Thesis for the Degree of Master of Science


Year 1997
Päivi Annala

Application of Principles of Statistical Process Control to Quality Control Measurements in a Paper Mill

The objective of this thesis was to study the application of the principles of statistical process control to the quality control measurements at the Rauma Paper Mill of UPM-Kymmene. The theoretical part describes the main principles and the most important tools of statistical process control as well as their application to evaluate and monitor the dispersion of the measurements. In the experimental part, the dispersions and capability indices of the quality control measurements in a paper mill were determined by statistical methods.

The total process variation consists of true process variation and variation inherent in measurements. By determining the share of measurement-inherent variation of the total process variation we can establish the suitability of the measurement in measuring the quality concerned. If the capability index of the measurement is not sufficient for measuring the quantity, the dispersion caused by the measurement has to be decreased. The significance of the various dispersion sources can be defined by means of different methods of experiment planning. As the variation inherent in measurements is reduced, the total process variation will also be reduced.

The evaluated measurements included paper testing by the Paper Lab, the most important manual measurements in the pulp laboratory and pulp testing by the Pulp Expert. The dispersion of the evaluated measurement was determined using the long method of the gage repeatability and reproducibility (GRR). This produced not only the dispersion of the measurements but also estimates of the significance of various sources of dispersion. The capability index of the measurements was also determined. The statistical method of experiment planning was used to evaluate the significance of the different sources of variation in the freeness measurements performed in the pulp laboratory.

The measurement-inherent dispersions proved to be greater than expected, with dispersion remaining at under thirty percent in only a few measurement. The indices of capability varied considerably in the evaluated measurements due to differences in the determination of specification limits. Further studies should focus on the development of the measuring instruments and methods as well as on universal determination of the specification limits. The experimental part of the thesis also established the limits for monitoring the reliabilities of measurements by means of the control charts used in the statistical process control.

This info last modified 26 Sep 2017 - contact webmaster