Hello dear JMP community
I will first explain the test method (no need to comment on whether it is right or wrong because this is what I got). The purpose is to detect a leakage.
Measurement process
- Part is mounted to a pressure station
- Pressure station rotate a vacuum pump x times to create a vacuum (atmospheric pressure is approx. 1000mbar)
- Pressure p1 is sampled at time t1
- After 15 seconds p2 is sampled at t2
Specification
- 700 mbar ≤ p1 ≤ 750 mbar
- p2 - p1 ≤ 4 mbar
Situation
If there is a leakage then the result can differ a lot from normal observations.
Challenges
This cause two challenges:
- This have a very high impact on my Cpk and Ppk, which i have a requirement for.
- I hoped I could use the continonus response to model my system but this is just creating to much noise and creating a lot of "outliers". When shifting to a binary response then my Power analysis for one sample proportion require a lot of sample to detect a difference in scrap.
Anyone with some good ideas to how I could solve this puzzle?
