Modelagem do rendimento no desdobro de toras de Manilkara spp. (sapotaceae) em serraria na nova fronteira madeireira do estado de Roraima, Brasil

Abstract

The yield in sawing logs has intrinsic relationship to sustainability in the use of forest resources in the Amazon and is of paramount importance to the business, as it enables improved planning, optimization and control of the production of the sawmill. The aim of this study was to estimate the yield of sawing logs Manilkara spp., quantify the waste generated, evaluate whether there is a difference in the yield between the diameter classes and adjust models to estimate the yield in lumber and volume of hollow logs. Seventy-one logs were sampled and grouped into diameter classes, and was made the cubing Smalian method for calculating the volume of logs and calculated the volume of lumber for determining yield and volume of waste generated in the sawing of logs Manilkara spp. The average yield in sawing logs of Manilkara spp. was 30,1% and showed no statistical differences in yield between the diameter classes studied and among the hollow logs and non-hollow logs. Among the classes studied, the class 5 was the one that represented the maximal yield. To estimate the yield of sawing lumber, seven linear models and five nonlinear models were tested, using as independent variable the smaller diameter of the log (Dpf), length of the log (L) and volume of log (Vt). To estimate the volume of the hollow logs, were tested eight linear models and four nonlinear models, using as independent variable the smaller diameter of log (Dpf), hollow diameter (Do), length of the log (L) and volume of log (Vt). All relationships between variables were analyzed based on the values of the correlation of Pearson. The model selection was based on the highest adjusted coefficient of determination (R2ajust), lower standard error of estimate (Syx%) and homogeneous distribution of the residuals values. Among the tested models to estimatte the yield of sawing lumber, the best model was the number 7, of double entry, because the simple-entry models did not obtained good statistics. To estimate the volume of hollow logs, the models with the best statistics presented were the model 2 and model 10, of double entry; and model 3, of single-entry, because of the convenience in the collection of the independent variable.