Engineering wind-farm wake models: beyond ten minute averages

摘要

Wind farm data with a time resolution beyond 10 min averages are used to discuss various short-time generalizations of engineering wind-farm wake models. We work with averaging times down to 15 sec and fit the models to single-event production patterns, i.e. we do not do any sector-averaging. The average quality of the model fits increases with increasing time averaging and saturates at a time resolution around 10 min as most short-term fluctuations are then averaged out. The fitted wake parameters turn out to be independent from the amount of time averaging and are significantly smaller than those values obtained from standard sector-averaged model fits. Relaxing the assumption of a homogeneous free wind up-wind of the farm and including cross-wind variation of the wind speed leads to a good improvement of the model fits while the inclusion of longitudinal non-stationarity, where the wake models are formulated in a co-moving reference frame, only leads to a rather small increase in mean fit quality.