ECN publicatie: facebook
Intermittence and scale separation above the North Sea
Barth, S.; Peinke, J.
Gepubliceerd door: Publicatie datum:
ECN Windenergie 12-6-2007
ECN publicatienummer: Publicatie type:
ECN-M--07-047 Conferentiebijdrage
Aantal pagina's: Volledige tekst:
5 Download PDF  (359kB)

Gepresenteerd op: European Wind Energy Conference 2007, Milan, Italy, 7-10 mei 2007.

A general better understanding of wind turbulences is of great interest for wind energy applications for both predictions of mechanical loads and power output. Many meteorological simulations, which are used as input for these predictions rely on the assumption that there is a so-called spectral gap, i.e. a separation of wind properties on large scales and small scales and that all the extreme fluctuations occur on time scales smaller 10 minutes. Using 10 minutes averages like in the the IEC standard 61400-12-1 for estimating power curves of wind turbines one tries to eliminate these extreme fluctuations, which enables for using faster and simpler models for flow properties on larger scales. While the 10 minutes have widely been used as a separation (averaging) scale there has been little support for this somehow arbitrary chosen scale in standard analysis, e.g. power spectra. In this paper we show an analysis that gives a good reason to select 10 minutes as separation scale allowing for simpler models on larger scales and more sophisticated models on smaller scales. The statistics of wind speeds measured at several heights up to 100m above the North Sea have been analyzed by means of velocity increments (velocity changes within a certain amount of time). The continuous measurement for a period of 6 months with 0.85Hz allows for considering scales ranging from 1.18s to several days on a solid statistical basis. While the probability density functions (PDFs) of velocity increments on small scales are very intermittent, i.e. they are non-Gaussian and heavy-tailed corresponding to high risks for large gusts, they become only Gaussian on scales in the order of days. This scale dependence of the PDFs is known from isotropic, homogenous turbulence as realized in laboratory experiments, e.g. free-jets or wake flows and can be quantified by an easy to calculate shape parameter using superstatistics. However, while the shape parameter for laboratory flows is a monotonic decaying function of the scale the atmospheric data considered here show a local maximum in the shape parameter at a time scale of 600s, which corresponds to the length scale of the controversial discussed spectral gap. This special behavior is not apparent in the data's spectra. Thus this increment analysis can be used for a more detailed investigation of scale separation to distinguish between small scale turbulence and larger structures within atmospheric winds.

Terug naar overzicht.