AbstractsBusiness Management & Administration

CT log scanning for sawing optimization with regard to the aesthetic quality of wood

by Lorenz Breinig




Institution: Universität Freiburg
Department: Umwelt und Natürliche Ressourcen
Degree: PhD
Year: 2015
Record ID: 1098334
Full text PDF: http://www.freidok.uni-freiburg.de/volltexte/2015/9994/


Abstract

Wood is a renewable material with favourable properties both in technical and aesthetic aspects. The natural origin of wood, however, also entails a considerable variability of its properties including the occurrence of internal wood features such as knots. With log breakdown in sawmills, the properties of the roundwood are largely retained in the resulting sawn products. Thus, controlling the sawing process in a way that the properties of the sawn products are in line with the demands for their particular use is a decisive factor for value creation for the sawmill industry. This can be achieved by individual breakdown optimization of each log which in turn requires scanning for internal log properties. X-ray computed tomography (CT) has been identified as the most suitable technique for this purpose. Since knots are the most frequent internal wood feature and thus most often determine the quality of sawn timber, the accuracy of automated knot detection in the CT image of a log is critical for the performance of optimizing its breakdown. Therefore, it is relevant to estimate both the achievable knot detection accuracy and the impact of knot detection errors on sawing optimization. In the scope of this thesis, first, a basic part of a knot detection algorithm was evaluated in terms of size measurement accuracy. The results of this evaluation, together with the results of comparable investigations, suggest that a size measurement accuracy slightly better than 5 mm can be attained. Furthermore, in a sensitivity analysis, the impact of different knot detection errors on the performance of sawing optimization by adjustment of log rotation was studied by means of sawing simulations. Errors in knot diameter, dead knot border position and knot rotational position were tested in different magnitudes as systematic and random errors. Knot diameter errors were found to most severely affect the realization of value optimization potential. At higher levels, errors in dead knot border position had a considerable impact, too. In addition, it was observed that other parameters of the roundwood conversion process, such as the price differentiation with product grade, had a major influence both on the effectiveness of sawing optimization and on the sensitivity to knot detection errors. When wood is used for visible applications, its aesthetic quality, i.e. the visual appearance of the wood surface perceived by a human observer, is the key quality aspect. Specifically optimizing log breakdown for the perceived visual quality of the sawn products might thus hold some potential for further added value when they are designated for visible use, such as floorboards or panelling. Hence, it was investigated within this thesis, whether a classification of board faces based on feature-pattern variables can be found that is consistent with visual perception of their appearance. The results indicated that the manual appearance sorting could be approximated quite well by the variable-based classification after pregrouping of the boards according to…