Assessment of clinker phases in cement production with multivariate model based on X-ray diffraction and X-ray fluorescence

A method for the assessment of the relative abundance and composition of clinker phases is presented. The tested population consists of 269 clinker samples from industrial facilities. Data from X-ray diffraction (XRD) and X-ray fluorescence (XRF) techniques are correlated with multivariate modelling. An empirical algorithm is presented to fit the XRD phase relative abundance (R2 of 0.84 for C3S; 0.89 for C2S; 0.92 for C3A; 0.93 for C4AF). A semi-empirical algorithm is also presented to assess the composition of clinker phases. The validation strategy includes a comparative assessment with traditional models based on random data. The model shows better fitting (e.g. R2 of 0.84 for C3S) to the experimental XRD data than traditional models (e.g. R2 of 0.59 and 0.54 for C3S with the Bogue and Taylor model, respectively) and values closer to the bisector (e.g. R2 of 0.78 for C3S with the present model, compared against −0.01 and −5.85 with the Bogue and Taylor model, respectively). Thus, starting from the relative abundance of oxides, the relative abundance and composition of the silicate and aluminate phases could be accurately determined. Factors affecting phase abundance and composition could be distinguished and independently controlled, enabling the delivery of improved technical and sustainability-related performance of cement.