Determination of a robust process for a union nut. Why D-optimal DoE did not solve random impact behavior?

The quality of injection molded parts is sensitive to multiple factors in the production process. To produce components in a substantially constant quality, it is necessary to find a robust process setting which is insensitive to disturbances. A robust process can be determined by applying a D-optimal factorial experimental design using process optimization software like CQC^® (Continuous Quality Control). This paper describes the task to find a robust process for a component, which has already been manufactured by injection molding in an industrial scale. However, unknown disturbances lead to random scrap. The component examined was a union nut with an internal thread made of polypropylene. The nut was tested for failure in a tailored impact test. As expected, the component was damaged at the weld line, however, fissure length or breakage pretended to occur incidentally. To identify process influences on the impact behavior systematic injection molding tests were performed, based on CQC^® methodology. Mold temperature, nozzle temperature, hot runner temperature, injection rate, back pressure, holding pressure, holding time, and cooling time were varied. 1600 parts were tested. Surprisingly, significant process influences on weld line strength were not determinable, due to both, minor reproducibility of the impact test and a further disturbance variable. The latter was identified to be the interaction of the hot-runner gate position and the cavity position within the 16-cavity mold. Higher drop height at ejection and weld line location at the part’s bottom induced a pre-damage of the upper 8 parts as they hit the conveyor belt.