Researchers have developed a novel multimodal auto-regressive transformer surrogate model designed to enhance the efficiency of geological carbon storage operations. This model effectively simulates variable well perforation and injection strategies while accounting for geological uncertainties. By processing multiple input modalities, including the geomodel, relative permeability functions, and control variables, the transformer predicts key metrics such as saturation, pressure, and CO2 mass. The surrogate model was successfully integrated into a Markov chain Monte Carlo data assimilation procedure, demonstrating significant uncertainty reduction for critical metaparameters like fault permeabilities. AI
IMPACT This model could lead to more efficient and safer geological carbon storage operations by better predicting and managing injection strategies under uncertainty.
RANK_REASON Academic paper detailing a new AI model for a specific scientific application. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
