Status and perspectives of CFD modelling of gas-liquid flows in the multi-fluid framework

There is an increasing request to use CFD-methods for simulations on medium and large scale industrial applications, e.g. in chemical engineering, energy techniques and nuclear safety. For most of such applications the Euler-Euler two or multi-fluid approach is the only feasible one. Gas and liquid phases are represented by interpenetrating fields and the information on the interface gets lost during the averaging process which is applied to obtain the balance equations. To close these equations the corresponding local phenomena at the gas-liquid interfaces have to be considered by closure models. As recently discussed by Lucas et al. (2016) there is not yet consensus achieved in the community regarding the most appropriate closures which limits the reliability of CFD-simulations using the Euler-Euler approach. A so-called baseline model concept was proposed in that paper. Since the closure models have to reflect the local phenomena a case by case tuning is not meaningful and instead a fixed set of closure models should be defined for certain flow conditions and applied to different cases without any modification.

Different flow morphologies as bubbly flows, droplet flows and segregated flows with large interfaces have to be distinguished. These different approaches require different closure models. In addition for poly-disperse bubbly flows it may be necessary to divide the gas phase into sub-phases reflecting bubbles of different size respectively. At HZDR a baseline model for poly-disperse bubbly flows basing on the inhomogeneous MUSIG (iMUSIG) approach (Rzehak and Krepper, 2016) and a model for segregated flows basing on the AIAD model (Porombka and Höhne, 2016) have been established. Especially the baseline model for poly-disperse flows with fixed model formulations and model parameters was validated on a large number of experiments (more than 150) for different flow geometries, flow rates and material systems. There is already an acceptable agreement for many cases, but for some also clear deviations occur. It is the scientific challenge to identify the main reasons for these deviations and figure out a better model for the corresponding phenomenon. The baseline model strategy will be illustrated by the recent developments to improve the modelling of bubbly flows and a general strategy how to develop better models will be presented.

In many flow situations interfaces may vary over a large range of scales combining dispersed and segregated morphologies. To handle such flows the innovative GENTOP concept was developed (Hänsch et al., 2012). It combines the iMUSIG and AIAD approaches and allows also simulating transitions between the different morphologies. The well validated baseline models are thus part of GENTOP. Recently the concept was applied for a simulation of a boiling pipe which includes flow pattern transitions (Höhne et al., 2017). The second part of the lecture will report about these developments which aim to extend the range of applicability of CFD simulations.

The above mentioned approaches were first implemented and tested in the commercial CFD-code ANSYS-CFX. Presently a similar framework is established for the OpenSource code OpenFOAM. A GitLab based version control system allows a high level quality assurance and has a high potential for international co-operation. Joint efforts can be done to qualify the code system.