Emprego de ferramentas estatísticas para avaliação da usinabilidade de ligas de alumínio

Abstract

The main objective of the present work is to study the effect of the input variables: mechanical properties of the workpiece (tensile strength, hardness and elongation) and cutting conditions (cutting speed, feed rate, depth of cut and lubri-cooling system) over the output variables: cutting forces, power consumption, cutting temperature, surface roughness (Ra, Rz and Rq parameters) and toolholder vibration in turning of aluminum alloys: 1350-O, 6082-T4, 6262-T6,7075-T73 e 7075-T6. This set of alloys covers a large range of properties of the aluminum alloys. A 2k factorial planning and a Central Composite Design (CCD) were used to determine the machining test conditions. After ANOVA and Regression analysis of the results, 2nd order models were generated, relating the input and the output variables for each aluminum alloy and also globally (considering the properties of the alloys). Surface Response Method SRM, level curves and gradient vector generated after the models were used to present and analyze the influence of the input variables and their interactions on the output parameters. Validated models used in a multi-response optimization process (genetic algorithm) identified the levels of the input variables that simultaneously optimize all the output parameters. The most important contribution of this work on the machining of aluminum alloys lies on its efficiency of describing the behavior of each machining characteristic (output variable) as a function of the input variable interactions within the range studied. Many researches on machining of aluminum alloys describe, in an isolated form, the behavior of the machining characteristics (output parameters) as functions of a few or only one factor, limiting the conclusions about the machinability of these alloys. In the present work it was observed that generally the machining output parameters varied individually according to the literature, however, new behavior, due to the interaction of the input variables, so far not described by previous work was presented.