Fully coupled modeling of long term cement well seal stability in the presence of CO2

• Published in: Energy procedia Vol. 4; pp. 5162 - 5169
• Main Authors: Wilson, James C., Benbow, Steven J., Metcalfe, Richard, Savage, David, Walker, Colin S., Chittenden, Neil
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 2011
• Subjects: Cement; CO2; Coupled modelling; Risk assessment; Well seal; Well seal; Coupled modelling; Risk assessment; Cement; CO2
• ISSN: 1876-6102; 1876-6102
• DOI: 10.1016/j.egypro.2011.02.493

Evidence from experimental investigations, natural analogue studies and demonstration projects support the assertion that significant quantities of CO2 will not leak from a properly selected and managed CO2 storage complex. However, site operators must nevertheless justify the expectation of secure CO2 containment to regulators and other stakeholders. Secure containment will depend upon well seal stability inter alia. This paper presents fully-coupled geochemical models of well seal stability, focusing on cement degradation. The models have been developed with the ambition of reducing the simulated chemical system to the simplest possible configuration that is able to reproduce key features from experimental and field observations. Simplifying the models in this way makes them suitable for inclusion in larger scale system-level models of the storage system, where the use of highly-detailed chemical models can be impractical. Two sets of models were constructed in order to simulate: (1) a cement carbonation experiment (9 days in duration); and (2) field observations of cement degradation from the ‘SACROC’ site (30 years of reaction time). Although some model input parameters are uncertain, the experimental system was successfully simulated and the model was subsequently ‘up-scaled’ and applied to the SACROC core data. Over longer time-scales (102–103 years), the effects of parameter and thermodynamic data uncertainties on model output are more significant than at shorter time-scales. Although further work is required to bound the effects of these uncertainties, this paper illustrates the practicality of constructing simplified fully-coupled models to explore the significance of uncertainties by carrying out sensitivity calculations.