Application of multivariate statistical analysis to raw materials for ceramic industries

Abstract

This paper shows an application of multivariate statistical techniques (hierarchical cluster analysis and principal component analysis) to seven ceramic variables (loss on ignition, firing linear shrinkage, total linear shrinkage, flexion rupture module, absorption of water, apparent porosity and apparent density) in samples belonging to three mines which provide raw materials for ceramic industries of the Santa Gertrudes/Cordeirópolis/Rio Claro Province. With the application of cluster analysis, the samples were divided into three groups, which have similar geological characteristics. The apparent density (DA) was the main cause of the variability (0,832 %). The loss on ignition LOI accounted for 0,776 % of total variance. It was observed that a narrow relationship exists among identified groups and the geological characteristics based on field works.