
arXiv (Cornell University), Год журнала: 2023, Номер unknown
Опубликована: Янв. 1, 2023
Generalisability and the consistency of a posteriori results are most critical points view regarding data-driven turbulence models. This study presents progressive improvement models using simulation-driven surrogate optimisation based on Kriging. We aim for augmentation secondary-flow reconstruction capability in linear eddy-viscosity model without violating its original performance canonical cases e.g. channel flow. Explicit algebraic Reynolds stress correction (EARSCMs) $k-\omega$ SST obtained to predict secondary flow which standard fails capture. The is achieved by multi-objective approach duct quantities, numerical verification developed performed various test cases. testing new guarantee that preserve model. Regarding generalisability models, unseen demonstrate significant prediction flows streamwise velocity. These highlight potential enhance fluid simulation while preserving robustness stability solver.
Язык: Английский