Effectiveness of 200% Visual Inspection to Identify Defects and Defectives

[beaser3]

Can anyone tell me what the increase in effectiveness is if you go from 100% visual inspection to 200% visual inspection?

We have a part that gets inspected 100% by the operator prior to getting packaged. Our customer found 2 defective parts out of a shipment of 500 pieces. We are not able to prevent the problem from happening at all and the suggestion at our corrective action meeting was to put have another operator "audit" the parts prior to boxing them. I know that this would not eliminate our chances of a defect but would it even help increase the detection number?

Thanks in advance!

 

[somashekar]

Re: Effectiveness of 200% inspection


I cannot say much about how effective it would be, but I guess that the first operator will feel relaxed since now there is a second operator to do the same stuff and the second operator will feel relaxed since his stuff is already been done by the first operator ...

 

[TShepherd]

Re: Effectiveness of 200% inspection

Hello,

What I have been taught and what I have experienced since 1967 is that 200% inspection is less effective than 100% and 300% inspection is again less effective than 200% inspection and so on and so forth.

The reason being is that each added 100% inspection is looking for fewer and fewer non-conforming parts and the probability increases that a non-conformance will get through.

I believe that you had said that your discussion for resolution was an audit after 100% inspection was being considered, in which case the probabilty is even greater that an additional 100% sort would fail.

You might consider mistake proofing through a vision system or fixturing element, however not knowing the part perhaps this is not feasible.

Tom

 

[ScottK]

Re: Effectiveness of 200% inspection

I think you are far better off making that 100% inspection more effective through engineering controls or, better yet, eliminating the problem altogether.

But even before that - did you sell the part at 0 defects? Clearly the process is not capable of 0 defects and additional inspection is adding cost.

 

[jasonb067]

Re: Effectiveness of 200% Visual Inspection

How many parts were inspected to get to the 500 which were sold?

I am reading that you are trying to convince yourself or others that 200% inspection is not going to be worthwhile. Is that the case?

 

[beaser3]

Re: Effectiveness of 200% Visual Inspection

We ran a total of 518 pieces to get the 500 that were sold. 8 out of the 18 rejected were for the same defective condition that the customer found. I don't necessarily want to convince anyone that this is not the way to go but I don't want everyone to walk away from the table thinking we are making a large improvement. I am also concerned about the added labor cost considering that these are low volume/low profit jobs to begin with. The group has already decided that we cannot prevent the reject from being produced so we have moved to detection. I would be more comfortable with doing a final box audit vs. putting another operator at the press if they will provide the same end result. Does that make sense?

 

[TShepherd]

Re: Effectiveness of 200% Visual Inspection

Clarification

In my response I was not saying that 200% inspection is not worth while but rather was refering to the effectiveness - which decreases from a statistical and practical view.

If it is the only alternative until a better resolution is found and it placates the customer (until another non-conforming parts arrives) it may be the right thing to do.

I get an inkling that the team may be looking to reduce the Detection Number of the RPN on the FMEA based on 200% inspection (not sure) - however it shouldn't unless the customer buys it.

 

[ScottK]

Re: Effectiveness of 200% Visual Inspection

We ran a total of 518 pieces to get the 500 that were sold. 8 out of the 18 rejected were for the same defective condition that the customer found. I don't necessarily want to convince anyone that this is not the way to go but I don't want everyone to walk away from the table thinking we are making a large improvement. I am also concerned about the added labor cost considering that these are low volume/low profit jobs to begin with. The group has already decided that we cannot prevent the reject from being produced so we have moved to detection. I would be more comfortable with doing a final box audit vs. putting another operator at the press if they will provide the same end result. Does that make sense?

I don't know what industry you're serving but maybe it's time to manage the customer's expectations and tell them that the process is simply not capable of producing perfect quality and set a limit of, say 1% defective where you will refund them for all defectives but not write a corrective action response for 1% or less by lot.
If the rate is over 1% you will define actions to bring it back below 1%.

 

[jasonb067]

Re: Effectiveness of 200% Visual Inspection

I would say that based on 10 being defetive and 8 being found you have a 80% effective visual sort criteria. Keep in mind this is one data set and one is not a trend.

So, any secondary sort could only expect worse results for the reasons stated above, "i do not need to look so hard because that guy up there is looking to".

If this is a geometric feature I would think that a simple gage in the hand of the first inspector would be more effective from a cost and effectiveness stand point.

Is something like that just not possible?

 

[bobdoering]

Re: Effectiveness of 200% Visual Inspection

I don't know what industry you're serving but maybe it's time to manage the customer's expectations and tell them that the process is simply not capable of producing perfect quality and set a limit of, say 1% defective where you will refund them for all defectives but not write a corrective action response for 1% or less by lot.
If the rate is over 1% you will define actions to bring it back below 1%.

OR...you might track the number of rejects you are finding, do an attribute chart. If you see a jump in defects found, react accordingly.

Make sure you have effective sorting area - lighting, inability to mix parts, etc.

Using you customer as the sorter of last resort is always a problem. Why can they sort better than you? It is even worse if the defect causes an automated system to jam, creating irritation and downtime. They may have the right to charge back the part value and downtime, overhead, etc.