Response surface methodology for advanced manufacturing technology optimization: theoretical fundamentals, practical guidelines, and survey literature review
Lucas Guedes de Oliveira,Anderson Paulo de Paiva,Pedro Paulo Balestrassi,João Roberto Ferreira,Sebastião Carlos da Costa,Paulo Henrique da Silva Campos
DOI: https://doi.org/10.1007/s00170-019-03809-9
IF: 3.563
2019-06-18
The International Journal of Advanced Manufacturing Technology
Abstract:Process optimization normally involves the combination of mathematical and statistical techniques which can be approached by distinct ways. Despite the fact that different methods can be found in the literature, the response surface methodology raised as one of the most effective ways for performing process optimization, by combining design and analysis of experiments, modeling techniques, and optimization methods. However, practical guidelines for response surface methodology and critical analysis of its applications are quite scarce. Thus, this paper aims to present the theoretical principles and practical guidelines for carrying response surface methodology as well as to provide empirical evidence on its critical aspects for manufacturing optimization. In order to accomplish with this objective, 49 papers published in the (IJAMT) from 2014 to 2017 were investigated and reproduced, allowing the analysis of 123 response surfaces. Surprisingly, more than 75.29% of the models have presented a saddle shape. The practical meaning of this finding is that the stationary point is not a suitable solution for the optimization of those surfaces. Besides, multiple response surfaces are more commonly found in the literature than individual ones. From this amount of papers, 71.88% of the works investigated have presented significant correlation with their peers and 87.61% have convexity incompatible with the optimization direction. Most of the optimization solutions have found outside of experimental region which reveals a preponderant neglect of the nonlinear constraints involving the definition of the experimental region. It was also verified that the proportion of functions in saddle format corresponds to 96.86% of the models estimated in flat regions. Moreover, it was found that the number of center points is commonly changed and all the manufacturing processes investigated are driven by at most five control parameters. Finally, considering the theoretical principles, the practical guidelines, and the obtained results, a follow-along example involving the optimization of AISI H13-hardened steel turning with PCBN wiper, previously published by the authors in IJAMT, was revisited by using response surface methodology. The results corroborated the proposed framework suitability.