Pulp property development Part I: Interlacing undersampled pulp properties and TMP process data using piece-wise linear functions

Thermo-Mechanical Pulp (TMP) refiners served in this work as an example of how to combine oversampled and undersampled variables on a common timeframe for further analysis. To get more information from the refining process, temperature measurement arrays in the refining zones were studied alongside process measurements such as motor load, production rates, plate gaps, dilution waters, pulp properties and manually measured blow-line consistency. The undersampled data set consisted of 63 laboratory samples obtained at a Swedish TMP mill, which were tested for tensile index, mean fiber length and Somerville shives content. The pulp samples were obtained at five different periods during three months to cover a large dynamic operating window. The data set was expanded using a piece-wise linear approach. The measurements inside the refining zone were shown to be important variables when interlacing the undersampled pulp properties with the oversampled process data set, consisting of 350 000 samples. Use of an extended entropy model provided a palette of information about the process conditions inside the refining zone. Particularly, the residence time and the consistency in the refining zones were essential for the pulp property development, as a link between the refining segment pattern used and the current state of refiner operation.