Premium
Offline prediction of process windows for robust injection molding
Author(s) -
Yu Shengrui,
Zhang Yun,
Yang Ding,
Zhou Huamin,
Li Juncong
Publication year - 2014
Publication title -
journal of applied polymer science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.575
H-Index - 166
eISSN - 1097-4628
pISSN - 0021-8995
DOI - 10.1002/app.40804
Subject(s) - process (computing) , process window , computer science , molding (decorative) , process variable , design of experiments , reliability (semiconductor) , mechanical engineering , engineering , mathematics , statistics , operating system , power (physics) , physics , quantum mechanics
Process parameters play a highly significant role in the final quality of parts produced using dynamic injection molding. Many researches have made great efforts in obtaining an optimum set of process parameters for improving molded part qualities with various optimization methods. However, this work has failed to provide sufficient information to adjust process parameters in the face of variable environmental conditions and various injection machines to ensure robust, high‐quality injection moldings. Current conditions are too cumbersome and require technologists to perform repeated, detailed optimization procedures on the mass production plant floor. An offline method for prediction of process windows is proposed in this article. The process window is significant for robust manufacturing, and optimization of process parameters. Considering that it is an irregular region in a multi‐dimensional space respecting to process parameters, numerical simulations based on DOE method were designed to offline build relationships between process parameters and part qualities. Then the simulation results were classified as positive or negative class, thereby yielding simulation sample data. Finally, the process window was verified using an SVM classifier and a set of simulation samples. Injection molding of an experimental production plate using various process parameters was conducted to verify the reliability of the predicted process window. The results show that, within tolerable deviations, the predicted window of experimental parts is in accordance with verification experiments. The proposed method demonstrates an ability to rapidly obtain a suitable set of process parameters for achieving consistency in part quality with low cost and high efficiency. © 2014 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2014 , 131 , 40804.