A highly versatile tool chain for robust computational fluid dynamics simulations

Abstract

The challenges of Computational Fluid Dynamics require careful preparation of models in order to ensure successful and correct calculations with reasonable resource consumption. This considerate preparation complicates the simple and wide use of a single simulation package for multiple different problem domains. This work introduces a robust tool chain which allows for a versatile application of Computational Fluid Dynamics calculations to greatly differing problem domains. Generic concepts ensuring and enhancing robustness and versatility are at the core of the thesis. These generic concepts are applied in the construction of a concrete Computational Fluid Dynamics tool chain, which stretches from problem definition to final evaluation. The performance of the software is of great importance in order to allow for application on a wide variety of different hardware, thus performance is crucial for ensuring a high degree of versatility. To put the versatility and robustness to the test, two applications from very different scientific disciplines are selected. On the one hand simulations are carried out to analyze water cooled heat sinks in both a consumer and an industrial application, which represents a classic engineering task. On the other hand the tool chain is also applied to medical input data in the form of Magnetic Resonance Imaging scans, which are processed to construct models for biomedical stress studies. This provides access to physical properties (e.g. velocity profiles and shear stress) otherwise inaccessible in situ. The applicability of the tool chain to these thematically different topics illustrates not only the claims of both versatility as well as robustness, but also that performance of the software does not get degraded by using the developed generic approach.