Physics-guided Analysis and Optimization of Clearance-induced Flow Dynamics in Centrifugal Pumps Using the Taguchi-Grey Relational Method
This study proposes a physics-guided optimization framework that integrates dimensional tolerance modeling, computational fluid dynamics, and Taguchi–Grey relational analysis to predict and improve the hydraulic performance of centrifugal pumps under realistic manufacturing variability. The framework establishes a quantitative link between geometric tolerances and flow-field evolution, thereby clarifying how clearance-induced deviations influence pressure redistribution, flow stability, and hydraulic energy transport. Four critical geometric tolerance parameters were evaluated using an L9 orthogonal array design. The numerical results identified an optimized blade-back clearance of 2.285 mm, corresponding to a hydraulic head of 19.588 m, an average flow velocity of 1.895 m/s, and a water horsepower of 1.824 kW. Analysis of variance further revealed that rear-shroud tolerance was the dominant factor governing pressure uniformity and hydraulic energy dissipation. Overall, the proposed framework provides a physics-based, tolerance-informed, and manufacturability-aware approach to hydraulic optimization by linking dimensional tolerance propagation, CFD-resolved flow behavior, and statistical performance ranking for clearance-sensitive pump design.