It is important to conduct research on geological processes to create novel computational technologies and addressing key energy and environmental concerns. Unconventional hydrocarbon recovery, geological carbon storage, and geothermal recovery are geological processes playing an important role in for the energy transition.

Despite recent advances in modelling techniques, geofluid processes face many challenges remaining to be solved due to the complexity of these systems and the requirement of fast simulation and optimization to accelerate decision-making process. Novel simulation approaches are required to resolve local details affected by multiple interacting factors, including rock heterogeneity, multiphasic flow, and coupled physics. Traditional physics-based modeling approaches towards such processes are often computationally expensive and require significant efforts for software/model development and maintenance. Meanwhile, data-driven approaches, such as reduced order modeling, machine learning, and deep learning, open new doors to efficiently predict fluid flow in porous media with strong nonlinearity and have gradually received widespread attention in the context of geological applications.

The aim of this Special Issue is to bring together original research and review articles discussing the algorithm development and application of data-driven modeling that are integrated with physics-based reservoir simulation field or experimental data related to geological processes. We are especially interested in submissions including geological processes such as unconventional hydrocarbon recovery, geological carbon storage, and geothermal processes. We hope that this Special Issue improves our current understanding of the topic to accelerate the decision-making in subsurface process management.

Potential topics include but are not limited to the following:

• Data-driven models to characterize geologic porous media with multi-scale heterogeneity
• Data-driven models to predict state variables (e.g., pressure, temperature, composition, and saturation) that evolve in both time and space as fluid is injected or produced from and to porous media
• Data-driven models to predict multiphase fluid injection or production rates (e.g., time series)
• Data-driven approaches to accelerate physics-based numerical flow solvers as preconditioners through tight coupling
• Data-driven modelling approaches to predict fluid flow in large-scale reservoir models
• Data-driven modelling approaches to predict fluid flow in fractured porous media with double or multiple porosities or permeabilities
• Real-time reservoir monitoring via transfer learning, meta-learning, or reinforcement learning
• Physics-informed machine learning, which is constrained by the governing physics of fluid flow in porous media
• Scalable optimization with data-driven modelling to manage oil and gas reservoirs
• Data-driven modelling applications for control and placement optimization
• Uncertainty quantification, history matching, and data assimilation based on data-driven modeling
• Novel AI-assisted pressure transient analysis (PTA), rate transient analysis (RTA), and decline curve analysis (DCA) approaches for unconventional reservoirs
• Data-driven physics-based surrogate modeling for fast and reliable production prediction or optimization