Turbulent shear flows in naval applications are characterized by vastly different lengths and time scales associated with rotor tip vortices and the vortical structures shed from the ship, and additional phase from water drops and water vapor. To tackle the modeling challenges, we propose a novel methodology that combines a vorticity preserving method and a new approach to LES turbulence modeling that is grid-spacing-independent and discretization-order-independent, called Explicitly Filtered Large Eddy Simulation (EFLES). The EFLES concept has been shown to work for both single-phase and two-phase flows, which is important for the Navy because of the spray from the surrounding sea water. For single-phase, it was found that the subgrid scale (SGS) model did not make any difference in accuracy, provided that an SGS model is used. For two-phase flow, however, the type of SGS model may matter. The advantage of EFLES is that one could use very large filter-widths since the LES grid is not related to the filter width. Hence, one could capture predictability and accuracy independent of the discretization order and number of LES grid points, which is an enabling force due to the significant reduction in computational cost.
Benefit: The successful completion of the overall project will lead to the development and validation of a methodology and a computational tool for vortex-dominated two-phase flow for naval applications. The methodology, that couples vortex-preserving capability and grid-spacing independent as well as discretization order independent turbulence model, is significantly more compute-efficient than the current state of the art technology and provides a breakthrough in predicting flows with vastly disparate length scales. The naval relevant wake interaction problems cannot be treated by conventional methods, such as free-wake, vortex particle method (due to turbulence), and vorticity-velocity formulation (due to the presence of particulate phase which requires compressible formulation). Even conventional LES fails to model growth of small scale turbulent structures in a computationally tractable manner, because varying grid resolution to conform to changing length scales renders conventional LES solution unreliable as the method is grid-spacing dependent. Moreover, conventional CFD suffers from numerical dissipation on coarse grids. The new computational tool will exactly counteract this dissipation by using adaptive vortex confinement in the vortex source region and grid-spacing independent subgrid modeling in the transitional and turbulent regions. The discretization-order independent property of the method also enables accurate and reliable prediction of practical CFD codes, which are commonly of second order. This enabling force is also desired in many non-naval applications involving growth of small scale structures and multi-scale interactions in single and multiphase flows, such as particle dispersion (volcanic flow, urban CFD), atmospheric flow (cloud formation), plume signature, etc.
Keywords: discretization-order independent, discretization-order independent, Explicitly Filtered LES, Large Eddy Simulation, ship-airwake rotorwake interaction, Filtered Coarsened DNS, grid-spacing independent, Adaptive vorticity confinement
