3。 Numerical experiment results

3。1 FE model

To develop green products [12], non-steel substitution ma- terials such as aluminum, composite, and plastic have been used。 While thermoplastics have lower mechanical strength and dimensional stability than steel, they offer huge potential, due to both a lower density and a higher possibility for func- tional integration。 Car fender is made of short fiber- reinforced plastic, Noryl GTX810。 Material-properties of this material, as obtained from MoldFlow data, are given in Table

1。 In order to reduce the mold-making cost and improve cool- ing-performance  and  applicability,  a  cooling  channel using

Figure 5。 Simulation model in MoldFlow environment。

an array of baffles is adopted in the plastic injection molding process (Figure 5)。

3。2 The combination between CAE-Integration tools

In order to reduce time and cost, a framework that com- bines MoldFlow with integrated tool, namely I-Sight is pre- sented in Figure 6。 The combination of process parameters based on DOE techniques, the number of simulations, read- ing as well as storing data, and optimization of response val- ues are carried out by integration tools。

The procedure is based on a few steps。 Simulation models are first designed in the MoldFlow environment with boun- dary conditions such as material properties, cooling channel properties, and process parameters。 The analysis process is then performed sequentially with the variation of input  para-

Figure 6。 The integration between CAE-integrated tools。

Table 2。 Levels of process parameters。

Factor Parameters –1 0 +1

A Mold temperature (°C) 60 80 100

B Melt temperature (°C) 280 300 320

C Packing time (s) 8 10 12

D Packing pressure (% Injection pressure)

60

80

100

E Cooling time (s) 10 15 20

meters to obtain response values。 The responses are then calculated and stored in a text file。 New loops for numerical experiments are completed until all necessary data, including clamping-force and warpage values, are obtained。 Objective functions are then built up and significance is verified with the RSM methodology。 The final step is for optimal solutions and optimal parameters to be obtained following NGSA    II-

based optimization searching。 The intent is for the application programming interface (API) language to be applied and make all work automatically。

3。3 Simulation, approximation, and analysis

In this research, the experimental plans are generated using the stipulated conditions using the Box-Behnken experimen- tal designs with 46 runs。 Box-Behnken experimental method is one of the effective designs based on multi-dimensional sphere and all the design points lie on a same sphere with at least three or five runs at the center point [13]。 During simu- lation, mold-temperature and melt-temperature ranges are determined, based on recommended values coming from MoldFlow。 Packing pressure is set to be a fixed percentage of the maximum injection pressure during the packing process。 Packing time and cooling time ranges are given from the molding process conditions。 Parameters and their levels    are