Otimização do processo de torneamento de aços endurecidos ABNT 52100 usando modelos de confiabilidade

Abstract

This paper deals with a modeling and optimization study of an ABNT 52100 hardened steel turning process using a reliability model. When applying the simultaneous optimization in the variables, one of quality and another of reliability, from the control arrangements with weight distribution, a set of Pareto-optimal solutions was generated. However, it was necessary to apply a method to prioritize the best response. Thus, this research was developed with the objective of applying the Ordinary Least Squares (OLS) method to the Surface Roughness regression function. The maximum likelihood method (MV) was applied as the most suitable method, since most of the studies use the Ordinary Least Squares (OLS) method for modeling the regression function reliability variables. When applying multiobjective optimization through the Normalized Global Criterion Method (MCGN), Pareto-optimal solutions were generated. In order to find out which the optimal solution of the set of Pareto-optimal solutions was, another method, the Global Percentage Error (EPG) method, was applied, generating an optimal point with the following configurations: The Cutting Speed (Vc) had the value equal to 209.43 m / min, the feed rate (Fn) equal to 0.0402 mm and the cut depth (Ap) equal to 0.11745 mm, resulting in optimized values of the following response variables, a roughness ratio (Ra) equal to 0.2866 mm and Tool life (T) equal to 16.5712 min. These results were the best obtained through simultaneous optimization, for better decision making, than the results obtained with the individual optimization of each response variable.