Results and Discussion

For each concrete, 5000 cycles (corresponding to about 3 years at 25 ºC) of hydration were executed using the microstructural model. The "final" degrees of hydration achieved using the model are summarized in Table VII. For ultimate degree of hydration, the values in parentheses indicate the values estimated for 28 days of hydration based on equations developed by Waller et al. [1]. Except for the w/c=0.3, s/c=0.3 concrete, the 3 year values are consistently higher than the 28 day values, as would be expected. Also, a reduction in w/c ratio below 0.55 is seen to significantly reduce the achievable hydration, due to both a decrease in the available pore space to be filled by hydration products and self-desiccation effects. The presence of silica fume is seen to further reduce the ultimate achievable hydration, as extra water is incorporated into the pozzolanic reaction products and is thus unavailable for participating in the cement hydration reactions.

Table VII also gives the volume fraction of the reactive silica fume that had reacted after the 5000 cycles of hydration, with the values in parentheses being the volume fraction of all of the silica fume, including that 10% of the silica fume which was initially assigned to be inert. The values for the higher silica fume additions can be compared to those measured via NMR analysis by Sun and Young for densified with small particles (DSP) pastes [14]. For silica fume replacements between 18% and 48%, these researchers determined that between 54% and 60% of the silica fume had reacted after 180 days, in general agreement with the model values of between 37% and 66% for the lower w/c ratio concretes.

4 provides a comparison of the experimental and model adiabatic temperature rise curves for a series of concretes without silica fume. By adjusting the induction time and the initial system temperature in the model, a reasonably good agreement is obtained between the model and experimental results. These results suggest that the model could be employed to simulate the adiabatic response of a wide variety of concretes, provided that the component materials had been characterized as performed in this study.

For the concretes containing silica fume, as shown in Figures 5 through 7, the agreement is also quite reasonable. Interestingly, for the higher w/c ratio concretes, there is some deviation between model and experiment at intermediate times, as was observed for the system containing only silica fume and calcium hydroxide in Figure 3. Because of the large amount of water (chemically bound and gel) incorporated into the pozzolanic C-S-H gel, the hydration effectively terminates at a much lower degree of hydration when curing is conducted under sealed conditions. If an H/S ratio of 2.1, as opposed to 3.9, is used in representing the pozzolanic C-S-H stoichiometry, the model predicted temperature rise significantly exceeds the experimental values at times greater than 30 hours, as sufficient water remains to continue the hydration. One of the complications in the modelling of cement-based materials is that the C-S-H gel (primary and pozzolanic) creeps and polymerizes over time [12, 15], expelling gel water which can then participate in further hydration reactions. However, based on these results, for the purposes of predicting the early temperature rise behavior of a concrete, the use of a constant stoichiometry to represent the C-S-H appears to be reasonable.