Wind farm wake modeling based on deep convolutional conditional generative adversarial network

Modeling of wind farm wakes is great importance for the optimal design and operation farms. In this work a surrogate modeling method parametrized fluid flows proposed wake modeling, based on state-of-the-art deep learning framework i.e. convolutional conditional generative adversarial network. Based data generated by high-fidelity large eddy simulations, novel model developed. The developed first validated against results show that it achieves accurate, efficient, robust prediction turbine flow, at all streamwise locations including both near far wake, spanwise velocity components, cases with different inflow profiles. Then an extensive parametric study carried out generalizes well to unknown flow scenarios. Furthermore, case investigated model. are then compared showing can predict (including fields) very well.