The problem of predicting the future performance of cement-based systems is complicated by a poor understanding of the behaviour of cement systems at long ages, as well as of the complex interactions which can occur between cement and waste components - active as well as inactive - and with cement blending agents including fly ash, glassy slags and natural pozzolans. The progress achieved in developing a predictive capability is reviewed. Considerable success has been achieved in modelling the chemically-related features of cement based systems in terms of pH, Eh, and of element partition between solid and aqueous phases. The behaviour of model radwaste elements - iodine and uranium - has been studied in depth and indicates that both sorption and precipitation occur. U(VI), in particular, forms solubilitylimiting compounds, e.g. uranophane. But in general, presently-available data are inadequate to predict many cement-waste stream teractions; future progress in modelling is likely to rely heavily on additions to the data base.
The repository environment will also condition chemical exchanges in cement-based systems. Progress is being made in predicting the impact of CO2, a common ground water component, on the performance of cement systems.