Global Optimization of Wake Steering for Large-Scale Wind Farms Using Generalized Serial Refine Method

Abstract

Wake steering via active yaw control has proven to be a highly effective approach for boosting power generation in wind farms. However, determining the optimal yaw angles for each turbine in large-scale wind farms is a complex challenge due to the high dimensionality of the optimization problem. In this paper, we propose the generalized serial-refine method that is capable of finding the global optimal solution to the yaw optimization problem of large-scale wind farms. We demonstrate the global optimization capability of this method by comparing its performance to five other conventional optimization algorithms (including sequential least squares quadratic programming (SLSQP), elitist genetic algorithm (EGA), particle swarm optimization (PSO), differential evolution (DE), and hybrid DE-SLSQP. In the two benchmark wind farm cases tested herein, the GSR method achieves the highest power gain with excellent efficiency and robustness in both cases. In particular, for the deliberately designed diamond-shaped wind farm case, while other algorithms are easily trapped at the local optimal solutions, the GSR algorithm effectively reproduces the monotonically decreasing pattern for the optimal yaw angles for each row of tandem arranged wind turbines, demonstrating its ability to find the global optima. Leveraging the global optimization capabilities of the GSR method, we further explore how wind direction and speed influence the effectiveness of yaw control. Our analysis reveals that active yaw control is most beneficial under wind directions that produce significant wake effects. Power gains diminish as wind speed increases, with yaw control becoming ineffective beyond 15 m/s, where wake losses are negligible.  This study not only introduces an advanced tool for wind farm wake steering optimization but also provides actionable insights for tailoring yaw strategies to diverse environmental conditions, ultimately improving the efficiency of large-scale wind farms.

Keywords: Wind farm flow control, Optimization algorithms, Wake steering, Wake model