Cpk, Subgrouping, and Can Attribute Charts Monitor Ongoing?

[gtg168w]

I am tasked to put a new inspection method in place at a plastic injection molder who currently relies heavily on 100% inspection. We are ISO9001, and have automotive (TS16949) and aerospace (AS9001) customers.

1 - If Can I use a c=0 acceptance sampling plan, lets say c=0 and AQL=0.065%? I know AQL=0.065% requires larger sampling sizes per lot, but it is a lot smaller than the current 100% sample size.

1a - Can a c=0 AQL=0.065% sampling plan be used without a CPK calculated for that process? I plan on using attribute charts (p-chart) for each process no matter what just because of their ease of use on the shop floor.

2 - CPK - For multicavity tools and for AIAG PPAP compliance, what's the best way to subgroup my sample pieces? I believe the PPAP manual says 100 samples in subgroup sizes of 4 or 5 samples? I am assuming it is best to group by cavity, this way the within-subgroup variation is smaller, resulting in better cpk indices?

2a - Do capable processes via CPK require ongoing monitoring? If so, I believe attribute charts would no suffice for this. Can attribute charts be used to monitor process control, but not used to revalidate the cpk? I know some companies are starting to require CPK validation yearly or every 2 years. Does this require on going SPC data every run? Or can I just make a "maintenance" schedule for my processes and know that by next month I need to repeat a capability study for this part/tool to meet the customers validation reqs?

3 - Do I have to ask the customer if they approve of my newly implemented inspection plan? I don't want to raise any unnecessary red flags. Of course, our goal of zero defects and 100% customer satisfaction is still the same.

4 - If a process is not capable after a capability study, say cpk is 1.0-1.3 or is below 1, can c=0 AQL=0.065% still be used? Or should I revert to 100% inspection?

 

[Jim Wynne]

I am tasked to put a new inspection method in place at a plastic injection molder who currently relies heavily on 100% inspection. We are ISO9001, and have automotive (TS16949) and aerospace (AS9001) customers.

[1 - If Can I use a c=0 acceptance sampling plan, lets say c=0 and AQL=0.065%? I know AQL=0.065% requires larger sampling sizes per lot, but it is a lot smaller than the current 100% sample size.

The type of sampling plan you're referring to is for lot acceptance. Is that what you want to use it for (as opposed to some kind of in-process inspection)?

1a - Can a c=0 AQL=0.065% sampling plan be used without a CPK calculated for that process? I plan on using attribute charts (p-chart) for each process no matter what just because of their ease of use on the shop floor.

Here is seems that you are referring to using a sampling plan inappropriately. If you're dealing with customer-designated key characteristics, you might have to do variables charting. If there are no customer-designated critical characteristics, and no mandated Cpk level, you're probably free to do whatever you want.

2 - CPK - For multicavity tools and for AIAG PPAP compliance, what's the best way to subgroup my sample pieces? I believe the PPAP manual says 100 samples in subgroup sizes of 4 or 5 samples? I am assuming it is best to group by cavity, this way the within-subgroup variation is smaller, resulting in better cpk indices?

For multi-cavity tools, each cavity should be qualified separately. Because customer requirements vary widely in these situations, it's usually best to discuss it with the customer to make sure you know what's expected. Some customers have supplier quality manuals wherein this sort of thing might be covered.

The PPAP default is a minimum of 100 measurements and 25 subgroups. This might not always be prudent or appropriate.

2a - Do capable processes via CPK require ongoing monitoring? If so, I believe attribute charts would no suffice for this. Can attribute charts be used to monitor process control, but not used to revalidate the cpk? I know some companies are starting to require CPK validation yearly or every 2 years. Does this require on going SPC data every run? Or can I just make a "maintenance" schedule for my processes and know that by next month I need to repeat a capability study for this part/tool to meet the customers validation reqs?

There are a lot of questions in there. Again, it depends a lot on customer requirements. When there are critical characteristics, some customers expect ongoing SPC, and for Cpk to be held to a given value (or greater). Other customers don't seem to care much so long as there's something in the PPAP package that looks like serious statistics. Unless there are explicit requirements to the contrary, you are again mostly free to do what you want so long as you can, at any given point, demonstrate that the process is operating at least as well as the PPAP "significant production run."

3 - Do I have to ask the customer if they approve of my newly implemented inspection plan? I don't want to raise any unnecessary red flags. Of course, our goal of zero defects and 100% customer satisfaction is still the same.

Significant changes to the control plan might require customer notification or approval. I've had experience with customers who would pitch a big hissy if there were only minor changes without their knowledge. On the other end of the spectrum there could be other customers who never even look at the control plan beyond confirming that there is one.

4 - If a process is not capable after a capability study, say cpk is 1.0-1.3 or is below 1, can c=0 AQL=0.065% still be used? Or should I revert to 100% inspection?

The default (AIAG) expectation is that there will be a special control plan for those situations, and increased inspection is required, possibly 100% inspection. If the process is not capable of achieving and maintaining the Cpk level required by the customer, you shouldn't be submitting PPAP until the customer is aware and has approved your recovery plan.