A Hybrid GNN-FEM Framework for Phase-Field Fracture Simulation. Physics-Preserving Hybridization for Generalizable Surrogate Modeling
Researchers have developed a novel hybrid framework combining Graph Neural Networks (GNNs) with Finite Element Methods (FEM) to accelerate phase-field fracture simulations. This approach integrates a GNN surrogate into the conventional staggered scheme, replacing the phase-field update at each load increment while retaining FEM for mechanical equilibrium. This selective surrogate strategy aims to reduce computational costs while maintaining accuracy and achieving strong generalization across diverse problem settings through dimensionless feature design and physics-informed loss functions. AI
IMPACT This hybrid approach could significantly speed up complex physical simulations, enabling more widespread use of advanced modeling techniques in engineering and materials science.
