Prediction of subsurface flow and transport is essential in many energy and environmental applications such as enhanced hydrocarbon recovery, geological storage of CO₂, and groundwater remediation. Owing to the intrinsic spatial heterogeneity of the subsurface environment and the non-linearity of governing equations of fluid flow, prediction of subsurface flow using high-fidelity computational fluid dynamics techniques becomes challenging in terms of computational complexity and cost. Data-driven and machine learning tools can potentially tackle these challenges by offering a computationally efficient alternative to physics-based models.

This session aims to gather researchers and professionals being involved in the advancement of machine learning-based proxy models with the focus on forward and inverse problems related to subsurface flow and transport. We highly encourage studies on scientific machine learning frameworks such as physics-constrained deep learning algorithms, which incorporate scientific computing and data-driven models in subsurface flow problems.