Characterising wind fluctuations at the Horns Rev offshore wind farm

1. Characterising wind fluctuations at the Horns Rev offshore wind farm Claire Vincent and Gregor Giebel (Risø-DTU) Pierre Pinson (DTU-IMM)

2. Introduction • Episodes of severe wind variability are observed at Horns Rev. • Particularly serious for offshore wind farms, due to the high concentration of turbines within a small area. • Fluctuating power supply is a challenge for TSOs and wind farm operators • Technical considerations – grid management, security of supply • Financial considerations – fluctuations must be balanced with reserve power • Episodes of severe wind variability are observed at Horns Rev. • Particularly serious for offshore wind farms, due to the high concentration of turbines within a small area. • Fluctuating power supply is a challenge for TSOs and wind farm operators • Technical considerations – grid management, security of supply • Financial considerations – fluctuations must be balanced with reserve power Examples from 2000 and 2001

3. Methodology • Define an objective measure of ‘variability’ • Adaptive spectral analysis • Where is there enhanced spectral intensity? • Apply the objective measure to long time series of wind speed from Horns Rev • 4 year time series of 10 minute observations of wind speed • Observations of wind direction, pressure, temperature, rainfall. • Characterise the types of atmospheric conditions in which variability tends to be enhanced • Conditional spectra • Correlations between variability and other meteorological variables

4. Observational Data • 62 metre meteorological mast • 4 years of 10 minute wind speed observations, from top-mounted cup anemometer. • 10 months of 12Hz wind speed observations from sonic anemometer mounted at 50 metres. • Measurements of wind speed and temperature at several heights, pressure, precipitation, wind direction, water temperature.

5. The Hilbert-Huang Transform • Method of adaptive spectral analysis first described in 1998 by Huang et al. • Using the Hilbert transform, the instantaneous frequency and amplitude of a time series can be calculated • However, the idea of ‘instantaneous frequency’ only makes sense for a ‘mono-component’ signal. • Huang’s innovation: 1. Decompose the data, so that each component is a mono-component signal. 2. Use the Hilbert transform to calculate the instantaneous frequencies and their amplitudes at each time.

6. Empirical Model Decomposition Fastest oscillations in the time series Quick response to non-stationarities in the data Original time series

7. Amplitude Modulation

8. Frequency Modulation Period: 2.4 hours Frequency: 1/2.4 = 0.4 cycles per hour

9. The Hilbert Spectrum Episode of enhanced variability

10. Results: Seasonality of wind fluctuations More intense fluctuations in winter and autumn

11. Results: The effect of wind direction Land Wake Sea

12. What does the wind ‘see’ from different directions? • Fetch: • Land sector (southeast to northeasterly flow) • Very long fetch to Greenland (north westerly flow) • Wake of Norwegian mountains (northerly flow) • North sea to England (southwest to northwesterly flow) • Synoptic conditions: • Storm tracks and fronts tend to approach from the west / southwest

13. Results: Variability and Pressure Tendency Falling pressure Rising pressure

14. Results: Variability and Precipitation Rmax at time t is the maximum observed 10 minute rain amount within the interval [t – 90mins, t + 90 mins] Rain is observed in discrete 0.25 mm quantiles. Contours show probability density function of variability for each Rmax value.

15. Results: Variability and Stability 10 minute cup anemometer data: 4 year time series 12 Hz sonic anemometer data: 10 month time series Most unstable Most stable

16. Summary of Results • Define a ‘severe variability event’ as being an event: • with total variability on timescales of 1-3 hours in 95th percentile • with duration > 3 hours. • 218 events in the years 2000-2003 • 83% occurred in the direction sector 180 – 330 degrees • 89% occurred in unstable or neutral conditions • In 2003, 76% recorded some rain during the event • 53% occurred when the 3 hour pressure tendency was rising • 67% occurred in the months September - February

17. A ‘classic’ event

18. Conclusions • Climatological patterns have been seen in large amplitude wind fluctuations at Horns Rev • The majority of episodes are associated with onshore flow, in rainy, unstable conditions • The predictability of wind fluctuations is linked to our ability to accurately forecast the wind field in these conditions • Further work • modelling the wind field in severely variable conditions • Installation of rain radar at Horns Rev in 2009 to track precipitation cells approaching the wind farm (Radar@SEA Danish project)

19. Acknowledgements • Measurement data was supplied by Vattenfall as part of the Danish Public Service Obligation (PSO) fund project ‘HRENSEMBLE – High Resolution ENSEMBLEs for Horns Rev’ (under contract PSO-6382), which is gratefully acknowledged. • The work was partly supported by the PSO project ‘Mesoscale atmospheric variability and the variation of wind and production of offshore wind farms’ (under contract PSO-7141). • The measurements from the Blåvandshuk meteorological station were provided by the Danish Meteorological Institute. • Coordinates of the Horns Rev II wind farm were kindly supplied by Dong Energy.