Mar 01, 2024
Einladung zum Statusvortrag von Sebastian Strönisch
Abstract:
Analysis, optimization and uncertainty quantification of the aerodynamic behaviour of turbomachinery components is a fundamental part of the current industrial design process and requires the repeated use of compute-intensive CFD simulations. Thereby, accelerating the solution process without compromising on accuracy is a major goal in the development of numerical flow solvers and design optimization frameworks.
Related work shows that graph neural networks are able to provide good estimates of flow quantities while maintaining the geometric accuracy of numerical meshes in a fraction of the time of a classical CFD.
Here, a state-of-the-art hierarchical graph framework is adapted and the training process distributed to apply the GNN to different turbomachinery design tasks on industrially relevant mesh sizes of around 2.5e06 points. This revealed a number of challenges and limitations that are discussed in this talk and need to be addressed in the future.
Betreuer: Prof. Dr. Wolfgang Nagel