Multi-Level Decoupled Optimization of Wind Turbine Structures

Jin Woo Lee, Efstratios Nikolaidis, and Vijay Devabhaktuni (University of Toledo)

CHALLENGE - This paper proposes a multi-level decoupled method for optimizing the structural design of a wind turbine blade. The objective is that cost should be reduced substantially in order to make wind energy competitive compared to other energy sources. One way to address this issue is to develop and use computational tools for designing optimized blades.

SOLUTION -  The NREL 5MW reference wind turbine blade is used for testing in this study. PreComp is the low-level solver. Using the laminate designs of each section, this code calculates those structural properties required for the high-level solver. The high-level solvers are smoothness calculator, FAST and stress calculator. FAST runs a transient wind turbine simulation and calculates blade bending moments in flap- and edge-wise, axial force and tower-to-blade clearances of each blade. Inputs to FAST are a wind profile and a wind turbine model. The latter includes blade structural properties, which are calculated in the low-level optimizations. This study uses modeFRONTIER to perform the optimization: SIMPLEX algorithm is used in the optimizations for both the high and low-levels. 

















BENEFITS - In comparison of the single- and multi-level methods in terms of the smoothness of the designs that they produce, the multi-level method got an average of 2.25 wins more than the single-level. The multi-level method finds optimum designs with lower mass than those of the single-level method in 3 out of 4 cases. The proposed method is suitable for optimization problems that require computationally expensive analyses and involve many design variables.