Simultaneous Optimization of Impeller Blade Loading Distribution and Meridional Geometry for Aerodynamic Design of Centrifugal Compressor

Abstract

The present optimum design method has been advanced for simultaneous optimization of impeller blade loading distribution and meridional geometry. This is based on an aerodynamic design method and a genetic algorithm. The aerodynamic design method consists of two parts: a meridional viscous flow analysis and a two-dimensional inverse blade design procedure. In the meridional viscous flow analysis, an axisymmetric viscous flow is numerically analyzed on a two-dimensional grid to determine the flow distribution around the impeller and diffuser. Effects of blades onto the axisymmetric flow field are considered by a blade force modeling. In the inverse blade design procedure, 3-D impeller geometry can be obtained from the result of meridional viscous flow analysis and the predetermined blade loading distribution. In the optimization procedure, the total pressure ratio and adiabatic efficiency obtained from the meridional viscous flow analysis are employed as objective functions. As a constraint of the optimization, mass flux distribution at the impeller trailing edge is introduced in the evaluation procedure, in order to suppress the boundary layer development near the shroud, especially under low flow rate condition. Total performances and three-dimensional flow fields of centrifugal compressors have been analyzed by 3D-RANS simulations to certify effectiveness of the present design method. The 3D-RANS simulations and the flow visualization have been applied to a conventional centrifugal compressor and optimized design cases. From the analysis results, the performance enhancement of optimized designs is confirmed under low flow rate condition including design point. In addition to that, it is revealed that the constraint works effectively on the performance improvement. As a result, construction of the simultaneous optimization using the aerodynamic design method and the genetic algorithm is successfully achieved.