CO2 storage in subsurface formations, such as saline aquifers, has emerged as a promising alternative to reduce greenhouse gas emissions and mitigate their impacts on global warming [1]. Ensuring the reliability of CO2 leakage risk assessment remains critical in the context of carbon capture and storage (CCS) technologies and raises significant concerns for predicting long-term behaviours. Various potential sources of leakage, such as geological faults and fractures or mineralogical changes of the rock matrix due to the acidification of the medium, may compromise the integrity of the seal and caprock. On the one hand, fault damage zones characterized by a fracture network can become highconductive flow pathways, where the permeability needs to be investigated to assess fault-related leakage rates [2]. On the other hand, the mineral reactivity of the reservoir structure with the injected CO2 may result in mainly carbonate dissolution under acidic conditions, locally impacting the flow paths, porosity and permeability [3].

Investigating the effects of these two phenomena is, therefore, crucial to ensure reliable management of CCS facilities. However, several uncertainties are associated with modelling these geochemical and structural mechanisms at the reservoir scale. Subsurface uncertainties arise from missing geological features and data sparsity regarding macro-properties distributions such as permeability and porosity. This necessitates sensitivity analyses in terms of the macroscopic description of the porous structures to ensure reliable management of CO2 storage in natural reservoirs. However, the parametrization of geological distributions is often left to the user's discretion [4].