Numerical methodology to predict and analyze cavitating flows in a Kaplan turbine

Abstract

A computational methodology to predict the appearance and the evolution of the cavitation phenomena in a scale model of a 5–blades Kaplan turbine was developed. The cavitating flow was modeled by a homogeneous approach with a barotropic state law and the k − ω SST turbulence model. RANS simulations were performed, near the Best Efficiency Point, on two computational domains (D1 and D2) comprising one periodic interblades channel consisting of the guide vanes, the runner blade and the cone (domain D1) or taking into account the entire draft tube (domain D2). At first, non–cavitating simulations were carried out in domain D1 in order to match numerical conditions to the experimental ones (similar mass flow Q, head H and mainly the torque T) via an iterative procedure. From this reference operating point obtained in free–cavitation regime, cavitating simulations were performed on both computational domains by reducing the Thoma number, σ. Two different inlet boundary conditions were tested: the classical one, imposing the mass flow rate Q; the new one, fixing the total pressure and, consequently, keeping the machine head H constant during the cavitation drop, as in experiments. In all the performed calculations, the outlet boundary condition is given by a constant static pressure. Computed torque and efficiency drop curves were compared to available experimental data. The best results were obtained with the computational domain D2 applying a constant total pressure on the inlet. The torque and efficiency evolutions were well–predicted with the proposed calculation methodology, and the numerical cavitation structures agreed with experimental observations. Analyses of the blade loading during cavitation breakdown are also proposed in the article. Unsteady simulations are under investigation to improve the prediction and the analyses of more developed cavitating regimes.