An online gauge for process analysis in the pulp industry: Integrating spectroscopic and physical measures of pulp quality to predict fitness for purpose
To sustain its leadership in the world marketplace, the Canadian pulp industry must continue to increase pulp quality, while it improves manufacturing efficiency, reduces energy consumption and decreases impact on the environment. Properties of a manufactured pulp vary with production parameters and characteristics of the feedstock. To produce pulps of the highest quality at lowest cost, a manufacturing process must continually optimize its conditions to fit the feedstock. This requires real-time process-analysis methods that measure the pulp in a production stream and predict its fitness for final purpose. Conventional wet-chemical methods now in use do not offer the possibility for real-time assessment. Work in our laboratory has shown that spectroscopic measurements, when fused with physical data on fibre characteristics, can accurately predict the properties of a finished paper product from the instantaneous measure of an in-process pulp. We propose to bring our spectroscopic methods to the production line, together with a machine learning approach to data processing that integrates spectroscopic information with known production parameters and real-time measures of fibre properties to realize an industry hardened, fully automated, online gauge of pulp and final-product paper quality attributes for deployment in a working pulp mill.