In recent years, high-pressure die-casting magnesium components have been gaining currency worldwide because of the excellent properties that magnesium alloys can offer to meet new product requirements. With the increasing application of magnesium parts worldwide, many research and development projects have been carried out to advance HPDC technology. However, truly optimized mold design and production of defect free castings remains a challenge for die casters. For many HPDC magnesium products, especially those specified for porosity-free and high cosmetic requirement, the challenge not only comes form a lack of a deeper understanding of how molten magnesium alloys fill the mold cavity and form defects, but also from improper preliminary part design. This study proposes a virtual prototyping system that integrates several effective soft and hardware tools for both the part and mold-design engineer to evaluate part manufacturability. Also, investigated in this study are the major causes of those defects that are the predominant cause of rejection of thin walled, leak-free magnesium parts requiring highly cosmetic finishes.
A Neural Network Based Defect Prediction Approach for Virtual High Pressure Die Casting
