A detailed study was carried out to understand the dynamics of pressure variations at different points in the injection-molding system. Thus, hydraulic, nozzle, and cavity pressures were evaluated, in addition to the pressure gradient in the cavity. Both steps and pseudorandom binary sequences (PRBS) were employed to obtain and compare dynamic models describing these variables. Subsequently, these models were employed to evaluate and select optimal controllers for the different variables.

Some aspects of injection‐molding dynamics were studied using a laboratory injection‐molding machine operated under the control of a microprocessor‐based servocontrol system. Two types of experiments were performed: deterministic tests which introduced step changes in the servovalve opening and stochastic tests using pseudo‐random binary sequence (PRBS) perturbations of the servovalve. Deterministic models were written for the hydraulic and nozzle pressures which were in good agreement with the experimental data. A stochastic transfer function‐noise model was obtained for the nozzle pressure, but an adequate model was not found for the hydraulic pressure. The agreement between the nozzle pressure stochastic model and the corresponding step test model was satisfactory.

Simple pseudo‐steady state relations between the hydraulic and nozzle pressures of an injection molding machine were presented and verified experimentally. A simulation study was performed to evaluate the performance of simple controllers using dynamic models developed for the hydraulic and nozzle pressures. The controllers chosen were the discrete proportional, proportional‐integral (PI), and proportional‐integral‐derivative (PID) types, tuned according to the ITAE criterion. The control of hydraulic pressure simulation showed that the PI controller had the best overall performance, whereas the result of nozzle pressure control loop simulation showed that the PID controller performance was better than that of the PI controller. All the controllers, in both loops, gave responses that were about an order of magnitude more rapid than the open loop response.

No abstract