Modeling of Alteration Behavior on Blended Cementitious Materials

Abstract

A “realistic alteration model” is needed for various cementitious materials. Hypothetical settings of mineral composition calculated based on the chemical composition of cement, such as Atkins’s model, have been used to estimate the alteration of cementitious material. However, model estimates for the concentration of certain elements such as Al and S in leachate have been different from experimental values. In a previous study, we created settings for a mineralogical alteration model by taking the initial chemical composition of cementitious materials from analysis results in experiments and applying their ratios to certain hydrated cement minerals, then added settings for secondary generated minerals in order to account for Ca leaching. This study of alteration estimates for ordinary portland cement (OPC) in groundwater showed that the change in Al and S concentrations in simulated leachate approached values for actual leachate[1]. In the present study, we develop an appropriate mineral alteration model for blended cementitious materials and conduct batch-type leaching experiments that use crushed samples of blast furnace slag cement (BFSC), silica cement (SC), and fly ash cement (FAC). The cement blends in these experiments used OPC blended with blast furnace slag of 70 wt.%, silica cement consisting of an amorphous silica fine powder of 20 wt.%, and fly ash of 30 wt.%. De-ionized water was used as the leaching solution. The solid-liquid ratios in the leaching tests were varied in order to simulate the alteration process of cement hydrates. The compositions of leachate and minerals obtained from leaching tests were compared with those obtained from models using hypothetical settings of mineral composition. We also consider an alteration model that corresponds to the diversity of these materials. As a result of applying the conventional OPC model to blended cementitious materials, the estimated Al concentration in the aqueous solution was significantly different from the measured concentration. We therefore propose an improved model that takes better account of Al behavior by using a more reliable initial mineral model for Al concentration in the solution.