Aerodynamic reconfiguration and multicriterial optimization of centrifugal compressors - a case study

This paper continues the recent research of the author, with application to 3D computational fluid dynamics multicriterial optimization of turbomachinery parts. Computational Fluid Dynamics has been an ubicuous tool for compressor design for decades, helping the designers to test the aerodynamic parameters of their machines with great accuracy. Due to advances of multigrid methods and the improved robustness of structured solvers, CFD can nowadays be part of an optimization loop with artificial neural networks or evolutive algorithms. This paper presents a case study of an air centrifugal compressor rotor optimized using Numeca's Design 3D CFD suite. The turbulence model used for the database generation and the optimization stage is Spalart Allmaras. Results indicate a fairly quick convergence time per individual as well as a good convergence of the artificial neural network optimizer.