CTMP Process Optimization Part II: Reliability in Pulp and Handsheet Measurements

Abstract

The objective of this investigation is to analyze and model pulp and handsheet properties using two different approaches based on a reduced set of binomial coefficients for small data sets and constrained linear functions for larger data sets. The properties are linked to process data from the daily CTMP operation in a CD-82 refiner. From a laboratory test program perspective, the test series was extensive and covered 80 pulp samples at different process conditions. In this paper, external variables (dilution water feed rate, specific energy and plate clearance) are compared with internal variables (consistency and fiber residence time) as predictors. It is shown that internal variables as predictors seem to outperform the external variables when estimating CSF Freeness, sheet density, tensile strength, tensile index, elongation to rupture, tensile energy absorption, tensile energy absorption index, tensile stiffness, tensile stiffness index, tear strength, tear index, short-span compressive test index, ISO brightness, Scott-Bond, Z-strength, shives(>= 0.3mm), long fibers and fines. It is important to cover all important dynamics to select the best possible models. Here, the absolute differences between the measured and estimated pulp and handsheet properties are ranked in ascending order, followed by a selection procedure before the polynomial fitting. Moreover, three different sets of chip mixtures were analyzed, and the results shown in this paper indicate that only one model is required for each pulp and handsheet property when using internal variables as predictors. Finally, at a given threshold of shives, it is shown that both Scott-Bond and sheet density can be optimized by changing the consistency and fiber residence time in the refining zones individually. This opens for implementation of new control strategies based on soft sensors.