Treffer: Development of a fast cloud-based simulation workflow for the complete aerodynamic evaluation of aircraft

References: [1] Gernot Kurt Boiger, Darren Sharman, Bercan Siyahhan, Viktor Lienhard, Marlon Boldrini, and Dominic Drew. A massive simultaneous cloud computing platform for openfoam. In 9th OpenFOAM Conference, online, 19-20 October 2021. ZHAW Zürcher Hochschule für Angewandte Wissenschaften, 2021.2 ; The field of Computational Science is facing an increasing demand for data-intensive investigations. Engineering tasks such as parameter-, sensitivity- and optimisation studies need ensemble computing to an ever-increasing extent. At the same time, the field of artificial intelligence (AI) is pushing for ever more extensive, numerically derived learning-, testing- ,and validation data. With the cloud software KaleidoSim, we can run hundreds of numerical simulations simultaneously [1] and generate large amounts of data in a short time. Although KaleidoSim supports various simulation tools, this study uses only OpenFOAM. In this work, we developed tools and routines to speed up, simplify and automate studies containing hundreds of simultaneous simulations in the cloud. To test our toolbox, we conducted a complete 360° aerodynamic analysis of various airborne vehicles. The study included 420 OpenFOAM simulation cases. Each case was a steady-state, Reynolds Average Stress (RAS) turbulence model-based, single-phase flow simulation on a 1.5 million cell hexahedral finite volume grid. Drag and lift coefficients were calculated for each case.We used a combination of Python and KaleidoSim Application Programming Interface (API) routines to develop the toolbox. The Python based graphical user interface (GUI) allows switching between different CAD models so that multiple aircraft can be compared. The GUI also enables mesh sensitivity analysis to identify optimised meshes for each aerodynamic shape. Based on this, we performed a series of mesh sensitivity analyses using snappyHexMesh and CfMesh grids. This work proved that a combination of cloud computing via KaleidoSim-based API routines and Python scripting can speed up ...