Abstract
In plastics injection molding, factors such as process conditions, mold design, polymer properties, and machine specification, will have significant influences on the part quality. This thesis presents the conclusive results on the study of correlations among the processing parameters, volumetric shrinkage, and linear dimensions. With the advent of Computer Aided Engineering systems, engineers are able to tune processing parameters, and predict part quality which were accomplished by the trial and error method before. However, there are still concerns on the accuracy of the CAE software nowadays. In this thesis, the first part is to investigate the volumetric and linear shrinkage of injection molded parts by comparing the results on the CAE to the results on the experiments. To search for processing windows, a design of experiment based on Taguchi Method was adopted for selecting 18 parameter sets. Then, the experimental results were employed for verifications of the CAE predictions on Shrinkage. The second part is to develop a model based upon Back Propagation Neural Networks for learning the mechanics of the process. After obtaining the model parameters, extra 8 processing parameters were tested experimentally to verify the accuracy of the model. The model can predict the shrinkage without doing the CAE simulations again though the discrepancy between the simulations and experiments were still observed. The applications of the results are very promising because the reduction on CAE simulation time is necessary at the moment.