Parametric Optimisation of CNC Face Milling Process Parameters for Surface Roughness and Material Removal Rate on Mild Steel Using Taguchi L9 Orthogonal Array and ANOVA

Abstract

Surface roughness and material removal rate are the two most critical performance metrics in CNC milling operations, determining workpiece quality and machining productivity respectively. This paper presents a systematic experimental study of face milling of mild steel (IS 2062 Grade E250) on a Vertical Machining Centre (VMC) to simultaneously optimise surface roughness Ra and material removal rate (MRR) using the Taguchi L9 orthogonal array methodology. Three process parameters — spindle speed (500, 1000, 1500 rpm), feed rate (75, 150, 225 mm/min), and depth of cut (0.5, 1.0, 1.5 mm) are studied at three levels each using a three-factor L9 design, requiring only nine experiments compared to 27 for a full factorial design. Signal-to-Noise (S/N) ratio analysis, Analysis of Variance (ANOVA), and main effects plots are used to identify optimal parameter settings and quantify the percentage contribution of each parameter. ANOVA results identify spindle speed as the dominant factor for surface roughness Ra (56.8% contribution), followed by feed rate (29.2%), while feed rate dominates MRR (62.4%). The optimal parameters for minimum Ra are spindle speed 1500 rpm, feed rate 75 mm/min, and depth of cut 0.5 mm, achieving Ra = 1.58 µm in confirmation experiments — a 44.4% improvement over the initial baseline setting. A regression model for Ra is developed with R² = 0.938. The conflicting optimal settings for Ra and MRR are resolved using a weighted grey relational grade approach, yielding a compromise setting that reduces Ra by 32.1% while increasing MRR by 83.2% relative to the worst-case initial condition.

Cite as:

Wakurde P B. (2026). Parametric Optimisation of CNC Face Milling Process Parameters for Surface Roughness and Material Removal Rate on Mild Steel Using Taguchi L9 Orthogonal Array and ANOVA. Advancement in Mechanical Engineering and Technology, 9(2), 32–39. https://doi.org/10.5281/zenodo.21820065