Multilevel optimization of profile of splitter blades in the impeller of a centrifugal compressor

In this research a multilevel optimization on the profile of splitter blades of a turbocharger compressor is performed using genetic algorithm in order to improve its performance. Successive corrections of profile at hub, midspan and shroud of splitter blades, with the objective of decreasing incidence losses at the leading edge and adjustment of blade loading at shroud, results in an impeller having improved splitter blades. The impeller flow filed analysis shows the optimization has been successful in reducing flow leakage at the shroud region and as well as losses in leading edge region. Although numerical simulations predict0.5% decrease in pressure ratio at design point, but 2.2 points improvement in isentropic efficiency is calculated. Based on the optimization results a new impeller is designed and manufactured and tested on a turbocharger test bed. Experimental results approve the simulation prediction results on the expected improvement in performance