The Relationship Between Timber Chemical Compositions and Mechanical Properties

Timber is widely used both in modern and historic construction, with ageing being the most serious risk in the structural mechanical strength. A non-destructive test to predict timber mechanical properties is urgently needed and many studies indicated that chemical compositions have close relationships to the mechanical properties. Hence, this study focusses on the relationships between timber chemical composition and the corresponding mechanical strength, providing contributions to the non-destructive testing of timber mechanical properties. Non-destructive testing has a wide prospect both in evaluation of modern timber construction and historic timber framed building conservation. Heat treatment is a method to cause changes in timber, whilst FTIR is the technique for analysing its chemical compositions. Static and dynamic mechanical properties were tested by a 3-point bending and a dynamic thermal mechanical analysis (DMTA) facility, respectively.
The changes of timber mechanical properties are the results of various combined chemical compositions. In general, condensation and cross-linking reactions play an essential role in timber strength improvement. The static bending mechanical properties, modulus of rupture (MOR) and modulus of elasticity (MOE) increase, whilst in the dynamic mechanical properties and Tan δ decrease, which indicates an increase in elasticity and/or decrease of viscosity. Pyrolysis reactions in hemicellulose and lignin, lead to a decrease in the timber static mechanical properties and increase in Tan δ of the dynamic ones. Both static bending mechanical properties (MOR and MOE) and dynamic bending mechanical properties (storage modulus, loss modulus and Tan δ) can be predicted by the peak areas of the normalised FTIR spectrum. The coefficients of determination (R-square) of all the regression models are between 0.62 and 0.9, which indicates that the models are functional.
In studies of timber accelerated ageing, the changes of each peak area during heat treatment are regressed by a model where temperature and treatment period as independent variables. The model shows that new pine can be treated in a two-steps heat treatment, which involves an air and a vacuum step to obtain similar chemical compositions of 580 years old real timber by heat treatment. The R-square of the model is more than 0.7 and thus, shows effective regression.