TABLE OF CONTENTS "A man's got to know his limitations."
CLINT EASTWOOD FROM THE MOVIE MAGNUM FORCE
Ideally you will find one of the two tried-and-true templates for response surface methods, the central composite design (CCD) or Box-Behnken design (BBD), well suited to your experimental needs. However, these standard RSM designs may not conform to multifactor operating constraints. For example, as shown in Figure 7-1, parts made in a hypothetical injection molding machine come out defective when temperature and pressure fall outside of the specified zone.
The pressure needed to fill an injection mold depends on the viscosity of the molten plastic the higher the temperature, the less viscous the plastic. If not done right, the process fails, either by under-filling the mold (short shot) or by putting in too much, creating parts with flash. (A good example of applying DOE to flash reduction can be seen in a report from Lin and Chananda, 2003) two students at Kettering University who successfully applied the tools of DOE to make better replicas of their school mascot, the "Bulldog.")
As illustrated in Figure 7-1, multifactor linear constraints must be imposed to prevent bad things from happening to the plastic parts at certain combinations of the two key processing factors. When both temperature and pressure are:
Set at low levels, they cause a short shot of plastic into the mold.
Set at high levels, they cause an excess of material, which creates flash along the edge of the part.
