About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
Frontiers in Solidification: An MPMD Symposium Honoring Jonathan A. Dantzig
|
Presentation Title |
Design of Light Wind Turbine Parts by Simulation Based Machine Learning |
Author(s) |
Youness Bami, Can Huang, Emir Subasic, Felix Weber, Jannik Zimmermann, Vitali Züch, Juergen Jakumeit |
On-Site Speaker (Planned) |
Juergen Jakumeit |
Abstract Scope |
The steady increase in the size and power of modern wind turbines has led to the need to reduce the weight of their castings. Shape optimization leads to thinner parts that are lighter and solidify faster. Faster solidification often leads to better local material properties, which in turn enable further size reductions. By integrating casting simulation with property prediction into shape optimization, this link between size and properties can be used to design lighter parts.
To achieve this goal, multiphase casting simulation is combined with a microscopic diffusion-driven growth model for eutectic grains in nodular cast iron to calculate microstructure parameters and estimate local material properties of wind turbine components. Based on many precomputed simulation data, a hybrid machine learning approach is trained to predict grain structure parameters and mechanical properties. First promising results of this simulation-based machine learning (SMiLe) approach are reported and compared with simulation and experimental results. |
Proceedings Inclusion? |
Planned: |
Keywords |
Iron and Steel, Machine Learning, Modeling and Simulation |