P
Hello
Attached is raw data from our process.
A row is a lot....15 data points per lot. Statistical information for each lot is also there.
Current spec is average must be below 5.5
We have not had any failed lots for this requirement.
Customer is considering spec limit of 7.6 with CPk 1.33 for lot to replace current requirement. When I compared the Ppk of the data set with this, Ppk was 1.30...so I thought we would be ok. I looked at Ppk for our current process, by using lot means against the 5.5 average and Ppk was much lower, so I predicted we would be more capable to this spec.
HOWEVER - looking at individual lot performance, if the spec was retro-activated, we would have several lots that would fail......so practically we are not better off. I don't understand. What am I looking at wrong here?
**Note, looking at the data (for capability analysis), you will see that it is neither normally distributed, or in control. Special cause for out of control are known process shifts. Given that stability is a requirement for capability, the predictions had grain of salt attitude, but I did not expect it to be that off.
Also attached are the capability charts of each example above.
Any advice would be appreciated.
Attached is raw data from our process.
A row is a lot....15 data points per lot. Statistical information for each lot is also there.
Current spec is average must be below 5.5
We have not had any failed lots for this requirement.
Customer is considering spec limit of 7.6 with CPk 1.33 for lot to replace current requirement. When I compared the Ppk of the data set with this, Ppk was 1.30...so I thought we would be ok. I looked at Ppk for our current process, by using lot means against the 5.5 average and Ppk was much lower, so I predicted we would be more capable to this spec.
HOWEVER - looking at individual lot performance, if the spec was retro-activated, we would have several lots that would fail......so practically we are not better off. I don't understand. What am I looking at wrong here?
**Note, looking at the data (for capability analysis), you will see that it is neither normally distributed, or in control. Special cause for out of control are known process shifts. Given that stability is a requirement for capability, the predictions had grain of salt attitude, but I did not expect it to be that off.
Also attached are the capability charts of each example above.
Any advice would be appreciated.
Attachments
-
40.9 KB Views: 144
-
108.4 KB Views: 212
-
110.9 KB Views: 216