원심 펌프의 임펠러에 대한 혹등고래 결절 형상 적용 및 최적화

Abstract

In this study, bio-mimetic tubercle shapes of humpback whale are applied to the impeller of a commercial centrifugal pump based on RANS simulations and design optimization. In specific, a few geometric parameters of tubercles are employed as the design variables of the trailing and leading edges of the impeller. Design optimization is performed based on a sampling approach and the radial-basis artificial neural network. As the objective function, a combination of the turbulent kinetic energy (TKE) and pump efficiency is used based on the effects of tubercle shapes found in previous studies. Pump performance and flow fields are compared between the optimized and original impeller designs. The results show that the optimized tubercle design at the impeller trailing edge affects the vortical structures significantly. Tubercles generate counter-rotating vortices in wake region of the impeller, which leads to vortex breakdown and a decrease of turbulent fluctuation by about 25%. Also, a trade-off between the TKE and pump efficiency is observed. In contrast, the optimized tubercle design at the impeller leading edge results in relatively smaller (about 3%) reduction of the TKE. While the leading edge tubercles suppress flow separation at the suction side significantly, the effect is localized near the leading edge. For both leading and trailing edge optimization, the changes of the pump efficiency and head are found to be relatively minor.