Project Grant 2615649

The National Science Foundation Office of Integrative Activities awarded the University of Texas at Austin $364,836 on September 1, 2026, under the Geosciences program (CFDA 47.050) to develop graph neural network-based digital twins for subsurface multiphase flow modeling. The project will create artificial intelligence methods that integrate observations and models to improve understanding of fluid movement through geological formations, particularly where complex geology and limited measurements hinder traditional approaches. The research will produce efficient surrogate models capturing multiphase flow in complex geological formations, validated through physical experiments. Transfer learning will adapt the models to new operating conditions and geological settings with minimal additional data. The project will establish a scalable framework for uncertainty-aware prediction, data assimilation, and scenario exploration, with resulting methods applicable to groundwater contaminant transport, geothermal energy, and related subsurface geoscience problems. The award will fund student and researcher training at the intersection of artificial intelligence and geosciences, with AI models, software, and benchmarks openly shared to support industry applications. Performance runs through August 31, 2029, at the University of Texas at Austin in Austin, Texas. No sub-awards are planned.