Forced Outage

A forced outage is the unplanned shutdown of a power station, a generating unit, a transmission line or a distribution line because something has gone wrong. The term is used across the electricity industry to separate failures from scheduled work. A plant taken offline for its annual overhaul is on a planned outage. A plant that trips because a boiler tube bursts, a fuel supply fails or an operator makes a mistake is on a forced outage.

Forced outages matter because a power grid must match supply and demand at every moment. Planners know when planned outages will happen and can schedule around them. Forced outages arrive without warning, and when many happen at once, as in the Texas freeze of February 2021 or Winter Storm Elliott in December 2022, they can force grid operators to cut power to millions of people. This article explains how forced outages are defined, how they are measured using the North American Electric Reliability Corporation's Generating Availability Data System (GADS), how metrics such as the forced outage rate (FOR), equivalent forced outage rate (EFOR) and EFORd are calculated, how these numbers are used in planning and capacity markets, and what the major cold-weather events revealed.

Definitions

The basic idea

Wikipedia defines a forced outage as the shutdown of a power station, transmission line or distribution line when it is unavailable to produce or carry power because of an unexpected breakdown. The typical causes are equipment failures, disruptions in the fuel supply chain and operator error.

The US Nuclear Regulatory Commission's glossary gives a fuller definition. It describes a forced outage as the shutdown of a generating unit, transmission line or other facility for emergency reasons, or a condition in which equipment is unavailable because of an unanticipated breakdown. Under the NRC's wording, an outage is "forced" if it could not reasonably be delayed beyond 48 hours from when the problem was identified, even if there was a strong commercial reason to delay. The NRC's examples include:

• failures of mechanical, fuel handling or electrical equipment or controls within the generator's responsibility
• a failure of a mine or fuel transport system dedicated to the power station, causing a fuel shortage that cannot be managed economically
• inadvertent or operator error
• limitations caused by fuel quality

Scheduled outages for inspection, maintenance or refuelling are not forced outages.

Forced, maintenance and planned outages in GADS

In North America, generator outages are recorded according to the GADS Data Reporting Instructions published by NERC. These rely on definitions from IEEE Standard 762, "Definitions for Use in Reporting Electric Generating Unit Reliability, Availability and Productivity." GADS classifies outages by how urgently the unit has to come off line.

The definitions above follow the descriptions in NYISO and NERC training materials. A NERC training deck sums up the dividing line between forced and maintenance outages: a forced outage is one where the unit cannot keep running through the following weekend without significant risk to equipment.

GADS also records deratings, which are partial outages. A derating means the unit is running but cannot reach its full output. Forced deratings are coded D1, D2 and D3, in parallel with U1, U2 and U3. They matter because a 600 MW unit stuck at 400 MW for a week has lost a third of its capacity, even though it never actually shut down.

Outages on the network

The same idea applies to transmission and distribution lines. A line that trips because of a lightning strike, a fallen tree or a failed insulator is on a forced outage. Distribution reliability is usually measured with customer-focused indices rather than outage rates. Examples are SAIDI (the average total interruption duration per customer) and SAIFI (the average number of interruptions per customer). This article focuses on generators, where the "forced outage rate" family of metrics is used.

A short history of outage data

Utilities have long tracked how often their plants break down, because spare capacity has to be sized for those failures. In the United States, the Edison Electric Institute ran a program collecting generating-unit outage data from the mid-1960s. Wikipedia reports that this collection began in 1963. NERC's Generating Availability Data System replaced that program, and utilities began reporting under GADS guidelines on 1 January 1982.

For three decades GADS reporting was voluntary. It became mandatory for conventional generating units of 50 MW and larger from 1 January 2012, and for units of 20 MW and larger from 1 January 2013. Wikipedia reports that GADS holds operating histories for more than 5,800 generating units, which it says represents 71% of installed US and Canadian generating capacity. NERC has since added reporting systems for wind and solar plants, whose outages are recorded differently because a wind farm is made up of many individual turbines.

GADS collects three kinds of data:

• Design data, describing the unit and its equipment
• Performance data, such as generation, fuel and start-ups
• Event data, describing each outage and derating, its timing and its cause

From these, NERC and utilities calculate standard performance indexes.

How forced outage rates are calculated

The time categories

GADS divides every hour in a reporting period into categories. NERC's Appendix F, "Performance Indexes and Equations," gives these definitions:

A derating is turned into "equivalent hours" by multiplying its duration by the size of the reduction and dividing by the unit's net maximum capacity (NMC):

EFDH = Σ (derating hours × size of reduction in MW) ÷ NMC

For example, a 600 MW unit derated by 200 MW for 30 hours adds 30 × 200 ÷ 600 = 10 equivalent forced derated hours.

Forced outage rate (FOR)

The forced outage rate is the basic measure. Wikipedia describes it as the probability that a unit will not be available for service when required, calculated from forced outage hours as a share of the time the unit was either running or on forced outage. Ignoring the small categories of pumping and synchronous condensing hours, the GADS formula reduces to:

FOR = FOH ÷ (FOH + SH) × 100%

The denominator includes only time the unit was running or broken. Time in reserve shutdown or planned outage is left out, because the unit was not being asked to run.

Equivalent forced outage rate (EFOR)

FOR ignores deratings. The equivalent forced outage rate adds them:

EFOR = (FOH + EFDH) ÷ (FOH + SH + EFDHRS) × 100%

again leaving out the minor pumping and synchronous condensing terms. NERC describes EFOR as the hours of unit failure, both outage hours and equivalent derated hours, as a percentage of the relevant hours. Deratings during reserve shutdown appear in the denominator so that the numerator and denominator cover the same periods.

Forced outage factor (FOF)

The forced outage factor measures forced outage hours as a share of all period hours:

FOF = FOH ÷ PH × 100%

FOF tells you how much of the year a unit spent broken. FOR tells you how likely it is to be broken when it is needed. A peaking plant that runs only a few hundred hours a year can have a low FOF and a high FOR at the same time.

Demand-based rates: EFORd

A plant that runs constantly and a plant that runs only on the hottest summer afternoons are exposed to failure in different ways. If a peaker breaks down during a month when it would not have been called anyway, that failure does not hurt the grid much. The equivalent forced outage rate on demand (EFORd) tries to count only the outage time that occurred when the unit would actually have been needed.

The GADS formula is:

EFORd = (FOHd + EFDHd) ÷ (SH + FOHd) × 100%

where FOHd is the number of forced outage hours that occurred during periods of demand, and EFDHd is the equivalent forced derated hours during demand. If periods of demand are recorded directly, FOHd can be counted directly. If not, NERC estimates it using a demand factor, f:

FOHd = f × FOH

f = (1/r + 1/T) ÷ (1/r + 1/T + 1/D)

where:

• r = average forced outage duration = FOH ÷ number of forced outage occurrences
• T = average reserve shutdown time = RSH ÷ number of attempted starts
• D = average demand time = SH ÷ number of actual starts

EFDHd is EFDH − EFDHRS if reserve shutdown events were reported. If not, it can be approximated as fp × EFDH, where fp = SH ÷ AH.

The f-factor comes from a simple model in which a unit cycles between being needed and not needed, and between working and broken. It is a probability-style estimate of how much of the forced outage time overlapped with demand. NERC notes that "demand" can be defined in other ways for special studies, such as by weather condition, load level or energy price.

There are capacity-weighted versions of all these metrics. Weighted EFOR (WEFOR), for example, weights each unit's figures by its capacity, so that a 1,000 MW unit counts more than a 50 MW unit when computing a fleet average.

A worked example (hypothetical)

The following numbers are invented to show how the formulas work. They are not data from any real plant.

A 500 MW gas-fired unit has the following record for a 365-day year (8,760 hours):

Check: 6,000 + 2,000 + 300 + 460 = 8,760 hours, so every hour is accounted for.

Forced outage factor. FOF = 300 ÷ 8,760 = 3.42%.

Forced outage rate. FOR = 300 ÷ (300 + 6,000) = 300 ÷ 6,300 = 4.76%.

Equivalent forced outage rate. EFOR = (300 + 120) ÷ (300 + 6,000 + 0) = 420 ÷ 6,300 = 6.67%.

EFORd. First the f-factor:

• r = 300 ÷ 6 = 50 hours, so 1/r = 0.0200
• T = 2,000 ÷ 40 = 50 hours, so 1/T = 0.0200
• D = 6,000 ÷ 38 = 157.9 hours, so 1/D = 0.00633

f = (0.0200 + 0.0200) ÷ (0.0200 + 0.0200 + 0.00633) = 0.0400 ÷ 0.04633 = 0.863

FOHd = 0.863 × 300 = 259.0 hours. Because all derating happened while the unit was in service, EFDHd = EFDH − EFDHRS = 120.

EFORd = (259.0 + 120) ÷ (6,000 + 259.0) = 379.0 ÷ 6,259.0 = 6.06%.

In this example, EFORd (6.06%) is lower than EFOR (6.67%) because the model estimates that about 14% of the forced outage time fell in periods when the unit would not have been running anyway.

Effect on capacity value. Many capacity markets discount a unit's rated capacity by its EFORd. In PJM's capacity market, unforced capacity (UCAP) is calculated as installed capacity × (1 − EFORd). For this hypothetical 500 MW unit: UCAP = 500 × (1 − 0.0606) = 469.7 MW. A unit with worse reliability would be credited with less capacity and paid less in the capacity market.

Pooling across a fleet

Averaging EFORd across many units is not as simple as averaging the percentages. NERC's Appendix F reproduces a sample from IEEE 762 comparing three pooling methods on the same set of units. Pooling the underlying hours gave 7.92%. Averaging the individual units' EFORd values weighted by capacity gave 8.02%. NERC states that it uses one specific method (its "Method 2," which calculates f-factors from the group's total hours and starts) for all of its EFORd calculations, because it smooths out units with very few service hours and avoids dividing by zero. When comparing published fleet statistics, it matters which method was used.

Why forced outages happen

Forced outage causes fall into a few broad groups.

Mechanical and electrical failures. In thermal plants, boiler tube leaks, turbine problems, generator faults, pump and fan failures, control system faults and transformer failures are common causes. Older units and units that cycle on and off more than they were designed for tend to suffer more.

Fuel supply. A plant can be in perfect condition and still be unable to run if it has no fuel. Natural gas plants that buy fuel on interruptible pipeline contracts are particularly exposed when gas demand for heating spikes. Coal piles can freeze, and fuel deliveries can be disrupted by weather or transport problems.

Weather. Extreme cold freezes instrument lines, valves and water systems. Extreme heat reduces output and can strain cooling systems. Ice on wind turbine blades shuts turbines down.

Human error. Operator mistakes and maintenance errors cause trips and failures.

External events. Transmission faults, fires, floods and other external events can force units off line, although GADS distinguishes some of these as "outside management control" so they can be excluded from certain metrics.

Cold weather: Winter Storm Uri, February 2021

What happened

In February 2021, an arctic air mass and freezing precipitation hit Texas and the south-central United States. The joint inquiry by staff of the Federal Energy Regulatory Commission (FERC), NERC and the regional entities studied the event from 8 to 20 February 2021, which it called "the Event," with the most severe period from 15 to 18 February.

The inquiry found that 1,045 individual generating units experienced 4,124 outages, derates or failures to start, of which 604 were natural gas-fired generators. The preliminary findings put the generation lost to unplanned cold-weather outages at 61,800 MW. For more than two consecutive days, from 15 to 17 February, the Electric Reliability Council of Texas (ERCOT) averaged 34,000 MW of unavailable generation, including units already on outage. That was 49%, or nearly half, of ERCOT's all-time winter peak load of 69,871 MW.

Load shedding

To keep the grid from collapsing, the three affected balancing authorities ordered controlled firm load shed (rotating blackouts) totalling more than 23,400 MW:

The inquiry called this the largest controlled firm load shed event in US history. Measured in megawatts of load lost, it was the third largest outage after the August 2003 northeast blackout and the August 1996 west coast blackout.

The timing in ERCOT shows how fast forced outages can pile up. ERCOT lost 15,000 MW of generation in five hours, from 10:00 pm on 14 February to 3:00 am on 15 February (central time). With grid frequency falling, operators ordered 1,000 MW of load shed at 1:20 am on 15 February, 10,000 MW by 4:30 am, and 20,000 MW by 7:00 pm. ERCOT's grid is largely separate from the rest of North America. Its direct current ties to SPP and Mexico allow a maximum import of only 1,220 MW, so it could not import much help.

Causes

The inquiry's breakdown of the 4,124 outages, derates and failures to start was:

Together, freezing and fuel issues accounted for 75% of the unplanned outages and derates. The most common freezing problems were frozen instrumentation, such as sensing lines and transmitters, and icing on wind turbine blades. Of the 1,293 fuel-related events, 1,121 (87%) were natural gas fuel supply problems. Those affected 357 individual gas-fired units: 185 in ERCOT, 141 in SPP and 31 in MISO. Even the "mechanical and electrical" failures increased as temperatures fell, so the inquiry concluded they were also related to the cold.

The gas supply problem had its own cause. US dry gas production fell to 65.4 billion cubic feet per day on 17 February, a 28% decline from 90.8 Bcf/d on 4 February. It was the largest monthly decline in US gas production on record. In Texas, Oklahoma and Louisiana combined, production fell by more than half. Wellheads and processing plants froze, and some lost electric power, which shows how tightly the gas and electric systems depend on each other.

Many plants in the southern US are built without enclosing buildings, so that boilers and turbines do not overheat in summer. That leaves equipment exposed in a deep freeze. The inquiry noted that freeze protection typically involves heat tracing, insulation, wind breaks, temporary heating and other measures.

Consequences and recommendations

The human cost was severe. The Texas Department of State Health Services, in its final report released at the end of December 2021, counted 246 deaths related to the storm in Texas. About two-thirds were from cold exposure, including 158 from hypothermia.

The FERC-NERC team made 28 preliminary recommendations in September 2021, nine of them designated as key recommendations, including changes to NERC reliability standards on generator winterisation. The inquiry also noted that this was the fourth time in ten years that cold-weather generation outages had threatened the bulk power system. It listed earlier peaks of unplanned outages of 29,700 MW in 2011, 19,500 MW in 2014 and 15,800 MW in 2018.

Cold weather again: Winter Storm Elliott, December 2022

Less than two years later, Winter Storm Elliott brought extreme cold to much of the eastern United States from 21 to 26 December 2022. The FERC-NERC final report, released in September 2023, found:

• 1,702 generating units experienced 3,565 outages, derates or failures to start, including 825 gas-fired units.
• At the worst point, coincident unplanned outages, derates and failures to start totalled 90,500 MW. That was 13% of the anticipated winter generating resources in the Eastern Interconnection, and the largest of the five cold-weather events studied since 2011.
• Including units that were already out of service, more than 127,000 MW was unavailable, about 18% of the US portion of anticipated resources in the Eastern Interconnection.
• Controlled firm load shed totalled 5,400 MW, the largest in the Eastern Interconnection's history.

These three categories made up 96% of all events. Natural gas fuel problems made up 20% of all outage megawatts and 83% of fuel-related outages. Gas production fell sharply again because of wellhead freeze-offs and processing problems. Power Engineering reported freeze-related production declines of 23% in the Marcellus shale and 54% in the Utica shale.

By fuel type, the 1,702 affected units were 47% natural gas, 21% wind, 12% coal, 3% solar, 0.4% nuclear and 17% other (oil, hydro and biomass). Those percentages count units, not megawatts. Wind farms and small gas turbines are numerous, while nuclear plants are few and large.

The report made recommendations including completing the cold-weather reliability standard revisions begun after Uri, an independent review of cold-related mechanical and electrical failures, and new rules for natural gas infrastructure reliability. FERC Chairman Willie Phillips said: "Someone must have authority to establish and enforce gas reliability standards." Gas production and pipelines are largely outside FERC's and NERC's reliability authority, which is why the report called on Congress and the states to act.

Comparing the cold-weather events

The figures come from the joint inquiries and are measured slightly differently from event to event, but the direction is clear: correlated forced outages during extreme cold have become a central reliability concern.

Fleet-wide trends

Cold snaps are the dramatic cases, but forced outage rates across the whole year also matter. NERC's 2026 State of Reliability report, released on 24 June 2026 and covering 2025, reported, according to Power Engineering's summary:

NERC put the historical norm for conventional generation at 7% to 8%. It said the 2025 increase was "not tied to a single event but was distributed across most months of the year." The extra unavailable energy from the increase was put at 39.8 TWh for coal and 19.1 TWh for combined-cycle units. NERC's previous report, covering 2023, had already described conventional and wind forced-outage metrics as historically high, with wind WEFOR at 18.9% in 2023.

Several explanations have been offered for rising outage rates, including the ageing of the coal fleet, plants cycling on and off more often to balance variable renewables, and reduced maintenance spending at plants expected to retire. These are explanations discussed in industry coverage rather than conclusions that can be pinned on any one factor. NERC's own framing is that reliability risk is "shifting away from single-contingency or localized disturbances" toward problems spread across the system.

Reading outage statistics carefully

Forced outage numbers are easy to quote and easy to misread. A few distinctions explain most of the apparent contradictions between reports.

Units versus megawatts. "Share of affected units" and "share of lost megawatts" are different measures. In Elliott, wind farms were 21% of affected units, but a wind farm and a 1,000 MW nuclear unit each count as one unit. Natural gas fuel problems were 20% of outage megawatts but 83% of fuel-related outages. Both figures are correct; they answer different questions.

Coincident versus total. Elliott's 90,500 MW was the peak of new unplanned outages happening at the same moment. Adding units that were already out of service for other reasons gives more than 127,000 MW unavailable. Headlines sometimes mix the two.

Preliminary versus final figures. The September 2021 preliminary findings on Uri gave 61,800 MW of lost generation. Later FERC and NERC documents, including the Elliott materials that compare the two storms, cite slightly different totals for 2021. Event studies are revised as data is cleaned, so it is good practice to cite which report a number comes from.

Fleet coverage. GADS covers most but not all capacity. Wikipedia's figure of 71% of installed US and Canadian capacity reflects the units in the database at the time it describes. Small units below the 20 MW reporting threshold are not required to report.

Excluded causes. Some metrics remove outages "outside management control," such as certain transmission or fuel events, so a plant's EFORd in one market can differ from its raw GADS EFOR. ISO New England's modified EFORd is one example.

Pooling method. As NERC's Appendix F sample shows, the same set of units can give a pooled EFORd of 7.92% or 8.02% depending on how individual results are combined.

Weighted versus unweighted. NERC's fleet figures, such as the 9.2% for 2025, are capacity-weighted (WEFOR). An unweighted average would give a small peaker the same influence as a large baseload unit.

How forced outage data is used

Resource adequacy planning

Grid planners must decide how much generating capacity is needed so that the chance of running short is acceptably low. Probabilistic studies model each unit as sometimes available and sometimes on forced outage, using its historical forced outage rate. They then calculate metrics such as loss of load probability (LOLP) and loss of load expectation (LOLE), and compare them with a reliability target chosen by the planner or regulator. Higher forced outage rates mean more reserve capacity is needed for the same reliability target.

A key weakness of traditional planning was that it treated unit failures as independent, like separate coin flips. Uri and Elliott showed that in extreme weather, failures are strongly correlated: the same cold freezes many plants, and the same gas shortage hits many gas units at once. Planners have been moving toward methods that account for weather-dependent outages.

Capacity markets

Grid operators with capacity markets pay generators for being available, not just for producing energy. To compare units fairly, they discount each unit's capacity by its forced outage history. In PJM, unforced capacity is UCAP = ICAP × (1 − EFORd). A PJM training example shows a 100 MW unit with a 5% EFORd yielding 95 MW of UCAP. NYISO and MISO use similar outage-adjusted formulas with different capacity bases. Some operators exclude certain outage causes considered outside the plant's control. ISO New England's modified EFORd, for example, excludes certain GADS cause codes considered "outside management control."

Some markets have moved to effective load-carrying capability (ELCC) methods for at least some resources. These estimate how much a resource actually contributes during the hours when the system is most stressed, which can capture correlated failures better than an average outage rate.

Benchmarking and maintenance

Plant owners use GADS data to compare their units with similar units elsewhere, to find recurring failure modes from cause codes, and to plan maintenance. A unit with frequent boiler tube failures, for example, may justify a tube replacement during the next planned outage. Starting reliability (the share of attempted starts that succeed) is an important metric for peaking units, which may be called on only a few times a year.

Regulation and standards

The Uri inquiry's key recommendations included changes to NERC reliability standards on cold-weather preparedness for generators, and the Elliott report called for those cold-weather standard revisions to be completed and their implementation monitored. Forced outage data is the evidence base for such rules.

Preventing and managing forced outages

The practices that reduce forced outages are well known, even if they cost money.

• Winterisation. Heat tracing, insulation and enclosures for instrument lines, valves and water systems, plus wind breaks and temporary heating, as described in the Uri inquiry. Many failures in 2021 and 2022 involved small components such as sensing lines rather than major equipment.
• Firm fuel. Firm pipeline transportation contracts, on-site fuel oil storage for dual-fuel gas turbines, and protected coal handling reduce exposure to fuel failures.
• Maintenance. Condition monitoring, predictive maintenance and well-planned overhauls catch problems before they force a shutdown.
• Operator training and procedures. Clear procedures and drills reduce human error, especially during unusual conditions.
• Gas–electric coordination. Because gas production, processing and pipelines depend on electricity, protecting gas facilities from load shedding and improving communication between gas and electric operators reduces the chance that blackouts deepen gas shortages.

When a forced outage happens, grid operators rely on operating reserves: spinning reserve (units already synchronised that can increase output quickly) and non-spinning reserve (units that can start quickly). If reserves run out, operators call on emergency measures such as imports, demand response, public appeals to conserve power, and, as a last resort, controlled load shedding. Rotating blackouts are designed to protect the grid as a whole from an uncontrolled collapse that could take far longer to recover from.

Common misconceptions

"A forced outage means the plant failed because of poor management." Not always. A forced outage can result from fuel supply failures, extreme weather beyond design conditions, or external events. GADS records cause codes partly so that these can be separated.

"Forced outage rate is the share of the year a unit was broken." That is closer to the forced outage factor. The forced outage rate measures broken time against the time the unit was either running or broken, which makes it a measure of reliability when needed. The two can differ a lot for units that run rarely.

"EFOR and EFORd are interchangeable." EFORd counts only outage time during periods when the unit would have been needed. For baseload units that run most of the time, the two are close. For peaking units, they can differ substantially.

"One fuel type caused the Texas blackouts." The FERC-NERC inquiry found failures across gas, coal, nuclear, wind and solar units, with freezing issues the largest cause and gas fuel supply problems the second. Natural gas-fired units accounted for the largest number of affected units, partly because gas is the largest part of the generation fleet in the region.

"Reserve margins protect against any failures." Reserve margins are sized assuming failures are mostly independent. When extreme weather makes failures correlated, more capacity can fail at once than a standard reserve margin covers, as happened in 2021 and 2022.

"Rotating blackouts are a failure of the grid operator." Controlled load shedding is an emergency tool to prevent a much worse uncontrolled blackout. The FERC presentation on Uri stated that ERCOT operators made the correct decision to shed firm load to recover frequency and prevent further outages.

Frequently asked questions

What is the difference between a forced outage and a planned outage? A planned outage is scheduled in advance, often months ahead, for maintenance, inspection or refuelling. A forced outage is unplanned and happens because equipment has failed or cannot safely keep running. Under GADS, forced outages are those requiring removal from service before the end of the next weekend.

What is a "good" forced outage rate? It depends on the technology and the metric. NERC described 7% to 8% as the historical norm for the weighted EFOR of conventional generation, with combined-cycle gas units much lower than coal units in recent data. Comparing units only makes sense within the same type and using the same metric.

Why do capacity markets use EFORd instead of EFOR? Because EFORd focuses on outages that occurred when the unit was needed, which is what matters for reliability. A peaking plant that breaks during the off-season is less of a problem than one that breaks during a heat wave.

Does a derating count as a forced outage? Not as a full outage, but forced deratings are counted in the "equivalent" metrics (EFOR, EFORd) by converting them into equivalent full-outage hours.

Is a failed start counted as a forced outage? Yes. In GADS, a startup failure (SF) is recorded when a unit cannot start or synchronise when called on, and its hours are added to forced outage hours (FOH) along with U1, U2 and U3 outages. That is why the f-factor in the EFORd formula uses both attempted starts and actual starts: the gap between them is a measure of starting reliability. A peaking unit that is called a few dozen times a year and fails to start several of those times will show a poor EFORd even if it runs well once it is going.

Who collects forced outage data in North America? NERC, through the Generating Availability Data System. Reporting has been mandatory for conventional units of 20 MW and larger since 1 January 2013. Grid operators such as PJM, NYISO, MISO and ISO New England use GADS data in their own calculations.

Were renewable plants affected in Uri and Elliott? Yes. In Uri, wind turbine blade icing was one of the most common freezing problems. In Elliott, wind made up 21% and solar 3% of the units that had outages, derates or failures to start. Wind and solar output also varies with weather even when equipment works, which is handled differently from forced outages in planning.

Why did natural gas supply fail during the storms? Wellheads and gathering lines froze, processing plants were disrupted, some gas facilities lost electric power, and pipeline capacity was stretched by heating demand. In February 2021, US dry gas production fell 28% in under two weeks.

Summary

A forced outage is an unplanned loss of a generating unit or a transmission or distribution element because of a failure or emergency. In North America, generator outages are recorded in NERC's Generating Availability Data System, which grew out of an Edison Electric Institute program in the 1960s, began in 1982 and became mandatory in 2012–2013. From GADS data come the forced outage rate, the equivalent forced outage rate that includes deratings, and the demand-based EFORd used in capacity markets, where PJM credits a unit with its installed capacity times (1 − EFORd). Those metrics underpin reserve planning, but Winter Storm Uri in 2021 and Winter Storm Elliott in 2022 showed that extreme weather can cause correlated forced outages far beyond what planners assumed: 61,800 MW and 90,500 MW of unplanned generation losses respectively, record load shedding and, in Texas, 246 deaths. Since then, NERC has reported rising fleet-wide forced outage rates, with conventional generation's weighted EFOR reaching 9.2% in 2025. Reducing forced outages depends on winterisation, firm fuel, good maintenance and closer coordination between the gas and electric systems.

Sources

• Forced outage (Wikipedia)
• Generating Availability Data System (Wikipedia)
• US NRC Glossary: Outage (forced)
• NERC GADS Data Reporting Instructions, Appendix F: Performance Indexes and Equations (2023)
• NERC 2023 Conventional GADS Training Module 05: Outage Event Reporting
• NYISO: Forced outage definitions
• FERC: February 2021 Cold Weather Grid Operations, Preliminary Findings and Recommendations (September 2021)
• FERC: The February 2021 Cold Weather Outages in Texas and the South Central United States
• NERC: Winter Storm Elliott report release (September 2023)
• Power Engineering: FERC-NERC final report, natural gas struggled the most during Winter Storm Elliott
• Power Engineering: NERC, conventional generation outages rose as reliability risks become more systemwide (2026 State of Reliability)
• Power Engineering: NERC, generation forced-outage rates continue to increase
• Texas DSHS: February 2021 Winter Storm-Related Deaths, final mortality surveillance report
• NPR: Texas officials put the final death toll from last year's winter storm at 246
• PJM: Capacity performance education (UCAP and EFORd)
• ISO New England: Updates to EFORd forced outage rate calculations