Choosing Nonconformities to Report

[QualiTEE]

Our current procedure calls for reporting/recording all nonconformities so we can capture our scrap costs and to watch for trends. Current management is looking to only record nonconformities that will cost >=$x to fix. Editing to add: and exclude certain categories from being eligible to formally reject (i.e.: tooling broke, part needs to be remade).

Just curious to see how common this is. How are your companies handling the decision on when to write a nonconformity?

 

[Jean_B]

Sage advice:
Don't when there's not a specification, but have a system that can sort out such candidates for making into specification.
If there's a specification there's an NC. Else you would have to differentiate must (spec) and should (guideline).
Avoid formalizing prescient selection like costing >= to fix, the 'important' ones, the dangerous ones. You will only know after the fact whether its true, or you'll convince yourself through confirmation bias that the only NC's worth it are the ones you're registering, because those are the only ones you have data for to show its true.
Do allow for an NC guide into a won't do further work for somesuch reason, but ensure it is trendable as death by a thousand cuts does exist.
Auto opt-ins (common sense) that don't need further rule-selection because their obviously important regardless of nitty picky details are not a bad thing, if they save cost and don't lose you resolution in analysis.

 

[Tagin]

Current management is looking to only record nonconformities that will cost >=$x to fix. Editing to add: and exclude certain categories from being eligible to formally reject (i.e.: tooling broke, part needs to be remade).

That seems like a really bad idea. It's one thing to make a decision to absorb the cost of occasional low-value nonconformities; it's another to make a decision to not classify them as nonconformities at all when the standard says you shall control nonconforming outputs.

 

[yodon]

I think both @Jean_B and @Tagin have given excellent advice, but my curiosity is piqued: what's management rationale for only recording the NCs > $x? Is the NC process overly burdensome? Is it not so burdensome but you have so many they're looking to pare the number down?

 

[QualiTEE]

I agree with everything you all said. I'm not on board with what management is suggesting.

Management claims the rejection-writing is a burden and takes too much time (costs too much money), plus they are looking to reduce high scrap rate. Editing to add: We have some issues with customer supplied material coming in out of spec. Our procedure states we write rejections up in these cases so we can track repair costs and watch for potential trends. Management does not think we should write anything up for this because the error is not ours and in some cases we won't be doing a charge-back to the customer.

 

[QualiTEE]

That seems like a really bad idea. It's one thing to make a decision to absorb the cost of occasional low-value nonconformities; it's another to make a decision to not classify them as nonconformities at all when the standard says you shall control nonconforming outputs.

But how do you absorb the costs on occasional low-value nonconformities without technically classifying them as an exception?

 

[Tagin]

But how do you absorb the costs on occasional low-value nonconformities without technically classifying them as an exception?

That's part of the disposition of NC material. First, you identify it as NC, then segregate it, then decide what to do with it (i.e., it's disposition). One of the disposition options, depending on your documented process, could be to scrap NC material (and absorb the cost) that is less than X$ unit cost (or similar criteria).

For that kind of low-value scrap, you might decide that your analysis is limited to monitoring that, for example, total scrap$/month is less than XX$. If its more than than XX$ then you do a deeper dive to see what's happening.

But it needs to be founded on risk: low cost does not always equal low risk; a circuit breaker in a product might be low-cost but high-risk if they are NC.

All of this starts, though, with identifying these items as NC, and then giving yourself choices in your NC process that are risk-appropriate.

 

[Ninja]

Reading the above, it sounds like all are agreeing but somehow not...

If a NC is identified, you shall control NC outputs.
Mgmt is asking to report (assumed "to us") only those of high enough value to do something about.
Both are easily done...control them all, report only those meeting X criteria, disposition the rest without bothering mgmt. Everyone (including the standard) get what they want...

 

[QualiTEE]

We're a small company of 24 people: 11 mfg personnel, 1 Mfg Mgmt, 1 QC, 1 sales, etc.

Instead of saying "report", maybe I should say "document". Management does not want to document the nonconformity; no formal rejection written to document the nonconformity itself or the disposition. Mfg labor would go to the original job or to an indirect scrap code with no traceability. Maybe that's "control" in a way, but I fail to understand the effectiveness of that method.

 

[Ninja]

A curiosity...I don't think it's "playing word games" ...

How are you declaring the NC on product? Is it really a NC, or is it a process that simply doesn't run at 100% (none of them do)?
0.5% scrap from a process can simply be "yield loss"...I wouldn't call it a NC in the first place...

It was normal for us to trash the first 50' and last 75' of the process...it was not an NC, it's just the process...

A process with yield loss costs over a certain amount...call it what you want, but fix it before picking a name...

Just thinking out loud...