See the second page of ERCOT’s latest reliability assessment.

Under the top table describing the reserve margins, they have a table labelled “Range of Potential Risks.” Something between the 4th and 5th columns is what happened this week. Add 2,000 MW of demand above ERCOT’s adjustment here (they assumed a record of 67,200 MW) and 10,000 MW from planned thermal outages, likely for maintenance, going into the cold front.

Take out another 1,600-1,900 MW from lower wind production than in the baseline, and reserves are below 2,000 MW. That’s where Texas was Sunday night.

At that point, the reserve margins are running so thin that it only takes a few plants failing to trip the grid as appears to have happened. Large base load thermal plants, once tripped offline, either due to cold or due to the grid going unstable and the plant tripping to protect circuitry, take hours to bring back online, longer in the cold.

ERCOT compounded the error by cutting power to the Permian Basin substations that power the national gas pipelines that feed the power plants, throughput went from 16 billion cubic feet to four, forcing more plants offline.

While there were mistakes made (too many planned outages, running too close to the edge Sunday night), ERCOT’s own planning document suggests this event was almost inevitable.

Demand was forecast to go above 74,000 MW Monday morning and Monday night. Wind production went almost to zero Monday night (650 MW) right when demand would have peaked if not for the blackouts. That’s a 10 GW deficit from where we were Sunday night. Even if Texas had all its thermal plants running like in the summer, the grid still would have collapsed without planned rolling blackouts.

The only way blackouts would have been avoided is if Texas had built 10 GW of gas (or not prematurely retired coal) in the past 5 years instead of building 20 GW of wind and solar.