Metamodel-based generative design of wind turbine foundations

Wind turbines play an integral role in energy transition agendas. The optimized design of wind turbine foundations is a complex and intricate task that requires iterative running computationally-intensive time-consuming finite element models. However, given the popularity these structures over past two decades, there wealth data from designs projects can be used for data-driven modeling structures. Given demonstrated accuracy success metamodels as alternative approach other simulation-based problems, this study aims to develop generative-design framework optimization using metamodel, complementary step more accurate modeling, reduce overall time without compromising accuracy. To end, first, random forest method multi-output metamodel based on set historical data. Then, metaheuristic method, i.e., NSGA II, adopted optimize process developed metamodel. In case study, foundation was designed proposed output assessed terms ultimate bending moment. results indicate provides significant gain (i.e., 99.93%) 1.75% percent error). Besides, conducted also offers designers better understanding importance each variable how certain variables influence moment-rotation behavior