Visual Inspection Effectiveness

[kwalityguy]

I have a defect that happened on 1 part out of 10,000 pcs shipped over a 2 year time period. It is a faint line in a dull gray sheet metal part. Our customer wants to know why we didn't visually detect this. I'm saying visual inspection is not capable of detecting a 100 ppm defect. Anyone else go to battle about this? How did you handle it?

 

[Sidney Vianna]

Our customer wants to know why we didn't visually detect this. I'm saying visual inspection is not capable of detecting a 100 ppm defect.

I don't understand what the ppm has to do with the ability to detect the defect. The likely reason for failing to detect the problem is part of so called "human factors".

 

[Mike S.]

Visual inspection is very rarely 100% effective. Now, it may be 100% effective or virtually so for gross stuff, like differentiating a car from a truck, or a similarly easily differentiated characteristic, but in most practical situations that employ it, visual inspection has varying degrees of inherent error. You may need to explain that to the customer. I certainly have to do it from time-to-time.

 

[kwalityguy]

I don't understand what the ppm has to do with the ability to detect the defect. The likely reason for failing to detect the problem is part of so called "human factors".

I believe ppm is directly related to visual inspection. If the line problem occurred at a rate of 900,000 ppm (9 of 10 parts have the line defect), visual inspection should detect a good portion of these. But if the defect rarely occurs (just once every 2 years), visual inspection becomes very ineffective.

 

[Sidney Vianna]

I believe ppm is directly related to visual inspection. If the line problem occurred at a rate of 900,000 ppm (9 of 10 parts have the line defect), visual inspection should detect a good portion of these. But if the defect rarely occurs (just once every 2 years), visual inspection becomes very ineffective.

Basically, you are saying that the inspector would only react to a defect if s/he observed it a number of times. If the inspector does not have visual acuity, the light conditions are inappropriate, the criteria for rejection is unclear, etc... s/he will miss the defect 100% of the time.

 

[Miner]

It's more of a vigilance issue. If the inspector sees a defect frequently, they remain vigilant. If they rarely see a defect, they begin to relax and pay less attention.

 

[Sidney Vianna]

It's more of a vigilance issue. If the inspector sees a defect frequently, they remain vigilant. If they rarely see a defect, they begin to relax and pay less attention.

Typical example of "human factor".

 

[Tagin]

It is a faint line in a dull gray sheet metal part.

Its not clear from your post if the inspectors saw it but let it pass, or did not see it and let it pass. If the former, it could be a "calibration" issue, if the inspector(s) did not consider this to be a defect, yet the customer did consider it a defect; if the latter, then it could be vigilance/human factors, as Sidney noted, or for such a small defect rate perhaps the inspected % of lot is inadequate to achieve the necessary confidence of acceptable defects.

In any case, it seems like there is a disconnect between the (previously unstated?) customer expectations for maximum defect rate vs. your inspection criteria that will need to be ironed out.

 

[Jim Wynne]

PPM Has nothing to do with it, or anything else for that matter. An inspector or operator missed one piece after 10,000 pieces of presumably "good" ones had been produced and shipped. I say "presumably" because we don't know how many the customer might have missed, or might have seen and figured it wasn't something worth making a commotion about. The only prudent way to handle this on the customer end is to notify the supplier as a courtesy and get back to work. On the supplier end, it should be enough to express regret, replace the "bad" part, and move on.

 

[AgnieszkaSz]

Visual inspection in manufacturing cannot be 100% effective. The simpler the task, the quicker human mind gets bored. I heard an opinion that maximal effectiveness is 40% - an inspector has limited time to perform inspection and repetitive tasks are boring to death. The mind wanders. If you don't assess such kind of answer too risky, just say that the company decided to take the risk of this particular failure because of economical reason - it is cheaper to accept this risk than to invest in 100% proof inspection system.