Point-particle drag, lift, and torque closure models using machine learning: Hierarchical approach and interpretability

Point-particle closure models that are utilized in Euler-Lagrange simulations play an important role in replicating true dynamics of particle-laden flows. The accuracy of these point-particle models depends on how well they incorporate the local microstructural information of neighboring particles. The current work presents a physics-based hierarchical machine learning approach for developing robust N-body closures. The inclusion of ternary interactions, in addition to binary interactions, enabled by the hierarchical approach leads to improved predictions.