The disposal of fly ash (particularly class F fly ash) is increasingly becoming a problem, and the way to solve this problem safely and economically is by utilization. Due to the unique behavior of fly ashes, it has not been widely used in soil modification applications. To improve or increase fly ash utilization, this study seeks to develop a generalized theory that can adequately predict the behavior of fly ash modified soils and to quantify the uncertainties in the behavior of the ash, which hinders its wide application in soil modification.
For this study, laboratory tests were performed on mixtures made from three different fly ashes of different chemical compositions. The laboratory tests were designed such that all the necessary properties needed for the analyses were addressed as well as data gaps observed in the literature. The laboratory testing provided engineering (index, deformation, and strength) properties and the necessary physicochemical properties of mixtures at various fly ash percentages for all the mixtures were determined.
Three mixture theory models were used to predict engineering properties based on the properties of individual constituents. The predictive accuracies of three models were assessed, and the model that best predicted actual results was selected for modification. The need for modification is because of deviations in the model predictions. The selected model was that of Voigt (1889).
Upon physicochemical analysis, it was observed that the chemical composition of the fly ashes play a significant role in the behavior of the modified soils, particularly that of calcium oxide (CaO). A relationship was found between CaO and the sum of oxides (SiO2+Al2O3+Fe2O3) in the fly ashes. As a result, the selected model was modified based on the two chemical components (CaO and the sum of oxides) of the fly ashes. A modification term ( ) was defined as a ratio between the CaO and the sum of oxides raised to an experimental index (x). The term varied with respect to engineering properties. The variation was found to be dependent on the experimental index (x). The modified model performed well in predictions of laboratory data and data from the literature.
Due to the improvements observed in the modified model’s predictions, it is believed that it can be widely applicable to a wide variety of fly ash-modified soils.