Understanding the effects of machinability properties of Incoloy 800 superalloy under different machining conditions using artificial intelligence methods

Abstract

Incoloy 800 is frequently used in high-temperature applications as it has the ability to retain good metallurgical stability at elevated temperatures. Due to the nature of the applications used for, parts made from Incoloy 800 usually require different machining processes such as milling and turning. Therefore, the current study aims to investigate the milling performance of Incoloy 800 under different cutting parameters (75–150 m/min and 0.075–0.15 mm/rev) and cooling conditions namely dry, flood, Minimum Quantity Lubrication (MQL) and Cryogenic (Cryo)+MQL. It was observed that all machinability metrics improved in the MQL+Cryo C/L environment. It is noticeable that the surface roughness value improved by 30% in this environment. In addition, a model based on artificial neural networks (ANN) and particle swarm optimization (PSO) was proposed to analyze the results and predict optimum cutting parameters. It appears that Cryo+MQL strategies are the best option for all cutting parameters. It was found that the estimations for surface roughness, flank wear, and cutting temperature with the proposed ANN architecture are achieved with overall relative error of 6.08%, 12.38%, and 8.32%, respectively. The proposed model resulted in good performance between the experimental test data and the predicted values. The developed model made the most efficient predictions for the MQL+Cryo cutting environment. It was observed that the estimations of the different input parameters in the MQL+Cryo cutting environment present a relative error of 8.36%, 1.46%, and 2.38% for surface roughness, flank wear, and cutting temperature, respectively. By utilizing the predictive capability of the trained ANN model, the optimization of the input parameters was carried out with the PSO technique. Thus, with the developed PSO-ANN model, promising findings were obtained in overcoming important handicaps such as time and cost in experimental studies. © 2024 Elsevier Ltd