A pragmatic approach to wind farm simulations using the dynamic wake meandering model

This paper presents a computationally efficient method for using the dynamic wake meandering model to conduct simulations of wind farm power production. The method is based on creating a database, which contains the time and rotor-averaged wake effect at any point downstream of a wake-emitting turbine operating in arbitrary ambient conditions and at an arbitrary degree of wake influence. This database is later used as a look-up table at runtime to estimate the operating conditions at all turbines in the wind farm, thus eliminating the need to run the dynamic wake meandering model at runtime. By using the proposed method, the time required to conduct wind farm simulations is reduced by three orders of magnitude compared with running the standalone dynamic wake meandering model at runtime. As a result, the wind farm production dynamics for a farm of 100 turbines at 10,000 different sets of ambient conditions run on a normal laptop in 1h. The method is validated against full scale measurements from the SmOla and OWEZ wind farms, and fair agreement is achieved. Copyright (c) 2014 John Wiley & Sons, Ltd.