A Closed-Form Correction for the Spalart–Allmaras Turbulence Model for Separated Flows

Abstract

The field inversion and machine learning (FIML) approach is leveraged to obtain a closed-form correction for the Spalart–Allmaras turbulence model to improve predictions of separated flows. Based on field inversion results obtained using the first-generation FIML Classic approach, a simple and compact closed-form expression is chosen to be used as correction model. The thus obtained correction model is optimized using the second-generation FIML Direct approach. Training and validation cases consist of a selection of airfoils in a wide range of flow conditions as well as the flat plate. The correction model and results for the training and validation cases obtained with the augmented turbulence model are presented, demonstrating the improved flow predictions.