Critical Characteristics - Capability for a Minimum Specification

[vardar3]

We have to report capibility information on criticle characterisitics to our customer. One of the characterisitics is a 750 lb minimum pull-out test. Can I complete capability on 100 pcs sample that includes a parameter that only has a minimum?

 

[Kales Veggie]

Re: Capability for a minimum specification

Yes. You can calculate capability:

Cpk (lower) = (Average - 750) / (3 standard deviations)

(make sure it meets all the tests for a valid calculation, such as normality)

 

[Jim Wynne]

Re: Capability for a minimum specification

If Cpk is an indication of how well centered the process is between the specification limits (and it is) how can you determine a meaningful index number if there is no "between"?

Your answer is the standard one in these cases, and it's what is usually done, but that doesn't make it correct. The problem is that when customers demand a capability study and a minimum Cpk result, they almost never understand what they're asking for, and why on its best day Cpk is only marginally useful, and at worst (which is most of the time) it's completely useless.

Trying to explain this to a customer is usually an exercise in frustration, but it's best to try to explain it anyway before resorting to fiction writing.

 

[TPMB4]

Re: Capability for a minimum specification

Being new to this Quality role I've looked at Cpk and capability studies. In my ignorance I've looked at it in respect to what we do and it never seemed to make sense to me as a relevent measure of a process.

I've tweaked the figures to see what effect it had and TBH with the process I did this to it never seemed to make much difference to the figure unless you changed it beyond what the process would do. If that makes sense. It could of course be that our process is highly capable (if that is the correct term).

As I said it was done "in my ignorance" so perhaps I did something wrong although I don't see how as it was a simple excel form that you entered figures into and it did all the calcs for you. Kind of like ones from this forum. I reckon if you don't know what the statistics mean then they are meaningless. If John is right then the upper and lower is needed. I can see his point.

Someone once told me that good statistics never lies...Can't remember the rest of that quote as it says something about bad statistics or statistics done without proper understanding. Kind of like politicians do with the information their mandarins give them.

 

[bruceliu222]

Good, thanks.

 

[whtan02]

Re: Capability for a minimum specification

If Cpk is an indication of how well centered the process is between the specification limits (and it is) how can you determine a meaningful index number if there is no "between"?

Your answer is the standard one in these cases, and it's what is usually done, but that doesn't make it correct. The problem is that when customers demand a capability study and a minimum Cpk result, they almost never understand what they're asking for, and why on its best day Cpk is only marginally useful, and at worst (which is most of the time) it's completely useless.

Trying to explain this to a customer is usually an exercise in frustration, but it's best to try to explain it anyway before resorting to fiction writing.

Then, what should be the correct way or practise?
Plotting xbar control chart & R-chart will do?
Thanks.

 

[Bev D]

Re: Capability for a minimum specification

A control chart with rational subgroups is a good start to assess stability.
but a control chart (aside from the rationale subgroup issue) only tells you the stability not the capability. stability is a prerequisite for capability.

A good next step is plot the raw data in multi-vari form against the spec limits. This will give you the best assessment of the capability.

Your customer on the other hand is probably only looking for a check in the box on soem form you must submit...in which case Kales approach is not too bad to satisfy your customer...it just won't be all that informative to you.

 

[bobdoering]

My favorite read of the PPAP manual is 2.2.11.5 NOTE

 

[Jim Wynne]

My favorite read of the PPAP manual is 2.2.11.5 NOTE

 

[bobdoering]

"When this is not true, using this analysis may result in unreliable information." Priceless.