Robust Optimization Design of Centrifugal Pump Impeller Considering Geometric Uncertainty

Abstract:

Aiming to address the problem of random performance fluctuations of centrifugal pump impellers caused by geometric uncertainties during design, manufacturing, and operation, a robust optimization design method of centrifugal pumps considering geometric uncertainties was proposed. A parameterized model of blade geometric uncertainty was constructed based on the Gaussian process-principal component analysis (GP-PCA) method, and the influence of blade geometric deviations on head and efficiency was quantified by using the non-intrusive polynomial chaos (NIPC) method. Combined with the radial basis function (RBF) neural network surrogate model and the non-dominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ), a robust optimization design of the centrifugal pump was carried out by taking the inlet blade angle, outlet blade angle, and blade wrap angle as design variables. The results showed that under random geometric deviations of ±3%, both the efficiency and head of the prototype impeller were decreased under design conditions, with the decrease being more pronounced under low flow rate conditions. After robustness optimization, the head and efficiency were significantly improved compared with the prototype pump, and the head variance and efficiency variance were greatly reduced. This improved the impeller's robustness to random geometric deviations while enhancing its hydraulic performance, and had certain guiding significance for the stable and reliable operation of the pump.