Multiphase flows are very common in industrial applications. Therefore it is important to investigate them in detail to be able to predict the behavior of such systems and optimize them for best performance. Verified simulation tools and methods can provide a detailed insight at reasonable time and cost frame. Computational fluid dynamic (CFD) can give a full spatial and temporal representation of multiphase systems. Using CFD, different multiphase phenomena such as mass transfer, heat transfer, bubble or droplet breakup, coalescence etc in complex geometries can be investigated and interpreted to help improving the process.
Detailed multiphase simulations (e.g. volume of fluid - VOF) can help for better understanding of the underlaying phenomena (heat and mass transfer) and developing closure models for being applied to less detailed multiphase simulation approaches (e.g. Euler-Euler) to be used for simulating big and complex geometries.
To refine all the needed details in a multiphase simulation the discretization (mesh) should be fine enough to capture all the flow and the phase structures. This is usually computationally very expensive, especially for big geometries (e.g. and industrial geometry). To overcome this drawback a dynamic meshing strategy can be used. A coarse mesh is used for the whole geometry at the starting time and is refined it just in the places where it is needed (e.g. at the phases interface). Since in a multiphase simulation the interface between phases is changing over time this solution based mesh refinement should be also done dynamically.
Big geometries are usually simulated on multiple CPU cores in parallel to reduce the overall calculation time. However, by using a dynamic meshing the calculation load of each processor core varies during the time, since the details needing refinement, e.g. interfaces in sub-domains assigned to the processor cores, are not the same for all processor cores . To overcome the unbalanced load distribution between processor cores, the geometry can be repartitioned and reassigned to the processor cores to meet a specified imbalance tolerance.
In our research group, using OpenFOAM® as the base code new models, libraries and solvers are developed and implemented for modeling of different types of multiphase flows on different scales. The codes and models are validated against experimental fluid dynamic approaches (using e.g. LDV, PIV or spacialy resolved Raman imaging) and used in simulating wide range of applications form lab scale to industrial apparatuses.