A Statistical Model for Predicting Auto-clave Expansion of Portland Cement

Abstract

The present study aims aimed to investigate factors affecting the soundness of Portland cement (in terms of autoclave expansion test). These factors are C3S, C2S, C3A, C4AF, fineness (in terms of specific surface measured by Blaine method), the minor oxides MgO, free CaO, SO3, and the variables obtained from the chemical analysis of cement like silica modulus (SM), alumina ratio (AR), loss on ignition (LOI), insoluble residue (IR), and lime saturation factor (LSF). The autoclave expansion prediction models were built by using multiple linear regression analysis and based on (40) different cement samples taken from (7) different Iraqi cement factories, Indian cement, and Kuwaiti cement. (29) of the samples were ordinary Portland cement while the other (11) samples were sulphate resisting Portland cement. It was found that the multiple linear regression is very suitable for predicting the autoclave expansion of Portland cement. It was also found that the increase of fineness of cement, LSF, and LOI decreases the autoclave expansion, while the increase in the other factors increases the autoclave expansion. The correlation coefficients of the proposed models were (0.71002 and 0.98338) for the first model, (0.84366 and 0.98789) for the second model, and (0.85593 and 0.98872) for the third model, with and without intercept respectively