Acceptable Outgoing Quality Level (AOQL) aka Acceptable Outgoing Quality (AOQ)

R

Randy Feliciano

I have seen formulas for AOQ and is at a lost on which one to use.
The inclusion of Probability equation(the one using factorials) is quite laborious.
Others that are quite direct seem contradicting.

Can any one help me compute for the AOQ in a valid direct way using only any or all of the folllowing parameters/entities:
1.) Sample size and lot size;
2.) No of lots inspected and rejected;
3.) no of reject units.

Thanks guys
 
E

e006823

Randy Feliciano said:
I have seen formulas for AOQ and is at a lost on which one to use.

Can any one help me compute for the AOQ in a valid direct way using only any or all of the folllowing parameters/entities:
1.) Sample size and lot size;
2.) No of lots inspected and rejected;
3.) no of reject units.
There is some shareware available at (broken link removed)
that will allow you to "take an existing sampling plan and use the software to evaluate it including calculating and displaying OC curves, AQL, LTPD, AOQL, ASN curves and much more. You also specify the desired protection and the program will generate a list of sampling plans that might be used."


Bob
 
Last edited by a moderator:
R

Randy Feliciano

AOQ or AQL

Thanks for the response.

I hope someone out there will take time just like you did.
Thank you.
 

Tim Folkerts

Trusted Information Resource
Randy Feliciano said:
I have seen formulas for AOQ and is at a lost on which one to use.
The inclusion of Probability equation(the one using factorials) is quite laborious.
Others that are quite direct seem contradicting.

Can any one help me compute for the AOQ in a valid direct way using only any or all of the folllowing parameters/entities:
1.) Sample size and lot size;
2.) No of lots inspected and rejected;
3.) no of reject units.

Thanks guys

The more I think about this, the less value I see in AOQ and the more I see in AOQL (Average Outgoing Quality Limit).


The calculation of AOQ assumes that the defect rate is known and is constant. That requires a process that is in control. However, I concur with Pyzdek's "Quality Engineer Handbook" when he says "... acceptance sampling should never used for processes in statistical control." Basically all you are doing is fishing around for defects to imrpove the overall quality. You ought to just improve the process to begin with.

If it's not in control (or you just don't know if it is in control), then you might expect large variations between lots. If you don't know the true fraction defective, then you can't calculate the AOQ. YOu can sample to look for the bad lots, but you don't know enought to calculate any meaningful overall AOQ.

It's a Catch-22.
If you know the quality, then you shouldn't sample and/or find AOQ.
If you don't know the quality, then you can't calculate AOQ.

However, you can always calculate AOQL. You can try all the different defect rates that might occur and see which is the worst. Then at least you can say "I don't know how good my process is, but I can assure you that after sampling, then the defect rate will average no more than x%."

Tim


P.S. From Pyzdek, the equation for AOQL is
AOQL = y [ 1/(sample size) - 1/(lot size) ]

where y is related to the acceptance number, c.

c y
0 0.368
1 0.841
2 1.372
3 1.946
4 2.544
5 3.174
10 6.535
 
P

Puzzle

Tim,

That is rather excellent.

Not having had to calculate an AOQL before, what figure would be considered good/excellent.

I have to assume the lowest possible.

Am I correct?

Chris
 
Top Bottom