Timeseries models¶
Farm performance is analysed by cross-comparison of four timeseries models: one built from the measured production, three computed by the digital twin.
Actual (100% availability)¶
What the turbines produced, and what they would have produced had every turbine been available: measured SCADA production, with the energy lost to stops, partial performance and data gaps estimated.
Metrics: Actual, Actual (100% availability), Energy (full performance), Full-performance slots, Missing-data loss, Out-of-environmental-spec loss, Partial-performance loss, Stopped loss, Wake gain from stopped turbines
Expected¶
What the farm is expected to produce over the period, at 100% availability: the digital twin run on the period's wind, every turbine operating as designed. The Expected levels of the waterfall are read from it.
Metrics: Expected (model total), Expected energy (full performance), Expected gross (current period), Freestream (current period), Internal wakes loss (current period), Wakes loss (current period)
Expected (as operating)¶
What the farm is expected to produce given how the turbines actually operated, yaw misalignment included, at 100% availability. Not used by the waterfall: it feeds the yaw misalignment loss and the recoverable energy.
Metrics: Yaw misalignment loss (current period)
Expected (long term)¶
What the farm is expected to produce in a wind year representative of the long term: the digital twin run on the long-term climate of the site, every turbine operating as designed.
Metrics: Expected gross (long term), Expected with site-specific performance (long term), Internal wakes loss (long term), Wakes loss (long term), Yaw misalignment loss (long term)