Automation of CAE-Workflows for Transient Fluid-Flow- and Aero-Acoustic-Simulations Using AI Techniques
Thursday, July 1, 2021 12:00 PM to 12:20 PM · 20 min. (Africa/Abidjan)
Stream#3
HPC Workflows
Information
Contributors:
Abstract:
Transient and fine grained Fluid-Flow- and Aero-Acoustic-Simulations for the design and improvement of devices of practical use such as rotating machinery (fans, turbines etc.) are demanding regarding computing time and resources - even despite making use of HPC architectures and HPC ready solvers. In addition, two aspects are still under development or object of research: First, the automation of optimization loops for transient simulations requires reliable and continuous monitoring and results interpretation throughout the whole simulation process where intelligent data analysis techniques are required. Second, such processes require holistic workflows which connect the geometries to be optimized - typically stored as parameterized CAD datasets - with the respective Fluid-Flow- and Aero-Acoustic Simulation results in order to provide CAE datasets for the production lines. The latter requires intelligent parameterization of the CAD data and intelligent workflow mechanisms for connecting the CAD datasets of the prototypes - the devices' positives - with the simulation grids - typically the fluid filled space within or around the devices, i.e. the devices' negatives. This presentation provides insight in recent developments and research regarding application cases from the above context with a focus on axial flow fans. The ongoing project is a collaboration of a mixed consortium of engineering companies (including the one of the author) and public research entities.
Abstract:
Transient and fine grained Fluid-Flow- and Aero-Acoustic-Simulations for the design and improvement of devices of practical use such as rotating machinery (fans, turbines etc.) are demanding regarding computing time and resources - even despite making use of HPC architectures and HPC ready solvers. In addition, two aspects are still under development or object of research: First, the automation of optimization loops for transient simulations requires reliable and continuous monitoring and results interpretation throughout the whole simulation process where intelligent data analysis techniques are required. Second, such processes require holistic workflows which connect the geometries to be optimized - typically stored as parameterized CAD datasets - with the respective Fluid-Flow- and Aero-Acoustic Simulation results in order to provide CAE datasets for the production lines. The latter requires intelligent parameterization of the CAD data and intelligent workflow mechanisms for connecting the CAD datasets of the prototypes - the devices' positives - with the simulation grids - typically the fluid filled space within or around the devices, i.e. the devices' negatives. This presentation provides insight in recent developments and research regarding application cases from the above context with a focus on axial flow fans. The ongoing project is a collaboration of a mixed consortium of engineering companies (including the one of the author) and public research entities.
