Abstract

Within the European context, CO2 storage operations shall address the potential impacts of largescale CO2 storage through risk assessment. The key risks identified for this onshore CO2 storage site were the migration through sub-seismic blind faults.To quantify the CO2 migration along such sub-seismic blind fault, the impact of the fault is investigated both for its mechanical and flow contribution.In this study, a blind sub-seismic fault was assumed near a CO2 injection well. The study area is in the eastern part of the Paris Basin and targeting the Lower Triassic Sandstone formation which is deemed adequate for CO2 injection. The arbitrary geometry of the fault (with limited throw c.a. 1 m), was assumed across the expected migration pathway of the injected CO2. The fault is assumed to extend vertically between the storage aquifer and a control aquifer.The flow-geomechanics model coupled with an uncertainties management computes the failure probability, i.e. the probability of mechanical failure in the fault vicinity. Such probability of failure is characterized by low to very low probability of occurrence which requires a large number of simulations to enable its evaluation. Each failure scenario models the CO2 migration from a storage horizon to a control aquifer when altering the mechanical parameters of the fault damage zones and fault core. To limit CPU requirements associated with the flow-geomechanics model, it is replaced by a fast surrogate model in order to evaluate the numerous simulations needed to evaluate the probability of failure. They show that limited CO2 migration is occurring within the fault but no breakthrough in the control aquifer. The injection induces limited pressure disturbance resulting in small ground level heave in the order of a millimeter at the end of the 30-year injection period at an average rate of about 0.8 Mtpa.For each geological layer, several uncertain mechanical parameters were considered to illustrate the approach: the permeability, the young modulus and Poisson ratio of the core and damaged zones of the fault, the shale content within the fault core, the critical gas saturation and capillary entry pressure of the fault damaged zones. The key response of interest was the safety factor based upon the Mohr-Coulomb failure criteria.An initial local sensitivity analysis identified that only four of the 18 uncertain input variables have a significant influence on the response of the model. A Sobol design of experiments was then launched with the four remaining parameters in order to construct a surrogate model. Gaussian process surrogates were chosen because of their capability of giving an approximation of the prediction error, which is useful to assess the quality of the surrogate along the domain of interest. Furthermore, the error of prediction can also be taken into account when computing probabilities. From this surrogate, it was concluded that there is a strong influence of the mechanical integrity of the base of the sub-seismic blind fault (Sobol Sensitivity Indices). Based on the surrogate model of the mechanical safety factor (elastic behavior i.e. no loss of integrity), a Monte-Carlo experiment was used to evaluate the probability of mechanical failure of the fault at the base of the storage cap rock.The most influential parameter on the mechanical safety factor is the anisotropy ratio between horizontal stresses. Within the considered domain of variations of the parameter, the unsafe domain (mechanical failure) may be reached at the upper bound of the uncertain domain and the failure probability was estimated to be lower than 10-3 for this case. A confirmation run with the detail numerical model at the upper bound of the uncertain domain confirmed that the mean behavior is inside the safe domain. Consequently, the uncertainty of the surrogate model was further reduced and showed no failure within the uncertain domain (failure probability below 10-6 for this case): the mechanical failure of the base of the sub-seismic blind fault is very unlikely.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call