Wind Availability: Why 97% Uptime Can Still Lose the Windiest Days
Wind availability is usually reported as a share of time: how many hours the turbines were ready to run. But wind farms are paid for energy, and energy is concentrated in the windiest hours. A turbine that is down for a day of strong wind loses far more than one down for a calm day, yet both count the same in a time-based figure. Energy-based availability weights each hour of downtime by the energy it cost. The two measures can tell very different stories about the same month, and NERC GADS Wind uses both views: hours for monthly performance, and potential lost energy for outage events. Owners who track only one are missing half the picture.
Ask a wind farm operator how the site performed and the first answer is often an availability figure: 97%, 98%, comfortably above the guarantee in the service contract. It sounds definitive. It is also silent on the question an owner cares about most — how much energy did that 2% or 3% of downtime cost? The answer depends almost entirely on when the downtime happened. And a time-based availability figure, by design, does not care when.
An hour down at 4 m/s and an hour down at 14 m/s count the same. They do not cost the same.
Who this matters to
Wind asset managers reconciling budget against actual, owners negotiating or enforcing service agreements with turbine manufacturers, and compliance teams reporting NERC GADS Wind, where both hours and lost energy appear.
Two ways to measure the same downtime
- •Time-based availability. The share of time the turbines were available to generate. Turbine-hours available divided by total turbine-hours in the period, usually with agreed exclusions. It is simple, easy to audit, and the basis of most turbine service agreements.
- •Energy-based (production-based) availability. The share of potential energy the turbines actually captured. Energy produced divided by energy produced plus energy lost to unavailability, where lost energy is estimated from the wind each turbine would have seen and its power curve. It answers the commercial question directly, at the cost of needing an estimate of what was lost.
The international standard for wind availability, IEC 61400-26-1:2019, defines an information model from which both time-based and production-based indicators can be derived. It replaced earlier technical specifications that treated the two separately. Neither measure is wrong. They answer different questions.
Where the two numbers drift apart
- •When the downtime happens. Wind power rises steeply with wind speed, so the energy in a month is concentrated in its windiest hours. Downtime during a windy spell costs many times more energy than the same hours of downtime on a calm day. Time-based availability counts them equally.
- •Derated is still available. A turbine running at reduced output because of a temperature limit, a component warning or icing is usually counted as available in a time-based figure. It is still losing energy, and only an energy-based measure sees it.
- •Contract exclusions. Service agreements commonly exclude some categories of downtime from the guaranteed figure — certain grid events, owner-requested stops, force majeure, sometimes waiting time. A contractual availability can therefore sit well above what the turbines actually delivered.
- •The reference problem. Lost energy has to be estimated, and the estimate is only as good as its wind reference. A stopped turbine's own nacelle anemometer reads differently from a running one, so lost energy is often better estimated from neighbouring turbines or a met mast. A careless reference can flatter or exaggerate the loss.
Same uptime, different months: an illustrative case
Numbers illustrative. A 150 MW wind farm with 50 turbines has 36,000 turbine-hours in a 30-day month. In both scenarios below, turbines are unavailable for 1,080 turbine-hours, so time-based availability is 97.0%. The potential energy for the month is 45,000 MWh. The only difference is when the downtime fell.
| Scenario | Time-based availability | Energy-based availability |
|---|---|---|
| Downtime during calm spells (average 0.4 MW lost per turbine-hour) | 97.0% | 99.0% — 432 MWh lost |
| Downtime during a windy week (average 2.7 MW lost per turbine-hour) | 97.0% | 93.5% — 2,916 MWh lost |
The same time-based figure hides almost seven times as much lost energy. At an illustrative $40 per MWh, the calm-weather month cost about $17,000; the windy-week month cost about $117,000. Both would pass a 97% time-based guarantee.
How NERC GADS Wind sees it
- •Performance: hours. GADS Wind monthly performance data is reported quarterly at sub-group level, and it is primarily hours-based. Turbine-hours are the number of turbines in the sub-group multiplied by the hours in the month — ten turbines in a 720-hour month is 7,200 turbine-hours — divided among the states the turbines were in.
- •Events: energy. Under the 2024 reporting instructions, a plant-level event begins when at least 20 MW of the plant's total installed capacity is lost to a forced outage, and each event carries a Potential Production MWh Loss: the capacity lost in each interval multiplied by the interval's duration, at the finest granularity available and no coarser than five minutes.
- •Unavailable is not the same as underperforming. NERC's own guidance also draws the line between unavailability and underperformance. Lack of wind and efficiency losses do not make an event: icing that halves every turbine's output is not an event, but icing that shuts 20 MW of turbines down is. Consistent reporting depends on classifying each of those situations correctly, interval by interval.
Where Ellume Bridge fits
Ellume Bridge automates NERC GADS reporting. It classifies each event into the correct cause code using the plant's own data, weather and operating context — telling a forced outage apart from a resource-unavailable period — computes hours and energy figures by arithmetic rather than AI, and routes every event through human review before filing, with a full record of who decided what and why.
A check you can run this quarter
Take the last twelve months and calculate both availabilities for each turbine. Rank the turbines by the gap between the two: the largest gaps show where downtime is landing in the windiest hours, or where derating is hiding inside “available” time. Then check when planned maintenance is scheduled, and read the exclusions in your service agreement against what actually happened.
Frequently asked questions
- What is energy-based availability?
- The share of potential energy a wind farm actually produced: energy produced divided by energy produced plus energy lost to unavailability. It weights each hour of downtime by the energy it cost.
- Which measure should an owner track?
- Both. Time-based availability is simple, auditable and usually what contracts guarantee. Energy-based availability shows what downtime actually cost. The gap between them is often the most useful number of all.
- Does NERC GADS Wind use energy-based measures?
- Monthly performance reporting is primarily hours-based, with some capacity and generation values. Outage events carry a Potential Production MWh Loss calculated from the capacity lost in each interval.
- Does icing count as a GADS Wind event?
- Not on its own. According to NERC's examples, icing that reduces output while turbines keep running is not an event; icing that shuts down at least 20 MW of turbines is.
