27 th Symposium on Naval Hydrodynamics Seoul, Korea, 5-10 October 2008 Vortical and Turbulent Structures Using Various Convection Schemes with Algebraic Reynolds Stress-DES Model for the KVLCC2 at Large Drift Angles Frederick Stern, Farzad Ismail, Tao Xing, Pablo Carrica IIHR-Hydroscience and Engineering The University of Iowa Iowa City, IA 52242-1585 ABSTRACT Vortical and turbulent structures for the KVLCC2 at a range of drift angles are studied using various convection schemes coupled with algebraic Reynolds stress detached eddy simulation (ARS-DES) turbulence models. The convection schemes and the turbulence models are evaluated quantitatively using rigorous verification and validation (V&V), including comparisons with available EFD data. For 0˚ drift, the integral forces are most accurately predicted by the fourth order (hybrid) interpolation scheme (FD4h) coupled with ARS but the local quantities (velocities and turbulent quantities) are best predicted by the second order TVD scheme with Superbee limiter coupled with ARS (TVD2S-ARS). For 12˚, the local quantities and the integral forces and moments are most accurately predicted by TVD2S-ARS. The vortical structures are also least dissipated when computed with TVD2S-ARS at 12˚ and 30˚ drift angles. Turbulent structures are analyzed using ARS- DES model with a pure FD4 scheme. The turbulent kinetic energy (TKE) and Reynolds stresses peak near the separation point at the bow and within a certain distance to the vortex core of all the vortices. Near the bow, turbulent structures are similar to those for a separated turbulent boundary layer and the recirculation region of a backward-facing step flows. Overall Reynolds normal stresses have similar distribution and magnitude compared to TKE. uw and vw are an order of magnitude smaller than uv , which has the same order of magnitude with the normal stresses. TKE and Reynolds stresses reach a local maximum right after the vortex breakdown points for helical vortex tubes and are intensified along the vortex core further downstream, likely due to the enhanced unsteady oscillation caused by the helical instability, which is consistent with previous studies on vortex breakdown for flows over a delta wing. INTRODUCTION Ship flows are challenging to computational fluid dynamics (CFD) due to unique physics and application conditions, ranging from resistance and propulsion to general six degree of freedom ship motions and maneuvering. Of interest herein are vortical and turbulent structures for ship flows at large drift angles, which cast challenges to develop robust and accurate convection schemes and advanced turbulence models. Most finite difference/volume convection schemes suffer from artificial diffusion and phase errors. In ship hydrodynamics, these errors cause undesirable effects such as smoothed and/or shifted free surface waves and also under prediction of vorticity, yielding inaccurate prediction of integral quantities such as forces and moments, and local quantities such as propeller plane velocities. Very high order accurate (3 rd order and above) convection schemes can be used to reduce numerical diffusion and phase errors. Di Masscio et al. (2008) evaluated several convection schemes on a NACA0012 airfoil and DDG51 surface combatant and showed that high order schemes are more accurate than low order schemes but at the expense of losing computational robustness. To achieve stability without compromising the solution accuracy, most ship hydrodynamics computations rely on second order convection schemes as witnessed in Tokyo 2005 CFD Workshop (Hino, 2005). However, different second order convection schemes have different magnitudes of numerical diffusion and phase errors, thus each scheme may produce results that vary significantly (Jasak et al., 1999). More importantly, the accuracy of the convection schemes depends on the grids used,