Adaptive boundary-constrained mesh deformation and data-driven aerodynamic optimization of compressor blade rows

The mesh generation of computational domains is a relatively time-consuming step in computational fluid dynamics (CFD) simulations of multi-blade-row compressors. To reduce the time consumption caused by repeated mesh generation, mesh deformation methods have gradually been applied to the aerodynamic optimization of compressors. However, current methods often have difficulty in addressing disturbances on rotor/stator interfaces and periodic interfaces in multi-blade-row compressor configurations, which impacts the accuracy of CFD evaluations in aerodynamic optimization. To address these challenges, an adaptive boundary-constrained mesh deformation (ABCMD) method for multi-blade-row compressors is proposed. The ABCMD method performs mesh deformation based on radial basis function interpolation, while adaptively constraining the deformation magnitude of boundary meshes to reduce boundary disturbances. Integrated with a prescreening surrogate-assisted differential evolution algorithm and the ABCMD method, a data-driven aerodynamic optimization platform is developed and validated on the STAGE35 axial-flow compressor. The results demonstrate that, compared to conventional mesh boundary constraint methods, the ABCMD method reduces boundary constraint points and mesh deformation time by 82% and 84%, respectively, while maintaining computational error within 0.167%. After optimization, the adiabatic efficiency at the design point and the surge margin are increased by 0.75% and 0.16%, respectively. Compared to conventional optimization methods, the mesh generation time and the overall optimization time are reduced by 93% and 23%, respectively. The ABCMD method can accelerate the mesh deformation process while maintaining high accuracy in CFD simulations. Moreover, the developed optimization platform can realize the rapid aerodynamic optimization of multi-blade-row compressors.