Multi-point optimization on the diffuser of an axial flow pump

Abstract

For many pump applications, it is necessary to satisfy the performance requirements in more than one operating point. The conventional single-point design method which would cause a sharp decrease in the off-design point cannot fully meet such requirement. In this paper, an approach of the pump diffuser optimization is used to satisfy the performance in two points simultaneously. The three coefficient of the quadratic polynomial which is used to control the three inlet blade angles corresponding to the hub, shroud and the stream surface between (span wise=0.5) are selected as design parameters. Head, efficiency and power of the pump in the two selected point are selected as objective functions. The objective functions in the two selected points are in relations of trade-off. Design of experiments (DOE), steady CFD simulation, response surface method (RSM), Neighborhood Cultivation Genetic Algorisms (NCGA) are used to solve this problem. The DOE theory is applied to reduce the number of tests, three-dimensional simulations are performed to predict the pump performance, the RSM (response surface method) is used to correlate the pump performance to the intermediate variable, NCGA is used to search the pareto solutions along the response surface. The multipoint design optimization method is proved effective in searching the pareto solutions to satisfy the given requirements.