Effects of interface model on performance of a vortex pump in CFD simulations

That the predicted head of a vortex pump is higher than that measured experimentally is very common in simulations of turbulent flow in such pumps. To identify why, reported here is a study of the turbulent flow of water in a vortex pump with a specific speed of 76 and fluid domains with 1/8-impeller and whole-impeller geometrical models and smooth walls using the 3D steady Reynolds-averaged Navier–Stokes equations, the standard k–ɛ model, and a scalable wall function in ANSYS CFX 2019 R2. The results show that the aforementioned phenomenon is related to the choice of interface model. With the 1/8-impeller model, the head predicted by the frozen rotor model agrees with the experimental head. By contrast, the transient rotor model provides a reasonably accurate head against the experimental head but requires huge computing resources and overestimates the pump efficiency, and the stage model is unsuitable for predicting the head of the pump. The flow patterns in the vaneless chamber and impeller predicted with the 1/8-impeller model are more uniform because of artificial fluid mixing on the interface than those predicted with the whole-impeller model by using the frozen rotor model, and the flow patterns predicted with the whole-impeller model by using the transient rotor model are in between. The hydraulic performance of the pump is predicted with the 1/8-impeller model and frozen rotor model at various viscosities, and the flow-rate, head, and efficiency correction factors are determined and correlated with the impeller Reynolds number.