Regression Models for Prediction of Properties of PVC Compounds Considering the Effects of Additives Dosis

Abstract

In this work, multivariable linear regression model for predicting certain properties of PVC compounds with applications in the construction industry were proposed. For this purpose, 24 experiments were carried out, by applying a combined factorial design 2 . The concentrations of a) a primary heat stabilizer (ET), b) a pseudo- component prepared with fixed concentrations of two process aids and an impact modifier (PCA), c) a commercial calcium carbonate (CCa) used as filler , and d) a commercial n titanium dioxide (TiO) used as a pigment/ UV stabilizer, were selected as independent variables. The dependent variables were: Â module of elasticity, yield stress, fracture stress, heat deflection temperature (HDT), melting time, melting torque , stability time , torque stability, bulk density (DAP) and flow time . Excepting for the fracture stress and the torque stability, independent variables and/or their second and third order interactions chosen, explained at least 70% of the variability of the investigated properties. The analysis of the coefficients of the regression equations indicated that the variable with the highest effect on the mechanical properties was the concentration of PCA.

Keywords

• Multivariable linear regression
• PVC compounds