Statistical Strategies for Decision-making Regarding the Quality of Particleboards with Glass Wool

Abstract

This study evaluated multivariate statistical strategies to select critical properties for the performance of particleboards bonded with urea-formaldehyde (UF) adhesive modified with glass wool residues. Panels were produced with six different proportions of glass wool incorporated into the UF adhesive (0%, 3.34%, 4.93%, 6.52%, 9.49%, 12.35%). These panels were characterized by physical, mechanical, fire-retardant, and acoustic properties. Three statistical tools were applied: hierarchical cluster analysis, principal component analysis (PCA), and Pearson correlation. PCA explained 80.3% of the total variance, revealing distinct patterns among treatments, especially at the lowest and highest filler contents. The correlation matrix showed the interdependence between the rheological properties of the adhesive and the final composite performance. Glass wool as a filling material, in the proportion of 3.34% of the adhesive, provided the best performance among the panels, as it promoted balance between mechanical, physical and acoustic properties. Up to the limit of 6.52% glass wool contributed to improving fire resistance without significant changes in mechanical strength but reduced dimensional stability due to changes in adhesive rheology. The combination of multivariate analyses provided a robust approach to identify key attributes and guide the formulation of panels with enhanced technical and functional performance.