Zonal Eddy Viscosity Models Based on Machine Learning

A zonal k − ω model is constructed, with the zones created by training a decision tree algorithm. The training data are optimized, model coefficient fields. Coefficient data are binned, with each bin assigned a particular coefficient value. The zones are parameterized by training the machine learning model with a local feature set. The features are coordinate invariant flow parameters. It is shown that this model gives superior performance, compared to the base model, in the incompressible adverse pressure gradient (APG) flow test cases. The correction produced by the machine learning algorithm is self-consistent; i.e. once the solution converges, the zones remain fixed.