Abstract:

A quality modeling system is described to support manufacturing process development and quality control. The system relies on the main effects relating the process variables to the quality attributes. The extensive simplex method is used to derive the feasible universe of process settings satisfying all specification limits. Three different types of feasibility are analytically derived including 1) the global feasible set that establishes the extreme limits of feasibility by allowing all the process variables to vary simultaneously within their allowable range, 2) the local feasibility, which shows the immediate feasibility for each process setting holding other process variables at their current setting, and 3) the controllability that is indicative of the range that may be obtained for each quality attribute while holding other quality attributes at their current value. The described system explicitly considers both modeling uncertainty and uncontrolled variation; the specification limits may be automatically tightened by the magnitude of the confidence intervals and process standard deviations to ensure a desired level of confidence and robustness. An example is provided for an injection molding process with four process variables and three quality attributes.