MULTIVARIATE PROCESS CHARACTERIZATION AND QUALITY CONTROL FOR INJECTION MOLDING

摘要

Process characterization provides a model of process responses as a function of process factors, which is useful for process optimization and quality control. In this paper, four Design of Experiments (DOE) are implemented for a thin wall molding process, including two fractional factorial designs, a D-optimal design, and a supersaturated fractional factorial design. The capability of the DOEs are subsequently analyzed with respect to the estimated main effects and defect prediction capabilities. The results indicate that fully saturated designs are satisfactory for process characterization, but all critical process factors should be investigated. Experimental designs having confounded process factors were found preferable to experimental designs of similar size that were not confounded, but investigated fewer factors.