Thread plug gauge failed calibration - Must I remeasure everything measured with it?

[Philip Holland]

Have been receiving several go and no go gauges back from calibration that are out of calibration by virtue of having grown. Last audit, the auditor randomly picked a gauge that had failed the last calibration. Well, I had replaced the offending member but had not documented any further action.
It is my belief that as a quality engineer, I should be able to make a judgement as to whether an audit of product bought off by this gauge is warranted. Thread gauges are used in IQA and to verify that correct inserts are installed in molded product.

------------------

 

[Brian Dowsett]

Phillip,
I'd say you're quite right to say that as a quality engineer you can make the decision about product bought off before the gauge was replaced. What reasonable auditors are usually looking for is some documented evidence that you went through that thought process. This could be as simple as a one liner attached to the calibration report saying something like "Consideration given to previous product, "growth" of gauge only represents X% of process tolerance and would not result in out of spec product being released -therefore no action taken on past production"

Regards

Brian

 

[Wayne]

Summary

Have been receiving several go and no go [thread] gauges back from calibration that are out of calibration by virtue of having grown. ...I had replaced the offending member...

Which member grew? I also assume that it was GO gage because it is most likely to be gauled and grow.
https://www.ring-plug-thread-gages.com/ti-gg2-Gage-Gauling.htm

... as a quality engineer you can make the decision about product bought off before the gauge was replaced.

The decision process would have been... There was a GO/NOGO Thread Work Plug Gage used to inspect product. Only product that the gage passed were shipped to the customer. The GO member grew during he production run. The GO member is smaller than the NOGO member. The GO member checks the maximum material condition. When the GO member grew it reduced the manufacturing tolerance, and may have failed some good product. It would not have allowed any rejected product to pass inspection. There is no need to do a product recall. The only required action was the action taken: replace the offending member.

What reasonable auditors are usually looking for is some documented evidence that you went through that thought process.

 

[Helmut Jilling]

Have been receiving several go and no go gauges back from calibration that are out of calibration by virtue of having grown. Last audit, the auditor randomly picked a gauge that had failed the last calibration. Well, I had replaced the offending member but had not documented any further action.
It is my belief that as a quality engineer, I should be able to make a judgement as to whether an audit of product bought off by this gauge is warranted. Thread gauges are used in IQA and to verify that correct inserts are installed in molded product.

------------------

Post #2 and #3 are correct, as is your comment. As an engineer, you are expected to make a judgement whether product acceptance was affected. An automatic recall is not required. However, this product would immediately be classed as suspect product. As an auditor, I would expect you to demonstrate that you did make such an assessment (a note of explanation on the calibration file or in your nc product log would be good examples). I would also consider the logic you describe. If it is valid, then your activities and decision are approrpiate. If you are in error, you would be looking at an automatic major nc.

 

[Jim Wynne]

I would also consider the logic you describe. If it is valid, then your activities and decision are approrpiate. If you are in error, you would be looking at an automatic major nc.

You might want to clarify what you mean by "in error." If the system requirements are based on an objectively erroneous implementation of the standard, then an NC is in order. But if the system is properly documented and an error in judgement was made (or someone deliberately skirted the system) then the audit findings might be different.

 

[Helmut Jilling]

You might want to clarify what you mean by "in error." If the system requirements are based on an objectively erroneous implementation of the standard, then an NC is in order. But if the system is properly documented and an error in judgement was made (or someone deliberately skirted the system) then the audit findings might be different.

OK, to clarify. Assume the Quality Engineer evaluated whether the out of cal gage had an effect on product quality. He clearly evaluated the situation and recorded the results. He has met the requirements of the standard. As post #3 pointed out if the gage grew in one direction, it would not result in defective product being accepted and shipped, but rather the result would likely be that good product was probably rejected. In this case, the customer is protected, and no recall would be needed. There is no nonconformity.

However, let's assume scenario #2. Same situation as above, but the Engineer made an "error" in judgement when he made his evaluation. He assumed there were no defective parts shipped, but in fact his analysis was flawed and the gage growing was capable of passing defective product, then there would be a nonconformity because he didn't do a recall. The fact that he attempted to do the right thing, and meet the requirements, but was wrong, does not wipe out the nonconformity.

Please remember, I agreed with the writer, that a recall is not automatic. Only if nc product was shipped, does the customer have to be notified. But, merely attemting to do the right thing and keeping records does not determine whether there is a nc. It is based on whether the process is effective.

 

[Jim Wynne]

OK, to clarify. Assume the Quality Engineer evaluated whether the out of cal gage had an effect on product quality. He clearly evaluated the situation and recorded the results. He has met the requirements of the standard. As post #3 pointed out if the gage grew in one direction, it would not result in defective product being accepted and shipped, but rather the result would likely be that good product was probably rejected. In this case, the customer is protected, and no recall would be needed. There is no nonconformity.

However, let's assume scenario #2. Same situation as above, but the Engineer made an "error" in judgement when he made his evaluation. He assumed there were no defective parts shipped, but in fact his analysis was flawed and the gage growing was capable of passing defective product, then there would be a nonconformity because he didn't do a recall. The fact that he attempted to do the right thing, and meet the requirements, but was wrong, does not wipe out the nonconformity.

Please remember, I agreed with the writer, that a recall is not automatic. Only if nc product was shipped, does the customer have to be notified. But, merely attemting to do the right thing and keeping records does not determine whether there is a nc. It is based on whether the process is effective.

I'm not sure we disagree here. I don't think that a single instance of error is necessarily evidence that the overall effectiveness of the system is lacking. Stuff happens, as they say, and no amount of documentation and indoctrination will prevent stuff from happening. It can be kept to a minimum, and systems can be optimized, (which doesn't mean "perfected") but when humans are involved...
I would think that before a major NC could be applied, an auditor would have to show that the issue in question is systemic and likely to occur again in similar circumstances unless corrected. In other words, if the system allows for error when error can reasonably be anticipated and prevented, then I agree that there's a significant problem.

 

[Helmut Jilling]

I'm not sure we disagree here. I don't think that a single instance of error is necessarily evidence that the overall effectiveness of the system is lacking. Stuff happens, as they say, and no amount of documentation and indoctrination will prevent stuff from happening. It can be kept to a minimum, and systems can be optimized, (which doesn't mean "perfected") but when humans are involved...
I would think that before a major NC could be applied, an auditor would have to show that the issue in question is systemic and likely to occur again in similar circumstances unless corrected. In other words, if the system allows for error when error can reasonably be anticipated and prevented, then I agree that there's a significant problem.

Certainly, this would not be a total system failure. However there are several definitions that require a major nonconformity. One that is an automatic Major NC is:

"Any noncompliance that would result in the probable shipment of nonconforming product. A condition that may result in the failure or materially reduce the usability of the products or services for their intended purpose."

The situation described in this thread would fall under this. A gage that is out of spec, when discovered, without the correct reaction response, would be an automatic major. A witness auditor would flag an auditor if he failed to make that connection. If the incorrect reaction response was caused by an error in judgement, it would not preclude the major. It would be similar to someone doing an incorrect inspection caused by setting the checking fixture incorrectly.

 

[Jim Wynne]

Certainly, this would not be a total system failure. However there are several definitions that require a major nonconformity. One that is an automatic Major NC is:

"Any noncompliance that would result in the probable shipment of nonconforming product. A condition that may result in the failure or materially reduce the usability of the products or services for their intended purpose."

The situation described in this thread would fall under this. A gage that is out of spec, when discovered, without the correct reaction response, would be an automatic major. A witness auditor would flag an auditor if he failed to make that connection. If the incorrect reaction response was caused by an error in judgement, it would not preclude the major. It would be similar to someone doing an incorrect inspection caused by setting the checking fixture incorrectly.

I guess we do disagree then. If I can show that my system is effective (i.e., if I can show multiple instances where the recall system worked as designed) and there is a single instance of a human making a mistake, there should be no reasonable presumption that system is not effective. I will agree that the hypothetical here would be unusual; most often when things like this occur it's because the system is flawed somehow, and a major NC would be justified and should be expected. Nonetheless, if I could show that my system is optimized and subject only to unpredictable human error, then I would fight an NC to my dying breath, and I would either prevail or have a new registrar.

 

[Helmut Jilling]

I guess we do disagree then. If I can show that my system is effective (i.e., if I can show multiple instances where the recall system worked as designed) and there is a single instance of a human making a mistake, there should be no reasonable presumption that system is not effective. I will agree that the hypothetical here would be unusual; most often when things like this occur it's because the system is flawed somehow, and a major NC would be justified and should be expected. Nonetheless, if I could show that my system is optimized and subject only to unpredictable human error, then I would fight an NC to my dying breath, and I would either prevail or have a new registrar.

Well, I don't know what else to say. I agreed with you that it is not a system failure. Normally, a single lapse is a minor. But if nonconforming product ships as a result, it is a major. I quoted the rule in my reply.

However, I don't make the rules, and neither do registrars. When you become certified, you agree to comply with the standard, and the rules that go with them. You can argue, or, you can even change registrars, but we don't make the rules. If you shop for a registrar that will bend the rules to suit your preferences, well...I wouldn't give a lot for that kind of an organization.