Large eddy simulation of cavitating and non-cavitating flow

Doctoral thesis, 2008

Most marine configurations are subject to a number of complex and important hydrodynamic phenomena affecting the overall performance of the vessel, both regarding resistance and propulsive efficiency. In this thesis, the possibility of numerically simulating some of these phenomena is investigated. Due to the very wide range of flow scales present, i.e. the Reynolds (Re) number is high, in most marine applications, turbulence becomes an important factor. Another important issue, especially regarding propulsion, is cavitation. Cavitation occurs because the pressure in the flow is lowered below the vapour pressure and the liquid starts to boil. To be able to numerically predict these phenomena, using techniques that can be useful for the industry within a reasonable time frame, advanced physical modelling is needed. The models used in this thesis are incorporated in the incompressible Navier-Stokes equations (NSE) that forms the basic formulas for naval hydrodynamics. In principle, these equations can be solved for all scales in a Direct Numerical Simulation, but because of the high computational cost of for this technique, it will not be of use for industrial applications within a foreseeable future, instead a model for the turbulent scales is used. In this thesis a number of methodologies are investigated, with the main focus on Large Eddy Simulation (LES), but also using Reynolds Averaged Navier Stokes (RANS) and Detached Eddy Simulation (DES). These three methodologies, and subsets of them, are tested and validated on a number of configurations that are of interest for marine applications. The cases considered are the circular cylinder, the axisymmetric hill, a generic submarine hull, the Darpa Suboff configuration AFF1 and AFF8 and a number of hydrofoils in cavitating flow conditions. For the cavitating flow the same methodologies are used as in the single phase flow condition, but with the difference that the phase interface and the mass transfer from one phase to the other needs to be treated. The phase interface is taken care of using a Volume of Fluid (VOF) approach, where the interface is tracked using a volume fraction equation. The mass transfer process needs to be modelled since it occurs at length scales that are not supported by the continuum assumption, on which the NSE is based. The mass transfer models are incorporated in the NSE as source terms in the continuity equation and the volume fraction equation. The overall results of the simulations are very encouraging, including many physical phenomena that also can be seen in experiments. The results from the different methodologies and models are discussed, and possibilities and advantages are related. The main conclusion of the work is that LES will be a very useful tool for marine applications, especially for transient phenomena such as cavitation.