The standard also covers swells, transients, harmonic distortion, voltage unbalance, and flicker. A manufacturing process does not care which category tripped it. A programmable controller that drops out at 70% voltage for ten cycles produces the same restart whether the cause was a lightning strike two feeders away or a breaker operation at the substation.

Utility reliability reporting draws a different line. IEEE Std 1366, the guide for distribution reliability indices, classifies any interruption of five minutes or less as momentary and keeps it out of SAIDI and SAIFI, the two indices utilities report. Momentary events get their own index, MAIFI, which far fewer utilities track or publish.

That line explains the official numbers. The [U.S. Energy Information Administration](https://www.eia.gov/todayinenergy/detail.php?id=66744) reported that the average customer served by utilities using IEEE reporting went without power for 11 hours in 2024, nearly twice the prior decade's average. Major events, including Hurricanes Beryl, Helene, and Milton, accounted for 80% of those hours.

EIA describes SAIDI as the duration of non-momentary interruptions, and its guidance says utilities following the IEEE method report interruptions lasting longer than five minutes; most reporting utilities use that method, and some use their own. Without major events, customers routinely lose about two hours a year, and the average customer saw 1.5 interruptions in 2024.

EIA's 2025 IEEE-reporting series, in the Electric Power Annual released October 21, 2026, records an average of 269.2 minutes, about 4.5 hours, and 1.34 non-momentary interruptions per customer, including major event days. The 2024 figures in the same series were 662.6 minutes and 1.531 interruptions.

These reliability statistics do not measure voltage sags. Under IEEE-based reporting, SAIDI and SAIFI exclude momentary interruptions of five minutes or less. For a plant, those are brief events that can trip a process.

## Momentary interruptions: two-thirds of the cost

The source of the momentary-versus-sustained split is [Understanding the Cost of Power Interruptions to U.S. Electricity Consumers](https://eta-publications.lbl.gov/bibcite/reference/22961), published by Lawrence Berkeley National Laboratory in 2004 after the Northeast blackout of 2003. Kristina Hamachi LaCommare and Joseph Eto built an end-use framework from three inputs: customer surveys on the value of uninterrupted service, utility interruption records, and EIA counts of customers by type and region.

Their base case put the national cost of power interruptions at about $80 billion a year in 2002 dollars. Momentary interruptions, defined as those under five minutes, accounted for $52 billion of it. Sustained interruptions accounted for $26 billion. The authors' explanation was that momentary events are far more frequent, and that for commercial and industrial customers the cost is set by the downtime that follows, which can run far longer than the event itself.

By sector, commercial customers bore $57 billion (73%) and industrial customers $20 billion (25%), with residential losses near $1.5 billion. The industrial cost per outage per customer was much higher, but there were only 1.6 million industrial customers against 14.9 million commercial ones. Sensitivity runs put the total anywhere from under $30 billion to over $130 billion.

The report did not price power-quality events. The authors wrote that they could not develop an estimate for them with the information available, and they called for national collection of interruption and power-quality data. No national annual total for sag and transient costs has appeared in the sources reviewed for this guide since.

LBNL updated the sustained-interruption half in 2018. Using better utility reliability data, LaCommare, Eto, Laurel Dunn, and Michael Sohn put sustained interruptions at $44 billion a year in 2015 dollars, with a modeled uncertainty range of roughly $35 billion to $50 billion. Commercial customers carried 70% of that, industrial 27%, and residential 3%.

One step in that update, accounting for customers with backup generators, lowered the estimate 18%; the $44 billion already reflects it. The 2018 work covered interruptions over five minutes only and did not revisit the momentary half.

LBNL has priced momentary interruptions per event since. Its January 2015 meta-analysis updated the models used by the Interruption Cost Estimate Calculator, first released in July 2011. Drawing on 34 utility survey datasets collected from 1989 to 2012, it estimated momentary-interruption costs of $12,952 per event for medium and large commercial and industrial customers (over 50,000 kWh a year) and $412 for small customers, in 2013 dollars.

LBNL's February 2026 ICE Calculator 2 report updates that work. It combines 15 survey activities across 30 service territories, with nearly 10,000 responses collected between December 2022 and June 2025, and includes a momentary scenario of up to five minutes alongside 2-hour, 8-hour, and 24-hour outages.

The weighted survey mean for a momentary interruption among non-residential respondents was $1,001 per event; the report's modeled average is $548 per event in 2025 dollars. Non-residential pools commercial and industrial customers of all sizes. The 2015 and 2026 studies differ in customer groups, survey periods, dollar bases, and models, so their averages should not be read as a cost trend. The 2026 report publishes per-event costs rather than a new national annual total.

## What one second costs: the EPRI survey

EPRI's 2001 survey provides historical estimates of brief-outage costs for industrial and digital-economy establishments. The report is [The Cost of Power Disturbances to Industrial and Digital Economy Companies](https://restservice.epri.com/publicdownload/000000003002000476/0/Product), done by Primen for EPRI's Consortium for Electric Infrastructure to Support a Digital Society and dated June 29, 2001. The survey reached 985 establishments between February and April 2001.

Three sectors were covered: the digital economy, continuous process manufacturing (paper, chemicals, petroleum, rubber and plastics, glass, primary metals), and fabrication and essential services (all other manufacturing plus utilities and transportation).

Those sectors held about 2 million establishments, 17% of the U.S. total, and roughly 40% of GDP. Respondents were given hypothetical outage scenarios and asked to estimate their costs line by line, a method called direct costing. EPRI's survey counted any complete loss of voltage as an outage, including losses of a fraction of a second.

The estimated net cost per event, averaged across all three sectors, was $1,477 for a one-second outage, $2,107 for three minutes, and $7,795 for one hour. A recloser event, two one-second drops a few seconds apart, was estimated at $2,848. A one-second outage lasts 0.03% as long as a one-hour outage and was estimated to cost almost 20% as much.

The reason is restart time. Respondents estimated that a one-second outage stopped or slowed their activities for 8.9 minutes on average. A recloser event or a three-minute outage disrupted activities for about 14 minutes, and a one-hour outage for 71.6 minutes.

The cost elements explain where the money goes. For a one-second outage, equipment damage ($554) and idled labor ($433) outweighed lost production ($284) and spoiled materials ($99). For a recloser event, equipment damage was $2,011 of the $2,848 total, which the report said was presumably from the strain of stopping and restarting so quickly.

Averages hide the spread. Almost three-quarters of establishments (74%) reported no estimated cost from a one-second outage, and half reported none from a three-minute outage. At the other end, 5% of establishments estimated losses of $20,000 or more from a one-hour outage, with individual values as high as $1.5 million. Continuous process manufacturers averaged $12,654 for a one-second outage and $14,746 for one hour.

Size and equipment mattered more than sector. Establishments using 5 gigawatt-hours or more a year averaged $32,000 for a one-second outage and more than $59,000 for one hour. Among industrial sites, those with variable speed drives reported one-hour outage costs about eight times those of sites without them, with electronic controls and computer-assisted production systems showing similar differences. The survey compared groups; it did not isolate a cause.

Respondents averaged 3.9 outages a year, and 49% of those outages lasted under three minutes. Outages of an hour or more were 20% of the total. The report noted that brief outages were often missing from official statistics; California utilities at the time were not required to report outages under three minutes to the state commission.

Added up, the average establishment's estimated annual outage loss was $23,318: $61,828 in continuous process manufacturing, $49,328 in fabrication and essential services, and $10,598 in the digital economy. The three sectors together came to $45.7 billion a year. In fabrication and essential services, the annual outage cost exceeded the average annual electric bill.

Because EPRI's outage figures include momentary events, they overlap with LBNL's $52 billion and cannot be added to it. The two studies used different methods and populations and should be quoted separately.

## Voltage sags: disturbances that sustained-outage statistics miss

EPRI's 2022 brief summarizes three benchmarks of U.S. power quality. The first Distribution Power Quality project monitored 277 sites at 24 utilities from 1993 to 1995, and a second phase used utilities' permanent monitors in 2002. The third, the Transmission-Distribution Power Quality report (TPQ-DPQ III, 2014), examined data from seven utilities across the continental United States between 2009 and 2012.

The brief is [Societal Costs of Power Quality Disturbances](https://restservice.epri.com/publicdownload/000000003002024890/0/Product), a July 2022 PQ TechWatch publication. The index used is SARFI, the system average rms variation frequency index, which counts events per site per year below a stated voltage threshold; interruptions count along with sags.

EPRI's TPQ-DPQ III study, using data collected from seven utilities during 2009 to 2012, reported an all-sites average of 15.7 sag and interruption events per year below 70% of nominal voltage and 48.5 below 90%, with 8.7 below 50% and 3.4 below 10%. All sites means transmission, subtransmission, distribution, and low-voltage monitoring points together.

The brief also counts events against equipment immunity curves: 17.0 a year outside the ITIC envelope for information technology equipment and 11.7 outside the SEMI F47 envelope, which EPRI describes as the number of events a process built to each curve could be expected to shut down for. Whether a given plant trips on any of them depends on its equipment, the event's duration, how many phases it affects, and how the plant is connected.

The earlier phases were in the same range: 23.4 events below 70% in the 1993 to 1995 study and 13.7 in the 2002 study.

Those counts cannot be set in ratio against EIA's 1.5 interruptions per customer; the populations, years, and methods differ. What they show is that the sustained-outage statistics plants quote leave the sag count out entirely.

The mechanism is structural. Pacific Gas and Electric's power note on short-duration voltage sags explains that most substations feed several circuits, so a fault on any one of them depresses voltage for every customer on the bus until the fault clears. PG&E's illustration is a customer who sees one or two interruptions a year and more than 20 sags.

Most sags are shallow and short. EPRI's data show that most events below 90% of nominal keep at least 50% of nominal voltage and last under half a second. Single-phase sags are the most common (53%), followed by two-phase (27%) and three-phase (20%). Three-phase sags are the most expensive to mitigate.

EPRI's brief calls the voltage sag the most significant power-quality event for industrial customers, and it also reports voltage swells and transients. Swells to 110% to 115% of nominal occurred about 2.2 times per site per year. Oscillatory transients, typically from capacitor-bank switching, reached 1.05 to 1.10 per unit roughly 35 times a year.

EPRI notes such transients can reach 4.0 per unit when a utility bank interacts with a smaller customer bank, and that the metal oxide varistors that protect against them erode with each high transient and may eventually fail.

Equipment immunity is the other half of the equation. [SEMI F47](https://store-us.semi.org/products/f04700-semi-f47-specification-for-semiconductor-processing-equipment-voltage-sag-immunity), first published in 1999 and current as F47-0706 (Reapproved 0812)E, requires semiconductor processing tools to ride through sags to 50% of nominal for 200 milliseconds, 70% for 0.5 seconds, and 80% for one second.

A SEMI task force has been revising it since 2019 because fabs still report tool trips from three-phase sags, which the current standard does not require testing for and which SEMI says can account for as many as 20% of events.

## What the sags and surges themselves cost

EPRI's 2001 survey asked a separate set of questions about power-quality phenomena, meaning everything other than a complete loss of voltage. Respondents reported an average of 8.3 power-quality problems a year that they were aware of. Thirty percent reported none, 20% reported 10 or more, and 2% reported 100 or more.

The symptoms were the familiar ones. Among establishments that had problems, every one reported power strips or breakers tripping, 93% reported lights flickering, 71% reported computers locking up or restarting, and 34% reported employees getting shocks. Damage to electronic controls was more common in the two industrial sectors than in the digital economy, and abnormal motor behavior was most common in continuous process manufacturing.

Annual losses from power-quality problems averaged $3,406 per establishment, with 69% reporting no cost at all. One percent lost $20,000 or more. The largest single report was $3 million from a millwork plant using 30 gigawatt-hours a year that logged more than 100 events and damage to computers, motors, and dynamic brakes. Fabrication and essential services averaged $9,643 a year, continuous process manufacturing $2,859, and the digital economy $599.

Consumption was the strongest predictor. Establishments using 5 gigawatt-hours or more averaged $59,601 a year in power-quality losses.

Across the three sectors the total was $6.7 billion, of which $5.7 billion fell in fabrication and essential services. Extrapolating to all U.S. businesses, on the assumption that unsurveyed sectors lose 25% to 50% as much per establishment, gave $15 billion to $24 billion a year.

The report did not break those costs down by phenomenon, because, in its words, most respondents could not make the distinction. It also described customer perceptions of power quality as a rough estimate or an underestimate, since the symptoms are often invisible until something fails. Respondents reported that every establishment had at least one UPS, 87% used surge protectors, 13% had backup generators, and the average site had invested $1,300 in mitigation, more than $2.5 billion across the three sectors.

EPRI adjusted the 2001 figures for inflation in a 2020 report (3002019395) and summarized the result in its 2022 brief. The brief's summary does not separate outage losses from power-quality losses, and the 2020 report itself was not retrievable for this guide, so those updated figures are not reproduced here. The brief does list the sectors it found most affected by power-quality events: lumber and wood products, food, chemicals, primary metals, paper, petroleum and coal products, and the Postal Service.

## Harmonics, unbalance, and flicker: measured, rarely priced

EPRI's benchmark measured harmonics; none of the sources reviewed here prices them separately. TPQ-DPQ III found voltage total harmonic distortion at the seven utilities well below the 5% limit that applied before 2014.

IEEE 519-2014 introduced an 8% voltage distortion limit at points of common coupling at 1 kV or below; the limit above 1 kV through 69 kV stayed at 5%. IEEE 519-2022 is the current edition, and its limits apply at the point of common coupling rather than at every measurement point inside a plant.

Voltage unbalance on the distribution systems studied stayed below 1.6%. EPRI reported sampled flicker values below 0.6 and compared them with low-voltage compatibility levels of 1.0 for short-term flicker and 0.8 for long-term flicker.

EPRI's brief adds a caution for the years since. It says inverter-connected distributed generation and power-electronic loads may be expected to decrease grid stiffness and increase harmonics, and that high-frequency resonance conditions apparently tied to inverter switching have been observed on some systems since the study was published. Harmonic heating of transformers and motors, nuisance tripping, and capacitor failures are the usual plant-level consequences, and none of the sources reviewed here puts a national dollar figure on them.

The plain statement is that the $6.7 billion and the $15 billion to $24 billion already include whatever harmonic and unbalance costs respondents could see in 2001, pooled with sags and transients, and that none of the sources reviewed for this guide separates them.

## What the national estimates include and exclude

Three recent figures circulate as "the cost of power outages": two dollar estimates and one reliability statistic. Each one counts something narrower than its headline suggests.

Oak Ridge National Laboratory announced a county-level analysis of major outages on March 4, 2026, [Analysis shows power outages cost US electricity customers billions](https://www.ornl.gov/news/analysis-shows-power-outages-cost-us-electricity-customers-billions). It found that major outages, defined as events affecting 30,000 customers or 15% of a county's customers, whichever is less, cost an average of more than $67 billion a year from 2018 through 2024 and $121 billion in 2024 alone.

Commercial and industrial customers averaged $6,031 per outage in 2024. ORNL reported 4,666 major outages in 2018 and 6,533 in 2024, and the average event lengthened from 9.6 to 11.8 hours. The release does not state a dollar base year.

ORNL built those figures on LBNL's Interruption Cost Estimate Calculator, which computes business costs as lost production plus direct interruption costs minus energy savings. The calculator's own surveys include momentary interruptions; the major-outage restriction belongs to ORNL's analysis, and its lead researcher said small, frequent outages would be added later.

LBNL's $44 billion (2018) covers sustained interruptions only. EIA's 11 hours (2024) is SAIDI, which under IEEE-based reporting excludes momentary interruptions. The ORNL study counts major outages only. For the brief events that can trip manufacturing processes, the national annual totals in the sources reviewed here remain the 2001 EPRI survey and the 2004 LBNL framework; LBNL's 2015 and 2026 work adds per-event costs for momentary interruptions and nothing for sags.

For the cost of the downtime that follows any of these events, see our [cost of unplanned downtime](https://reliamag.com/topics/articles/cost-unplanned-downtime-manufacturing) article. The dollar figures there are per hour of lost production; this guide separates per-event estimates from annual totals.

## At a Glance

## Calculate Your Own

National averages say little about any one plant. The EPRI survey shows why: 74% of establishments reported no estimated cost from a one-second outage, and a handful of sites carried most of the losses. The number that matters is yours, and it takes a few inputs.

**Annual cost of short events = N × (D × M + L + S + E + R)**

N is the number of events a year that trip your process. A power-quality monitor at the service entrance records electrical events; matching that log against process-trip records is what tells you which events stopped production, and local faults and large motor starts belong in the count too. Some utilities can provide SARFI for your substation at the threshold your equipment drops out at.

D is minutes of process downtime per event, measured from the trip to full-rate production; EPRI's respondents estimated that activities stopped or slowed for 8.9 minutes after a one-second outage. M is contribution margin per minute of production. L is idled labor per event, S is scrap and spoiled material, E is equipment damage averaged across events, and R is restart costs such as purges, cleaning, and quality checks.

Count each cost once. Estimate contribution margin only on production that will not be recovered. Add only event-related costs that the margin estimate does not already include, and count each labor, material, repair, and restart cost a single time.

If production is made up later, count the incremental cost of recovering it rather than treating all delayed output as lost sales. LBNL's direct-cost method for the ICE Calculator works the same way: lost production net of what is made up, plus incremental costs, minus incremental savings.

Worked example, with every input assumed. A plant's trip log shows 12 process-tripping events a year, an assumed figure. Restart to full rate takes 25 minutes and none of the output is recovered. The $400-per-minute production margin is calculated after deducting normal direct labor, so lost production is $10,000 per event.

The wages deducted in that margin calculation remain payable during downtime, so they are added back once: thirty operators idle for 25 minutes at $35 an hour cost $437.50. The other cost entries exclude those wages. Scrap averages $1,500, equipment damage averages $200 across events, and restart consumables and checks cost $300. If your margin figure leaves continuing labor undeducted, omit the separate wage line instead of adding it.

Per event: $10,000 + $437.50 + $1,500 + $200 + $300 = $12,437.50. Times 12 events: $149,250 a year. That total is the arithmetic of the assumptions above, and your own inputs will move it. The first input to work on is N: use the event and trip records to identify which equipment needs additional ride-through capability before conditioning anything else.

## How We Evaluated

Independent editorial analysis based on publicly available government data, peer-reviewed research, and the original pages behind widely quoted benchmarks.

Reliable does not sell scheduling software, staffing services, or consulting and has no commercial interest in routing readers toward any particular vendor. Reliable does not accept payment for rankings. Vendors may sponsor enhanced listings with additional detail, but editorial rankings are independent. [Read our editorial policy](https://reliamag.com/editorial-policy/).

Sources are tiered. National laboratories and federal agencies (LBNL, ORNL, EIA) and consensus standards (IEEE, SEMI) come first. EPRI is an industry-funded research institute whose members are electric utilities; its 2001 survey was conducted by its affiliate Primen, and its monitoring benchmarks come from member utilities. A utility customer bulletin (Pacific Gas and Electric) is used for one mechanism only.

Each figure stays tied to its own population, method, and dollar year. Nothing is summed across studies, and our own arithmetic is labeled as arithmetic.

## Honest Limitations

The two studies that put national annual totals on momentary interruptions and power-quality events are dated 2001 and 2004. The EPRI outage costs came from respondents' estimates for hypothetical scenarios, its power-quality costs from respondents' own experience, and EPRI's brief describes those power-quality cost perceptions as rough estimates or underestimates.

LBNL's 2015 meta-analysis and its February 2026 ICE Calculator 2 report measured momentary interruptions per event, the latter for non-residential customers as a pooled class. No source reviewed here publishes a newer national annual total for momentary interruptions, or separate national totals for voltage sags or harmonics. EPRI's 2020 inflation adjustment changes the dollars without changing the data, and the report behind it could not be retrieved.

The sag-frequency benchmark rests on seven utilities monitored from 2009 to 2012, averaged across every voltage level monitored rather than distribution feeders alone. EPRI itself flags that inverter-based generation and loads may have changed the harmonic environment since. Sag exposure at a specific plant depends on feeder length, substation configuration, lightning density, and whether supply is overhead or underground, none of which an all-sites average captures.

The 2001 survey's national extrapolation assumes that unsurveyed sectors lose 25% to 50% as much per establishment as the surveyed ones. That assumption produces the $104 billion to $164 billion outage range and the $15 billion to $24 billion power-quality range, and the report says the extrapolation is valid only if the assumption holds.

EPRI's per-site event counts and EIA's per-customer interruption counts come from different populations, years, and methods. This guide presents them side by side to show what each measures and draws no ratio between them.

Harmonics, unbalance, and flicker have measured levels but no separate national cost figure in any source reviewed. Statements about their plant-level effects in this guide describe mechanisms; none of them is a measurement.

## Frequently Asked Questions

### How much do momentary power interruptions cost U.S. industry?

Lawrence Berkeley National Laboratory's 2004 framework estimated that interruptions under five minutes cost U.S. electricity customers about $52 billion a year in 2002 dollars, two-thirds of an $80 billion total, with industrial customers bearing about $20 billion of all interruption costs. LBNL's 2018 update covered sustained interruptions only. Its February 2026 ICE Calculator 2 report puts a momentary interruption at a modeled $548 per non-residential customer event in 2025 dollars, a per-event figure rather than a new national total.

### What does a one-second power outage cost a manufacturing plant?

Respondents to EPRI's 2001 survey of 985 industrial and digital-economy establishments estimated a net average cost of $1,477 per one-second outage in the 2001 survey's dollars, driven by about 8.9 minutes of stopped or slowed activity, equipment damage, and idled labor. Continuous process manufacturers averaged $12,654, and sites using 5 gigawatt-hours or more a year averaged $32,000. Three-quarters of establishments reported no estimated cost at all.

### What sag and interruption frequencies did EPRI's monitoring study find?

EPRI's TPQ-DPQ III benchmark, drawn from seven U.S. utilities between 2009 and 2012, found an all-sites average of 48.5 sag and interruption events a year below 90% of nominal voltage and 15.7 below 70%. EPRI counts 11.7 of those events a year outside the SEMI F47 immunity envelope; whether a given process trips on them depends on its equipment and how the sag reaches it.

### Why can a brief power disturbance cause a much longer production stop?

Because the process has to restart. EPRI's respondents estimated that a one-second outage stopped or slowed activities for about 8.9 minutes. Equipment damage accounted for most of the estimated recloser-event cost; the report suggested rapid stopping and restarting as a possible explanation. Sites with variable speed drives, electronic controls, and computer-assisted production reported far higher costs than sites without them, and an interrupted process can also scrap work in progress.

### Are harmonics included in power-quality cost estimates?

Only in pooled form. EPRI's 2001 survey asked respondents for a single annual power-quality loss covering sags, surges, transients, and harmonics together, because most could not separate them. No source reviewed for this guide publishes a national dollar figure for harmonics alone, though EPRI has measured distribution voltage distortion and found it below the pre-2014 IEEE 519 limit.

### Does the EIA's 11 hours without power include momentary interruptions?

No. EIA reports SAIDI, which measures non-momentary interruptions, and under the IEEE 1366 method most reporting utilities use, an interruption of five minutes or less is momentary. The 11-hour average for 2024, the 4.5 hours for 2025, and the interruption counts that go with them cover sustained outages only, for customers served by utilities using IEEE reporting.

## Related Guides

* [Hidden Factory: Microstops and Lost Capacity](https://reliamag.com/topics/guides/hidden-factory-microstops-lost-capacity)
* [Industrial Downtime Cost Benchmarks: What Published Studies Actually Show](https://reliamag.com/topics/guides/industrial-downtime-cost-benchmarks)
* [Electric Motor Failure Statistics: Causes, Failure Rates, and Survey Data](https://reliamag.com/topics/guides/electric-motor-failure-statistics)
* [The Aging Transformer Fleet: What the 40-Year Claim Really Says](https://reliamag.com/topics/guides/aging-transformer-fleet-statistics)
* [Equipment Breakdown Insurance Loss Statistics: Causes and Cost Shares](https://reliamag.com/topics/guides/equipment-breakdown-insurance-statistics)

## Sources

* Lawrence Berkeley National Laboratory: [Understanding the Cost of Power Interruptions to U.S. Electricity Consumers](https://eta-publications.lbl.gov/bibcite/reference/22961) (LBNL-55718, 2004), [Berkeley Lab Study Estimates $80 Billion Annual Cost of Power Interruptions](https://newscenter.lbl.gov/2005/02/02/berkeley-lab-study-estimates-80-billion-annual-cost-of-power-interruptions/) (news release, February 2005), [Berkeley Lab Estimates Sustained Electric Power Interruptions Cost the U.S. Approximately $44 Billion Annually](https://emp.lbl.gov/news/berkeley-lab-estimates-sustained) (October 2018), [Updated Value of Service Reliability Estimates for Electric Utility Customers in the United States](https://eta-publications.lbl.gov/sites/default/files/lbnl-6941e.pdf) (LBNL-6941E, January 2015), and [ICE Calculator 2: Final Report for Phase 1 and 2 of the National Initiative to Update the Interruption Cost Estimate Calculator](https://eta-publications.lbl.gov/sites/default/files/2026-03/ice-calculator2_phase_ii_final_report.pdf) (February 2026)
* Electric Power Research Institute: [The Cost of Power Disturbances to Industrial and Digital Economy Companies](https://restservice.epri.com/publicdownload/000000003002000476/0/Product) (Primen for CEIDS, June 2001) and [Societal Costs of Power Quality Disturbances](https://restservice.epri.com/publicdownload/000000003002024890/0/Product) (PQ TechWatch, July 2022, reporting TPQ-DPQ III and the 2020 inflation update)
* Oak Ridge National Laboratory: [Analysis shows power outages cost US electricity customers billions](https://www.ornl.gov/news/analysis-shows-power-outages-cost-us-electricity-customers-billions) (March 4, 2026)
* U.S. Energy Information Administration: [Hurricanes in 2024 led to the most hours without power in the United States in 10 years](https://www.eia.gov/todayinenergy/detail.php?id=66744) (Today in Energy, December 1, 2025) and [Electric Power Annual, Table 11.1, Reliability metrics of U.S. distribution system](https://www.eia.gov/electricity/annual/table.php?t=epa_11_01.html) (2025 data by reporting method, released October 21, 2026)
* IEEE: [IEEE Std 1159-2019, Recommended Practice for Monitoring Electric Power Quality](https://standards.ieee.org/ieee/1159/6124/) (disturbance definitions) and [IEEE Std 1366-2022, Guide for Electric Power Distribution Reliability Indices](https://standards.ieee.org/ieee/1366/7243/) (the five-minute momentary threshold used in reliability reporting)
* SEMI: [SEMI F47, Specification for Semiconductor Processing Equipment Voltage Sag Immunity](https://store-us.semi.org/products/f04700-semi-f47-specification-for-semiconductor-processing-equipment-voltage-sag-immunity) (F47-0706 (Reapproved 0812)E, current edition) and [Revision to SEMI F47-0706 (Reapproved 0812), Voltage Sag Immunity Task Force proposal](https://downloads.semi.org/web/wstdsbal.nsf/b8865fa87d9e7b57882579fb005c3cd7/bcb5fe4f9e88b79588258aaf0025f693!OpenDocument) (three-phase sag share and task force history)
* Pacific Gas and Electric: [Short Duration Voltage Sags Can Cause Disruptions](https://www.pge.com/assets/pge/docs/contact-us/report-an-issue/Short_Duration_Voltage_Sags.pdf) (customer power note)