An injection moulding press can look superb for an hour and disappointing over a month. Within a single shift on one resin lot, shot-to-shot weights barely move. Change the resin lot, let the barrel heater drift, swap the operator who sets the hold pressure, and the average wanders. Short-term capability indices such as Cp and Cpk, built from tight subgroups, only see the shot-to-shot noise. They are blind to the week-to-week wandering.
Process performance indices, Pp and Ppk, use a different sigma. Instead of estimating spread from within-subgroup ranges, they calculate the standard deviation from every individual measurement in the study, taken over a long period. That overall sigma includes both the shot-to-shot noise and the wandering between lots, shifts and setups. It describes what the customer actually received across the whole run.
Both kinds of index use the same specification limits and the same arithmetic structure. The only thing that changes is which sigma goes into the denominator. Once you understand that, every other difference between capability and performance falls out of it.
Standard Procedure
- 1Identify the sources of variation that only appear over time: resin lots, setups, operators, ambient temperature.
- 2Decide the study window so that those sources have a chance to act, typically several weeks or many lots.
- 3Measure individual parts throughout the window and keep them in time order.
- 4Use the full dataset, not just tight subgroups, when the question is long-term performance.
Lesson Summary
- ✓Short-term indices see only within-subgroup noise; long-term indices see everything that happened during the study.
- ✓Same formulas, same specification limits, different sigma.