Linking pulp variation to TMP operation by better selection and treatment of process data.

The article presents a study about the use of various systematic method for better selection and treatment of processing thermomechanical pulp (TMP) data to increase the usefulness and accuracy of partial least squares black-box models for pulp quality in Canada. The study is conducted in a certain TMP newsprint mill in the eastern part of the country. The pulp freeness, average fiber length and fines content were combined to formed a single statistical model. The study reveals that methods that uses pretreated data were better, whether evaluated using statistical metrics or qualitative tests.