CPFD Software

CPFD Software Presents Machine Learning Research for CFD Simulation Acceleration

CPFD Software presented research, titled “Graph Neural Network Reduced-Order Models for Multiphase CFD: Fluidized Bed and Hydraulic Fracturing Applications,” at the ASME 2026 Fluids Engineering Division Summer Meeting. The study introduces a framework that uses machine learning to accelerate high-fidelity multiphase Computational Fluid Dynamics (CFD) simulations generated with the company's Barracuda Virtual Reactor software. The approach demonstrates prediction speedups of 80–105× for transient flow fields and up to three orders of magnitude for isolated snapshot predictions, supporting potential digital twin applications.