Rapid Fault Leakage Screening for CO₂ Storage Under Uncertainty: Application to a Faulted Miocene Aquifer in the Malay Basin

Hariharan Ramachandran , Iain De Jonge-Anderson, Ana Widyanita, Uisdean Nicholson, Andreas Busch, Sebastian Geiger, Florian Doster

Geological CO2 storage is critical for mitigating climate change, yet the risk of leakage along faults poses significant challenges. Faults function as structural traps or preferential CO2 migration pathways depending on their sealing capacity. This study applies a computationally efficient rapid fault leakage analysis tool that combines a vertically integrated reservoir model with an upscaled fault leakage function based on steady-state flow approximations, conceptualising faults as zones of increased vertical and reduced horizontal permeability. The methodology is applied to faulted Miocene clastic aquifers in the Malay Basin, representing a complex, structurally heterogeneous setting typical of mature Southeast Asian storage prospects. The analysis evaluates 600 well locations to identify optimal injection sites, incorporating both deterministic scenarios (fully sealing versus leaking faults) and comprehensive uncertainty quantification across fault property heterogeneity. Comparison of sealing and leaking scenarios establishes physical controls governing leakage behaviour, while uncertainty quantification using 10,000 Monte Carlo realisations across 71 parameters (fault-specific permeability, capillary entry pressure, and transmissibility for 23 faults, plus reservoir properties) identifies the dominant controls on leakage outcomes. First, leakage sensitivity concentrates on a small fault subset: only 5 of 23 faults exhibit Spearman correlation with outcomes, with a single fault dominating system response. Second, k-medoids clustering identifies three distinct regimes, with high-leakage scenarios comprising 44% of realisations, the modal outcome rather than a rare event under current parameter uncertainty. Third, cluster membership is predominantly controlled by Fault 23 permeability, while Fault 1 transmissibility controls lateral access to the northern fault array, a cascade pathway that demonstrates the value of multi-fault uncertainty quantification. The findings demonstrate that targeted characterisation of 5-7 critical faults can effectively collapse predictive uncertainty, enabling transition from high-uncertainty screening to confident operational planning. The tool's computational efficiency enables rapid assessment supporting early-stage site screening, risk assessment, and targeted data acquisition, thereby accelerating the identification and prioritisation of secure CO2 storage sites.