AbstractsEngineering

The overall objective of this thesis was to create a system for characterizing the material properties of wood, analogous to those used to specify other materials. In order to obtain homogenous materials prediction methods were evaluated. Therefore, one objective was to evaluate the scope for using structural wood properties to predict mechanical properties (Study I) and further develop the functions (Study IV). Another objective was to evaluate the effects of two different silvicultural regimes on the wood structure characteristics and mechanical properties of different wood tissue types at different heights in the tree, and reasons for these differences (Study II). In addition, a simple, fast and reliable light microscopy method for analyzing microfibril angle was developed, and its utility was assessed (Study III). Further objectives were to evaluate the utility of biorthogonal partial least squares regression models, using near infrared NIR spectral data, for predicting material properties and classifying wood as juvenile or mature wood (Study V). The papers (I, II, IV and V) are also used in this thesis as a basis to discuss how to use structural properties and near infrared spectroscopic data to characterize the material properties of trees in the forest. In Study IV, in total 200 clear wood samples from 24 Scots pine (Pinus sylvestris L.) trees in 18 stands were used in the analyses, while 100 samples from eight trees in two stands were used in studies I – III and V. The samples (especially those used in Study IV) spanned most of the ranges and combinations of structural and mechanical property values found in Scandinavian forests, thus diminishing the risk of random covariance between predictors and properties of interest. The ability to predict a mechanical property generally increased for each of the three examined resolution levels of determination of structural properties with r2 values often reaching 0.9 at the highest resolution (Study I and IV). Silvicultural regimes clearly have great potential to influence material properties, in terms of both differences in average levels and trends in vertical and radial directions within trees (Study II). The method developed for determining microfibril angles provided results that were almost identical to those obtained using a commonly applied method, but it was easier to perform and thus cheaper (Study III). The modelling of near infrared spectra was able to predict both mechanical and structural properties at high accuracy (Study V). The possibilities with proposed application model modelling material properties of individual annual rings are great combining, e.g., a combination of several material properties according to a property table. Thus, wood could be matched, according to its specific material properties, to appropriate products, and thus increase the income for all parts throughout the forest-wood chain.