Skip to main content
Scenario Library
Energy & UtilitiesFebruary 13–17, 2021

Winter Storm Uri: The Grid Failure the Signals Foretold

In February 2021, a polar outbreak pushed the Texas grid within minutes of total collapse. More than 4.5 million customers lost power, wholesale prices sat at the $9,000/MWh cap for days, and hundreds of people died. The atmospheric warning signs appeared weeks earlier — but the failure lived in the seams between weather, gas logistics, and electricity markets.

AtmosphericEnergyLogisticsEconomicBehavioral

This brief is a retrospective, illustrative analysis of publicly documented events. WeaveCast was not in operation during these events, and no customer outcome or endorsement is implied. Figures are drawn from public reporting and official post-event reviews.

What Happened

Winter Storm Uri brought sustained sub-freezing temperatures across the entire state of Texas. Electricity demand surged toward record winter levels at the same time that generation of every type — gas, coal, nuclear, and wind — was forced offline by the cold.

  • More than 4.5 million Texas customers lost power, many for multiple days in sub-freezing conditions.
  • Roughly half of ERCOT’s generating capacity was unavailable at the worst point; grid operators later said the system was minutes from an uncontrolled, weeks-long blackout.
  • Wholesale power cleared at the $9,000/MWh price cap for about four days, bankrupting retailers and co-ops.
  • The State of Texas officially attributed 246 deaths to the storm; independent estimates ran higher.
  • Natural gas production in Texas fell by nearly half as wellheads, gathering lines, and processing plants froze — and some gas infrastructure lost the electricity it needed to operate, creating a feedback loop.

The Signals Were There

Each of these indicators was publicly observable before the event. The lead time shows how far in advance the signal existed.

AtmosphericLead time: 4–6 weeks

A sudden stratospheric warming event in early January 2021 disrupted the polar vortex — a well-known precursor of severe mid-latitude cold outbreaks. Forecasters flagged the risk of an extreme cold intrusion into the southern U.S. weeks before the storm.

EnergyLead time: 10 years

The 2011 Texas cold event produced a joint FERC/NERC report recommending winterization of generators and gas infrastructure. A decade later, most recommendations remained voluntary and unimplemented — a standing structural vulnerability.

LogisticsLead time: Years (structural)

Gas-electric interdependence was documented and growing: compressors and processing plants increasingly ran on grid power, meaning rolling blackouts would cut the fuel supply of the very plants needed to end them.

EconomicLead time: Months

ERCOT’s seasonal assessment modeled an extreme winter peak well below what materialized, and the market’s energy-only design offered weak incentives for winter firming — visible in public planning documents.

BehavioralLead time: 1–2 weeks

Simultaneous residential heating demand across all of Texas — a population behavior pattern in extreme cold — was predictable from the temperature forecast alone, yet demand forecasts undershot it.

Why Siloed Analysis Missed It

Every individual signal was known to somebody. Meteorologists tracked the polar vortex. Gas analysts knew the wellheads were unwinterized. Reliability engineers had the 2011 report on their shelves. What no one operationally maintained was the joined-up picture: an extreme cold forecast, multiplied by unwinterized gas supply, multiplied by gas-electric interdependence, multiplied by simultaneous statewide heating demand. Each silo rated its own risk as manageable. The compounded risk was catastrophic.

Through a Cross-Domain Lens

How an analysis system built for domain fusion — the approach WeaveCast operationalizes — would frame the same problem.

  1. 1

    Treat the stratospheric warming event as a trigger to re-price downstream risk: when a polar-outbreak precursor fires, automatically re-evaluate exposed infrastructure in its probable path.

  2. 2

    Fuse the atmospheric forecast with infrastructure metadata — which generators and gas facilities in the cold footprint are winterized, and which failed in 2011.

  3. 3

    Trace the dependency graph: grid load forecasts that account for gas system electrical dependence would have shown the feedback loop before it began.

  4. 4

    Surface the compound probability explicitly: a transparent reasoning trace showing weather × fuel supply × demand interaction is exactly the picture that siloed dashboards could not produce.

The takeaway

Uri was not a forecasting failure — the cold was forecast. It was a synthesis failure. The risk lived in the interaction of four domains, and no institution owned the interaction.

Drawn from the FERC/NERC joint inquiry into the February 2021 cold weather event, ERCOT public statements, the Texas Department of State Health Services mortality count, and contemporaneous public reporting.