"Visual Quality Control Inspectors are able to find every defect" 1

[Julius_B]

Hello everyone,

I am curious to find out if you would agree with the statement "Visual Quality Control Inspectors are able to find every defect" (e.g., in an end-of-line quality control) and if you think that a discrepancy between the perception and reality of the human inspection accuracy exists in practice.

I am looking forward to insightful discussions!

Best
Julius

 

[weldstan]

Errors will always occur some/most of the time, Lighting issues, inspectors thought processes during inspection, distractions, experience, visual accuity, knowledge, etc. All and more play a part in inspection acuity.

 

[Julius_B]

In my experience, inspector accuracy and especially the inter-inspector accuracy (i.e., how often do inspectors agree with each other on the evaluation of the same part) is overestimated by superiors. So do you think that the issue and the extent of the issue of inspector inconsistencies is widely known and accounted for in the industry practice?

 

[IRstuff]

The statement is nonsense at its face value; no human is 100% accurate and no human can even sustain 100% alertness for more than a few minutes.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[SWComposites]

Yep, based on my experience with aerospace QC inspectors, I agree with IR, that statement is rubbish, and likely written by someone either clueless or delusional.

 

[geesaman.d]

Any visual inspection operation is only good for a given certainty, say 90% or 95%. If you want higher certainty, you need to layer the inspection steps. Naturally, 100% only happens at infinite inspection steps.

That said, in cases of one-off components, a properly detailed and structured visual inspection will miss very little. The challenge comes with quantity.

 

[3DDave]

A visual inspection should follow a checklist so missing things is far less likely. Consider that commercial pilots use visual inspections of the instrument panel based on a checklist it's reasonable to rely on it.

 

[Gr8blu]

Think about what you're really saying. If the inspection is visual, the defect has to be visible to be perceived. How do you visually check the surface finish of a cylinder bore or the concentricity of a piston to a cylinder once the engine is completely assembled (which is where an end-of-line QC check would come into play)?

Converting energy to motion for more than half a century

 

[IRstuff]

Much depends on how much the inspector needs to check; a single thing that's relatively easy to inspect might come close, but something that has, say, 300 features to inspect, even with a checklist, is likely to see errors leak through.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[SJones]

No, I don’t think that a discrepancy exists for the 6 or 7 reasons stated above, notwithstanding the possibilities of hidden imperfections (which is why VT is supplemented by other NDT methods).

Steve Jones
Corrosion Management Consultant

http://www.linkedin.com/in/drstevejones

All answers are personal opinions only and are in no way connected with any employer.

 

[Julius_B]

If the design of the visual quality control is objectively great (e.g. each inspector looks for only one defect class, regular retraining, regular and short breaks for inspectors, ...) and no real feedback on the quality assessments is collected, how will I ever know how accurate the inspection really is?

 

[Dhurjati Sen]

I can comment on the oil and gas industry where inspection is quite important. But the inspectors or the inspection engineers miss out sometimes, because they are overworked or they do not know where to look into. Two sets of eyes are always better for the first case, while for the second, there is no substitute for experience.

DHURJATI SEN
Kolkata, India

 

[SJones]

Do you mean 'reliable' rather than "accurate"?

Steve Jones
Corrosion Management Consultant

http://www.linkedin.com/in/drstevejones

All answers are personal opinions only and are in no way connected with any employer.

 

[Julius_B]

I am referring to accuracy in terms of how many parts classified by the visual inspection as OK are indeed OK and do not contain any missed errors. @SJones

 

[SJones]

Reliability then. Google “Reliability of nondestructive testing”. Should keep you going for a while.

Steve Jones
Corrosion Management Consultant

http://www.linkedin.com/in/drstevejones

All answers are personal opinions only and are in no way connected with any employer.

 

[r6155]

All NDEs require a visual exam, obviously. Example: Now the inspector needs to visually interpret the radiograph.
Remember: the inspector only decides if the product passes or not, but does not reject it. If it does not pass, it must be analyzed by the designer who rejects or approves it.

Regards

 

[SWComposites]

Julius - the "reliability" is going to be highly dependent on the inspector training, inspector motivation and workload, details of the visual inspection, difficulty of the visual inspection, etc.

One thing you could do is set up a controlled experiment; place a number of known defects on some parts, without any indication of where they are, include some no defect parts in the mix, and have a number of inspectors perform inspections multiple times; from that you should be able to get an estimate of the "probability of detection". Note that this will probably be optimistic as in real productions inspectors will be pressured to speed up the inspections, get tired, etc.

 

[FacEngrPE]

Inspection programs need the same attention to their design, that the rest of your product needs.
The way the results of inspection feed back to manufacturing is just as important as the inspection itself, as it is not enough to reject discrepant parts, you should be adjusting the process real time so that production of bad parts is minimized.

 

[0707]

No, visual inspectors, sometimes are able to detect visible defects.

 

[Tmoose]

The OP has not said what "defects" are expected to be, or need to be found.
A lot more context is needed up front to raise the OP to a discussable level.

On a rough cast surface some pretty significant shrinkage and linear ( cracks) defects may be missed by a well done visual inspection.
On a fully machined surface the odds of detection are much better.

Is anyone Expecting a visual inspection to detect a machined diameter that is .002" oversize?

So my take on the original question (as stated )is similar to IRstuff's and SWcomposites'. Rubbish.

I do wish Julius_B Good luck whipping his discussion into useable shape.
As stated it has a LONG way to go.