A Feasibility Study of Wind Farm Yaw Angle Optimization

Abstract

Wind farm energy production optimization has received significant attention in recent years. Much of this effort had been focused on optimizing positions of wind turbines within a wind farm domain during the design and planning stage. Optimization of wind turbine positions can reduce wake interactions of upstream turbines. In addition to optimizing turbine positions to reduce wake interactions, prior studies have shown that optimizing yaw and pitch angles can improve energy production as upstream wakes yaw away from downstream turbines. However, yaw angle optimization at the wind farm level has been difficult due to lack of low-fidelity wake model for simulating yawed wakes. Recently, an analytical wake model capable of simulating yawed turbine wakes had been developed, which enable wind farm-scale yaw optimization. In this work, a binary quadratic programming model problem formulation has been developed to optimize yaw angles of wind farms. Yaw optimization of two position-optimized layouts available in the literature were performed to study the potential of yaw optimization. In particular, we set out to understand how layout and wind farm density affect yaw optimization potential. An optimized layout of 39 turbines with 40m rotor diameter in a 2km by 2km domain was used in this study. For this wind farm, yawing optimization only improved power production by ∼1.5% under favorable wind directions. However, as wind farm power density increases by increasing rotor diameter to 60m and 80m, power production improved by ∼5% and ∼10% respectively, under favorable wind directions. Finally, another yaw optimization was performed on an optimized layout with 48 turbines of 82m rotor diameter in a 4km by 4km domain using the proposed model formulation. Under favorable wind directions, yaw angle optimization improved performance by ∼4%. The results show that yaw optimization can improve power production in the same order of magnitude as layout optimization, and that it should be considered in addition and/or in tandem to layout optimization.

Wind Farm Yaw Optimization via Random Search Algorithm

“…The authors initially developed an BQP model [59], which allows turbines to yaw a predefined set of angles. The authors' experience developing and working with mathematical programming models for wind farm layout optimization problems suggested that this BQP model can be quickly solved using state-of-the-art solvers such as Gurobi and CPLEX, and perhaps outperform certain types of metaheuristic algorithms.…”

“…To the best of the authors' knowledge, the only existing parametric studies on wake steering is the work by Kuo et al [59], which illustrated that the potential of wake steering is a function of wind farm density. That work employed a binary quadratic programming (BQP) formulation.…”

“…Table 4. Power density and percent production improvement of layout from Grady et al [60] using binary quadratic programming (BQP) model [59] compared with proposed random search (RS) algorithm.…”

“…Percent improvement in power production by wind direction sector, comparing proposed random search (RS) algorithm and binary quadratic programming (BQP) model[59] for D = 60 m.…”