Otimização das condições de corte e simulação da confiabilidade da ferramenta no torneamento do aço ABNT 52100 Endurecido

Abstract

High-Speed Machining is recognized as one of the leading manufacturing technologies for achieving higher productivity rates and reduced production costs. In this context, optimizing cutting conditions emerges as an alternative to make it viable. Additionally, several strategies for optimizing surface roughness have been proposed in field hard turning. However, although the effects of tool wear on the surface roughness of the workpiece can be understood, this effect is often neglected by most of the strategies proposed. Consequently, the ability of the process to reproduce the results of the optimal solution in practice is negatively affected, since the surface roughness of parts machined with the same cutting edge deviates from the expected value as wear progresses. Additionally, analyzing the reliability of cutting tools in machining is essential for determining the appropriate time to replace them, thereby avoiding premature or delayed tool changes. Considering the presented scenarios, this study proposes a multi-objective optimization strategy aimed at maximizing tool life and minimizing surface roughness Ra, cutting time, and the total machining time and cost. This strategy is applied to evaluate the performance of a mixed ceramic tool during the dry turning of hardened ABNT 52100 bearing steel at high cutting speeds. Response Surface Methodology was used to obtain quadratic models for the evaluated responses. The effect of tool wear on surface roughness Ra was established during the tool life tests. As a result, the tool condition that generates the highest Ra values was identified, corresponding to the roughness values obtained under the experimental conditions. Dimensionality reduction of the multi-objective problem was achieved using Factor Analysis. The scores of the rotated factors retained in the analysis were used to derive the Mean Squared Error functions of the factors, which served as the objective functions for optimization. Multi-objective optimization was performed using the Normal Boundary Intersection method. The optimal setups defined by the optimization were able to achieve surface roughness Ra values between 0.28 and 1.05 μm, tool life between 11 and 19 minutes, cutting time ranging from 0.23 to 0.41 minutes, total machining time between 0.91 and 1.10 minutes, and machining cost per piece between R$3.55 and R$5.01. The variance of the tool's lifetime was obtained by applying Poisson regression to the square residuals of the model derived using ordinary least squares. In addition, a cost-effective method was proposed to simulate the tool's reliability to make its application viable. With the mean and variance of the tool life, the shape (β) and scale (δ) parameters of the Weibull distribution were estimated using a simple optimization method proposed with constraints. Confirmation experiments were conducted, which validated the effectiveness of the optimization and the tool reliability simulation.