Effect of Sampling on Setup Verification Requirements

M

Monte

Setup Verification

Hi all,

In a recent surveillance audit as I looked over the auditors' shoulder, I noticed an error in one of our work instructions. It referenced an obsolete software package which was used to "verify" setup. In later moving to revise this document with area manager we were unable to clearly understand how 4.9.4 applies in this process. In this particular area the product is cut to length before it is put into the kiln. There are several characteristics which are checked in this area, most relating to cutting: Length, Squareness of cut etc.
We use constant sampling of subgroups for these and several other characteristics and these are entered electronically. We use control charts for all these.
Now for the question. Since we are already doing sampling to the "Nth" degree (probably overkill) what more should we do for "setup verification"? We never just fire up the line and run a lot without a ton of data being collected, so what do we do? We have lots of evidence of "verification" right?

thanks,

monte :bigwave:
 
J

JRKH

Monte,

The purpose in 4.9.4 is to verify that you are performing the process the way that you say you will. Also that there are no loss of capability if something is changed.
If I understand you correctly, I would say you have it covered.

James
 
M

Monte

James,

Thanks, we thought we had it covered as well, but were cited in a surveillance audit 3 yrs. earlier for lack of evidence of "verification". So we used this old software which simply had a check of several setup items for each area. I don't really see that as fulfilling 4.9.4 since the glossary definition of setup verification mentions the use of SPC. It also recommends "last part off" comparisons. I agree with you that it seems we had it covered but our registrar disagreed. Is this one of those "rogue" auditor issues that may never come up again? (ie don't worry be happy) or since I am re-working the procedure anyway how do I satisfy verification other than current SPC compared to last run? If the last run was in "control" but barely in spec. I don't want to center my process around the "last part" do I?

Just wondering,

Thanks again James!!
:frust:
 
Top Bottom