Comparative study of biodegradable coolants on aluminum–lithium alloys using the design of experiment approach and machine learning model

Abstract

Environmentally friendly methods are essential to improve surface quality and machining performance while reducing negative environmental impacts. This has significantly affected the approach to cutting fluids for machining high-performance materials like third-generation aluminum–lithium alloys. In the current study, a comparative analysis of two types of lubricants, Trim E950 and glycol, was conducted. The five variables, namely, surface speed, feed rate, axial depth of cut, corner radius, and lubricant concentration, are selected to investigate surface finish and the heat-affected zone. A central composite design, a class of response surface methodology, was used to design the experimental runs with a 95% confidence level. Analysis of variance confirmed the goodness-of-fit, with feed rate and corner radius showing significant influence due to their higher F -values. SEM analysis indicated minimum surface irregularities at 8% glycol and 12% Trim E950 lubricant concentrations. The optimum Ra values obtained were 0.211 µm using Trim E950 and 0.21 µm using glycol. Similarly, the optimum heat-affected zone (HAZ) values achieved were 20.82 IACS for Trim E950 and 20.3 IACS for glycol. The Random Forest Regressor machine learning model was chosen to validate the experimental dataset. The model showed excellent predictive accuracy for Ra with R 2 score values of 0.98 and 0.97 for Trim E950 and glycol, respectively, whereas HAZ prediction exhibited moderate sensitivity to the training dataset size.