Recently, many Internet of Things (IoT) have been proposed and developing to industry and markets. They drive people to design and create an automatic production environment. Before executing automatic production, how to retain good quality for injection products is one of the crucial factors. To retain good quality, it is commonly using CAE to assist from original design to revision and to fabrication. However, even using CAE, it doesn’t guarantee the quality factors obtained from CAE can be applied to real experiments. In this study, firstly we have focused on what the major factors were to cause the difference between CAE simulation and real testing on the warpage quality. We further applied numerical simulation to decouple what the main driving forces are to make the difference between simulation and experiment. Results showed that in the original process setting, the warpage difference between simulation prediction and experiment is 0.34 mm. We further found out the major difference came from the injection filling response is too slow (delayed about 29%) and packing pressure is insufficient (23% lower) in real experiment comparing to simulation prediction. Moreover, after calibrate the machine response the warpage difference between simulation prediction and experiment is reduced to 0.12 mm. It is improved about 65% in accuracy. Moreover, we have introduced viscoelastic (VE) effect into this study. Result showed that in the presence of the VE effect the warpage difference between simulation prediction and experiment is further reduced to 0.04 mm. Obviously, the VE has a great impact in this case study. However, the influence of VE is great or not for other cases that should be further studied.