ANN model of brick properties using LPNORM calculation of minerals content

Abstract

Mineralogical composition of heavy clays is one of the most important properties when stadying raw materials in brick industry. Within this study, quantitative determination of minerals using LPNORM calculation was performed, using the first algorithm among the so-called norms that allows the introduction of a list of minerals and their configuration. This algorithm is implemented for the first time in practice, in order to calculate the minerals content in brick raw materials. The influence of minerals quantity, along with the firing temperature (800-1100 degrees C), and several shape formats of laboratory brick samples were investigated, and the acquired data were used to build Artificial Neural Network (ANN) model. ANN model was developed in order to predict the final products parameters, and its results have been afterwards compared to experimental data. ANN model, coupled with sensitivity analysis, was obtained with high prediction accuracy, according to coefficient of determination, r(2): 0.880-0.884 in compressive strength calculation, 0.954-0.960 for water absorption, 0.869 for firing shrinkage, 0.979-0.984 for water loss during firing and 0.907 for volume mass of cubes model.