Inline measurements and control of paper Web strength, ILSE

Abstract

The objective of this thesis was to investigate the feasibility of the patent [1] where the most defined input of the studied process is known as Elastic modulus. The implementation of inline measurement system would help a direct control of the paper quality online and therefore react to its instant changes by adjusting E-modulus online. To achieve this goal, different existing control techniques and assumptions were used on the process whereby the poperoller provides the energy to stretch the paper and the amount of stretch created into the paper is due to the speed difference between the last drying cylinder group and the poperoller. The simulation results proved that it is unsuitable to use the patent model whereby E-modulus is calculated based on theory of Young modulus, on the theory of stress to the strain. The quality specs that were used to train and validate the model are bursting strength; short compression test (SCTcd) in cross machine direction and concora medium test (CMT30). It was noted that the number of things might cause the low correlation or there may be the possibility of no-correlation at all. Indeed, it was found that the raw data give much more good prediction than E-modulus. This is because the information we normally want from the data is due to the contribution of individual variable in the process within the specified field of study. Therefore, it is undoubtedly known that good data gives good predictions, which result in the good model. One-month data was used for the modal analysis where two minutes was found to be the best time stamps to retrieve the online data in the system and therefore have the correlation with the measured paper quality in the laboratory. However there were no study done on the data storage system.The study trusted the storage system and proceeds to the analysis of the stored data. Since the result from data set are bad, it good to react on data storage in the paper factory and look on the sampling time that can be better than the chosen time before the end of Jumbo reel. Another important notice is about the direct use of raw materials usually used to increase the paper strength like starch. It is possible that the contribution of E-modulus will be assisted with other strength additives material for the recycled paper on the factory. Furthermore, the testing environments of the paper online are very different with the measured samples offline. This may cause low correlation together with the combination of other uncontrolled nonlinearities on the factory. To come up with those conclusive ideas; a research work was structured into three main parts. The first part was the identification of process variables and the feasible way of accessing the required data for process analysis. At this stage, a number of process variables, which are not directly measured, it was the time to investigate the way of measuring them during its second part of my research. The example was thickness. Referring to thickness testing, three types of sensors were reviewed but finally they were not implemented. Another technique with good assumptions was used due to its performance. It was a relation between already installed grammage sensor and thickness by assuming constant density of the paper. The final part of the thesis is devoted to PCA and PLS model design. The PCA model give both the correlation between process variables without taking into account the response variables and number of principal components, which represent the most of the variance in the data set and thus contribute to the dimension reduction of the process variables without losing too much information in the data set. The PLS simulations based on the PCA results proved the very low correlation between inputs where E-modulus is used and the qualities which thus makes the patent unsuitable for other paper mills