Rendimento no desdobro de toras de Andira parviflora Ducke. em serraria na Amazônia Central

Abstract

The sustainable of amazonian forestry wooden resources utilization has an intrinsic relationship with its yield and exploitation of timbers in the sawmill. It is of great importance for the management of this industries the knowledge about the yield, wich allow the optimization of the production of sawn wood. The objective of this study was to determinate the yield in the logging process of Andira parviflora Ducke., to verify possible yield differences among diameter classes, to verify the generated residues and to adjust mathematical models to estimate the future yield of lumbering. A total of 56 logs were sampled and grouped into 4 diameter classes represented by numbers 1, 2, 3, 4. For the log volume evaluation was held strict cubage by the Smalian method, next was evaluated the sawn wood volume for subsequently calculate the yield and the residues volume generated in the sawing process. The average yield of sawed wood in the processing logs of Andira parviflora Ducke. was 29.52% with 1,12% confidence level and presented statistical differences only between diametric class 1 in relation to the other 3 classes studied. Among the 4 classes studied, class 4 was the one with the highest yield value. 12 models were used to estimate the yield of sawn wood, being 6 linear and 6 non-linear models, being the yield in sawn wood (R) the dependent variable in function of the independent variables log diameter (D), log length (L) and log volume (Vt). All relationships between the variables were analyzed based on the Pearson correlation values. The quality of adjustment and selection of the models was based on the higher adjusted coefficient of determination (R²ajust), lower standard error of the estimate (Syx%) and homogeneous graphical distribution of the residues. Among the models tested to estimate yield in sawn wood of Andira parviflora Ducke., the best was the model 4, linear of double entry, but the model 3, linear of simple entry, also showed a good quality of adjustment. The nonlinear models 7, simple entry and 8, double entry, obtained adjustment quality similar to models 3 and 4, being these 4 models the most suitable for estimating the yielding of sawn wood of Andira parviflora Ducke.