Prediction Model Based on DoE and FTIR Data to Control Fast Setting and Early Shrinkage of Alkaline-Activated Slag/Silica Fume Blended Cementitious Material

This study aims to develop a material-saving performance prediction model for fast-hardening alkali-activated slag/silica fume blended pastes. The hydration process in the early stage and the microstructural properties after 24 h were analyzed using design of experiments (DoE). The experimental results show that the curing time and the FTIR wavenumber of the Si-O-T (T = Al, Si) bond in the band range of 900–1000 cm(−1) after 24 h can be predicted accurately. In detailed investigations, low wavenumbers from FTIR analysis were found to correlate with reduced shrinkage. The activator exerts a quadratic and not a silica modulus-related conditioned linear influence on the performance properties. Consequently, the prediction model based on FTIR measurements proved to be suitable in evaluation tests for predicting the material properties of those binders in the building chemistry sector.