From μCT data to CFD: an open-source workflow for engineering applications

The generation of high-quality volume meshes out of mu CT data can be a challenging task that is mandatory to perform CFD simulations of fluid flows within or around scanned objects. Despite the growing relevance of CT-based CFD simulations, there is still a lack of a standardized, free-to-use workflow. In this work, an open-source based workflow is presented, which covers all steps from the CT image stack to a volume mesh in the end. The software packages ImageJ, ParaView, and Blender are used for surface reconstruction and processing of mu CT data; resulting meshes are evaluated regarding geometric parameters, surface fidelity, and used memory. For volume meshing, the mesh generator snappyHexMesh implemented in OpenFOAM is used. A detailed description of this workflow is provided in the Supplementary Materials. We demonstrate how potential pitfalls and shortcomings can be avoided and how surface smoothing is especially important to preserve the surface area of scanned objects. We use CT data of a 10 ppi open cell foam to illustrate that data decimation can reduce the required time for the volume meshing by up to 45% and RAM up to 75%. The resulting meshes are used to simulate flow fields and heat transport with a surface heat source. The resulting temperature fields show that differences in the surface area and recesses in an unsmoothed mesh can affect the outcome of heat transport simulations, highly relevant for reactive CFD simulations. The results highlight the quality of our workflow's outcome, which can help engineers that rely on CT reconstruction processes, also for other applications.

Automated summary

In this study, an open-source based workflow for generating volume meshes from CT data is presented. The resulting meshes are evaluated to assess the quality and importance of the workflow, and potential pitfalls and shortcomings are demonstrated to be avoided.