Research

CFD and ML can form a mutually reinforcing workflow: ML can automate and accelerate CFD preprocessing, reducing the effort required to set up complex simulations. In turn, CFD provides high-fidelity training data that enables ML models to learn flow physics. Finally, ML can interact with CFD during simulation and optimization, or serve as a surrogate to replace it.