Abstract
The results of two three-dimensional CFD models of the 0.6 m diameter, 0.6 hub-to-tip ratio Impulse Turbine are compared qualitatively and quantitatively with experimental data for a geometrically similar turbine under similar conditions. One model uses hexahedral cells exclusively to discretise the computational domain, the other uses a hybrid meshing scheme, utilising hexahedral cells at the rotor blade and the guide vanes, tetrahedral cells in the bulk of the remaining computational domain and pyramidal cells to mediate the interface between these two types of cell. Quantitatively, there is a visible difference in the predicted non-dimensional pressure drop across the turbine. The main qualitative difference between the results of the two models is a large region of separated flow extending from the tip of the rotor blade, on the pressure side, to roughly the 50% radial station. This is predicted using the hybrid scheme and that is essentially absent in the hexahedral model. Simple design calculations are used to illuminate the flow regime presented to the rotor by the guide vanes and conclusions are drawn as to the performance of each grid in predicting performance and providing insight into the fluid flow. Both Models produced results in reasonable agreement with experiment. The unstructured, hybrid grid predicted a pressure drop that was greater than the structured hexahedral grid and both grids underpredicted pressure drop to some extent. The hybrid, unstructured grid quality was higher than that of the structured all hexahedral grid. The hybrid, unstructured grid was somewhat easier to create than the structured, all hexahedral grid.