Improving feedstock quality control in formaldehyde-based resin and wood-based panel production through near infrared spectroscopy

Abstract

To assure the quality control of industrial processes, it is important to adopt reproducible and efficient methodologies. Spectroscopic methods, such as near infrared (NIR), are a good option as they are fast and may be used to indirectly estimate multiple physicochemical properties. In this study, NIR spectra of key feedstock samples used in the production of formaldehyde-based resin and wood-based panels, namely urea, ammonium sulfate, ammonium nitrate, sodium hydroxide, and acetic acid, were acquired. Multivariate data analysis was applied to establish the correlation between the spectra and the properties being measured. Quantitative models were then created using partial least squares regression to predict the concentrations of feedstock samples. This study presents quantitative models that were created by combining spectra measured on two probes, which achieved similar prediction results as single-probe based models. The performances of the best models were compared with the reference methods for each of the evaluated samples. For the samples under study, the proposed approach is suitable for routine analysis across multiple equipment configurations using the same quantitative model. NIR spectroscopy combined with chemometric models could be a valuable complement to support in-line raw material monitoring and plant digitalization in the wood panels industry.