Availability was never a production metric

Availability answers one question: for what fraction of the period was the turbine able to operate. It is the metric service agreements are written on, it is what gets reported upward, and it is genuinely useful for the thing it measures. What it cannot tell you is whether an available machine was producing what it should have been producing, because a turbine running continuously at four percent below its potential is available for every one of those hours.

This is the gap the industry's reporting inherited from its contracts. Warranty structures made availability the number both parties argued about, so the reporting chain was built to produce it accurately, at site level, month by month. Nothing in that chain was designed to detect a machine that never stops and never quite performs, and a metric does not acquire a capability because the business would find it convenient.

The losses that live in the gap are the ones with no fault attached. Static yaw misalignment costs energy every hour the turbine runs and raises nothing. Pitch calibration drift across a blade set does the same. A degrading main bearing produces measurable inefficiency for months before it produces an alarm. In every one of those cases availability is unaffected, which is why an operator can hold a strong availability figure and a weak production figure simultaneously and see no contradiction in their own reporting.

The consequence is commercial rather than technical. Availability is a supplier-facing metric: it apportions responsibility for downtime between operator and service provider. Production is an owner-facing metric: it determines revenue. When the only number an organization tracks closely is the first, the incentive structure quietly optimizes for the wrong quantity, and the fastest way to see this is to ask who inside the business would notice a two percent production loss with no downtime attached.

There is a fair counter-argument, which is that availability is objective and production expectation is modeled, and a contract cannot be written on a model that a supplier does not agree with. That is true, and it is a reason to keep availability contractual rather than a reason to stop measuring production. The two coexist: availability settles the service agreement, and a per-turbine production expectation tells the owner what the asset is actually returning.

The limit of production analysis is that it estimates rather than counts. An availability figure is a tally of hours; a production deficit is an inference from a model, and the honest form of it carries a confidence and the evidence behind it rather than presenting itself as a measured quantity. That distinction matters most in exactly the conversation where the number will be used, which is why every finding OpenTurbine produces carries its own evidence.

Something to check this quarter: take the site with the best availability in the fleet and rank its turbines against each other under matched conditions. If the ranking is flat, the availability figure was telling the whole story. If it is not, the difference between the top and the bottom of that list is production that was never on a report.