Complete methodology on generating realistic wind speed profiles based on measurements

Abstract

The  wind  speed  represents  the  main  exogenous signal applied to a Wind Energy Conversion System (WECS) and determines  its  behavior.  The  erratic  variation  of  the  wind  speed, highly  dependent  on  the  given  site  and  on  the  atmospheric conditions,   makes   the   wind   speed   quite   difficult   to   model. Moreover,  wind  modelling  for  medium  and  large  time  scales  is poorly   treated   in   the   present   literature.   This   paper   presents methods for generating realistic wind speed profiles based on real measurements.  The  wind  speed  profile  is  divided  in  a  low-frequency  component  (describing  long  term  variations)  and  a turbulence  component  (corresponding  to  fast,  high  frequency variations).  For  modelling  the  low  frequency  component  two methods  are  implemented  and  compared:  one  based  on  the  Van der  Hoven  spectrum  and  one  based  on  auto  regressive  models. The importance of the turbulence component is also analyzed and modelling methods based on white noise filtering are presented.