ISSN 3156-271X
Abstract
Chip removal is a critical factor in determining tool life in dry drilling, yet the chip reduction coefficient (CRC) which is a direct measure of chip removal remains underutilized in predictive process models. The present study develops and validates predictive models for CRC using Response Surface Methodology (RSM) as well as Artificial Neural Networks (ANN). Three input parameters (feed rate (0.03–0.37 mm/rev), cutting speed (106.36–173.64 m/min), and depth of cut (0.33–1.17 mm)) were investigated using a central composite design on mild steel workpieces. The 2nd order polynomial RSM model demonstrated excellent predictive capacity with R² of 0.9513, Adj. R² of 0.9075, and Adeq Precision of 15.603. ANOVA revealed that the interaction between feed rate and depth of cut (AC) was the most dominant factor (F = 60.32, p < 0.0001), followed by the quadratic effect of depth of cut (C²: F = 87.63, p < 0.0001). Numerical optimization yielded an optimal CRC of 1.80833 at feed rate 0.30 mm/rev, cutting speed 146.22 m/min, and depth of cut 0.50 mm, with a desirability of 93.1%. ANN achieved R² of 85.15% (Exp = 0.1264 + 0.8884 ANN). The models presented provide industry-ready tools for predicting chip morphology and optimizing parameters to prevent chip clogging and subsequent tool breakage.
Keywords: Chip Reduction Coefficient, Dry Drilling, Tool Breakage, Chip Evacuation, Response Surface Methodology, Artificial Neural Network