AQL - AOQL - What % equals a certain level of AQL

M

miss quality

Hello,
I am having issues with AQLs and need some explanation!
here we go: to what % equals a certain level of AQL (for example AQL 4.0)?
Some companies like to use AQLs = %, however, looking at a MIL STD 105 sampling rule, AQL 4.0 does not equal to 4% (depending on the sampling plan it varies from 7 to 8%).

this question is basically related to the following issue: products passing the incoming inspection but "failing" in production (or in 100% inspection) if limit is fixed at 4% for AQL 4.0.

thank you!
 
M

miss quality

Hi Mark, I am not that advanced, no! except if we can say AQL in incoming = LTPD in production in accepted lot? would that be true?
in other words: I have 1 lot that passes inspection level 2 normal sampling plan, AQL 4.0. In production, what % of defects is the lot allowed to have (or if I find 4 % of defects (so after 100% inspection in production), is my lot still in specs?

where can I find this info for any AQL level? is there a corrsepondance between AQL in incoming and % in production?
thanks again for your help, I appreciate!
 
U

Umang Vidyarthi

Hi Mark, I am not that advanced, no! except if we can say AQL in incoming = LTPD in production in accepted lot? would that be true?
in other words: I have 1 lot that passes inspection level 2 normal sampling plan, AQL 4.0. In production, what % of defects is the lot allowed to have (or if I find 4 % of defects (so after 100% inspection in production), is my lot still in specs?

where can I find this info for any AQL level? is there a corrsepondance between AQL in incoming and % in production?
thanks again for your help, I appreciate!

A sampling plan and an AQL are to be chosen in accordance with the "Involved Risk". AQL by itself does not indicate chances of acceptance/rejection of lots. It only relates to, what might be expected from a series of lots. It is mandatory to check the OC curve of the sampling plan to determine the "risk factor."

There is a very good website developed by Galit Shmueli that develops sampling plans, interactively with the user. This may help you select the right sampling plan in accordance with Mil Std tables.

Since you are new to the subject, I would suggest you go through the guidelines provided in Mil Std 105E, and some articles on the subject.

Umang :D
 

Attachments

  • MIL_STD_105E.pdf
    2.2 MB · Views: 1,794
M

miss quality

Hi Umang,

thank you for the reply and the link. My AQL levels are already defined. I took the 4.0 level as an example.
I am just trying to figure out if AQL 4.0 = 4% or more. if we look at the mil std sampling plan, % are usually equal to *1.5 to *2 the AQL level but I do not know if I can base myself on this, probably not?
in other words, if I supply a lot that passes incoming at 4.0, can the lot be rejected if it is found to have more than 4% of defects classified with an AQL 4.0 in production (= on the total quantity, not just a sample of the lot)?
thanks again
 
B

B.Read - 2010

Hi,

If you are looking for the answer to does a 4.0 AQL equal 4%, I beleive you have already reached your answer. The trick is understanding your process and developing a statiscally valid sampling plan based on you process performance. The AQL and LPTD are based on your sampling plan. The AQL is that percent defective with a 95% percent chance of acceptance. The AQL describes what a sampling plan will accept. The LTPD is that percent defective with a 10% chance of acceptance. The LTPD describes what the sampling plan will reject.

You mention your AQL is set for you. As your process matures and gets better, your sampling plan should change as well. This gets handy when it comes to cost of Quality discussion with management.

This is the link to a pretty good article about sampling plans www.variation.com/techlib/as-7.html

Sorry to confuse you more
 
M

miss quality

Hi B. and thank you for your reply.
I may be slow because I am still not sure about the answer. sorry!!!
Here is the case: I am yor customer and I received a lot. The lot passed the incoming inspection AQL level 4.0.
During production I decide to separate all the defects classified with an AQL 4.0 from the rest of the lot. At the end, I come up with 4% of defects in the lot.
Based on this, I decide to reject the lot.
Is this legit to consider then that AQL = %? If it is, based on what (text etc)? if not, same thing.

Looking at my sampling plan the % of rejects acceptable should be 21/315 = 6.67%.

Sorry for asking again.

Regarding ym AQL levels, I can not change them as they are inherent to my industry, not just my company (glass industry). if we could have the same sort of process control as pumps or plastic, it would be great but this is not going to happen before long!

thanks again for your help, I really appreciate it :)
 

Bev D

Heretical Statistician
Leader
Super Moderator
Hi B. and thank you for your reply.
I may be slow because I am still not sure about the answer. sorry!!!
Here is the case: I am yor customer and I received a lot. The lot passed the incoming inspection AQL level 4.0.
During production I decide to separate all the defects classified with an AQL 4.0 from the rest of the lot. At the end, I come up with 4% of defects in the lot.
Based on this, I decide to reject the lot.
Is this legit to consider then that AQL = %? If it is, based on what (text etc)? if not, same thing.

Looking at my sampling plan the % of rejects acceptable should be 21/315 = 6.67%.

I'm not completely clear about your question, but I think I understand one part of it (Please correct me if I've got this wrong): If you as the customer have an AQL of 4% for your supplier and then you 'find'(?) 4% of the lot to be defective you should not reject the lot back to the supplier. An AQL stands for ACCEPTABLE QUALITY LEVEL; the sampling plan is structured to accept a lot that 4% defective 95% of the time...so, in this example you got what asked for...

as for the 21/315 = 6.67%: I'm assuming that this refers to a sample size of 315 units with accept on 21 defects? if so, it's important to remember in this case that the defect rate in the sample will NOT = the defect rate in the lot...
 
M

miss quality

Hi B. thanks again for the reply :agree1:
you pretty much understood my question:
If you as the customer have an AQL of 4% for your supplier and then you 'find'(= collect all the defects seeing during production) 4% of the lot to be defective you should not reject the lot back to the supplier.
but then I guess if I have 4.1 %, then the lot could be rejected, couldn't it?

it's important to remember in this case that the defect rate in the sample will NOT = the defect rate in the lot...
so in that case, can we really compare AQLs (so a sampling plan) and acceptable level in the whole lot?

Thank you again for helping me, this is like chinese written with a roman alphabet to me :nope:
 

Bev D

Heretical Statistician
Leader
Super Moderator
but then I guess if I have 4.1 %, then the lot could be rejected, couldn't it?

Well, you could but the probability of accpetance or rejection isn't a 'cliff' it's morel like a gentle slope so the probability that your supplier 'accepted' or will accept a lot that is 4.1% defective is still pretty high, certainly greater than 90%. (That just seems a little unfair to the poor sypplier doesn't it?) That is one of the inherent weaknesses in the AQL system. For this reason I prefer a straightforward RQL - Rejectable Quality Level - system that will provide a 95% or even 99% probability of rejecting a lot that is some percent defective - this actually protects me as the Customer a lot better and is more intuitively obvious. I have posted on the appropriate formula to use in other threads if you are interested.
 
Top Bottom