Gas Assisted Injection Molding optimization with M.O.G.A.
Gas assisted injection molding is a processing technique used to produce hollow sections in plastic parts. This is achieved by injecting a gas into a tool cavity partially filled with resin, while the resin injection continues to prevent a hesitation of the flow front. After the resin injection is completed, the gas injection continues, forming a gas bubble that keeps the resin flow front moving until the resin skin reaches the end of the mold. Once the material has completely filled the article, and the resin skin is fully established, the gas bubble continues to be pressurized to avoid resin shrinkage. Finally, the article is cooled adequately to establish skin strength, and the gas is finally vented. The advantages of this technique over classic injection moulding are numerous, and include increased output, reduced resin consumption, reduced design constraints, and lower costs. However, to obtain a product with the desired properties, it is necessary to accurately control the size and location of the penetrating gas bubble, the gas pressure, and other process parameters such as temperature and timing of the gas stream process phases. All these parameters are important to determine the gas flow path and speed which affects the final product quality.
modeFRONTIER was coupled with CADMould to design the moulding process for a polycarbonate frame. Because the resin and gas injection points were pre-determined, modeFRONTIER had to control ten process variables:
- Resin injection time
- Resin filling percentage
- Resin melt temperature
- Gas injection times (3 injection stages)
- Gas injection pressure (3 values)
- Mold wall temperature
The goals of the optimization were two:
- Minimum ejection time (to maximize through-put)
- Maximum gas penetration in the part
Two constraints had to be satisfied to consider the process acceptable: the mold cavity had to be completely filled, and resin break-throughs had to be avoided. break-throughs had to be avoided.
The problem is clearly very complex: the space of the possible solutions is 10-d dimensional, and there are multiple, conflicting, objectives to be pursued. To tackle it efficiently, an initial exploration of the design space (set up using a Taguchi DOE algorithm) was followed by an optimization phase. Using our multiobjective genetic algorithm it was possible to find the Pareto frontier, and thus a set of optimal designs for different trade-offs.