Segmented Cutting Coefficient Based Chatter Modeling for High-Speed Micromilling of Ti6Al4V

Abstract

Miniature components with complex shape can be created by micromilling with high surface accuracy. However, for difficult-to-machine materials, such as Ti-alloys, failure of low flexural stiffness micro-tools is a big limitation. High spindle speeds (20,000 to 100,000 rpm) can be used to reduce the undeformed chip thickness and the cutting forces and hence the catastrophic failure of the tool can be avoided. This reduced uncut chip thicknesses, in some cases lower than the cutting edge radius, can result in intermittent chip formation which can lead to dynamic variation in cutting forces. These dynamic force variations coupled with low flexural rigidity of micro end mill can render the process unstable. Consequently, accurate prediction of forces and stability is essential in high-speed micromilling. Most of the previous studies reported in the literature use constant cutting coefficients in the mechanistic cutting force model which does not yield accurate results. Recent work has shown significant improvement in the prediction of cutting forces with velocity-chip load dependent coefficients but a single function velocity-chip model fails to predict the forces accurately at very high speeds (>80,000 rpm). This inaccurate force prediction affects the predicted stability boundary at those speeds. Hence, this paper presents a segmented approach wherein a function is fit for a given range of speed to determine the chip load dependent cutting coefficients. The segmented velocity-chip load cutting coefficient improves the cutting force prediction at high speeds. R2 value is found to be improved significantly (>90% for tangential cutting coefficient) which yields the better forces prediction and hence more accurate stability boundary. This paper employs two degrees of freedom (2-DOF) model with forcing functions based on segmented velocity-chip load dependent cutting coefficients. Stability lobe diagram based on 2-DOF model has been created for different speed ranges using Nyquist stability criteria. Chatter frequency ranges between 1.003 to 1.15 times the experimentally determined first modal frequency. Chatter onset has been identified via a laser displacement sensor to experimentally validate the predicted stability lobe.