Multivariate process analysis utilizing Six Sigma methodologies for the prediction of injection molded part quality

There are many aspects to the injection molding process which influence final molded part quality. Changes in these variables can be analyzed by a multivariate control system to quantify and predict the quality of parts manufactured by a highly complex process. This thesis investigates process characterization and the effects of environmental conditions to ultimately enable an automated on-line quality assurance methodology that can be implemented in a short period of time. A multivariate model was created for a process producing "good" parts and from these correlations newly molded parts were then compared to the model and deemed as "acceptable" or "unacceptable" based on multivariate statistics. The critical dimensions and short term mechanical properties of each part were then measured to verify the models prediction of final part quality.;The implemented research investigates the use of various imposed process faults, environmental conditions and the performance of two data analysis systems (statistical process control (SPC) compared to principal components analysis (PCA)) for the purpose of quality control. The results showed that multivariate analysis predicted ten of the twelve process faults and rejected three molding cycles (2% of the total) that appeared to produce acceptable products. It can also be concluded that environmental conditions influence final molded part quality to varying degrees and that it is necessary to control these conditions for applications with tight tolerances.