Towards Fast GNN Surrogates for CO2 Migration in Complex Geological Formations
Researchers have developed a graph neural network (GNN) surrogate model to simulate CO2 migration in complex geological formations. This data-driven approach aims to replicate key physical behaviors of multiphase flows, offering a faster alternative to traditional simulation methods. The model was evaluated on the SPE11A benchmark, demonstrating competitive forecasts for gas saturation and liquid-phase density, crucial for monitoring CO2 storage. AI
IMPACT This research demonstrates the potential of GNNs for accelerating complex scientific simulations, which could impact fields requiring detailed geological modeling.