Numerical simulations of fluid flow
Abstract
In this project we will do highly resolved computational fluid dynamics simulations (CFD) for several applications. Depending on the requirement on the resolution of the turbulent flow structures, we are using and developing all levels of turbulence modeling, i.e. highly resolved DNS (Direct Numerical Simulation), LES (Large Eddy Simulation), URANS (Unsteady Reynolds-Averaged Simulation), and hybrid LES/URANS models such as DES (Detached Eddy simulation). In cases when we are studying multiphase flow we use LPT (Lagrangian Particle Tracking) and VOF (Volume Of Fluid) methods. All of these methods require lots of computational resources, in particular when applied to realistic cases with high Reynolds numbers. Examples of application areas, that will be studied using the current computational resources, are given here:
1: Hybrid LES/URANS simulations of flow in the U9 Kaplan turbine model, with rotor-stator interaction through a sliding grid approach, at several operating conditions, and including clearance effects. This is particularly challenging due to the high Reynolds numbers and wide range of geometrical scales that need to be resolved. We have been part of the development of the functionalities for such simulations since 2007, and since a number of years we have a code that is validated and that runs efficiently in parallel.
2: Active flow control in swirling flow with rotor-stator interaction, using LES and hybrid LES/URANS models. Here we are doing fundamental research on how to mitigate large pressure pulsations by adding water jets. It is required to resolve the jets sufficently well, which is why we use LES and a well-established test-case at a slightly lower Reynolds number.
3: Coupling between 1D system simulations and 3D VOF simulations with free surfaces, using an LES turbulence model closure. As more of the dynamics of the flow is resolved, the interaction with the surrounding system needs to be taken into account. We are going to study transients between different flow conditions, and for that we are developing and testing the 1D-3D approach. The 3D part of the problem is very computationally demanding.
4: URANS simulations of cooling air flow in electric generators, with rotor-stator interaction through an overlap sliding grid approach. This case is special in terms of its highly complex geometry that needs to be resolved with large computational meshes. We are additionally evaluating the overlapGgi methodology with the aim of reducing the computational requirements.
5: Studies of transients between different flow conditions in a Francis turbine. This requires a detailed resolution of the flow details, using hybrid LES/URANS methods, and very long time scales related to those of the transient processes.
7: LES of diffuser flow with fibers. This is a fundamental study of fiber flocculation and rheology, requiring a very detailed resolution of the turbulence.
8: Fluid-structure interaction of the blood flow in the neck during whilpash motion. The interaction between the blood pressure and human tissues, at approximately the same density, requires a strong coupling that is extremely time consuming.
