Quantitative reactive transport modeling of Portland cement in CO 2 -saturated water Bruno M. Huet *, Jean H. Prevost, George W. Scherer Department of Civil and Environmental Engineering, Princeton University, Princeton, NJ 08544, USA 1. Introduction The context of this work is the modeling of leakage of carbon dioxide (CO 2 ) from sequestration reservoirs. Different pathways for CO 2 migration (Duguid et al., 2005) might lead to significant CO 2 release from the reservoir and thus failure in its main goal: keeping CO 2 underground and away from Earth’s atmosphere in order to limit global warming (Bachu and Adams, 2003). Engineers and scientists agree that depleted hydrocarbon reservoirs and deep saline aquifers are the two most appealing locations for injection, because, in the former settings, the cap rock was sufficient to contain oil and gas over geological time and, in the latter environments, potential sequestration capacity is enormous (IPCC, 2005). On the other hand, the wells used to extract petroleum are now sealed with cement, which may be subject to corrosion by carbonic acid created during injection of CO 2 . Cement corrosion might turn existing wells into high permeability pathways for CO 2 leaks (Gasda et al., 2004). The potential CO 2 flow up cemented or plugged wells is likely to be strongly influenced by the chemical reactivity of cement, which may lead to sealing or widening of fractures within plugging or completion cement. On one hand, the release of calcium may lead to the formation of calcium carbonate and clogging of the annulus. On the other hand, calcium depletion within the cement and dissolution of its hydrates may lead to the formation of a high permeability silica gel layer, which seems to have poor mechanical integrity at room temperature and pressure (Duguid et al., 2005). The present generation of reactive transport codes is highly challenged by the range of mechanisms that have to be accounted for in order to accurately assess long term integrity of CO 2 storage: multiphase equilibrium and aqueous geochemistry as functions of both temperature and pressure, multiphase transport by diffusion, advection and dispersion, and heat transfer. An exhaustive introduction to reactive transport modeling applied to hydro- geochemical systems can be found in Lichtner et al. (1996). Those codes need to be modular (Van der Lee et al., 2003) so that software maintenance cost can be decreased and various mechanisms can be easily turned on or off in order to better understand single or coupled non-linear effects. Such a code, Dynaflow TM , is presented here and used to assess the reactivity of well cement in contact with a CO 2 saturated brine. The purpose of the present transport module is to understand the cement degradation mechanism for the most critical CO 2 leakage pathways along a well. Duguid et al. (2005) presented recently the different leakage pathways of CO 2 up a well. There are two main scenarios for which cement degradation could theoretically decrease zonal isolation: (i) CO 2 vertical mass transfer through the bulk of the cement plug or cement sheath that has no defect or (ii) CO 2 radial mass transfer from an existing leakage pathway in which a CO 2 rich fluid flows at high rate. For the first scenario, CO 2 mass transfer occurs both by diffusion and advection. However it would require geological time to reach the surface because (i) the International Journal of Greenhouse Gas Control 4 (2010) 561–574 ARTICLE INFO Article history: Received 5 February 2009 Received in revised form 15 November 2009 Accepted 17 November 2009 Available online 25 January 2010 Keywords: Cement Carbonation Reactive transport modeling Wellbore integrity ABSTRACT A modular reactive transport model, Dynaflow TM , is used to simulate the reactivity of cement in CO 2 - saturated water of intermediate salinity (0.5 M). Methodology for coupling transport and geochemical modules is derived and its assumptions are discussed. The modules are coupled in a sequential iterative approach to accurately model: (1) mineral dissolution/precipitation (2) aqueous phase speciation and (3) porosity-dependent transport properties. Simulation results reproduce qualitatively the dissolution of cement hydrates (CH, C-S-H, AFm, AFt) and intermediate products (CaCO 3 ) that have been observed experimentally. However, when using a standard power law to relate effective transport properties to porosity, modeling and experimental results do not coincide; here, agreement between simulations and observations is obtained by modifying the functional dependence of effective diffusivity on mineralogy. Furthermore, for this particular system for which concentration gradients are the only driving force, the assumption of neglecting the mass balance of water or density changes might show its limits. Therefore, future work should investigate the likely need to account for reaction-driven advection. ß 2009 Elsevier Ltd. All rights reserved. * Corresponding author. Tel.: +1 609 258 16 19; fax: +1 609 258 27 60. E-mail address: bhuet@Princeton.EDU (B.M. Huet). Contents lists available at ScienceDirect International Journal of Greenhouse Gas Control journal homepage: www.elsevier.com/locate/ijggc 1750-5836/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijggc.2009.11.003