Historical events
The project picked six real-world energy events documented in public reports, and asked: when the methodology is run against the model calibrated to each event, does its trajectory line up with what the public reports describe?
The six events are split into two groups. The first three are physical grid events. The second three are electricity-market events that test the phase-2 extension.
Each event below has a thirty-second audio rendering of the model’s trajectory. The audio is the model’s behavior, not a recording of the real event. The model was calibrated from public reports about each event; the model is not the grid.
Listen to the events
Texas, February 2021 — winter storm Uri. Natural gas froze, then electricity generation collapsed.
30-second audio • two-coupled-system simulation
The most extreme cold snap in the February 2021 Texas storm — a 30°C temperature drop. The model traces natural-gas pipeline throughput collapsing, then electricity generation collapsing behind it.
important caveat
This is the sound of a computer model, not a recording of the actual event. The model uses numbers taken from public reports about each event. It is not a measurement of any real power grid.
European Union, May 2022 — REPowerEU. The bloc's plan to wean off Russian gas, pushing capital into alternative energy.
30-second audio • two-coupled-system simulation
The European Commission's May 2022 plan to wean off Russian gas (REPowerEU). The model traces the regulatory push, then capital flowing into alternative-energy projects in response.
This event is growth, not collapse. The audio is inverted so the same listening shape applies — louder = more change underway.
important caveat
This is the sound of a computer model, not a recording of the actual event. The model uses numbers taken from public reports about each event. It is not a measurement of any real power grid.
Japan, March 2011 — Fukushima accident. Nuclear plants went offline; fossil-fuel imports filled the gap.
30-second audio • two-coupled-system simulation
Japan after the March 2011 Fukushima accident, with most of the nuclear fleet (85%) taken offline. The model traces nuclear capacity dropping, then fossil-fuel imports rising to fill the gap.
important caveat
This is the sound of a computer model, not a recording of the actual event. The model uses numbers taken from public reports about each event. It is not a measurement of any real power grid.
Germany, 2022 — hours of negative wholesale electricity prices. Renewables overshot demand, so the grid paid customers to consume.
30-second audio • publicly-reported monthly counts
Germany 2022 — hours of negative wholesale electricity prices, month by month. Renewables generated so much power that the grid paid customers to take it. Source: Germany's federal regulator (BNetzA) annual monitoring report.
This event is excess, not shortage. The audio is calibrated so louder = more hours of negative prices.
important caveat
This is the sound of a computer model, not a recording of the actual event. The model uses numbers taken from public reports about each event. It is not a measurement of any real power grid.
California, 2000-2001 — the electricity crisis. Wholesale prices spiked over twelve times above their pre-crisis level.
30-second audio • publicly-reported monthly prices
The California electricity crisis of 2000-2001 — three years of monthly wholesale prices, from the pre-crisis baseline through the peak (over $370/MWh, more than 12× normal) and back. Source: FERC's 2003 Final Report on the crisis.
Louder = prices further from normal. The spike that hits in mid-2000 is audible.
important caveat
This is the sound of a computer model, not a recording of the actual event. The model uses numbers taken from public reports about each event. It is not a measurement of any real power grid.
Texas, February 2021 — the price view of winter storm Uri. Monthly wholesale electricity prices spiked sixty-six times normal.
30-second audio • publicly-reported monthly prices
Texas February 2021, viewed through the price lens — four months of average wholesale electricity prices that bracket the Uri storm. February's price was roughly 66× the surrounding months after prices hit the $9000/MWh cap. Source: ERCOT post-event report + the FERC/NERC inquiry.
Louder = prices further from normal. The February spike dominates the 30 seconds.
important caveat
This is the sound of a computer model, not a recording of the actual event. The model uses numbers taken from public reports about each event. It is not a measurement of any real power grid.
Recovery results
Phase A (physical grid events)
| Event | Public report source | Result |
|---|---|---|
| Texas February 2021 (Uri winter storm) | ERCOT public filings, PUC of Texas orders, DOE post-event analyses | The model’s trajectory recovers the qualitative shape of the event. |
| EU May 2022 (REPowerEU plan) | European Commission dashboards, IEA tracking reports | Recovers. |
| Japan March 2011 (Fukushima → FIT scheme) | METI public records, Institute of Energy Economics Japan | Recovers. |
Phase 2 (electricity-market events)
| Event | What the model looked for | Result |
|---|---|---|
| California 2000-2001 | Extreme spot-price instability (peak ≥ 10× baseline) | Detected at 12.7×. |
| Texas February 2021 (market lens) | Extreme spot-price instability (peak ≥ 10× baseline) | Detected at 66.8×. |
| Germany 2022 | Sustained negative-spot-price hours (≥ 100 in the year) | Detected at 196 hours. |
The first two are crisis-spike events (prices shoot up because of scarcity). The third is an excess-electricity event (prices go negative because there is more supply than demand). The methodology picks up both shapes without confusing one for the other.
The interpretation of what each event means — whether a market
“failed”, whether a transition “succeeded” — is left to the longer
written close-out documents
(POST_MORTEM.md,
phase_2/POST_MORTEM_PHASE_2.md).
The measurements on this page only report whether the model’s
trajectory recovered the documented qualitative shape.