Conclusions
The Matlab model of the Supergen Wind 5 MW exemplar wind turbine, which has been used by various researchers over the last decade, especially within the Supergen Wind Consortium is improved in speed. This is achieved through prewarping, discretisation and conversion to C in order. Discretising optimally prevents the Simulink solver from non-optimally performing numerical integration for simulating the stiff drive-train module, and thus improves the simulation speed. Discretisation is also an essential prerequisite for conversion to C. Subsequently converting the discretised model to C, and then to CMEX for simulation in Matlab/SIMULINK, further improves the simulation speed. It allows the discretised model to be duplicated and extended to constitute a wind farm model without increasing the simulation time exponentially in contrast to the original continuous model. The simulation results demonstrate that even with one hundred turbine models included in a wind farm, the simulation speed remains at 70s. It is important again to emphasise that each turbine model is neither simplified nor compromised. In the second part of the paper, the wind farm controller introduced in [11] is tested by application to a wind farm model of 30 turbine models, which is only possible because the turbine model has been improved in speed – in [11], the available wind farm model could only contain 10 turbines due to the computational effort and speed required for simulating each turbine model. The simulation results demonstrate that the wind farm controller performs satisfactorily when applied to the wind farm model of 30 turbines. The comparison of the simulation results to those from the situations in which the wind farm controller is applied to smaller wind farm models demonstrates that the fluctuation of the wind farm power output decreases as the wind farm size increases. The controller is therefore scalable and could be more valuable to larger wind farms.
