Step-Wise Multiple Linear Regression Model Development for Shrinkage Strain Prediction of Alkali Activated Binder Concrete

Among the recent discoveries for alternatives of portland cement (PC) concrete, Alkali-activated binder (AAB) concrete is prolifically being considered as the most eco-friendly and sustainable alternative. The present study evaluates the shrinkage behaviour for three different AAB mixtures containing fly ash and/or slag at different proportions which are activated by sodium hydroxide and sodium silicate. Multiple linear regression models are developed to predict shrinkage strains of ambient-cured AAB concrete as a function of age and percentage of fly ash in the precursor. The aim of this work is to come up with a generalized equation that can predict the shrinkage of various binary blended AAB mixes cured at room temperature. The predicted models are ranked based on RMSE and then compared with the experimental data. The correlations were found to be quite satisfactory (R² = 0.937) and can be used to estimate the shrinkage for similar AAB mixtures. It is observed that the proposed model agrees more closely with the experimental results from the present study.