Week 05 — Below Cut-In

Power curves, cut-in speed, and what a fleet does when the wind is merely weak.

Week 05 · depth · 24 August 2026

In weak wind a fleet is not producing less; parts of it are absent from supply.

Property

depth

Stance

Advances the claim

How

The event floor sits at 0.9% of capacity; the threshold sits at 12%.

Studio activity

Derive why the course's threshold is a threshold rather than a proportion.

There is a wind speed below which a turbine produces nothing at all, and it is not zero wind. Below that point the machine is not underperforming — it is not participating. The course carries no turbine specification and quotes no cut-in speed, because it has no source for one; what matters here is the shape of the response, not its calibration.

How this advances the thesis#

It explains the depth of the trough rather than its length. A drought is not a linear dimming of supply proportional to the weather; it is a threshold effect, and thresholds are what turn a mild meteorological anomaly into an operational event. This is the week the shape of the event stops being arbitrary.

In the session#

We work in the course’s own unit — percentage of combined installed wind and solar capacity — because that is the unit the definition is written in, and the conversion to wind speed would require a power curve this course does not have.

Take two numbers from the trace and the definition. Inside the reference event the combined output floor is 0.9% of capacity, a property of the synthetic trace. The course’s threshold is 12%, a boundary this course chose. The threshold sits roughly thirteen times above the floor.

Now derive the consequence. If supply fell proportionally with the weather, an event would approach the threshold gradually and spend most of its hours near it — and a threshold would be a bad instrument, because small errors in where you put it would change the answer a lot. Instead the trace collapses to under one per cent and stays there. The threshold is nowhere near the action.

That is why week 2 found what it found: moving the threshold from 12% to 10% or 14% barely changed the event, while moving the span destroyed it. A fleet whose machines stop rather than dim produces a deep, flat floor, and a boundary drawn anywhere in the empty space above that floor gives the same answer. Bring the calculation and the sentence that follows from it: what the response curve has to look like for a threshold definition to be robust.

Its relationship to the canonical event#

This is the week that justifies the form of the event. The definition names an output level rather than a weather condition, and it treats that level as a cliff rather than a slope. Both of those choices are only defensible if the underlying fleet behaves as a population of machines that switch off, which is what this week establishes. Week 2 chose where to put the line; week 5 is why a line is the right instrument.

What next week needs#

If wind can be absent, the natural hope is that solar covers it. Whether that hope survives winter is week 6.