During natural gas injection and withdrawal, faults may cause lateral leakage, leading to gas loss or migration. Thus, fault lateral sealing is critical for safe gas storage operation. This study uses the L gas reservoir as a case to analyze fault dip, reservoir thickness, and overburden thickness. Logging data are integrated into the fault-sealing triangle diagram, confirming that lithological contact and fault rock sealing dominate in this reservoir. A quantitative model evaluates fault lateral sealing capacity based on the original gas reservoir's anatomy. Dynamic development data reveal pressure variations across fault sides during injection-production phases. A numerical simulation incorporating time effects assesses dynamic sealing capacity, with end-of-production pressure differentials selected for analysis. The fault-sealing model developed from this data explains sealing mechanisms and factors (fluid pressure, tectonic stress, fluid property changes) influencing pressure differences. Machine learning-assisted history matching of 12 injection-production cycles achieves 89% prediction accuracy (R2 = 0.89), exceeding conventional SGR-AFPD regression methods by 32%. The study highlights lateral sealing capacity discrepancies among stages and clarifies reasons for pressure difference variations, providing a strong theoretical basis for future gas storage assessments.
Meng et al. (Wed,) studied this question.