Turbulent Closure for Sediment Transport Using Symbolic Regression Based on DNS Data
In session: MON 1.1 - Machine Learning I
Please login to view abstract download link
This work aims to improve the turbulence modelling in RANS simulations for particle-laden flows. Using DNS data as reference, the error of the model assumptions for the turbulence transport equation is extracted and serves as target data for a machine learning process called SpaRTA. The resulting corrective algebraic expressions are implemented in the RANS solver SedFoam-2.0 for cross-validation.