Prediction of longitudinal shrinkage and bow in Norway spruce studs using scanning techniques

Abstract

Straightness is one of the most important properties for making timber an attractive material for modern mechanized building. Several studies have shown that a lack of straightness is one of the main reasons for choosing materials other than timber in the construction industry. This paper presents a way to model moisture-induced bow from longitudinal shrinkage data predicted from an analysis of images of the surface of Norway spruce studs. For this study, eight studs (45 × 95 × 2500 mm and 45 × 120 × 3000 mm) of Norway spruce timber were selected. Bow in these studs was measured at two moisture contents below the fiber saturation point. The studs were then split into three slices 11 mm thick, and the surfaces of these slices were scanned to obtain color information and images of the tracheid effect. The slices were cut into sticks with dimensions of 10 × 10 × 200 mm. The longitudinal shrinkage coefficient of these sticks was measured. A multivariate model was created to model the longitudinal shrinkage coefficient data from the information in the images. The predicted longitudinal shrinkage data was used to model bow. The mean value of the measured longitudinal shrinkage was 0.0121 (SD 0.0123). The root mean square error of prediction (RMSEP) for the multivariate model was 0.0079, which is regarded as good. Thus, it was possible to model moisture-induced bow with good accuracy using the predicted longitudinal shrinkage data.