Optimization of Production Conditions in Cement Mill to Achieve Optimal Cement Properties at Early Ages

Abstract

Abstract A statistical model was evaluated for predicting the two properties of cement, setting time and the amount of water required to achieve standard consistency, from the production conditions in a cement mill. The evaluation was performed by the application of partial least square regression. The observation X-matrix was partitioned into two sub-matrices, one containing observed component compositions in the cement samples and the other containing observed process conditions. A model for predicting potential properties of the process condition was then evaluated. Potential setting time and potential amount of water required to achieve standard consistency were predicted from the observed variation in the process variables only. The component composition was kept constant and equal to their mean values. The predictions explained significant variation in both properties. Sensitivity analysis based on simulation, optimization, and prediction made it possible to study the influence of the grinding process on the properties. The characteristics of the cement like the superficial microstructure from thermogravimetric analysis were included in the investigation to explain more mechanistically or chemically the variation in the properties.