Sustainable machining of high-performance Ni-Fe-Cr superalloys: Experimental evaluation and machine learning modelling under duplex-nozzle nano-MQL

Abstract

Sustainable machining of Ni-Fe-Cr superalloys remains a critical challenge due to their high strength and poor thermal conductivity. Although vegetable-oil-based minimum quantity lubrication (MQL) has shown promising performance, its potential is further enhanced with nanoparticle additives. However, duplex-nozzle MQL and graphene oxide (GO) nanofluids have been scarcely explored for machining Ni-based alloys. Addressing these research gaps, the present study develops a groundnut-oil-based GO nanofluid and experimentally evaluates its performance in turning Incoloy 800HT under duplex-nozzle nano-MQL. Surface roughness, tool wear, power consumption, carbon emissions, and chip morphology were analyzed across varying cutting conditions. The GO nanofluid demonstrated superior lubrication and cooling capabilities, achieving surface roughness as low as 0.231 µm and reducing flank wear to 0.042–0.129 mm through improved heat dissipation. Cutting speed strongly influenced both wear and power consumption, which ranged from 310.3 to 694.5 W. Carbon emissions decreased with higher speed-feed combinations, and chip morphology indicated stable cutting under a duplex nozzle nano-MQL. Additionally, machine-learning algorithms, namely random forest (RF) and Gaussian process regression (GPR), were implemented for predicting and modeling machining responses. The RF yielded coefficient of determination (R²) values of 0.975, 0.963, 0.962, and 0.965 for Ra, VBc, Pc, and Ce, respectively. Whereas GPR provided higher accuracy with R² values of 0.99, 1.00, 0.99, and 0.99 for the same-machining responses. In addition, the prediction accuracy was verified through mean absolute error (MAE). Based on R² and MAE results, GPR outperformed RF, therefore, GPR is recommended for future application in machining research.