Abstract
According to the Taylor tool life equation, tool life is dependent on cutting speed (or spindle speed for a selected tool diameter in milling) and their relationship is quantified empirically using a power law exponent, n, and a constant, C, which are tool-workpiece dependent. However, the Taylor tool life model is deterministic and does not incorporate the inherent uncertainty in tool life. In this work, Bayesian inference is applied to estimate tool life. With this approach, tool life is described using a probability distribution at each spindle speed. Random sample tool life curves are then generated and the probability that a selected curve represents the true tool life curve is updated using experimental results. Tool wear tests are performed using an inserted (uncoated) carbide endmill to machine AISI 1018 steel. The test point selection is based on the maximum value of information approach. The updated beliefs are then used to predict tool life using a probability distribution function.
Original language | English |
---|---|
Pages (from-to) | 403+411 |
Journal | Journal of Manufacturing Systems |
Volume | 31 |
Issue number | 4 |
DOIs | |
State | Published - Oct 2012 |
Externally published | Yes |
Funding
The authors gratefully acknowledge financial support from the National Science Foundation ( CMMI-0927051 and CMMI-0926667 ). They would also like to thank M. Traverso and G. Hazelrigg for numerous helpful discussions.
Keywords
- Bayesian
- Decision analysis
- Milling
- Modeling
- Tool life
- Tool wear
- Value of information