It is widely recognized that melt temperature in injection molding has significant effect on part quality, repeatability and appearance. Control of melt temperature in molding of high-precision components depends on the non-linear heat transfer and fluid dynamic behavior governing the process. Such systems are difficult to represent completely when developing a standard control strategy, particularly if the material changes or set-point profiles are frequently modified. This research proposes a new strategy for combining two disciplines; model-based predictive control (MPC) and computational-fluid-dynamics (CFD). A CFD model of a three-heater zone injection molding machine (IMM) will act as a slave to the MPC algorithm by providing the necessary open-loop step response data required to dynamically reformulate a matrix which contains the discrete response of the controlled variable.; A CFD model of the heater zones was developed which included interactions between the three heater zones, barrel, melt and screw regions of an IMM. The model of the three heaters were excited under various simulated conditions, and their individual open-loop temperature responses were used within a predictive control strategy to attain new steady-state temperature profiles. Both the CFD-based MIMO and SISO MPC algorithms were able to provide improved control over the barrel and nozzle melt temperature when using temperature dependent, scaled open-loop data compared to the ARX-model response based on the initial open-loop data. A reduction in overshoot and settling time of the closed-loop temperature responses were obtained, thereby minimizing the potential for polymer degradation.
展开▼
