Numerical Simulation of Unsteady Turbulent Cavitating Flows

Computational Fluid Dynamics (CFD) is a very important tool for the study of complex fluid flows and the design of hydraulic fluid flow machinery. At the same time, experimental analysis is very difficult to perform. Thus, for a better understanding of the behaviour of such complex flows, including turbulence, unsteadiness and cavitation, a suitable knowledge of CFD is indispensable. Generally, the specific applications of CFD codes for solving this type of engineering problems are not well documented and a previous work for the acquirement of the CFD code capabilities is necessary. This work presents numerical investigation concerning complex unsteady flows, including turbulence and cavitation. The main objective was to acquire deeper knowledge about the software potentials for solving this kind of flows. To reach this objective several cases have been studied with a commercial CFD code (Fluent v6.1). The turbulence models being used were mainly the Spalart-Allmaras, the Standard k-ε, and the k-ω model [1], [2], [3], [4]. The utilized cavitation model was from Singhal, 2002 [5]. Based on long term considerations, this investigation aims at the application of the acquired knowledge and experience for further investigations relative to the cavitation phenomena in real fluid flow machines. Several steps were necessary to understand the suitable simulation process of unsteady turbulent cavitating flows. The case of an unsteady and turbulent, non-cavitating flow around a 2D circular cylinder was studied as a first step using different turbulence models at Reynolds numbers around the critical drag-crisis region. Compared with experimental data, the results are quite divergent, but similar numerical researches (J.S.Cox et al., 1997, [6], M.M.Zdravkovich, 1997, [7]) revealed comparable conclusions as does the present work. Mainly 3D effects are the cause of the non accuracy of the findings (e.g. P.D.Ditlevsen, 1996) [8]. As a second step, the cavitation phenomena has been studied in several applications. First, the full cavitation model implemented in Fluent (Singhal, 2002) has been tested comparing findings with corresponding experimental data. The first case was a steady, cavitating flow through a sharp-edged orifice by Nurick, 1976 [9]. Further, the unsteady turbulent flow around a 2D NACA 0015 hydrofoil has been simulated using the cavitation model. This work was based on the publications by Kubota, 1992 [10] and Berntsen et al., 2001 [11]. Results revealed that depending on the case, the cavitation model offers useful results, but only in a qualitative way. Accurate fittings with experiments are obtained only in few cases. The theoretical validity of the present cavitation model could be questioned. Future work consists of the prediction of damage caused by cavitation (comparing numerical results with experimental databases e.g. Escaler, 2001) using adequate software tools. The final goal is to apply the knowledge obtained to damage prediction in turbomachinery