Modelagem do rendimento no desdobro de toras de eucalipto cultivado em sistema agroflorestal

Abstract

This work fitted regression models in order to estimate the efficiency of wood processing in sawmills using trees from agroforest systems. Specifically, it intended to determine the yield of processed boards volume, to adjust models for measuring the processing efficiency and the time needed for processing wood as a function of dendrometric variables, considering two different products. The database was composed of information derived from 150 logs of eucalypt clones encompassing 5 diameter classes. The real volume of each log was derived through detailed log scaling according to Smalian s method. After log processing, the boards were measured to determine the volume of sawn wood. The time needed for processing and the log dimensions were used to fit efficiency models as a function of time and of dendrometric characteristics. The equations were evaluated based on the coefficient of determination, standard error of estimation and graphic distribution of residues. The equations to estimate the processing efficiency were more influenced by log volume and by the diameter of the largest side. The best equations to estimate the volume of the boards were the ones that included log volume, diameter of the thickest side and log length as independent variables. When considering both products together, the best equations were the ones that included log volume, diameter of the thickest side, diameter of the thinnest side and the type of product as independent variables. The equations used to estimate the time needed to saw the logs were influenced by diameter class, in the case of product 1 and by the volume of the logs in the case of product 2. The equations used to estimate the time needed to process both products together showed similar performance. The time needed was influenced by the dendrometric variables and their combinations.