Turbulence Modelling of the Flow Around a Prolate Spheorid

In numerous engineering disciplines, including naval and offshore engineering, the accurate prediction of turbulent flows is essential for cost-efficient and safe designs. For industrial applications, the turbulence must be modelled, and the development and understanding of appropriate turbulence modelling for a given type of problem follows an extensive set of case studies. The present thesis aims to be a contribution to that endeavor with respect to separating bluff body flows. This thesis evaluates the ability of a set of RANS, hybrid and LES turbulence models to predict the flow around the 6:1 prolate spheroid at 45 incidence at Re = 16000 based on the minor axis. The flow configuration under investigation is the topic of an ongoing series of DNS studies at NTNU Trondheim, where preliminary results at Re = 16000 were used for model validation. This bluff body flow is a challenging turbulence modelling test case, which contains complex features as laminar-turbulent transition, flow separation, unsteadiness and an asymmetric wake. Thus, the test case is well suited to investigate shortcomings of different models and to drive model development. All simulations were performed with the CFD software OpenFOAM, and the models were verified and validated (V&V) with the standardized procedures in ASME (2009) and Hills (2005). Because of issues with numerical stability, the V&V procedures could only be followed for RANS and hybrid models, whereas a less formal evaluation was performed for the LES models. The RANS models and hybrid models, which required numerical damping, exhibited a steady and symmetric flow about the meridional plane. The Smagorinsky LES model was able to predict the wake asymmetry, but it was less severe than that seen in the DNS study. Based on the stabilized and the undamped results prior to instability, unsteadiness and turbulence anisotropy at the near- body are assessed as critical aspects of the flow. These features of the flow are seen as essential in developing wake oscillations of sufficient magnitude to establish the wake asymmetry. Accordingly, hybrid models with an anisotropic RANS mode and LES models are recommended for modelling of this configuration and similar ones. Due to the shortcomings of this study and the continuous need for further V&V studies of bluff body flows, possible methodological improvements for a future V&V study of this flow configuration are discussed.