The use of calibrated mid-infrared Fourier transform (FTIR) spectroscopy for predicting the main bending strength parameters of thermally modified wood is described for the first time. The analysis is based on the handling of OSC-filtered FTIR data from control and treated wood samples, using partial least squares (PLS) regression. Sets of Norway spruce, Scots pine and Beech wood samples were thermally treated to 20 schedules, and then characterised in bending according to BS373:1957 in terms of the module of rupture, modulus of elasticity and six other bending strength parameters. The equilibrium moisture content at test, the nominal density, and the heat-induced weight loss are also reported. FTIR spectra were used to develop calibration models for the parameters listed, for each of the three wood species, and for the three species combined. The relationship between the laboratory-determined properties, and the FTIR-fitted ones was very good in all cases. All FTIR calibrations for individual sets had goodness of fit (R2Y) and accumulated goodness of prediction (Q2CUM) statistics above 0.98 and 0.96, respectively. FTIR spectra of independent samples were next obtained, and the bending strength parameters predicted using the calibration models. Predicted values from calibration models were in good agreement to those determined by actual testing. Except for three properties in pine, ratios of root mean square error of prediction (RMSEP) to standard deviation (SD) varied from 0.13 to 0.88. In addition, calibrations containing from 8 to 10 wavenumbers instead of spectral ranges 3500-2800 and 1800-550 cm-1 were also attempted. The reduction of predictors from 492 to the 8 to 10 more significant ones led in most cases, to small increases in the RMSEP, highlighting the possibility of using a simple, hand-held device for the prediction of mechanical and other physical properties of modified wood at a larger scale. In another experiment, sets of treated samples were used to investigate the anti-swelling efficiency (ASE) of modified Spruce wood using diffuse reflectance mid-infrared Fourier transform (DRIFT) spectra. R2Y and Q2CUM were both 1.00 for the calibration model. ASE predictions using DRIFT spectra of independent samples, was successfully accomplished, with a ratio of RMSEP/SD of 0.37.

Download document (338 kb)
free for the members of IRG. Available if purchased.

Purchase this document