Availability is a definition before it is a measurement
Technical availability, time-based availability and production-based availability answer different questions. Time-based availability divides by calendar time; production-based availability divides by energy; some contracts exclude grid outages, some exclude force majeure, some count curtailed hours as available and some do not.
IEC 61400-26 introduced a framework for this precisely because the industry could not compare figures meaningfully. Before comparing two availability numbers, compare the definitions behind them.
- Which time base: calendar time, contractual time or production time
- Which states are excluded: grid, curtailment, environmental, force majeure
- Whether the figure is time-based or production-based
The records behind the number
Whatever the definition, the calculation needs trustworthy records: when the turbine stopped, why, when work started, when it finished, and whether it returned to service fully. In many organisations those facts are reconstructed after the fact from operator logs and maintenance reports, which is where the drift enters.
If downtime reason and status transitions are captured as events during the month — by the people doing the work — the availability review becomes an interpretation exercise rather than an archaeology exercise.
Maintenance KPIs worth reviewing together
No single metric survives on its own. A falling mean time to repair with a rising repeat-failure rate is not improvement; it is the same defect being fixed faster and less permanently.
- Repeat failures per alarm code, per turbine model
- Work order age and closure distribution
- Planned versus corrective workload split
- Parts consumption and waiting-for-parts time
- Session phase durations: diagnosis versus repair versus documentation
Diagnosis time is usually the hidden cost
In corrective maintenance, repair time is often reasonably predictable and diagnosis time is not. Two interventions on the same defect can differ by hours depending on whether the technician had the alarm context, the historical record and a diagnostic path.
Measuring phase durations — travel, preparation, diagnosis, repair, testing, documentation — makes that variability visible, and it is usually the strongest argument for better knowledge capture rather than for faster wrench work.
Report the evidence, not just the score
An availability percentage with no traceable records behind it invites argument rather than action. A review that shows the specific turbines, alarms and interventions that produced the figure supports decisions: which defect to campaign on, which model needs an engineering review, which spares to stock.