Warpage 2。58 2。56 0。8 2。58 2。59 0。4

Cooling time 11。53 11。56 0。3 11。53 11。57 0。3

Stress 23。7 24。0 1。3 23。7 23。6 0。4

The cooling time is minimized when both mold temperature and melt temperature reach the minimum values。 Warpage is reduced when the packing time increases。 Table 5 shows the relative errors between predicted values and actual values of both RSM and RBF metamodels at their optimum design points。 The verification results show that the relative errors are very small。 Therefore, the fidelity of these metamodels is high, and the optimization results are reliable。

It can be seen that we can capture the relationship between warpage, cooling time, and process parameters via metamod- el with 81 numerical experiments。 The computing cost or the number of simulations is lower than the direct GA or direct hybrid optimization methods。 The number of simulations is surely lower than direct gradient-based optimization methods because it takes 155 runs for SQP algorithm to converge when solving the multi-objective optimization in case study 2。 Some remarks can be made in this case study。 When the relationship between process parameters and the responses is low non- linear, using RSM and RBF model is an efficient way for optimization。 Both RSM and RBF give the same optimization result。 This asserts the fidelity of the metamodel applied to low nonlinear optimization problem。 Therefore, direct optimization and other long run optimization methods are not investigated in this case study。

5。 Conclusions

This paper reviews the state-of-the-art of injection molding process parameters optimization。 The existence of many methods and techniques applied to molding parameters optimization shows that there is no perfect method to solve all opti- mization problems。 However, usable frameworks and appropriate guidelines can be generalized to facilitate the optimization process in injection molding design。 We introduced two robust general optimization frameworks for optimizing molding parameters including direct numerical optimization and metamodel based optimization。 The proposed frameworks are sys- tematic, general, but very flexible。 The selection of optimization method, metamodel, and optimization technique depends on the molding characteristic of the molded part, the experience, habit, and available tools of the designer。

The proposed optimization methods in this work can be considered as a paradigm for all optimization methods applied to simulation-based optimization in injection molding。 The frameworks that accelerate the optimization process and seam- lessly integrate the CAE simulation tools in injection molding and the optimizer were systematically introduced。 The opti- mization process is performed automatically。 As a result, the integrated tools those are used in conjunction with adequate optimization approach speed up the molding design process and ensure the quality of the molded part。