AQL (Acceptable Quality Level) Sampling Plan for Small Lots

[new_to_this]

Hello
I work in a very small company, we never had sampling plan. Now we got a contract for some 300 - 400 parts that would be produced in 4 to 5 lots. Now the owner wants me to come up with sampling plan. Can anyone please instruct me on where to start, what would be the best and simplest sampling plan, because I am totally lost? Here is what I know:
-I know that we will be using AQL 1.0
-Operators would be responsible (they will measure each part they produce to make sure is within spec)
- We will be using this plan to inspect the parts before we ship them.
Do I need to have Sampling work instruction and procedure? (We are ISO17025 accredited)
Also I found some OC curve calculators on Excel but I am not sure how is fraction defect (p) determined? Can anyone please help?
Thanks in advance for all your inputs.

 

[Bev D]

Re: AQL Sampling Plan for Small Lots

welcome! where in Indiana? I am from Huntington, now living in Maine. shame about Butler!

I assume you mean an AQL of 1%?

the easiest thing to do is to go to http://www.sqconline.com/
and use the mil std 105 calculator. Enter in 1% (it's the default) and the lot size. (no need to change any other parameters for now) hit submit and this will give you the plan. For example I used a lot size of 91-150 and the sample size is 13 units, accept the lot if there are no defects and reject the lot for 100% screen if there are 1 or more defects...
the OC curve is shown at the botom of the page.

 

[Tim Folkerts]

Re: AQL Sampling Plan for Small Lots

-Operators would be responsible (they will measure each part they produce to make sure is within spec)

If you are indeed measuring each part produced, then sampling isn't really what you need. Sampling is used when you are only measuring some of the parts and using that info as a way to estimate the overall quality.

In this case, I would suggest SPC -- probably the I-MR chart since the parts seem to be individually produced. This will give you a much better idea about the quality of the parts than sampling would. Since it seem you are already planning to collect the data, SPC would not add much to the time and effort involved.

If you want to double check the operators with a final inspection, then sampling would be appropriate there. You or some other expert would re-inspect a sample of the items to ensure that they really are as good as the operators think.

Tim

 

[brossbach1]

Re: AQL Sampling Plan for Small Lots

I hope this can help on what you are looking for. I have attach a sampling plan.

 

[bobdoering]

Re: AQL Sampling Plan for Small Lots

I work in a very small company, we never had sampling plan. Now we got a contract for some 300 - 400 parts that would be produced in 4 to 5 lots. Now the owner wants me to come up with sampling plan. Can anyone please instruct me on where to start, what would be the best and simplest sampling plan, because I am totally lost?

It would be handy to know the process to give you more detailed answers. For example, I am concerned with some small lot processes that the lot has so few parts that the run is done prior to warmup (period prior to reaching steady state) of the machine (or very little of the process is generated after warmup.) That offers less benefit from standard charting methodologies. Warmup is actually a special cause.

 

[CristianoP]

Re: AQL Sampling Plan for Small Lots

I hope this can help on what you are looking for. I have attach a sampling plan.

As you already know, if you calculate the probability of acceptance of a lot with a proportion of defectives = AQL, you get 1-alpha.

Using a free tool (like Gnumeric) I see:
sample size= 38896, acceptance number= 1 -> alpha(0.03%) = 0.99999234041342
which has nothing to do with alpha= 1% in the line: "Producer's Risk (Alpha) 0.01".

The best plan that fulfills your requirements (AQL, alpha, RQL and beta) is:
sample size= 387, acceptance number= 1, which gives
alpha(0.03%) = 0.0060488551827
beta(1%) = 0.09997418407881

http :// www .webalice.it/cristiano.pi/SSP003.tif - OBSOLETE BROKEN 404 LINK(s) UNLINKED - PLEASE HELP - REPORT POSTS WITH BROKEN LINKS


If you like double sampling plans, you could use this:

http :// www .webalice.it/cristiano.pi/DSP003.tif - OBSOLETE BROKEN 404 LINK(s) UNLINKED - PLEASE HELP - REPORT POSTS WITH BROKEN LINKS

Cristiano

 

[Hygins]

Re: AQL Sampling Plan for Small Lots

Hello,
my name is Hygins, I have been looking at different post and never got the courage to write back to any post.
Guess that today will be my first time here.
My company does work on small lots size too. I'm working on attribute sampling plan. Right now, I know which AQL (2,5) I want to set for the incoming control, but I don't really get the link between the alpha (consumer risk), the probability of acceptance and the AQL.

 

[Bev D]

Re: AQL Sampling Plan for Small Lots

hello Hygins!

to begin answering your questions:

alpha is related to Producer's risk not consumers risk. It is the probability that inspection will 'over detect' and therefore reject a lot that is at (or below) the Acceptable Quality Level (AQL). This happens when the sample has a non-representative number of defects in it compared to the true defect level of the lot. The producer will then end up re-inspecting or reworking a lot that has an acceptable quality level.

beta is the Consumers risk. This is the probability that inspection will not detect and therefore accept a lot has a defect rate at or higher than the Unacceptable Quality Level (also known as the RQL or LTPD). This results in the consumer - or other customer - receiving material that is too defective.

for incoming control you may want to specify both an AQL and an RQL (LTPD) level...what level is OK to accept and what level will you want ot reject back to your supplier(s)...

 

[Hygins]

Re: AQL Sampling Plan for Small Lots

Thanks Bev D! I got it now.