Modelagem da produtividade do baldeio de eucalipto com uso de Forwarder

Abstract

Regarding forest harvesting, Brazil has made significant progress in mechanization, but there is still a need to carry out studies with new variables influencing the extraction process with Forwarder. This work aimed to model the productivity in the extraction of wood in stands of Eucalyptus spp. using the Forwarder depending on the independent variables Average Individual Volume (VMI) of the forest, Extraction Distance (DE) and Terrain Slope (DT). 810 production units (UP) were used to cultivate Eucalyptus spp. of two units of a forestry-based company, located in the municipalities of Mucuri, Bahia, and Aracruz, Espírito Santo. The total volume of the areas corresponded to 4,602,717.69 m3 of wood. The harvesting system used in the operation consists of Cut- to-Length. All information used in the study was collected in the database provided by the company. For data analysis, only areas with production of barkless wood logs, measuring seven meters in length, were considered, with a margin of variation of 20 cm more or less. To model the Forwarder's productivity, VMI, DE and DT were used as independent variables, and as a dependent variable, the productivity of machine, expressed in cubic meters for each hour worked (m3h -1 ). The identification and exclusion of discrepant observations (outliers) were carried out before adjustment using a box plot. In order to obtain more homogeneous samples, the data aggregation technique was applied. Model adjustment was performed using multiple linear regression with a second-order (quadratic) model in Microsoft Excel 2016® software. The quality of the model was assessed using the correlation coefficient (R2 ), adjusted correlation coefficient (R2 adjusted), standard error of the estimate (Syx), standard error of the estimate in percentage (Syx (%)) and graphical analysis of the error. During the evaluation period, the operation achieved an average productivity of 59.94 m3h -1 . The results of the regression analysis indicate that the proposed model has a correlation coefficient (R2 ) of 0.6151 and an adjusted correlation coefficient (R2 adjusted) of 0.5876. The standard error of estimate (Syx) was calculated at 6.9274 m3h -1 and standard error of estimate in percentage terms (Syx%) was calculated at 12.50%. VMI was the variable with the greatest influence on Forwarder productivity. Taking as a basis the slope of 0° and the VMI class with the greatest representation in the study (from 0.100 to 0.200 m3 tree-1 ), it was observed that the increase in productivity can vary from 4.93% to 5.43% on average with relation to the previous class (< 0.100). It was observed that the higher the VMI, the greater the increase in productivity at higher DEs. It was noted that there is an inflection point in the curve when the VMI reaches 0.280 m3 tree-1 . The productivity modeling carried out in this study managed to explain 61.4% of the variations in Forwarder productivity. Given the complexity and number of variables selected, the model developed demonstrates an efficient forecasting capacity for productivity of extraction of woods using the Forwarder.