Optimizing drilling parameters for unidirectional glass fiber/nanoclay-epoxy matrix composites using gray relational analysis and response surface methodology

In this innovative study, the drilling performances of laminated composites containing different ratios (0, 1, 3, and 5 wt% ) of nanoclay were examined and optimized using the Multi-Criteria Decision-Making method. Control factors included material (NC), feed rate (f), cutting speed (Vc), drill bit (D), were considered and an appropriate experimental design was made. Thrust force (Fc), cutting tool temperature (T), and delamination factor (Df) were determined as quality characteristics. Delamination in the holes was measured using imaging analysis, and a two-dimensional (2D) delamination factor based on the nominal diameter was calculated. As the nanoclay ratio increased, the Fc increased by over 50% due to the increase in friction and material strength. Conversely, as the drill tip angle decreased, a relative decrease was observed in the Fc and T values. According to the quality characteristics, NC was found to be the most effective control factor, with 65.53% and 70.74% for Fz and T, respectively, while D was found to be the most effective control factor with 41.23% for Df. Using Gray Relational Analysis, the optimum drilling parameters were found to be a pure composite material, a drill bit with a 90° tip angle, 140 m/min cutting speed, and a 0.04 mm/rev feed rate. The results calculated with the mathematical models obtained using the Response Surface Method were tested with a series of verification experiments. The relative error values of the results obtained from these experiments and the results measured from the models were calculated as approximately 1% for Fz, nearly 0.54% for T, and almost 1.48% for Df. The results calculated with the mathematical models obtained using the Response Surface Method were tested with a series of verification experiments. The relative error values of the results obtained from these experiments and the results calculated from the models were calculated as approximately 1% for Fz, approximately 0.54% for T, and approximately 1.48% for Df.

Graphical abstract