OR/13/004 CO2 transport mechanisms

Rochelle, C A, Purser, G, Milodowski, A E, Noy, D J, Wagner, D, Butcher, A, and Harrington, J F. 2013. CO2 migration and reaction in cementitious repositories: A summary of work conducted as part of the FORGE project. British Geological Survey Internal Report, OR/13/004.

In a repository setting, CO₂ could be present in dissolved form (as CO_2(aq), HCO₃^-, CO₃^2- or complex ions depending upon local fluid chemical conditions), but if enough of it were generated to exceed saturation then it could also be present as a free phase. In terms of transport mechanisms therefore, it could migrate as:

• Diffusion of dissolved species along a chemical gradient.
• Downward advection of dissolved species due to density increase of water upon CO₂ dissolution (e.g. Kneafsey and Pruess, 2010; Pau et al., 2010).
• Advection of free-phase CO₂ along a pressure gradient, or buoyant upward advection.

In this investigation, we studied the impacts of CO₂ diffusion on NRVB cement via static batch experiments. The work focussed on the formation of chemical reaction fronts and the ensuing mineralogical and textural changes, rather than on measuring diffusion rates. The impacts of CO₂ advection were studied via flow experiments. These included flow of dissolved and gaseous CO₂ (being the more likely forms in a future repository setting). However, we also investigated the flow of supercritical CO₂ to scope cement reactivity under more extreme conditions.

The nature of CO₂ migration within the NRVB cement was also controlled to some extent by chemical reaction, a process that might not be apparent for other gases in different repository concepts (e.g. hydrogen migration through bentonite). Thus for example, CO₂ consumption through the formation of secondary carbonate minerals would set up steeper chemical potential gradients, enhance the apparent rate of diffusion, and possibly lead to faster migration of carbonate reaction fronts. Conversely, the formation of secondary carbonate minerals could block porosity, leading to significant reductions in permeability in the flow experiments. In terms of overall migration of CO₂ within cementitious systems therefore, it is necessary to consider chemical reaction, reaction kinetics and transport.

The aims of this study did not include quantification of diffusion properties. However, we were able to track the net effect of diffusion-controlled carbonation via the progression of carbonation fronts as a function of time. It should be noted that the fronts do not record the position of the leading edge of CO₂ ingress, as minor amounts of CO₂ reacted with cement in advance of the fronts. Instead, the carbonation fronts record the position where sufficient CO₂ had permeated the core to enable enough carbonation to occur to change the structure of the cement (including complete carbonation of the reactive cement minerals). The fronts migrated by a few mm over several weeks (Figure 12). However, the data from several of the 40 day long experiments show quite variable degrees of ingress of the carbonation fronts. Whilst some of the 40 day data are broadly in line with extrapolation of the 10 and 20 day data, some samples showed much slower ingress of the reaction fronts into the samples. We are unsure why this was the case, though sub-sampling from a heterogeneous cement block may have resulted in experimental samples of slightly different properties, and this might be one explanation.

Figure 12    Three optical microscope images with maximum ingress of reaction fronts highlighted. Slides (a) and (b) are from experiments using a ‘young’ porewater, and (c) from one using an ‘evolved’ groundwater.
Also shown is an illustrative plot of the average ingress of the leading reaction front into the cement over time. Different samples reacted for 40 days showed different degrees of CO₂ ingress.

Whilst many of the samples used in the static carbonation tests showed relatively uniform carbonation when sectioned, some displayed carbonation focussed along specific zones. These appear to be regions made from coarser grains, with the coarser porosity infilled by Ca(OH)₂. These were more reactive to CO₂ than the surrounding, finer-grained CSH-rich regions, resulting in non-uniform, complex reaction fronts. The cause of the original differences most likely relates to differential sedimentation of grains during the pouring and setting of the cement. The overall effect is to introduce a degree of uncertainty in the average position of the carbonation fronts, as different samples were affected to different degrees (i.e. small sub-cores from one part of the original large cylinder of cement may be affected, whereas others will not). That said, only a few of the samples used for the static tests showed these features, and none of the samples used for the flow tests appear to have been affected. The addition of certain (usually organic) additives to the cement may be one way to minimise differential sedimentation, but these additives may have deleterious effects on the solubility and sorption behaviour of some radionuclides and thus be unwanted components within a repository (Francis et al., 1997). It is therefore possible that differential sedimentation may be a feature of additive-free NRVB cement in a repository, and that diffusional transport may result in regions of carbonation having complex shapes.

Experiments using an advective flow of CO₂ resulted in relatively uniform alteration of the cement, with no preferential alteration along specific zones. Injection of CO₂ was along the central axis of the core sample. As a consequence of the fast reaction of CO₂ and flow rate used, this resulted in a cone-shaped leading edge to the carbonated zone (Figure 10). As per the static (diffusion-controlled) experiments, a reaction zone was observed, though this was thinner in the flow experiments. Once the apex of the cone-shaped leading edge of the reaction front reached the outlet side of the cement sample it appeared to focus CO₂ migration. The net effect was to leave small areas of isolated unreacted cement close to the outlet end of the cement sample.

Flow in both the batch experiments (diffusion) and flowing experiments (advection) was via an open pore network. The NRVB cement is designed to have a very high permeability in order to allow gas migration and prevent excess pressures building up within a repository. Though some permeability reduction was observed in the laboratory tests, the carbonated cement still remained permeable to gas — though to a lesser degree. Whilst we recognise that the potential for excess gas pressures will also be governed by the rate of gas generation (including gases other than CO₂), based upon the results of this study there is no strong evidence to suggest that permeability will decrease sufficiently enough to allow excess gas pressures to develop (at least for realistic gas generation rates). It should also be recognised that CO₂ will be consumed by the cement through carbonation reactions, a process which will act to reduce gas pressures. As such, mechanical disruption of the NRVB cement seems unlikely, and consequently the potential for gas transport along newly-formed cracks appears unlikely.

There are however, two caveats to the above:

1. We cannot rule out the possibility that much slower carbonation reactions (i.e. at rates more realistic of a repository timescale) might produce narrower reaction fronts with more efficient infilling of porosity, and consequent larger reductions of permeability. These may serve as more effective barriers to CO₂ flow and allow higher differential pressures to develop. Indeed studies of naturally-carbonated, naturally-occurring CSH phases indicate that some reaction fronts can effectively ‘armour’ centimetre-scale nodules of CSH phases from further carbonation for several thousands of years (Milodowski et al., 2009; Rochelle and Milodowski, 2013).

1. As well as forming anastomising networks of ‘chicken wire’ fabrics, CO₂-cement chemical reaction occasionally caused two types of slightly larger, but still very localised temporary shrinkage cracks to develop (Figure 13). These were mm-scale features, located at the active carbonation front, and were only observed for unconfined samples in the batch experiments. One set of shrinkage cracks was oriented approximately parallel to the flow direction and the other set were approximately perpendicular to it. Each of the small cracks was only a temporary feature, as secondary precipitation eventually sealed these as the carbonation front progressed through the sample. Thus an individual pathway may be temporary, but a longer-lasting, narrow, ‘dynamic’ zone of potential flow pathways associated with the actual carbonation front appears to have slowly moved through the cement.

Figure 13    Partially carbonated cement sample showing zones having different degrees of carbonation together with associated reaction fronts.

Issues of gas transport through cement are of interest to other industries as well as radioactive waste storage. Several relate to hydrocarbon/CO₂ extraction, gas storage and the underground sequestration of CO₂. A key concern is the behaviour and sealing of cements that line boreholes and provide a seal between the surrounding rock and the steel borehole liner. Whilst the migration of gases such as methane will be mainly subject to physical control, CO₂ will react with the borehole cement. There are an increasing number of borehole cement carbonation studies, including laboratory tests, recovery of actual borehole samples, investigation of natural analogues, and predictive modelling (e.g. Carey et al., 2007; Milodowski et al., 2011; Rochelle and Milodowski, 2013; Wilson et al., 2011). Of particular concern is the potential for CO₂ transport along imperfectly-sealed interfaces between rock-cement and cement-steel. Whilst CO₂ extraction and enhanced oil recovery (EOR) schemes may only consider borehole lifetimes of a few tens of years, those concerning underground storage of CO₂ have to consider timescales of 10 ka to 100 ka — significant on a repository performance assessment timescale. Many CO₂-cement reactions and transport processes appear common to both CO₂ storage interests and radioactive waste disposal, even though the permeability characteristics of the cements may differ significantly.