CFD Analysis of Ventilation Flow for a Scale Model Hydro-Generator

In 2006, the first Computational Fluid Dynamics (CFD) simulations of the ventilation of specific hydro-generator components were performed at the Hydro-Que´bec Research Institute (IREQ) and lately the entire ventilation circuit is being investigated. Due to the complexity of flow calculations, a validation process is necessary and for this reason a 1:4 scale model of a hydro-generator has been built at IREQ to get experimental data by means of particle image velocimetry (PIV). This paper presents 2D and 3D simulation results for the scale model obtained with a commercial CFD code and addresses the challenges associated with the application of CFD to hydro-generators. In particular, the effect of rotor-stator interface (RSI) types and configuration is analyzed to determine the approach that best suits this application. Two-dimensional calculations show that the steady state multiple frames of reference (MFR) solution is highly sensitive to the type (frozen rotor (FR) vs. mixing plane (MP)) and location of the RSI. A parametric study is performed where each interface configuration is compared to the transient case results. The MFR-FR interface model produces results that may vary significantly depending on the relative rotor position and the radial location of the RSI in the air gap. The MFR-MP interface model appears to be more coherent with reference values obtained from a transient case, since the radial velocity profiles in the stator are similar. Furthermore with an appropriate radial positioning of the interface, the windage losses are within 20%. Simulations of the complete 3D ventilation circuit revealed a maximum variation of 10% in both total ventilation flow rate and total windage losses, between the RSI configurations studied. However, the relative flow distributions, normalized with respect to the total flow rate, are unaffected by changes in RSI configuration. This paper focuses mainly on sensitivity studies to numerical settings, but this comparison still requires experimental validation before any final conclusions can be made.