The generation of hydrogen in-situ from hydrocarbon reservoirs has emerged as a carbon neutral technology for fossil fuel-based hydrogen production. This technology has been extensively investigated for heavy oil reservoirs through in-situ combustion gasification. This study proposes in-situ hydrogen generation from depleted gas reservoirs and assess graphite gravel packing for selective hydrogen production with underground carbon storage. The viability of this hydrogen generation process was accessed through process simulation, followed by experimental investigation and molecular simulation of the selective production of hydrogen through graphite. Equilibrium and kinetic models reproduced measured effluent fractions, confirming their reliability. The simulation outcomes reveal that higher temperature and steam-to-carbon ratio increase hydrogen yield/purity, whereas high pressure favors methanation. This necessitates elevated temperatures beyond the usual reaction temperature under reservoir conditions. Longer residence time and judicious catalyst loading improve conversion while limiting diminishing returns. Adiabatic simulation yields lower hydrogen purity than isothermal but better reflects field behavior. Reservoir mineralogy governs outcomes as quartz-rich rocks inhibit hydrogen production by steam reforming, while clays/feldspars reported elsewhere can be catalytic. The experimental results showed that graphite can be used as gravel pack in the production well to produce hydrogen and retain carbon dioxide underground. Literature report indicates that high compaction can further enhance separation significantly reducing the carbon emission associated with hydrogen production from fossil fuels.

Accurate identification of fracture geometry in hydraulic fracturing is essential for understanding fracture propagation, optimizing stimulation design, and predicting production performance. Distributed acoustic sensing, as a high-resolution near-wellbore monitoring technique, provides rich spatiotemporal data for real-time observation of fracture responses. However, reconstructing fracture geometry from distributed acoustic sensing measurements remains challenging due to high model dimensionality, ill-posed inversion processes and substantial computational costs. This study presents a fracture geometry inversion framework based on radial basis function, in which the fracture width distribution is represented using a small number of radial basis function modes. Owing to the intrinsic smoothness and symmetry of radial basis function, the method eliminates the need for explicit regularization terms, thereby simplifying the objective function and improving inversion stability. This approach significantly reduces the number of inversion parameters while enhancing both accuracy and physical consistency. Applications to a synthetic benchmark model and real field data from the hydraulic fracturing test site demonstrate that the radial basis function-based method consistently outperforms conventional fullparameter inversion approaches, in terms of fitting accuracy and computational efficiency. The proposed method provides a structurally informed and computationally efficient modeling framework for high-dimensional fracture inversion, offering a promising solution for real-time fracture monitoring and parameter estimation in hydraulic fracturing operations.