Optimal shape design using machine learning for wind energy and pressure

As energy consumption presents exponential growth, topics on sustainable design and energy generation are brought to the forefront. This paper presents procedures and methods suggest the initial designs for a shell pavilion that generates energy. Thus, optimal strategies incorporating a wind turbine that maximizes wind speed and structural integrity were proposed. Computational Fluid Dynamics (CFD) was used to find wind pressure around the thin shell pavilion. Also, the paper adopted the parametric design process to find an optimal form for the design. The main obstacle to running CFD simulation for parametric design is the heavy computational load and time. To achieve Multiple Objective Optimization (MOO) with less computational effort, we created a data-driven framework that engages Artificial Neural Network (ANN) and Genetic Algorithm (GA), which predicted both wind speed around the turbine and structural behaviors. These performance values were fed into optimization algorithm to find the form that satisfied the goal of this study. The results demonstrate that the proposed methods find various shapes optimized for different materials. Moreover, several optimal designs were identified based on various preferences. Especially by considering energy generation and structural integrity equally, the optimal pavilion can generate a wind speed of 14.73 m/s with a shell thickness of 1.27 cm when constructed using steel. The main contribution of this paper is integrating the Machine Learning method and advanced wind pressure calculation with CFD to find optimal designs that are more comprehensive than typical.