Texas grid freeze (2021) investigation

01 · The briefing

Begin with the question.

In mid-February 2021, a severe winter storm moved across Texas and the wider South Central United States. Electricity demand was at a record level just as many power plants went offline. The University of Texas Energy Institute reports that offline capacity rose by 24,600 MW in the 24 hours from noon on 14 February to noon on 15 February. Operators of the Electric Reliability Council of Texas (ERCOT) ordered firm load shed, meaning controlled outages, to keep the grid from a wider collapse. More than 4.5 million people in Texas lost power, and a joint review by the Federal Energy Regulatory Commission (FERC) and the North American Electric Reliability Corporation (NERC) counted at least 210 deaths. Explanations offered at the time pointed to different parts of the system, including wind and gas supply, the preparation of the plants and the way the grid was operated. The official reviews use different units and dates, and some of their figures do not reconcile, so part of the task is to check whether the headline numbers hold together. This investigation separates the failures the official reviews measured from the explanations they tested, and asks which failures the record supports. The question is not who is to blame, but what the measured record shows.

A large tree covered in snow and ice on a snowy lawn, with a small figure in red near the right edge. — Where it happenedFebruary 2021 Snow and ice on a tree in Austin during the February 2021 winter storm. It shows the weather, not the grid. Photo: David Kitto · CC0 · Wikimedia Commons.

The timeline

2021-02-11
The period the state later uses for storm-related deaths begins on 11 February 2021 and runs through 4 June 2021.
2021-02-15
ERCOT orders its first 1,000 MW of load shed at 1:20 a.m. and then another 1,000 MW, with operators having about nine minutes to prevent roughly 17,000 MW of units tripping on underfrequency relays.
2021-02-15
ERCOT system frequency reaches a low of 59.302 Hz at about 1:55 a.m., according to the UT Energy Institute timeline; the FERC and NERC narrative gives 59.304 Hz at 1:54 a.m.
2021-02-15
At 1:55 a.m. ERCOT orders a further 3,500 MW of firm load shed; the FERC and NERC narrative puts the total shed by then at 8,500 MW, including orders this timeline does not list.
2021-07
The University of Texas committee's timeline report is published, listing a 24,600 MW loss of offline capacity in 24 hours.
2021-10-27
The Texas Department of State Health Services reports 246 confirmed winter storm-related deaths as of this date.
2021-11
FERC and NERC publish their joint report, counting 23,418 MW of manual firm load shed and at least 210 deaths.

A researched reconstruction for learning. These reasoning steps simplify a complex record; no score can capture the full history.

02 · The evidence

What does the record show?

Read each exhibit before settling on an explanation. The source links below every exhibit lead to the record behind it.

Exhibit 01

FERC and NERC: the scale, and the 81 percent above design temperature

The joint FERC and NERC report, published in November 2021, states that more than 4.5 million people in Texas lost power and that at least 210 people died, mostly in connection with the outages. It reports a combined 23,418 MW of manual firm load shed, which it describes as the largest controlled firm load shed event in U.S. history. It also finds that 81 percent of freeze-related generating unit outages occurred at temperatures above the unit's stated ambient design temperature, and that some units still lacked winterization plans. It establishes the scale and the design gap. It does not assign each outage to a single cause.

Source record s1web.archive.org

Exhibit 02

UT Energy Institute: the 24,600 MW total and its listed parts

The University of Texas committee's July 2021 timeline report states that offline capacity rose by a total of 24,600 MW in the 24 hours from noon on 14 February to noon on 15 February 2021, citing increases of 13,000 MW for natural gas and 3,700 MW for wind. It lists reasons for plant outages, including weather-related outages of about 30,000 MW, fuel limitations of 6,700 MW and equipment issues of 5,600 MW. The increments it lists add to about 21,600 MW, not 24,600, a gap this case flags. It establishes the figures as stated. It does not show how much of each category was cause rather than effect.

Source record s2energy.utexas.edu

Exhibit 03

FERC and NERC narrative: frequency and the nine-minute window

The FERC and NERC narrative of the 15 February operations describes frequency falling as generating units failed. ERCOT ordered its first 1,000 MW of load shed at 1:20 a.m., then an additional 1,000 MW, and operators had only nine minutes to prevent about 17,000 MW of generating units tripping on underfrequency relays. The narrative says frequency then stayed below the generator trip level for more than four minutes. It establishes the sequence as the review reconstructed it. It does not separate the roles of cold weather, fuel supply and equipment failure in each unit's outage.

Source record s1web.archive.org

Exhibit 04

Texas health department: 246 confirmed deaths as of 27 October 2021

The Texas Department of State Health Services mortality report, dated 31 December 2021 with data current to 27 October 2021, counts 246 confirmed winter storm-related deaths across 77 Texas counties for 11 February to 4 June 2021. It identifies storm-related deaths from mortality surveillance forms, death certificates and verified informal reports. It establishes a state count to a stated date and by a stated method. The figure is higher than the FERC and NERC count of at least 210, which was produced at a different time and by different methods that the sources do not reconcile in detail. It counts deaths linked to the storm, not which grid failure caused each death.

Source record s3web.archive.org

03 · Your reasoning

Follow the strongest explanation.

Choose an interpretation, then compare it with the evidence. You can revise your choice. The goal is understanding, not a perfect score.

Reasoning step 01 of 04

Which factor does the FERC and NERC report most directly link to the size of the outages?Plants were not built or prepared for the temperatures they faced, since most freeze-related outages occurred above their stated design temperature.The report finds that 81 percent of freeze-related generating unit outages occurred above the unit's stated ambient design temperature, which points to preparation as a central factor.Extreme cold alone made the outages unavoidable, even for well-prepared plants.The finding that most freeze-related outages occurred above design temperature suggests that preparation mattered, so cold alone is not the best reading.Demand alone was the cause, since heating load exceeded what the grid could supply.Demand was at a record level, but the UT report's 24,600 MW loss of offline capacity points to supply losses as central, so demand alone is incomplete.A cyberattack disabled control systems at several major plants.The reviews cited here identify weather, fuel, equipment and frequency problems and do not report a cyberattack.

Evidence-based explanation

The best answer is a. The FERC and NERC report says 81 percent of freeze-related generating unit outages occurred at temperatures above the unit's stated ambient design temperature, and that some ERCOT units still lacked winterization plans. That points to preparation as a central factor. The report does not reduce the event to one cause, and the 81 percent figure describes freeze-related outages, not every outage in the event.

Weigh it against

Exhibit 01FERC and NERC: the scale, and the 81 percent above design temperature Exhibit 02UT Energy Institute: the 24,600 MW total and its listed parts

Reasoning step 02 of 04

What do the operator actions in the frequency record show?Operators could have avoided all load shed, and the shedding was ordered to save money.The UT and FERC accounts describe load shed as a step to avoid a catastrophic failure, so the cost explanation has no support in the record.Load shedding began only after the outages had ended and frequency had recovered.The sequence shows frequency falling and load shed being ordered as it fell, so this reverses the order of events.Load shed was a deliberate step to stop a wider collapse, taken while about 17,000 MW of units were at risk of tripping and operators had about nine minutes.The FERC and NERC narrative and the UT timeline both describe the shedding as a response to generation loss, which fits this reading best.Frequency held steady, and the outages came only from transmission lines failing.The record shows frequency falling to 59.302 Hz and generating units tripping, so transmission failure alone does not fit.

Evidence-based explanation

The best answer is c. The FERC and NERC narrative says ERCOT ordered 1,000 MW of load shed at 1:20 a.m., then another 1,000 MW, while frequency kept falling and operators had only nine minutes to prevent about 17,000 MW tripping. The UT report gives a low of 59.302 Hz. The shedding was a response to generation loss, not its cause, and the review's timings are its own reconstruction.

Weigh it against

Exhibit 03FERC and NERC narrative: frequency and the nine-minute window Exhibit 02UT Energy Institute: the 24,600 MW total and its listed parts

Reasoning step 03 of 04

How should the 210 and 246 death counts be used?As the same count, since both reports used the state's data for the same period.The FERC and NERC figure is 'at least 210' and the state figure is 246 as of 27 October 2021, with different methods and dates, so they are not the same count.Stated together with their dates and methods: at least 210 (FERC and NERC) and 246 confirmed as of 27 October 2021 (state health department), with both open to revision.This keeps each figure tied to its date, scope and method, which is what the two sources support.Set both aside, since death counts cannot be linked to grid failures at all.Counts cannot show which failure caused each death, but they remain the main measure of human cost, so ignoring them loses information.Use 246 as the final toll for February 2021 alone.The state count covers 11 February to 4 June 2021 and is a count to a stated date, so calling it final or limiting it to February is wrong.

Evidence-based explanation

The best answer is b. FERC and NERC report at least 210 deaths, while the Texas health department counts 246 confirmed storm-related deaths as of 27 October 2021, for 11 February to 4 June 2021, using death certificates together with mortality surveillance forms and verified reports. The two figures differ in date and method, so stating both with their dates is more accurate than picking one. Counts can be revised as records are reviewed.

Weigh it against

Exhibit 01FERC and NERC: the scale, and the 81 percent above design temperature Exhibit 04Texas health department: 246 confirmed deaths as of 27 October 2021

Reasoning step 04 of 04

How should the 24,600 MW total in the UT report be read?As confirmed, since the listed increments add up to the stated total.The listed increments add to about 21,600 MW as printed, so the stated total does not follow from them.As a figure the grid operator's public data would confirm, which the report cites directly.The cited text gives no operator data for this total, so the claim cannot be checked against it.As a figure to reject, because the storm had no effect on generation.The report lists weather-related outages of about 30,000 MW, so this contradicts the same report.As the report's stated total, reported with its caveat: its listed increments differ from it by about 3,000 MW.This keeps the report's own figure while flagging the unexplained gap, which is the most careful reading of the source.

Evidence-based explanation

The best answer is d. The UT report states a total of 24,600 MW in 24 hours, but the increments it lists (3,700 MW wind, 13,000 MW natural gas, 3,000 MW coal, 1,300 MW nuclear and 600 MW solar) add to 21,600 MW. The gap may reflect rounding or categories not shown, but the text reviewed does not reconcile it. Reporting the total with that discrepancy is the careful choice.

Weigh it against

Exhibit 02UT Energy Institute: the 24,600 MW total and its listed parts

04 · The findings

The chain behind the outcome.

The official reviews agree on the main pattern: a severe winter storm caused generating units to fail faster than supply could be replaced, and grid frequency fell to a level that forced load shed. The UT Energy Institute timeline reports a 24-hour loss of 24,600 MW of offline capacity, though the increments it lists add to 21,600 MW, and it attributes the outages to weather-related, fuel, equipment and frequency problems. The FERC and NERC report finds that 81 percent of freeze-related generating unit outages occurred above the units' stated design temperature, which makes preparation a central factor, while fuel supply and equipment problems remain significant. Operators shed 23,418 MW in total and had about nine minutes to prevent roughly 17,000 MW from tripping. Deaths were reported as at least 210 by FERC and NERC, and as 246 confirmed by the state health department as of 27 October 2021. The two counts were produced at different times and by different methods. The evidence supports a combined explanation: generation that was not prepared for the cold, along with fuel and equipment failures during record demand, and a grid that had little time to respond. No single cause accounts for the event.

Contributing causes and conditions

Severe cold during record electricity demand, which raised the load the grid had to serve while many plants were already failing.
Freeze-related outages of generating units, most above their stated design temperature, with some units lacking winterization plans.
Fuel limitations and equipment problems, which the UT report lists as 6,700 MW and 5,600 MW of outages.
A fall in grid frequency as generation dropped, which forced load shed of 23,418 MW in total.
A window of about nine minutes before roughly 17,000 MW of units could trip on underfrequency relays, which could have caused a wider blackout.

OutcomeTexas grid freeze (2021)

Take the method with you

Check whether a headline total matches its own components before repeating it, because a gap of several thousand units changes what the total means.
When a finding is stated as a share, such as 81 percent of freeze-related outages, read it as a claim about that subset, not about every failure.
Keep counts together with their dates, definitions and methods, because a death toll without them cannot be compared with another.

You can read findings at any time. Exploring every reasoning step marks this case as complete in your notebook.

Open another investigation

Why so many plants stopped at once

Begin with the question.

The timeline

What does the record show?

FERC and NERC: the scale, and the 81 percent above design temperature

UT Energy Institute: the 24,600 MW total and its listed parts

FERC and NERC narrative: frequency and the nine-minute window

Texas health department: 246 confirmed deaths as of 27 October 2021

Follow the strongest explanation.

Evidence-based explanation

Evidence-based explanation

Evidence-based explanation

Evidence-based explanation

The chain behind the outcome.

Case report

Contributing causes and conditions

Take the method with you

Sources & further reading

Field notes that use this case

Why so many plants stopped at once

Begin with the question.

The timeline

What does the record show?

FERC and NERC: the scale, and the 81 percent above design temperature

UT Energy Institute: the 24,600 MW total and its listed parts

FERC and NERC narrative: frequency and the nine-minute window

Texas health department: 246 confirmed deaths as of 27 October 2021

Follow the strongest explanation.

Evidence-based explanation

Evidence-based explanation

Evidence-based explanation

Evidence-based explanation

The chain behind the outcome.

Case report

Contributing causes and conditions

Take the method with you

Sources & further reading

Field notes that use this case

Related investigations