Unlocking potential: the role of chemometric modeling in pulp and paper manufacturing

Abstract The review focuses on application of chemometric modeling in raw material characterization and quantification of parameter in various steps of pulp and papermaking processes. These models are built by supervised or unsupervised machine-learning techniques. Chemometric models are predominantly developed with data from spectroscopic instruments like, Ultraviolet spectrophotometer (UV), Near Infrared spectrophotometer (NIR), Fourier Transform Infrared Spectrometer (FTIR), Raman Spectrometer etc. These spectroscopic data are large in size which are reduced by applying different dimension reduction techniques. Moreover, these data contain instrumental noise in most of the cases, and they are de-noised or smoothed by several preprocessing techniques for developing better performing models. Among the popular calibration techniques for classification, Soft Independent Modeling for Classification Algorithm (SIMCA), Discrimination Analysis (DA) are mentionable. Multiple Linear Regression (MLR), Principal Component Regression (PCR), Partial Least Square Regression (PLSR), Artificial Neural Network (ANN), Support Vector Machine (SVM) etc., are used for quantification of different physical, morphological and chemical properties of wood, pulp and paper properties. Based on chemometric models, sensors for online measurements of different parameters in pulping and papermaking processes are being developed in recent years. Through this review, better performing multivariate analysis based chemometric modeling techniques have been identified for determining different parameters by comparing the existing ones which could be used in different processes in pulp and papermaking industries.