Evidence room, reveal stage. The left column is frozen at and uses only ST1 and ST2. The right column reports what later reports from ERCOT () and FERC and NERC staff () established. Nothing in the right column was knowable on November 5, 2020.
The 2020 extreme winter scenario beside the February 2021 outages
In short: ERCOT's harshest 2020 scenario assumed 9,879 MW of thermal forced outages and left 1,352 MW of capacity for operating reserves. ST1 SARA At its single worst point, ERCOT later counted 52,037 MW of unavailable capacity. That count is not a thermal forced outage figure: it is measured at nameplate, covers all fuels, and includes outages that existed before the storm. ST1 SARA ST4 outage causes
The direction is clear: losses far exceeded the modeled assumption. The exact size of a real time shortfall cannot be read off by subtracting one figure from the other. No probability is assigned, no single cause is assigned to every outage, and this page makes no claims about named people. ST3 2021 report ST4 outage causes
As of What the 2020 assessment could support
Recomputed from the printed rows: Total Resources 82,513 MW minus Peak Demand 57,699 MW gives 24,814 MW before uses. Each column then subtracts its own adjustments. All four recomputed totals match the printed ones. ST1 SARA
| Row | Col 1 | Col 2 | Col 3 | Col 4 |
|---|---|---|---|---|
| Seasonal load adjustment | 0 | 9,509 | 0 | 9,509 |
| Thermal maintenance outages | 4,074 | 4,074 | 4,074 | 4,074 |
| Thermal typical forced outages | 4,542 | 5,339 | 4,542 | 5,339 |
| Thermal 95th percentile forced outages | 0 | 0 | 0 | 4,540 |
| Low wind output adjustment | 0 | 0 | 5,279 | 0 |
| [d] Total uses | 8,616 | 18,922 | 13,895 | 23,462 |
| [e] Capacity for operating reserves, normal operating conditions | 16,198 | 5,892 | 10,919 | 1,352 |
The extreme thermal forced outage assumption
Column 4 counts 5,339 MW of typical plus 4,540 MW of 95th percentile thermal forced outages, a total of 9,879 MW. Adding thermal maintenance outages of 4,074 MW gives 13,953 MW of thermal capacity out. The rows are historical averages for weekday winter mornings over three prior winters, plus fuel limitation derates in north Texas. Wind is handled separately and column 4 has no low wind adjustment. ST1 SARA
The extreme load columns add 9,509 MW, for an extreme winter forecast of 67,208 MW, "Based on the 2011 winter." ST1 SARA
What the 2011 precedent showed
"But those reserves proved insufficient for the extraordinary amount of capacity that was lost during the event from trips, derates, and failures to start." ST2 2011 report
In 2011, 193 ERCOT units failed or were derated, a cumulative loss of 29,729 MW over two days, not all at one time. With scheduled outages, about one-third of the fleet was unavailable at the lowest point. ST2 2011 report
What a reader could support on this date
- The reserve rows are planning estimates under "Normal Operating Conditions", not guarantees. ST1 SARA
- The harshest published column already left only 1,352 MW. ST1 SARA
- A cold event in 2011 removed far more capacity than any single SARA outage row, so outages beyond the modeled range were a documented possibility, not a forecast. ST2 2011 report
- Neither source gives a probability for an extreme storm, and none is assigned. ST1 SARA ST2 2011 report
Later What later ERCOT and FERC/NERC reports established
ERCOT: peak total unavailable generation (ST4)
The highest unavailable capacity from February 14 to 19, 2021 was 52,037 MW, on February 16 at about 8:00 AM. Nameplate MW, all fuels It counts outages and derates of every fuel type at nameplate rating, and it includes 7,650 MW of Existing Outages that began before February 8. ERCOT calls the figures preliminary. ST4 outage causes
| Category | MW | Share |
|---|---|---|
| Existing Outages (before February 8; planned, forced, mothballed) | 7,650 | 15% |
| Weather Related | 27,567 | 53% |
| Fuel Limitations | 6,130 | 12% |
| Equipment Issues | 7,457 | 14% |
| Transmission Loss | 1,459 | 3% |
| Miscellaneous | 512 | 1% |
| Frequency Related | 1,262 | 2% |
| Total, all categories | 52,037 | 100% |
| Excluding Existing Outages (derived) | 44,387 | 85% |
Categories come from generator responses to ERCOT information requests. "Weather Related" means explicitly attributed to cold weather in those responses. The categories do not show a single cause for every outage, and they are not split by fuel type, so the thermal forced share is unknown. ERCOT warns nameplate values for wind and solar are "generally much higher than the actual amount of power that would have been available". ST4 outage causes
FERC/NERC: a two day average, a different measure (ST3)
ERCOT averaged 34,000 MW of generation unavailable from 7:00 a.m. February 15 to 1:00 p.m. February 17, 2021, a window of 54 hours. Expected capacity MW, average This is an average over the window based on expected capacity, not an hourly peak, and it includes generation already on planned or unplanned outages. ST3 2021 report
Demand: observed peak versus estimated unconstrained demand (ST3)
- Observed: actual peak load of 69,871 MW on February 14, 2021. ST3 2021 report
- Estimated, not observed: peak load without load management of 76,819 MW on February 15, 2021. Load was being shed, so this demand was never measured as served. ST3 2021 report
- ERCOT shed 20,000 MW of firm load at the worst point. ST3 2021 report
- ST3 restated the SARA extreme column as 1,352 MW and said: "The SARA is not intended to predict the likelihood of any of these scenario outcomes." ST3 2021 report
Modeled assumption beside later observations, in MW
Unit and scope mismatch The 2020 bars are SARA planning MW for thermal units only. The ST4 bar is nameplate MW for all fuels, including existing outages, at one hour. The ST3 bar is an expected capacity average over 54 hours, not an hourly peak. They share an axis so the scale is visible, not because they measure the same thing. ST1 SARA ST3 2021 report ST4 outage causes
Compare the column 4 assumption with one later measure
Choose a later measure to open its source evidence, the mismatch list and the illustrative arithmetic. Without scripts, all three are shown below.
ST4 peak hour, all categories
9,879 MW
column 4 thermal forced outages, SARA planning MW
52,037 MW
nameplate MW, all fuels, includes existing outages
Illustrative only 52,037 minus 9,879 is 42,158 MW. Against all column 4 thermal outages it is 38,084 MW. This is not a thermal forced outage overrun and not a simulated reserve shortfall: the observed figure is not a thermal forced outage measure.
Source evidence for this comparison
"The highest amount of unavailable capacity during the period of February 14-19, 2021 occurred on February 16 at ~8:00 AM and was 52,037 MW." ST4 outage causes
"All generator outage and derate values reflected in the graphs are based on generator nameplate capacity" ST4 outage causes
"Based on the 95th percentile historical average of forced outages for December through February weekdays, hours ending 7 am - 10 am, for the last three winter seasons" ST1 SARA
- Units: nameplate versus SARA planning MW.
- Fuels: all, including wind and solar, versus thermal only.
- Status: includes existing planned, forced and mothballed outages versus forced only.
- Time: one observed hour versus a historical average of winter morning hours.
ST4 peak hour, without existing outages
9,879 MW
column 4 thermal forced outages, SARA planning MW
44,387 MW
nameplate MW, all fuels, outages starting on or after February 8, any cause
Illustrative only 44,387 minus 9,879 is 34,508 MW; against all column 4 thermal outages it is 30,434 MW. Removing existing outages narrows the scope gap but does not close it: the figure is still nameplate, still all fuels, and the categories are causes, not forced outage status.
Source evidence for this comparison
"Existing Outages: Generator outages or derates that started before the issuance of the Operating Condition Notice on February 8, 2021; includes ongoing planned and forced outages as well as seasonally mothballed units." ST4 outage causes
"Because wind and solar output is typically much lower than the specified nameplate capacity, the outage and derate MW values used for those units to develop this report are generally much higher than the actual amount of power that would have been available in the absence of the outage or derate." ST4 outage causes
ST3 two day average, shown separately
13,953 MW
column 4 thermal maintenance plus forced outages, SARA planning MW
34,000 MW
expected capacity, average over 54 hours, not an hourly peak
Illustrative only 34,000 minus 13,953 is 20,047 MW; against thermal forced outages alone it is 24,121 MW. The average includes planned and unplanned outages of all generation, so it is not a thermal forced outage figure either, and it does not describe any single hour.
Source evidence for this comparison
"averaged 34,000 MW of generation unavailable (based on expected capacity) for over two consecutive days, from 7:00 a.m. February 15 to 1:00 p.m. February 17" ST3 2021 report
"including generation already on planned or unplanned outages" ST3 2021 report
Why the subtraction is illustrative, not a simulated shortfall
| Dimension | ST1 column 4 model | ST4 peak hour | ST3 average |
|---|---|---|---|
| Rating basis | SARA planning MW; basis of outage rows not stated in excerpt | Nameplate | Expected capacity |
| Fuels | Thermal only | All, including wind and solar | All generation |
| Outage status | Forced, with maintenance as a separate row | Outages and derates of any cause, including existing outages | Includes planned and unplanned outages |
| Time | Historical weekday winter mornings | One hour, February 16 about 8:00 AM | Average over 54 hours |
- A real time reserve calculation also needs that hour's actual demand, imports, wind and solar output and operating reserve rules. The SARA table is a seasonal snapshot and these excerpts do not supply those inputs. ST1 SARA ST3 2021 report
- ST4 says its nameplate values overstate wind and solar losses, so the 52,037 MW figure is larger than the power actually lost. ST4 outage causes
- Demand was being curtailed, so neither the observed peak nor the estimate yields an exact supply gap that would have avoided blackouts. ST3 2021 report
Demand: modeled, observed and estimated
The observed peak exceeded the 2020 extreme forecast by 2,663 MW. The later estimate of demand without load management exceeds it by 9,611 MW, but that is an estimate of demand that was never served, on a different day. Neither difference, alone or added to the outage figures, gives an exact supply gap that would have avoided blackouts. ST1 SARA ST3 2021 report
Unknowns and limitations
- How much of the ST4 total was thermal forced outage in SARA terms. The categories are causes, not fuel types or outage status. ST4 outage causes
- The expected capacity equivalent of the ST4 peak hour. ST4 outage causes
- The rating basis of the SARA outage rows, which the ST1 excerpt does not state. ST1 SARA
- ST4 is preliminary and subject to change; it revised its peak total from an earlier figure. ST4 outage causes
- Whether any outage had one cause; the excerpts do not allocate overlapping causes. ST4 outage causes
- The exact shortfall that would have avoided load shed. ST3 2021 report
Sources
- ST1 SARA: ERCOT, Final Seasonal Assessment of Resource Adequacy, Winter 2020/2021, November 5, 2020. As of
- ST2 2011 report: FERC and NERC staff, report on the February 1 to 5, 2011 Southwest cold weather event, August 2011. As of
- ST3 2021 report: FERC, NERC and Regional Entity staff, The February 2021 Cold Weather Outages in Texas and the South Central United States, November 2021. Later
- ST4 outage causes: ERCOT, update on causes of generator outages and derates, April 27, 2021. Later