Multi-objective optimisation of drilling process variables of LM5 alloy using grey relational analysis

LM5 alloy, with its excellent durability against corrosion, is ideal for marine and aesthetic applications. This investigation aims to improve the drilling process variables of the LM5 aluminium alloy to fulfil the specific needs of the naval industries, including naval construction, offshore buildings, and undersea vehicles. In order to gain a greater insight into how drilling process parameters like feed rate (FR), spindle speed (SS), and type of drill material (DM) affect responses such as burr height (BH), surface roughness (SR), and thrust force (TF) while drilling. The LM5 aluminium alloy plate was made using a stir-casting method. Drilling studies were executed utilising a design-of-experiment (DOE) technique L9 orthogonal array (OA). Thrust force was measured with a piezo-electric dynamometer. The surface roughness was assessed with an SR tester, and the BH was determined with a vision measurement instrument. GRA simultaneously determines the ideal process variables for machining LM5 alloy with the lowest TF, SR, and BH. It is observed from the ANOVA table that spindle Speed is the most effective control factor for maximum GRG (58.31%), feed rate (29.9%) and Drill material (9.84%). As a result of GRA, the Spindle Speed was identified as the most critical control element. The statistical models are created using the multiple linear regression method and employed to determine response parameters due to multiple control factors. The model’s accuracy was successfully validated using confirmation tests. LM5 aluminium alloy is highly corrosion-resistant, making it ideal for marine and aesthetic applications. LM5 alloy plates were produced using the stir-casting method. Drilling experiments followed a design of experiments (DOE) approach using an L9 orthogonal array. Grey relational analysis (GRA) was used to identify optimal drilling conditions minimizing TF, SR, and BH. Non-linear regression models were developed to predict response variables based on control factors. Confirmation tests validated the accuracy of these models.