For many products, milling comprises a portion of the manufacturing process, whether on the product itself, as in many aerospace structures, or one of the components used in its fabrication, such as the mold in injection molded parts. In either case, the milling process must be cost effective, which motivates process optimization. In a simplistic view of high-speed machining, a user might consider it sufficient to minimize the computer numerically controlled (CNC) tool path time. However, minimizing tool path time requires other factors that limit the material removal rate to be considered. These limiting factors include: chatter (unstable cutting), dimensional tolerances, surface finish, and tool wear or breakage. Each factor has an associated cost. If these constraints are violated, the manufacturer must either correct the mistake or scrap the part, both of which carry additional costs. Also, in some instances unwanted process behavior can cause damage to the tool, which leads to necessary tool replacement. In the end, the considerations must be combined to realize maximized profit.;In this research a framework is developed to combine the limiting factors listed previously, as well as the uncertainties associated with each, into a profit optimization scheme so that informed decisions about parameter selections in the milling process can be made. This framework is based on decision of analysis, a combination of decision theory implementing Bayesian statistics and experimental analysis. Decision analysis provides methods of analyzing the milling process such that the decision maker can identify deterministic values, uncertain quantities or effects, and the aspects of the process that can be controlled such that the information can be combined appropriately. Once these considerations have been combined successfully, the effects of the user controlled quantities on the overall desirability of the process, measured in this research by profit, can be determined. In addition, the user can improve the results by performing experiments to add new knowledge of the system and diminish the effects of uncertainties.;This research represents the first steps toward making this framework a reality. Initially, the milling system is characterized and organized using decision analysis and its visualization tool, the decision diagram. The effectiveness of this organization is tested using a discrete optimization on a group of test parameters. Then, treatments are applied to one process limiter, stability, in order to develop a continuous optimization and enable calculations of the value of information (maximum value a user would pay for information gain) and value of experimentation (value a user places on the information obtained from a particular experiment). Finally, a method to update information or beliefs about the system's stability condition using a Bayesian approach is detailed and tested using a numerical example.;A second aspect of this research is the development of a new milling "super diagram". This diagram provides a simple way to display information about many process limiters in the same parameter space. The initial diagram presents the combined information from both surface location error and stability within a user selected range of spindle speed and axial depth of cut values. This diagram displays milling process information in a format relevant to the user and tailored to the user's tolerance requirements. (Full text of this dissertation may be available via the University of Florida Libraries web site. Please check http://www.uflib.ufl.edu/etd.html)
展开▼
