I would like to know what is the proper way to handle out of tolerance gauge in reference to holding and quarantined product. If a gauge has a calibration interval of one year how can you possibly go back and check product made up to a year ago. With more and more manufacturers not holding product in their warehouses for long periods of time how can they possibly go back and check their product against a out of tolerance gauge? How should the calibration lab notify the production floor?:( :(

That is one of the hardest problems you will face in regard to calibration. Your records should indicate where the gage is being used which should give you an idea of what has been measured with it. From there, you may be able to determine what has been sent out. Your system should allow you to evaluate the effect the OOT gage had on the final product. If, for example, you knew the gage was being used to measure a non-critical characteristic and maybe the product underwent a subsequent quality check, maybe you wouldn't need to do anything.

On the other hand, you may determine the gage was OOT past a "safety" point so you would have to notify the customer they "MIGHT" have some suspect parts.

As to the one year interval, I would assume you don't use the gage without verifying the state of calibration. For example, the date on the label is confirmed, the instrument inspected for dirt/damage and zeroed out before use. In our shop, if a gage were dropped , dirty, or wouldn't zero, the operators would take the gage to the lab. I am sure there are a some reasons why a gage could be off without giving you an indication but most should be found before use. I guess if you are still worried about the interval you might consider using a validation block daily before using the gage.

Dave

We have in place many of the things you have suggested. During our last audit, the auditor ask a question about a notice we send out to notify the department heads about a OOT gauge. The notice has two paragraphs asking the department head to choose one. The paragraphs ask them if the out of tolerance condition warrants the product to be quarantine or not. The auditor ask us how can you quarantine the product if it is already been shipped. I guess I'm looking for a way to word our out of tolerance notice that will satisfy everyone if that's possible.

You could add a sub-option to the quaratine option asking if the risk warrants notifying the customer of suspect product having shipped.

You gentlemen have made some good suggestions and thanks for your help. I will be meeting with my quality supervisor today and hopefully put this item finally to bed.

Umm, what about Gage R&R in regular intervals? Wouldn't that help catching a varitaion that would be due to out of tolerance gauge?

I agree this is the great black hole for calibration. When I worked at Pratt&Whitney Aircraft any gage used to check a part was serialized and that number was put on the paperwork for that part. There was no exceptions. Since then I have worked in the auto industry and it is wide open. Very often our parts are on a customers car before the gage is found to be bad. What we do is have the appropriate manager sign a statement to the effect that the out of tolerance gage did not affect the overall quality of the product. I am not sure this is valid or even good business, but it is what we do at the present time. Short of holding all parts until the gages that checked them are calibrated I am not sure there is a perfect answer to your problem.:cool:

Some really great suggestions, here. By the posts I can tell everyone are old pros at this stuff, but I will throw some things out there for consideration.

First, let's make the assumption that within a stable system, instruments should not be out of tolerance (O.O.T). The processes are fairly stable, instruments are in good working order, appropriate uncertainty ratios are assigned, intervals set correctly, etc. So, something is not working like it should.

Could you increase the tolerance? Buy more reliable instruments? Buy standards with less uncertainty? Could you establish say process limits, where the instrument may be O.O.T. by say 1 PSI, but set a process limit for 2 PSI?

So if an instrument is O.O.T., is the instrument drifting, standard bad, etc.? By what magnitude is the unacceptable error? The magnitude may not be great enough to significantly affect the product.

As mentioned earlier, possibly some Root Cause analysis to determine the O.O.T. cause. Shortened interval, intercomparison verifications, and limited calibrations may be alternatives also.

I realize this is 1st grade stuff and some of this things costs time and money, but if I had what I would deem as "critical exposure", I would do everything in my power to eliminate the O.O.T. involving customer notification, shipped product, etc.