On-going capability - Can we decide for ourselves how often to measure cpk?

[Dawn]

To be in compliance with on-going capability:
Can we decide for ourselves how often to measure cpk? Do we have to do a capability study on every order?
Can we use our sample size (6 pieces) maintain those numbers and when we have run 5 orders; we have 30 numbers to do a capability study and determine if our cpk is higher or lower? Will this be acceptable?

------------------

You can decide and document how often to measure Cpk yourself. You really need to decide on a frequency that suits you.
If you take 6 pieces over 5 runs and use this data to calculate Cpk, what actions do you take if you get an NG result? You've already shipped product.
Can you use data from initial capability studies, calculate control limits and run an X bar & R chart, taking a sample each time you manufacture product? If sample is in control each time it shows you don't have any major changes. You could then recalculate Cp & Cpk on a rolling basis using the last say 10 or 20 results.

[Marc]

Thanks, Tony. Your help is appreciated!

Following my experience, I am SPC supervisor and have more than 100 XR charts, it is not necessary to calculate too often your CP/CPK index.
In the past I had to send to our customers a certification about CPK for each lot we sent.
Then we have dimostrated them the state of control of our processes.
Now I update the CPK only in the case something is changed in the process, e.g. new raw material (new supplier or new parameters...), or speedy, or when there is a stable change in the position of the process.
When your process dimostrates a state in control the CPK index doesn't change and you don't need to update.
Usually I update this index once/twice per year (after 80/100 values) and I don't find big difference between the old and the last calculation.

------------------

[Dawn]

But does that satisfy the QS intent?

[Laura M]

Dawn, I believe it does. I used a similar strategy. If a process originally meets CPk requirements, mean centered, etc, and you remain in control, then you have no reason to believe your CPk would have changed. I think the 3rd edition of QS is a little clearer on this than the 2nd. Page 37 - "The supplier shall maintain......to accomplish this...ensure that the control plan and Process flow diagram are implemented."

I had a bit of a problem at an inital assessment (to the second edition) because the auditor didn't understand SPC and was looking for a procedure that stated exactly how often we recalculate. Our procedure at the time stated something like "when a process change is implemented, or control charts indicate....Cpk will be recalculated" Ended up reading alot like the 3rd edition.

Would welcome other thoughts on this. Our strategy was intended to meet the standard without putting ourselves out of business by performing unnecessary statistical analysis.

Laura hit the nail on the head when she referred to unnecessary statistical analyses. One of the most important things to keep in mind is customer and / or internal classification of characterisitics. If you do not have to chart, don't, unless you can save money by effecting improvements. Once a Cpk gets above 3.0 (in the 4-6 range), you really need to change how you are charting. Fewer samples may be needed, for example. Too many of us, in our zeal to demonstrate statisitical capability, have overdone it to the extent that we lost the positive value of good, effective control charting.