Análise estatística de processo visando aumento da vida útil do revestimento de rolo aplicador de amido na produção de papel

Abstract

Additives are used in a paper production process, including starch. For application, the starch goes through a cooking process, and it is used in the process in order to improve the physical strength and appearance of the papers, increases solids retention and water drainage from the production system and, consequently, optimizes the process. The application takes place in two stages: in the preparation of the cellulose mixture that will be pressed and dried to become paper and, later, on the surface of the already formed paper. In this second case, the application of surface starch occurs through the flow of the solution around the upper fixed press roller of the Speed Sizer set, which consists of a pair of rubber-coated cylindrical presses that rotate with a linear speed of 1000 m/min, which the paper to be coated passes through. The objective of this graduation work was to perform a statistical analysis of process data in order to identify the variables that affect the useful life of the rubber coating of the fixed press of the set, looking forward to increasing its durability. For this, thirteen process variables were considered, known as: pressure on both sides of the press, inlet and outlet temperature of the starch in the system, operating temperature of the press and its head, pressure and speed of the starch metering bar, flow and system starch inlet and outlet viscosity, traction and press rotation speed. Data for each of the thirteen variables mentioned above were collected three times a day, during 9 months and seven days, for four different press rolls operating on the same machine and under the same conditions. Through the methodology of CART Classification (or decision tree algorithm) and Ordinal Logistic Regression, the variables that have the greatest influence on the process were identified, and values were suggested for their operation, aiming to increase the useful life of the rubber and reduce the system-related maintenance costs. In addition, statistical process control tools were applied and it was possible to identify that the system does not behave as a statistically controlled system, presenting high variability and resulting in failures in the tests of the I-AM control charts used in the analysis.