Timber strength variation due to establishment density—can grading cope with it?

Abstract

Abstract High planting densities in coniferous stands usually lead to advantageous wood properties with high yields for strength graded timber in sawmills. In timber design, however, the focus is on mechanical wood properties, as structural engineers rely on the characteristic values for strength, stiffness and wood density defined for the strength classes. Almost 700 Douglas-fir boards from plots with establishment densities of 1000, 2000 and 4000 trees per hectare were graded with different strength grading machines before the boards were finally tensile tested. Three existing models including both the functional relationship and the class limits were applied to the tensile strength prediction: Modelknot based on knot size and wood density, ModelEdyn based on the dynamic modulus of elasticity (Edyn) only and Modelknot,Edyn based on Edyn and knot size were used to calculate the indicating properties (IPs) IPknot, IPEdyn and IPknot,Edyn. The ratio of achieved and required characteristic values for strength, stiffness and wood density was calculated for various T classes according to the European standard EN 338:2016. The focus was on two questions: (1) Do the samples from different densities meet the characteristic values of the respective T classes? (2) How does the complexity of the grading model affect the characteristic values of sawn timber with raw material planted at different initial densities? The results showed that the stiffness and wood density requirements of the T class profile according to EN 338:2016 were met for boards of all initial densities. Boards from higher and highest densities always exceeded the strength requirements, by on average $+$16 per cent and $+$36 per cent, whereas boards from low establishment densities fell short of the strength requirements by on average −15 per cent. Grading of multiple classes in one run also influenced the characteristic strength value. Applying the strength class combinations T15/T9, the negative deviations from the required strength value of the lower class T9 improved for boards coming from plots with an initial stand density of 1000 trees ha−1 from −14 per cent to −13 per cent (Modelknot), from −12 per cent to −2 per cent (ModelEdyn) and from −10 per cent to $+$6 per cent (Modelknot,Edyn). The non-compliance of the characteristic strength values at low initial density was only in a few cases below the limit value required for the test (90 per cent). The relative distance between the characteristic values within the three densities, nevertheless, was high. Complex grading models resulted in both better yields and low deviations from the requirements.