I know you people have headier topics to discuss but I need to make a plain english statement in a document concerning AQL and need a little help. Please evaluate this statement based on MIL-STD-105.

"An AQL of 1.0 means that a lot that is 1% defective will pass 95% of the time."

Thanks,
Glenn

Glenn -

The meaning you write may be true for some specific plan. However, not all plans will accept at 95% of the time.

Don't have my copy of MIL-STD-105 handy, but it has been supplanted by ANSI/ASQ Z1.4 anyway and it says the same as references below, I believe.

Reference ANSI/ISO/ASQC A3534-2-1993 Statistics- Vocabulary and Symbols - Statistical Quality Control , ANSI/ASQC Z1.4-1993 Sampling Procedures and Tables for Inspection by Attributes and, ANSI/ASQC S2-1995 Introduction to Attribute Sampling.

AQL is generally defined as "That level of quality acceptable to the consumer as a long term process average". In day to day terms, it means that the great majority of lots with quality equal or better than AQL will be accepted by the sampling scheme employed. They do not quantify the % acceptance of lots. For that you need to look at the OC curve of the specific plan selected.

Hope that helps.

Does anyone have a good scan of the tables?
MIL-STD-105E (http://Elsmar.com/pdf_files/MIL-STD-105E.pdf)

I have both the MIL-STD and the ANSI/ASQ Z1.4. After reviewing Dave's comment and re-examining the OC curves, I can see that for a given AQL the Pa for a series of lots varies with the sample size as well as the %defective (process average). So that, Code letter E for sample sizes 26-50 @AQL=1.0 says that if the pct def average is .077, the plan will accept 99% of the time but if the pct def average is .394, the plan will only accept 95% of the time. So, my "laymen's" statement turns out to be "The AQL is an index that determines the quantity of defects that are likely to be accepted by the sampling plan." So, less risk mean a lower AQL number and a lot more $$ to implement it. The business decision is a trade off.

This is a useful site if you cant be bothered with the tables.
Military Standard 105E (ANSI/ASQC Z1.4, ISO 2859) Tables (http://www.sqconline.com/)

Good link, cochranemurray! Thanks!

But i need details !!

Could you mention what are the details you need - otherwise no body can help you.

Before you do that, do us a favour. Go to the bar at the top of the page (blue in colour) and look for the 'search' function. Key in '105', set it to search for the whole forum and you'll be directed to the search engine. Key in 'mil std 105', 'search the forum' and you'll get about 2-3 pages of past discussions. I believe what you needed should be there!

Welcome and enjoy yourself.

Dear every body,

On the sampling plan also, please help me on below issue: I choose single sampling, inspection level II, AQL 0.65%
How can I know a confidence level for the lot to be accepted at customer site ?

Appreciate for your help.

Thanks...Vu Nguyen

Vu Nguyen,

First off, welcome. We're glad you found this site and hope you receive the information you want and perhaps yopu can contribute your own knowledge at some point.

For inspection level II, AQL 0.65%, for any lots up to 280 pieces, the table says to draw 20 pieces and accept with 0 defects (Line F in the table). Then in the back of the standard, you can look up "Table X-F" which lists probabilities of rejecting lots with varying quality - also known as the OC Curve. For example, lots with

0.0503% nonconforming will be accepted 99% of the time
0.256% nonconforming will be accepted 95% of the time
3.47% nonconforming will be accepted 50% of the time
15.0% nonconforming will be accepted 5% of the timeThe table lists a few other values, and you can also calculate thes directly if you really want. For other sizes of lots, you would have to look up the appropriate lines on the tables.


Tim F

Dear Tim,

I am gald to know you and thank you very much for your quick respond. I will look back the table you said to study more, if i have any issue that i not yet understand, i will ask again.

Best regards

Vu Nguyen

Dear Tim F.,

Thank so much for your instruction, I had looked on the table and have some more questions:
1. Within the table X-F, I found some columns that has same AQL, such as: 0.65, 2.5, 4.0, 6.5. How about this ?
2. The percentage of nonconformity (such as: 0.053, 0.256...) How to calculte this from the lot. Ex: Lot qty: 150, Sample size: 20, AQL 0.65, Inspection level II, single.

Best regards...Vu Nguyen

Is this standard lately been revised?

Dear Amanbhai,

I glad to know you, per your question, the standard i had used is Mil-STD-105D. Any idea please let me know.

Vu Nguyen

Is this standard lately been revised?

I'm not sure of the revision level / status of ANSI/ASQ Z1.4.

MIL-STD-105 is obsolete.

Dear Tim F.,
I had understood how to look on the table X-F, but how to calculate percentage of nonconformity ? ex: 0.0502% or 0.256% ? is it a percentage of nonconformity that we inspected from sample size ?
These percentages would be the "true" defect rate. Suppose you produce many parts and the actual defect rate averaged over all the parts produced was 0.256%. If you pack these into lots of 200 pieces, then you would expect those lots to pass inspection 95% of the time. You don't calculate this number based on the sample you test.

Thank so much for your instruction, I had looked on the table and have some more questions:
1. Within the table X-F, I found some columns that has same AQL, such as: 0.65, 2.5, 4.0, 6.5. How about this ?
Row F has several different sampling plans associated with it, all of which use a sample size of 20.
AQL 0.65: accept 0
AQL 2.5: accept 1
AQL 4: accept 2
...The higer defect rates you are willing to accept, the higher number of defects you will accept in a sample. The different columns in Table X-Fjust correspond to the different sampling plans.

2. The percentage of nonconformity (such as: 0.053, 0.256...) How to calculte this from the lot. Ex: Lot qty: 150, Sample size: 20, AQL 0.65, Inspection level II, single.
You don't calculate these percentages based on the sample you test. If you really need to know these percentages, you would have to inspect the entire lot.

These percentages are used for "what if" type calculations. If you produced many parts and the actual defect rate averaged over all the parts produced was 0.256% (and if you packed these into lots of up to 280 pieces), then you would expect those lots to pass inspection (normal, level II) 95% of the time.


Does that help?


Tim F


P.S. As I recall, there have been some minor changes in the text at the beginning of the standards, and some minor changes to the switching rules but the tables themselves haven't changed. I have the Z1.4 1993 version, and I know there is a more recent update (2003?). I'd have to check more carefully to know the exact differences between the specific versions of MIL-STD-105 and Z1.4

Dear Tim F.,

I got it, thanks so much.

Best regards...Vu Nguyen

This is a useful site if you cant be bothered with the tables.
Military Standard 105E (ANSI/ASQC Z1.4, ISO 2859) Tables (http://www.sqconline.com/)

thanks alot for the link.
I have farwarded that to relevant personnel in my company.