Article | Open

Reactive Transport Modeling of the Enhancement of Density-Driven CO2 Convective Mixing in Carbonate Aquifers and its Potential Implication on Geological Carbon Sequestration

  • Scientific Reports 6, Article number: 24768 (2016)
  • doi:10.1038/srep24768
  • Download Citation
Published online:


We study the convection and mixing of CO2 in a brine aquifer, where the spread of dissolved CO2 is enhanced because of geochemical reactions with the host formations (calcite and dolomite), in addition to the extensively studied, buoyancy-driven mixing. The nonlinear convection is investigated under the assumptions of instantaneous chemical equilibrium, and that the dissipation of carbonate rocks solely depends on flow and transport and chemical speciation depends only on the equilibrium thermodynamics of the chemical system. The extent of convection is quantified in term of the CO2 saturation volume of the storage formation. Our results suggest that the density increase of resident species causes significant enhancement in CO2 dissolution, although no significant porosity and permeability alterations are observed. Early saturation of the reservoir can have negative impact on CO2 sequestration.


It has been widely recognized that a significant reduction of CO2 emissions is necessary to maintain atmospheric greenhouse gas concentrations at around 450 ppm CO2 equivalent, thus limiting the effect of anthropogenic climate change1. Reduction targets suggest a 30% reduction of 1990-level by 2020 and even up to 80% by 2050 (commission of the European communities, 2007). The feasible technology to achieving targets is Carbon Capture and Storage (CCS). In general, injection of CO2 under supercritical conditions into geological formations such as saline aquifers has been proposed to sequester CO2 over a long period of time2,3. Deep saline aquifers are estimated to have the greatest capacity for CCS, 103–104 gigatons4.

The injection of a reactive substance such as CO2 into the saline formation results in chemical disequilibration and initiation of various chemical reactions. It is important to understand the direction, rate, and magnitude of such reactions, both in terms of their impact upon the host rocks’ ability to contain the injected CO2 and in the long-term stability of CO2 containment5. Structural, residual, solubility, and mineralogical trapping are the dominant mechanisms by which the injected CO2 is contained6,7,8,9. Among these, mineralogical trapping is the only mechanism that permanently sequesters the CO2 while others are regarded as storage processes10. This permanent sequestration ensures long term stability.

After injection, mass transfer at the interface between CO2 and brine occurs by molecular diffusion of CO2 into brine. A 2–3% density increase of brine-CO2 solution leads to gravitational instability. Under favorable conditions (greater than critical Rayleigh number, Ra), natural convection enhances dissolution further11. An increase in dissolved minerals further increases the density of the formation water which, in turn, enhances natural convection. In addition, precipitation and dissolution of minerals may lead to an alteration in porosity and permeability of the host formation. Consequently, convective mixing and, therefore, CO2 solubility trapping are influenced.

Several studies including a few laboratory tests have been performed regarding convective mixing from perturbed saturated boundary11,12,13,14,15,16,17,18,19. However, the geochemical effect to this end is modeled to a lesser extent. Ennis-King and Paterson20 investigated analytical and numerical methods and observed that the overall dissolution rate depends on the balance between the effects of permeability alterations and the consumption of dissolved CO2. They showed that the influence of ion concentrations (e.g., Ca+2, Mg+2) on the fluid density alters the plume structure and favors faster dissolved plume development. Ghesmat et al.21 used linear stability theory and direct numerical simulation to show that geochemical reactions stabilize the unstable diffusion boundary layer because of the consumption of dissolved CO2. Their results implied that more CO2 can be trapped through mineral interactions. Zhang et al. classified dissolution-diffusion-convection process into four stages: (1) dissolution-dominated period; (2) diffusion-dominated period; (3) early convection-dominated period; and (4) late convection-dominated period22. Islam et al.23 presented simulation results of convective mixing with reactions adding heterogeneity and geothermal effects. They concluded that at a fixed Damkohler number (Da), reaction orders make substantial difference of mixing over longer period of times. Geothermal gradient exhibited negligible impact. Ward et al.24 studied reactive flow in the limit of high Ra in which the domain considered was deep, shallow or of intermediate depth, and for which the Da was respectively large, small or of order unity. For large Da the rapid reaction rate limits the plume depth and the boundary layer restricts the rate of solute transfer to the bulk volume, whereas for small Da the average solute transfer rate is ultimately limited by the domain depth and the convection is correspondingly weaker. All these previous researches were very generalized, in terms of Ra and Da.

Fu et al.25 presented formation of rock-dissolution patterns that arose from a series of calcite dissolution reactions during convection. They used high-resolution simulations to examine the interplay between the density-driven hydrodynamic instability and the rock dissolution reactions and analyzed the impact on the macroscopic mass exchange rate. Their conclusion was the geochemical reactions terminate significantly earlier than the time when convective mixing stops. However, it should be clearly noted that geochemical effects may not always accelerate advection. A precipitation reaction such as that between the acidic brine and a rock formation rich in calcium feldspar promotes the deposition of solid calcite and kaolinite, removing CO2 from the liquid phase. Such a reaction may actually attenuate convection motion26. Dai and co-authors27 developed an integrated Monte Carlo method for simulating CO2 and brine leakage from carbon sequestration and subsequent geochemical interactions in shallow aquifers. Their results showed shallow groundwater resources may degrade locally by reduced pH and increased total dissolved solids. Ennis-King and Paterson20 and Fu et al.25 showed results of porosity and permeability changes from reactions and subsequent impacts on convection mechanism. However, in this work we show, by performing reactive transport modeling with speciation of a series of calcite and dolomite dissolution reactions, that dissolution enhancement causes local density increase and effects from porosity and permeability change are almost negligible even when the reservoir reaches near 100% dissolved CO2 saturation.

Results and Discussions

We first have conducted sanity test of chemical speciation. CO2 leak problem is solved considering chemical reactions of Eqs 3, 4, 5, 6, 7, 8 by following the semianalytical approach in Yang et al.28. It is noteworthy that coupled reactive transport calculations strongly depend on accuracy of geochemical database. Different geochemical databases and uncertainty of thermodynamics data can severely affect the results. For the transport part analytical solution for the scenario with leaky well was used. Figure 1 shows that CO2 stream in host carbonates after equilibrium dissolution can be increased by more than 1.5 folds. In our simulation the maximum equilibrium dissolved CO2 concentration of 0.97 mol/l is used in saturated CO2-brine interface at the top. In order to quantify saturation area of the aquifer, average concentration formulation, , is used. The direct numerical simulations have been carried out for Ra numbers of 1000 and 10,000, where the associated mean permeability values are 10 and 100 mD, respectively.

Figure 1: Equilibrium concentrations after breakthrough.
Figure 1

Figure 2 show time lapse of dissolved CO2 fronts from 20 years period to the time the reservoir domain becomes completely saturated. In the case of no reactions, because Ra is low, convection is diffusion dominated, resulting in very slow plume advancement. Even after 900 years CO2 propagation is still limited to upper 10% of the reservoir. However, reaction activities with calcite and dolomite make results completely different. By that time the entire porous formation becomes saturated with CO2. Whenever CO2 arrives at any particular domain after equilibrium reactions, CO2 stream in the form of is dissolved from resident rocks, increasing local density of the brine. The increased density drives more instability which, in turn, causes plume boundaries to advance further downward. For this specific reservoir and thermophysical conditions in 900 years the pore volume is chock-full with dissolved CO2. The process adds significant feedback, however, in a negative sense because of early shutdown of both solubility and mineral trapping processes. Reactions constantly accelerate motion of the fronts. Because convective mixing shows diffusion dominance and the medium is homogeneous the cells do not form wormholes. Therefore, reaction fronts are planar throughout the process. No bifurcation occurs. Figure 3 exhibit results of same Ra, however for heterogeneous permeability formation. Dykstra-Parson coefficient of 0.55 was applied in generating the distributions shown in Fig. 4. Initially, results do not vary much from its homogeneous counterpart other than the pattern of plume evolution. As convection proceeds in addition to heterogeneity effects concurrent geochemical reactions make noticeable difference in dissolution process. Because of local permeability variations and nonlinear flow dynamics from the beginning of plume development hopf-bifuraction of the cells occur and CO2 stream added from carbonates help spread initially laterally and then finally dense phase sinks down. Thus, compared to no permeability contrasts, average concentration of CO2 in the aquifer differs significantly with time. ~100% saturation is reached 50 years earlier. On the other hand, <20% saturation only by reverse buoyant flow conveys clear message of reaction effects on CO2 convection (see Fig. 5). We have also tested results of layered anisotropic heterogeneity . Mean permeability is too small to add any effect resulting in almost same results as homogeneous reservoir.

Figure 2
Figure 2

Concentration maps of Ra = 1000, homogeneous case; (a,c,e and b,d,f) show results of with and without reactions, respectively.

Figure 3
Figure 3

Concentration maps of Ra = 1000, heterogeneous case; (a,c,e and b,d,f) show results of with and without reactions, respectively.

Figure 4: log10k distributions of heterogeneous reservoir.
Figure 4
Figure 5: Average concentrations of dissolved CO2 in the reservoir for the case of Ra = 1000.
Figure 5

Figures 6, 7, 8 plot concentration contours for the case of relatively high mean permeability 100 mD (Ra = 10,000). As expected, high permeability allows dissolving more CO2 and consequently reaction effects become more pronounced. Among three cases, the most intense enhancement is observed in heterogeneous reservoir where CO2 saturates completely in 500 years. Ra is high enough to initiate convection in early times and for heterogeneity the convection cells get more dynamicity because of availability of local high permeable favorable paths. Thus CO2 dissolution swells, transporting more CO2 saturated brine in contact with carbonates. The concurrent convection and reaction processes drive fast spreading of dense dissolved phase in the reservoir both laterally and downward. Figure 9 shows concentration profiles of Ra = 10,000. In this case layered heterogeneity adds little more reaction effect over the homogeneous distribution. For no reactions case maximum saturation obtained is ~40%. The early saturation due to geochemical effects has a negative effect on CCS in the sense that much injected CO2 may remain as completely undissolved. As a result, potential of leakage through high permeable zones or abandoned wells can be greater. To our best knowledge, no experimental results of convection enhancement due to geochemical reactions are reported. This strongly suggests a potential future research.

Figure 6
Figure 6

Concentration maps of Ra = 10,000, homogeneous case; (a,c,e and b,d,f) show results of with and without reactions, respectively.

Figure 7
Figure 7

Concentration maps of Ra = 10,000, heterogeneous case; (a,c,e and b,d,f) show results of with and without reactions, respectively.

Figure 8
Figure 8

Concentration maps of Ra = 10,000, layered heterogeneous case; (a,c,e and b,d,f) show results of with and without reactions, respectively.

Figure 9: Average concentrations of dissolved CO2 in the reservoir for the case of Ra = 10,000.
Figure 9

Figure 10 displays porosity changes, based on volume fraction deviations, for the case of Ra = 10,000 and heterogeneous reservoir. Though dissolution observed has significantly been boosted by the reactions, maximum porosity alteration until the reservoir becomes CO2 saturated is only 0.02%. This slight increase is almost negligible and cannot affect hydrodynamics of the system by any means. By using porosity-permeability relation29, , respective permeability change is 0.06% which is also too small to distable fluid flow further.

Figure 10: Change in porosity (Ra = 10,000, heterogeneous formation).
Figure 10

Methods (reservoir model and governing equations)

Following the model development in Islam et al.30 the continuity and transport equations read

Here c, t, u, k, represent concentration, time, velocity, permeability, and the stream function. x and z are coordinates in lateral and vertical (positive downward) directions. Ra is the primary parameter that defines buoyancy driven flow. The bottom, lateral boundaries are impervious, and the top dense CO2-brine interface is perturbed with saturated dissolved CO2. Numerical solutions of Eqs. 1 and 2 are explained in details in Islam et al.31.

As dissolved CO2 comes into contact with rocks, it lowers the pH and triggers redistribution of carbonate aqueous species via the following reactions:

The chemical system involves eight reactive aqueous species (CO2(aq), H+, OH, , , , Ca+2, Mg+2), conservative species (Na+, Cl) which are not involved in above reactions, and quartz (SiO2) matrix which remains inert. A sequential approach is used to first solve c from the conservative transport equation (Eq. 2), followed by coupling with equilibrium reactions Eqs 3, 4, 5, 6, 7, 8. The solution methods discussed in De Simoni et al.32 and Yang et al.28 are adopted for reactive transport. All required thermodynamics data (reaction constants, Debye parameter, species valence, etc.) are obtained from database file of PFLOTRAN (, accessed on Dec 20th, 2015) code. The reservoir and thermophysical data, and initial concentrations used are reported in Table 1.

Table 1: Data used in simulation.


Convection from CO2-laden boundary is examined by coupling with reactive modeling of resident carbonate (calcite and dolomite) formations. Reactions involved were in locally equilibrium. We tested both homogeneous and heterogeneous cases with low and high mean permeability in term of Ra number. The sanity test delivered initial results of maximum equilibrium concentration of dissolved species. This showed dissolved CO2 stream from reactions can increase local density even more than one order of magnitude. Geochemistry based on reaction intensity can cause serious alteration of flow dynamics. Our numerical results provide clear message that upon affecting local concentrations density-driven convection is enhanced substantially. For instance, Ra = 10,000 and heterogeneous distributions show that reservoir reaches ~100% CO2 saturation in 500 years while only convective flow covers 40% upper area. Negligible porosity and permeability increase do not affect the hydrodynamics at all. The strong enhancement, in turn, adds negative impact on CCS in the sense that more free phase CO2 may exist in the storage formations, increasing the potential for leakage and prolonging the period for long-term monitoring.

Additional Information

How to cite this article: Islam, A. et al. Reactive Transport Modeling of the Enhancement of Density-Driven CO2 Convective Mixing in Carbonate Aquifers and its Potential Implication on Geological Carbon Sequestration. Sci. Rep. 6, 24768; doi: 10.1038/srep24768 (2016).


  1. 1.

    et al. Climate change 2007: Synthesis report: An assessment of the intergovernmental panel on climate change. IPCC Plenary XXVII, Valencia, Spain. (2007, Nov 12–17).

  2. 2.

    , & Aquifer disposal of CO2: Hydrodynamic and mineral trapping. Energy Conv. Mgmt. 35, 269–279, doi: 10.1016/0196-8904(94)90060-4 (1994).

  3. 3.

    Geochemical reaction modeling of carbon dioxide, CO2, in geological formations: Theory PhD thesis, University of Maryland, (2014).

  4. 4.

    , , , & Intergovernmental Panel on Climate Change. IPCC Special report on Carbon Dioxide Capture and Storage 53 (New York, 2005).

  5. 5.

    , & The underground sequestration of carbon dioxide: containment by chemical reactions in the deep geosphere. Geological Society, London, Special Publications 157, 117–129 (1999).

  6. 6.

    , & Experimental study of CO2 injection into saline formations. SPE J. 588–594, doi: 10.2118/110628-PA (2009).

  7. 7.

    , & Onset of convection in CO2 sequestration in deep inclined saline aquifers. J. Can. Pet. Technol. 48, 22–27, doi: 10.2118/09-08-22-TN (2009).

  8. 8.

    et al. Simulation of CO2 storage in saline aquifers. Paper presented at SPE/EAGE Reservoir Characterization and Simulation Conference, Abu Dhabi, UAE (2009, Oct 19–21).

  9. 9.

    & Minimizing risk of gas escape in gas storage by in-situ measurement of gas threshold pressure and optimised completion solutions. Paper presented at Europec/EAGE Conference and Exhibition, Rome, Italy (2008, Jun 9–12).

  10. 10.

    A geochemical compositional simulator for modeling CO2 sequestration in geological formations. PhD thesis, University of Utah (2010).

  11. 11.

    & Quantification of CO2 masses trapped through free convection process in isothermal brine reservoir. Int. J. Heat Mass Transf. 87, 128–137, doi: 10.1016/j.ijheatmasstransfer.2015.03.083 (2015).

  12. 12.

    , , & Onset of convection in a grvitationally unstable diffusive boundary layer in porous medium. J. Fluid Mech. 548, 87–111, doi: 10.1017/S0022112005007494 (2006).

  13. 13.

    & Laboratory flow experiments for visualising carbon dioxide-induced, density-driven brine convection. Transp. Porous Media 82, 123–139, doi: 10.1007/s11242-009-9482-2 (2010).

  14. 14.

    et al. High resolution simulation and characterisation of density-driven flow in CO2 storage in saline aquifers. Adv. Water Res. 33, doi: 10.1016/j.advwatres.2010.01.009 (2010).

  15. 15.

    , , & The effect of heterogeneity on the character of density-driven natural convection of CO2 overlying a brine aquifer. Adv. Water Res. 34, 327–339, doi: 10.2118/138168-MS (2011).

  16. 16.

    et al. Experimental study on effects of geologic heterogeneity in enhancing dissolution trapping of supercritical CO2. Water Resour Res. 51, 1635–1648, doi: 10.1002/2014WR015778 (2015).

  17. 17.

    , & Convective instability and mass transport of diffusion layers in Hele-Shaw geometry. Phys. Rev. Lett. 106, 104501, doi: 10.1103/PhysRevLett.106.104501 (2011).

  18. 18.

    & Role of convective mixing in the long-term storage of carbon dioxide in deep saline formations. SPE-84344, ATCE, Denver, Co, USA (2003, Dec 5–8).

  19. 19.

    et al. Convective dissolution of carbon dioxide in saline aquifers. Geophys. Res. Lett. 37, L22404, doi: 10.1029/2010GL044728 (2010).

  20. 20.

    & Coupling of geochemical reactions and convective mixing in the long-term geological storage of carbon dioxide. Int. J. Greenhouse Gas Control 1, 86–93, doi: 10.1016/S1750-5836(07)00034-5 (2007).

  21. 21.

    , & The impact of geochemistry on convective mixing in a gravitationally unstable diffusive boundary layer in porous media: CO2 storage in saline aquifers. J. Fluid Mech. 673, 480–512, doi: 10.1017/S0022112010006282 (2011).

  22. 22.

    , & Reactive transport modeling of effects of convective mixing on long-term CO2 geological storage in deep saline formations. Int. J. Greenhouse Gas Control 5, 241–256, doi: 10.1016/j.ijggc.2010.10.007 (2011).

  23. 23.

    , , & Effects of geochemical reaction on double diffusive natural convection of CO2 in brine saturated geothermal reservoir. Int. J. Heat Mass Transf. 77, 519–528, doi: 10.1016/j.ijheatmasstransfer.2014.05.040 (2014).

  24. 24.

    , , & High-Rayleigh-number convection of a reactive solute in a porous medium. J. Fluid Mech. 760, 95–126, doi: 10.1017/jfm.2014.594 (2014).

  25. 25.

    , , & Rock dissolution patterns and geochemical shutdown of CO2-brine-carbonate reactions during convective mixing in porous media. J. Fluid Mech. 764, 296–315, doi: 10.1017/jfm.2014.647 (2015).

  26. 26.

    & Geochmeistry of silicate-rich rocks can curtail spreading of carbon dioxide in subsurface aquifers. Nature Comm. 5, 5743–5748, doi: 10.1038/ncomms6743 (2014).

  27. 27.

    et al. Probabilistic evaluation of shallow groundwater resources at a hypothetical carbon sequestration site. Sci. Rep. 4, 4006, doi: 10.1038/srep04006 (2014).

  28. 28.

    , , & Semi-analytical approach to reactive transport of CO2 leakage into aquifers at carbon sequestration sites. Greenhouse Gases Sci. Technol. 5, 786–801, doi: 10.1002/ghg.1527 (2015).

  29. 29.

    Scale effect on porosity and permeability: kinetics, model, and correlation. AIChE J. 47, 271–275, doi: 10.1002/aic.690470206 (2001).

  30. 30.

    , & Double diffusive natural convection of CO2 in a brine saturated geothermal reservoir: Study of non-modal growth of perturbations and heterogeneity effects. Geotherm. 51, 325–336, doi: 10.1016/j.geothermics.2014.03.001 (2014).

  31. 31.

    , & Numercal investigation of double diffusive natural convection of carbon dioxide in a brine saturated geothermal reservoir. Geotherm. 38, 101–111, doi: 10.1016/j.geothermics.2013.07.001 (2013).

  32. 32.

    , , & A mixing ratios-based formulation for multicomponent reactive transport. Water Resources Res. 43, W07419, doi: 10.1029/2006WR005256 (2007).

Download references


A.I. and A.S. are partially supported by US Department of Energy (DE-FE0012231). C.Y. acknowledges the Center for Frontiers of Subsurface Energy Security, an Energy Frontier Research Center funded by the U.S. Department of Energy (DOE), Office of Science, Basic Energy Sciences (BES), under Award # DE-SC0001114.

Author information


  1. Bureau of Economic Geology, The University of Texas at Austin, TX, USA

    • Akand Islam
    • , Alexander Y. Sun
    •  & Changbing Yang


  1. Search for Akand Islam in:

  2. Search for Alexander Y. Sun in:

  3. Search for Changbing Yang in:


A.I. and A.S. designed research. All authors contributed to interpretation of results and writing.

Competing interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to Akand Islam.


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Creative CommonsThis work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit