HZDR baseline model strategy

The description of polydisperse bubbly flows via Eulerian-Eulerian CFD requires a dozen closure models for the interfacial forces, the liquid turbulence and various bubble-bubble interactions. For each of those closures a growing number of empirical and semi-empirical models exist in the literature. In the multiphase community there is no consensus about the use of those models nor their inherent model constants. Different research groups choose different sets of models even on a case-to-case basis, hindering the predictive power of CFD.

Goal of our baseline strategy is to establish an universally valid set of baseline models with reliable predictive abilities for all types of bubbly flow in any geometry. Baseline closure models are mechanistic models based on local flow conditions producing convincing results for a large set of validation cases.

In order to pursue this strategy a workflow for the fully-automated analysis of OpenFOAM simulations is regularly applied to an extensive database of CFD simulations. The fast and efficient production of large amounts of results, which are summarized in well-structured reports, allows us to facilitate new machine learning tools for the data analysis. A fuzzy logic controller has been introduced in order to quantify the agreement of our simulations with the available validation data as illustrated in the figures below. This novel quantification of the CFD prediciton quality for the first time allows the systematic evaluation of closure models and contributes to a sustainable baseline model development.

References

• Lucas, D.; Rzehak, R.; Krepper, E.; Ziegenhein, Th.; Liao, Y.; Kriebitzsch, S.; Apanasevich, P., A strategy for the qualification of multi-fluid approaches for nuclear reactor safety, Nuclear Engineering and Design 299 (2016), 2-11.
• Hänsch, S.; Evdokimov, I; Schlegel, F.; Lucas, D., A workflow for the sustainable development of closure models for bubbly flows, Chemical Engineering Science 244 (2021), 116807.