SWIFT: A computer program for stochastic wind field simulation

The problem

To assess and predict the effect of atmospheric turbulence on the dynamic response of wind turbines, we need a stochastic wind field simulation program. Such a simulator should recreate numerically the gross features of the wind, which include those properties which have the most significant impact on the blade loads and, therefore on the dynamic response of wind turbines.

For example, to ensure that the `rotational sampling effect' is captured properly (see Figure 1), the simulated wind field should have the proper auto- and cross-correlation structure, both in space and time.

An impression of the spatial 3 dimensional wind field as created by SWIFT is presented in Figure 2 for a high turbulence level and  large shear.

The approach

Ideally, the numerical realisation should satisfy the laws of fluid mechanics. However, such an approach would be sufficiently complex to prohibit almost any meaningful calculations or inferences. Instead, the simulation scheme is based on a simplified, empirical, statistical model of atmospheric turbulence.

The selected algorithm for generating a realisation of the stochastic wind field with the required statistical characteristics is the so-called `Spectral Representation Method'. The basic idea is to specify discrete values of the spectra and cross-spectra and then to calculate the Inverse Discrete Fourier Transform (IDFT) and so generate discrete time sequences which are regarded as sampled values of the continuous random signals required.
Special effort has been put into a highly efficient computational scheme.

The result

The experience, which has been obtained in many national and international wind energy research projects over more than 10 years, has shown that SWIFT (Simulation of WInd Fields in Time) is a reasonable compromise between accuracy, flexibility and computational efficiency. SWIFT has also proved to be a reliable program and has considerably improved the prediction capabilities of PHATAS. SWIFT is the standard stochastic wind field simulation program used by ECN for design assessments following recommendation by certification bodies like IEC, DNV, and Germanische Lloyd.

Details of the code are:

• All three components of the wind speed vector are generated on a polar grid, fully correlated if needed.
• The program is user friendly and `fool proof'. No special knowledge is required; defaults are provided for most of the input to SWIFT. For the use in certification calculations some input is predefined, ensuring that the generated wind field is compliant with the guidelines of Germanischer Lloyd and/or the IEC61400-1, edition 3 standard.
• Written in FORTRAN and running on various platforms ranging from PC under Windows to Unix work stations.

Publications

• SWIFT; Program for Three Dimensional Wind Simulation. Part I: Model description and program verification
  Winkelaar, D.;
  ECN-R--92-013  1992; 64p.
• Ontwikkeling van een stochastisch windbelastingmodel, deel 1: Windmodellering
  Winkelaar, D.;
  ECN-C--96-096; december 1996; 66p.
• Ontwikkeling van een stochastisch windbelastingsmodel, deel 2: Aero-elastische modellering
  Winkelaar, D. and Engelen, T.G. van;
  ECN-C--96-097, December 1996.
• Ontwikkeling van een stochastisch windbelastingsmodel
  Winkelaar, D. and Hendriks, H.B.;, deel 3: Validatie
  ECN-C--96-098, December 1996.