Modeling bending strength of oil-heat-treated wood by near-infrared spectroscopy

Abstract

Refractory wood species like American beech (Fagus grandifolia Ehrh.) and pin oak (Quercus palustris) are difficult to be treated by impregnation with wood preservative formulations or chemical modification agents. They need to be subjected to a preservative process that uses a heat treatment in a deep fryer with vegetable oil at 220 °C for 120 min. Such heat treatment influences several parameters related to the chemical and mechanical wood properties, such as the bending strength of earlywood (Ew) and latewood (Lw) growth rings. In this study, models to estimate bending strengths of Ew and Lw growth rings from near-infrared spectroscopy (NIRS) data have been proposed for oil-heat-treated wood samples of American beech and pin oak. The models were developed using the partition to latent structures (PLS) regression technique. For both species, the bending strength was significantly lower for the heat-treated wood compared to the untreated wood. In the case of beech, the mean bending strength was 112.87 ± 8.55 MPa for the untreated wood samples and 80.78 ± 4.79 MPa for the heat-treated wood samples. In the case of pin oak, the mean bending strength was 99.57 ± 5.34 MPa for the untreated wood samples and 83.69 ± 5.71 MPa for the heat-treated wood samples. All the PLS models were significant at p value < 0.0001. For the Ew tissues, the R2 with the validation data set was 0.78 for beech and 0.95 for pin oak. The corresponding root mean square errors of prediction (RMSEp) were 44.57 MPa for beech and 34.10 MPa for pin oak. The corresponding ratios of prediction to deviation (RPD) were 2.2 and 4.8 for beech and pin oak, respectively. For the Lw tissues, R2 was 0.86 and 0.92 for beech and pin oak, respectively. The corresponding RMSEp were 50.01 MPa for beech and 47.87 MPa for pin oak. RPD values were 5.8 and 3.8 for beech and pin oak, respectively. For both the Ew and Lw tissues and species, B coefficients of the response surface models (ranging from − 108.8 to 195.8) were statistically significant indicating that the models can be used to detect which wavelengths of the spectra have the highest impact on the effect of heat treatment on the bending strength.