Optimization Of Machining Parameters For Improved Surface Finish Using Taguchi Method And Grey Relational Analysis

Abstract

Broaching is a high productivity machining process where the quality of the surface produced is greatly affected by tool geometry and cutting variables. In order to achieve a consistent surface finish, the geometry of the teeth and the process parameters of keyway broaching should be optimized. The following study was performed to determine the effects of cutting speed (V), rake angle (α) and clearance angle (δ) on the arithmetic mean surface roughness (Ra) of a broached keyway where 27 experimental observations are provided. A complete 3³ factorial matrix was used, with cutting speed of 7, 8 and 9 m/min, rake angle of 10°, 12° and 14°, and clearance angle of 0.45°, 1.0° and 1.5°. Because the supplied dataset contains one reported Ra value for each of the 27 combinations, the analysis was reconstructed as a full-factorial dataset rather than as a reduced Taguchi orthogonal-array experiment. For the smaller-the-better quality characteristic, signal-to-noise ratios were calculated and main effects were evaluated. The measured Ra values ranged from 0.27 to 2.02 µm, with the minimum observed value of 0.27 µm obtained at 7 m/min, 10° rake angle and 1.5° clearance angle. Main-effect averages showed that the lowest mean Ra occurred at 7 m/min (0.684 µm), 12° rake angle (0.793 µm), and 0.45° clearance angle (0.737 µm). The additive Taguchi main-effect analysis therefore indicates 7 m/min, 10° and 0.45° as the highest-S/N setting; however, the corresponding measured experiment produced 0.78 µm, demonstrating that the main-effect prediction should not be treated as a validated optimum without independent confirmation and replication. A main-effects ANOVA assigned 8.40%, 0.31% and 2.75% of the total variation to cutting speed, rake angle and clearance angle, respectively, with none reaching statistical significance at the 5% level under the unreplicated error model. The results emphasize the need for replicated measurements and confirmation experiments before establishing a production-level optimum.