We are a plastic injection molder and recently had our ISO/TS Pre-Assessment. One of the OFIs the auditor suggested was "consider statistical methods for setup verifications". Can anyone elaborate on this? Any input would be greatly appreciated.

Thanks,
Billie Jean

We are a plastic injection molder and recently had our ISO/TS Pre-Assessment. One of the OFIs the auditor suggested was "consider statistical methods for setup verifications". Can anyone elaborate on this? Any input would be greatly appreciated.

Thanks,
Billie Jean

It's hard to say what the auditor might have meant. My guess is that he/she thought it might be a good idea to analyze your setup verification process in general, but I suppose the intention could have been some sort of analysis of individual setups. If you know what the auditor meant, post back and let us know, but your best bet is to ask the auditor if you're not sure what was expected.

It's hard to say what the auditor might have meant. My guess is that he/she thought it might be a good idea to analyze your setup verification process in general, but I suppose the intention could have been some sort of analysis of individual setups. If you know what the auditor meant, post back and let us know, but your best bet is to ask the auditor if you're not sure what was expected.
There are several methods for this. One simple method, if you already use SPC to control the process, is to verify that the setup subgroup is within the +/- 1 standard deviation band of your control chart.

If you do not use SPC, at each setup take a subgroup mean and verify that the confidence interval for the mean includes the nominal value of the specification. You will need to determine a statistically appropriate sample size.

It's hard to say what the auditor might have meant. My guess is that he/she thought it might be a good idea to analyze your setup verification process in general, but I suppose the intention could have been some sort of analysis of individual setups. If you know what the auditor meant, post back and let us know, but your best bet is to ask the auditor if you're not sure what was expected.
These are the comments I received from the auditor, "The OFI was in regard to TS 16949:2002 7.5.1.3 Verification of job set-ups "the organization shall use statistical methods of verification, where applicable". There also is a note that "last-off-part comparisons are recommended". I don't remember the exact details, but in the QS-9000 Glossary about the setup subject, I believe it suggested something like a Precontrol Chart, wherein the actual set up 1st piece product measurement was compared to pre-control limits set at 1/3, 2/3 and 3/3 of the spec. width, with the goal of being centered in the spec range, hopefully inside the 1/3 band. In a nutshell, the Standards are looking at closeness to "aim" (usually the spec mid-range), or as close to the last piece in the previous run/campaign, using some simple statistics to show "how close is close"."

See Miner's earlier post. I'll add the fact that the "where applicable" thing is significant, or should be. There are instances (especially in cases of hard tooling--stamping, injection molding, e.g.) where setup is less likely to have an effect than in processes such as machining or other processes where the conformance of the part is wholly dependent on the efficacy of the setup. In other words, a reasonable argument may be made in some cases that use of statistical methods isn't "applicable."

Pre-control and control charting would be sufficient to "monitor" how setup samples compare to production pieces. I am from a different school in that I believe the best method is to do a DOE showing that setup to setup variability is small. If it is not small, then use control charting.

I advise my clients to only use control charting for processes where variation is a concern. It does not make sense to control chart a process that is extremely stable.

We are a plastic injection molder and recently had our ISO/TS Pre-Assessment. One of the OFIs the auditor suggested was "consider statistical methods for setup verifications". Can anyone elaborate on this? Any input would be greatly appreciated.

Thanks,
Billie Jean
I think the auditor mentioned the firstoff inspection of the sample through the process capability , i think i confused u , if u tell the problem clearly i can suggest some ideas

In Quality Journey,
Senthil

I think the auditor mentioned the firstoff inspection of the sample through the process capability , i think i confused u , if u tell the problem clearly i can suggest some ideas

In Quality Journey,
Senthil
We are a small plastic injection molder. Right now we have over 600 molds and 12 machines. We have controlled set up sheets for each mold in each machine. On start up, the initial parts are scrapped until the press is cycled in and parts of an acceptable quality are being produced. The Quality Control Roving Inspector begins sampling after optimum flow is achieved and acceptable quality is repetitively being produced. Any nonconformities after start up are documented on a p-chart.

In our current process, other than p-charting (which is after the fact), I do not see where we do any set up verification.

Do you have any CC, or SC (dimensions) on the part, you have to monitor, and run SPC ? Do you have to take measurements, at first part inspection , or you have only visual requirements for the products ?
György

Do you have any CC, or SC (dimensions) on the part, you have to monitor, and run SPC ? Do you have to take measurements, at first part inspection , or you have only visual requirements for the products ?
György
We do have special characteristic dimensioning on some parts and we run cpk. Dependant upon what the requirements are, if dimensions are needed we will do a dimensional first part inspection, if not, then it is just visual.

We are a small plastic injection molder. Right now we have over 600 molds and 12 machines. We have controlled set up sheets for each mold in each machine. On start up, the initial parts are scrapped until the press is cycled in and parts of an acceptable quality are being produced. The Quality Control Roving Inspector begins sampling after optimum flow is achieved and acceptable quality is repetitively being produced. Any nonconformities after start up are documented on a p-chart.

In our current process, other than p-charting (which is after the fact), I do not see where we do any set up verification.
The statistical analysis would be to show how you KNOW that there is acceptable quality.

Presumably, your auditor gave you some idea of what might be improved as a result of using a statistical technique? Do you have a significant problem with set-up and start up scrap or obtaining steady state run conditions, for example?

Or is it just one of those 'off the cuff' comments because the company haven't (yet) considered applying statistics to set up verifications?? Is this a high priority activity for some dollar savings?

If the first situation is true - the auditor is off base - that's consulting!
If the second case exists, then unless management see that some benefit could be obtained from the application, then treat the comment with caution. My guess is it's simply a brain f*rt and feels it's justified as being an 'OFI' since you didn't have a 'good' answer for him........

Andy