Column Experiments: Results of Experiments and Modelling

K. Bateman Fluid processes Group, British Geological Survey, Keyworth, Nottingham, NG12 5GG, UK

k.bateman@bgs.ac.uk

P. Coombs Fluid processes Group, British Geological Survey, Keyworth, Nottingham, NG12 5GG, UK

D. J. Noy Fluid processes Group, British Geological Survey, Keyworth, Nottingham, NG12 5GG, UK

J. M. Pearce Mineralogy and Petrology Group, British Geological Survey, Keyworth, Nottingham, NG12 5GG, UK

P. D. Wetton Mineralogy and Petrology Group, British Geological Survey, Keyworth, Nottingham, NG12 5GG, UK

In order to describe the migration of an alkaline plume from the cementitious engineered barriers of a geological disposal facility for radioactive wastes, it is necessary to employ coupled chemistry and flow computer models. Although evidence from natural systems is useful to constrain reaction mechanisms and minerals to be incorporated into such models, time-dependent information is generally lacking. A series of laboratory column experiments was conducted as test cases to evaluate the capabilities of the coupled model PRECIP (Noy, 1990) to predict product solids and output fluid compositions over time. The experiments reacted minerals (calcite, quartz, albite, and a muscovite/quartz mixture) of importance to the radioactive waste disposal programmes in the UK, Sweden and Switzerland, with simplified 'young' (Na-K-Ca-OH) and 'evolved' (Ca(OH)2-saturated), synthetic cement porewaters. An experimental temperature of 70°C was chosen as a compromise between likely disposal temperatures and the elevated temperatures necessary in order to achieve results on a laboratory timescale. The coupled model PRECIP was used to provide'blind' predictive calculations.

Modelling

The model predictions for quartz reacting with the Na-K-Ca-OH fluid indicated dissolution along the length of the column. Product minerals, however, were restricted to close to inlet end, hillebrandite forming first followed by tobermorite. Predictions involving the Ca(OH)2-saturated fluid indicated less quartz dissolution. than for the simulation with the Na-K-Ca-OH fluid. The PRECIP modelling for albite reacting with the Na-K-Ca-OH fluid predicted strong dissolution close to the inlet end of the column. Tobermorite was formed first together with a very small amount of foshagite. Calculations involving the Ca(OH)2-saturated fluid also showed a vigorous reaction close to the inlet end of the column. In this case, hillebrandite formed in large quantities near the inlet, followed by foshagite and then tobermorite. Predictions for the reaction of calcite with both fluids showed dissolution to occur only immediately adjacent to the inlet end of the column, with the accompanying formation of small amounts of portlandite. The model predictions for the muscovite/quartz mixture with the Na-K-Ca-OH fluid showed dissolution along the entire length of the column. Hillebrandite was predicted in large quantities close to inlet end of the column. A small amount of foshagite was predicted to be found next, followed by tobermorite. The reaction for the muscovite/quartz mixture with Ca(OH)2 fluid was much simpler. Dissolution took place along the whole column but at about half the rate found with the Na-K-Ca-OH fluid. The products predicted to be found in this experiment were hillebrandite, and foshagite.

Experiments

In the calcite columns, dissolution of calcite was found when reacted with both fluids, accompanied by the formation of small amounts of portlandite. In the quartz columns reacted with the Na-K-Ca-OH fluid, most dissolution occurred in the first half of the column. Electron microprobe analyses of the secondary the CSH phases showed a considerable range in Ca:Si ratios. These ranged from approximately 0.5 to 1.6, equivalent to the CSH(I) group of minerals (Lea, 1970), suggesting the CSH phases evolved from hillbrandite through xonotlite, to okenite. Similar results were seen in the columns reacting albite with the Na-K-Ca-OH fluid, where the Ca:Si ratios varied from approximately 0.6 to 1.3. All analyses contained up to 1 wt % Al. The columns reacting muscovite/quartz mixture with the Na-K-Ca-OH fluid, showed a similar pattern to that of the albite columns. The experiments using the Ca(OH)2-saturated fluid generally showed the same pattern of reaction, with CSH and CASH phases being precipitated throughout the columns but with less dissolution of the starting materials.

Conclusions

Results of the modelling in this study have generally been in good agreement with the experiments in terms of the types of secondary mineral phases predicted to form. The predictions did not contain the all the variations of Ca:Si observed during mineralogical analysis, due to restrictions in the model database. However, they do demonstrate the evolution of CSH phases with different Ca:Si ratios over time and with distance along the columns.

Acknowledgements

This work was funded jointly by UK Nirex, Nagra and SKB, and is published with the permission of the Director of the British Geological Survey.

References

Lea, F. M., The chemistry of cement and concrete. Edward Arnold Ltd., London (1970).

Noy, D. J.,. Safety Studies Report, UK Nirex Ltd., NSS/R275 (1990).