Quantitative risk assessment for scenario of CO2 migration through fault within a large-scale geological storage site

Within the framework of the European directive on CO2 storage for instance, risk analysis of the CO2 storage behavior is a key element to establish the confidence with the storage integrity over time. The MANAUS (Methodology of ANAlysis Unified and of management of risks of geological Storage of CO2) aims at enabling the security of future storage developments and operations. The MANAUS project developed a rational and operational methodology of analysis and of management of risks for the CO2 geological storage within the French regulatory context. A methodological framework was built, starting from the identification of risks scenario deviating from the normal evolution of the storage. The generic methodology is applied to different potential sites of CO2 storage to assess one failure scenario for each site. The paper discusses the approach implemented for one of these scenarios aiming at quantifying the CO2 migration along a fault from a large-scale storage. To that end flow modeling and uncertainties management commercial codes were coupled to quantify the failure probability i.e. the probability of CO2 migration towards a control aquifer. Such probability of failure is characterized by low to very low probability of occurrence which requires quite a large number of simulations to enable its evaluation. Each failure scenario models the CO2 migration from a storage aquifer to a control aquifer when altering the flow properties of the fault zone. Due to the computation costs of the simulation model, surrogate models are computed based on a polynomial chaos approach and a limited set of simulations. A random variable of interest, for example the CO2 migration in the control aquifer, can be expanded on a polynomial base of the input random variables such as properties of the aquifers and fault. This surrogate model is assessed using a design of experiments and approximates the real model. Then, different fault failure analyses are performed on these response surfaces using either Monte-Carlo simulations or approximation methods such as FORM/SORM or 2 SMART. Different concepts for fault modeling were implemented to assess the impact of the modeling approaches (3D discretized fault with a compositional reservoir simulator v.s. 2D fault with a two phase flow model). The 3D fault model relies on the Shale Gauge Ratio concept for the fault core and includes the damaged zone around the fault Besides the Shale Gauge Ratio, the capillary and pore entry pressure models fitted to the available literature data control the CO2 migration within the fault. The 2D model proposes alternative relations between the pore entry pressure and van Genuchten capillary pressure parameters. In both model, intrinsic permeability and porosity into the damaged zone are assumed to progressively decrease with the distance from the fault core down to the host rock values. To establish the surrogate model for the 3D modeling approach, a reduced set of about fifty simulations was used. For the 2D model, a simpler approach was selected to verify the consistency of the simulation modeling approaches. Both fault models lead globally to similar results: no CO2 breakthrough but some pressure disturbance in the control aquifer. CO2 migrations into the fault zone never reach the control aquifer even for the most permeable assumptions in both 2D and 3D fault modeling approaches. The surrogate models, build from the 3D approach, show that pressure disturbance in the control aquifer appear in about 30% of the cases i.e. when the fault properties allow some pressure transmission between the injection zone and the upper aquifer. The computation methodology implemented to model this migration scenario may readily be applicable to estimate the failure probability of the different risks identified for a CO2 geological storage and its applicability of risk assessment methodologies