Revisiting field estimates for carbon dioxide storage in depleted shale gas reservoirs: The role of geomechanics

• Published in: International journal of greenhouse gas control 2021-02, Vol.105, p.103222, Article 103222
• Main Authors: Xu, Shiqian, Ren, Guotong, Younis, Rami M., Feng, Qihong
• Format: Article
• Language: English
• Subjects: CCS field estimates; CO2storage; Coupled hydro−mechanics; Flow in fractured porous media; Shale gas
• ISSN: 1750-5836; 1878-0148
• DOI: 10.1016/j.ijggc.2020.103222

[Display omitted] •Revised published field-estimates for CO2 storage in depleted shale gas reservoirs.•Incorporated geomechanical deformation and aperture dependent constitutive models.•BIC shows proposed model is equally plausible to the previously published models.•Forecasted storage capacity is revised upwards by two to three folds.•Forecasted injection rate is revised upwards by two to three folds. The prospects for CO2 sequestration in depleted shale gas formations are widely studied. Reported studies offer quantitative predictions of ultimate storage capacity as well as CO2 injection rate for a typical injection site or on an areal basis. Reported estimates for the ultimate storage capacity in various shales vary from 10 Gt to 50 Gt. In recent analysis, a numerical hydrodynamic model, neglecting geomechanical effects, was calibrated to historical gas field production data from various shales, and applied to estimate both injection capacity and rate. The projected site injection rates fell up to two orders-of-magnitude short of anticipated capture rates. It is hypothesized that geomechanical deformation, particularly within fractures, under depletion and pressurization can lead to significant variation in production and injection rates. This work revisits the field estimates by utilizing a coupled hydrodynamic and poromechanical model. History matching of field production data is performed using Ensemble Smoother with Multiple Data Assimilation (ES-MDA), and a Bayesian model-selection approach is applied. The results indicate that while both models are equally viable from the perspective of Bayesian model-selection, the coupled model estimates indicate a two-fold increase of capacity over those of the flow model. Moreover, the projected injection rates incorporating deformation are enhanced by two-folds at early times and by up to five-folds at later times.