Artificial-neural-network-based subgrid-scale models in the strain-rate eigenframe for large-eddy simulation of compressible turbulent channel flow

Artificial-neural-network-based (ANN-based) subgrid-scale (SGS) models for turbulent channel flow often suffer from instability and poor generalization. Here, we propose an ANN-SGS model based on the strain-rate eigenframe and apply it to large eddy simulations of compressible turbulent channel flow. Our results indicate that the newly proposed model can predict flow statistics more accurately than traditional SGS models, and it also exhibits generalization capability for both Reynolds and Mach numbers.