Arthur Cato
Thesis Title
Data-driven wall models for Large Eddy Simulation using symbolic regression
Research overview
The presence of solid walls can make Large Eddy Simulation computationally expensive due to the grid requirements to solve the small scales near them. The use of wall functions can decrease these requirements for numerical simulations; however, they are usually invalid or inaccurate for complex cases. To overcome some of the limitations of the existent algebraic wall models, the use of Machine Learning approaches based on high-fidelity data is studied. Particularly, a new methodology based on symbolic regression techniques using Gene Expression Programming is investigated.
Supervisors
Qualifications
M.Sc. Mechanical Engineering, University of Sao Paulo, Brazil (2021)
B.Eng. Mechanical Engineering, University of Sao Paulo, Brazil (2018)