What is the best approach to determining process parameters for injection molded plastic detail parts. There are multiple cavities (up to 64). and we have automated data collection systems on the presses. We can create a data glut, but how to best perform statistical analysis, and create logical parameters for the individual alarm settings is the concern.
scott.atkin 3/2/2001
I have run projects in the past directed at variation reduction. Several options exist.
One is to develop a direct measure for the variation such as deviation from target either absolute or as a percentage (value/target or value-target). Using this variation establisha limit and Variation DPMO for % reduction.
Another is to use a stability factor such as Q1/Q3 or difference between 95% point and 5% point as the measure of variation and then measure improvement as the % change in the stability factor.
lastly a measure of improvement could be established based on the Sum a Sqyuares analysis and changes theirin.
Unfortunately with some of these methods you are creating non-intuitive measurements that may be harder to explain to the line workers.
statpoint 2/20/2001
The obvious answer is 50%, but I don't think I'd do it that way. The proper measure of improvement would relate more to a measure such as DPMO, which is not a linear function of sigma. A 50% reduction in sigma might give a much greater reduction in the number of nonconformities.