Avaliação da madeira de Eucalyptus benthamii Maiden & Cambage por espectroscopia no infravermelho próximo, com vistas à produção de celulose

• Autor(a) principal: Baldin, Talita
• Data de Publicação: 2018
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Manancial - Repositório Digital da UFSM
• dARK ID: ark:/26339/0013000008hg0
• Texto Completo: http://repositorio.ufsm.br/handle/1/14730
• topic: Anatomia da madeira; Avaliação não destrutiva; Modelos de calibração; Qualidade da madeira; Química da madeira; Wood anatomy and chemistry; Near-Infrared Spectroscopy; Non-destructible technique; Prediction models; Wood quality; CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL

Analysis of wood quality is a routine practice in the cellulose industry. However, most traditional methods of analysis are time-consuming, expensive, and require descruction of trees being sampled. Therefore, the use of non-destructive techniques has become of great interest for characterization of wood quality. Additionally, non-destructive tecniques have great application in genetic improvement programs, since they allow selection of trees that are still in the stands. Near Infrared Spectroscopy (NIRS) is a non-destructible, fast, and reliable technique that can be applied in many different samples. NIR has been shown to be an efficient tool in determining the chemical, anatomical, physical, and mechanical properties of wood. The aimed of this study was to characterize the wood of Eucalyptus benthamii as to its anatomical and chemical constitution. Aditionally, the potential of using Eucalyptus benthamii wood to produce cellulosic pulp was estimated by NIRS. Evaluation of anatomical and chemical traits of the wood was carried out following traditional methodologies commonly used in laboratories. For the development of the calibration models 87 trees were used (75 of E. benthamii, 4 of E. dunnii, 4 of E. grandis, and 4 of E. saligna). A portion of the sample was used for analysis of ash, extractives, lignin, holocellulose, basic density, and cellulosic pulp yield. Another partion was milled and used to acquire the spectra, which were later correlated to laboratory values. Calibration of the model was determined by partial least squares (PLS) regression analysis. Selection of the best models was based on the following statistical criteria: coefficient of determination (R²), mean cross-validation error (RMSECV), residual forecast deviation (RPD), and number of latent variables (VLs). The values observed for anatomical and chemical traits of Eucalyptus benthamii wood remained within the range considered apprpriate for cellulose production. NIRS calibration models presented satisfactory adjustments for pulp yield (R2 = 0.83), wood density (R2 = 0.73), holocellulose content (R2 = 0.82), total lignin content (R2 = 0.74), and Klason lignin content (R2 = 0.82). The NIRS models developed in this study present a viable commercial tool for carachterization of samples of Eucalyptus benthamii wood for the cellulose industry. Noteworthy, CMPC - Celulose Riograndense has implemented these prediction models in their program to accelerate the genetic improvement of E. benthamii.