Determining c=0 AQL numbers

When using a zero acceptance sample plan, how does one determine the best AQL number to use (i.e. .10, 1.0, 2.5, 4.0, etc...)? I've read a lot about the Z1.4 standard, but can't find anything that gives insight as to how to pick the right AQL number. The examples always seem to just pick one, and then derive a sample size from that. Can someone help me know what factors go into picking the right AQL number?

Thanks.

Since the cancellation MIL-STD-105E, the availability of statistically sound, sampling tables to personnel performing in-process and end item product audits has been scarce. Most people performing these audits today, are still using tables from this canceled document or are using Contractor's Sampling Tables. The table below is approved for use by DCMA QA personnel
performing zero-based sampling. lf no AQL is contractually specified, an AQL of 1.0% is suggested.

I cut and pasted this from the Defense Contractor Management Agency.
The Table can be found at

http ://guidebook. dcma.mill 226 ltools _links_fi lelstat-sample.htm. You may need to add this link to your Controlled Doc.s for 3rd party audits. I know I received a minor from my CB for not having it controlled.
Mike

miketay62 said:

When using a zero acceptance sample plan, how does one determine the best AQL number to use (i.e. .10, 1.0, 2.5, 4.0, etc...)? I've read a lot about the Z1.4 standard, but can't find anything that gives insight as to how to pick the right AQL number. The examples always seem to just pick one, and then derive a sample size from that. Can someone help me know what factors go into picking the right AQL number?

Thanks.

Click to expand...

Sampling, and determining what AQL to pick, is all about risk tolerance....how much are you or you customer willing to take? What sampling plan fits your risk tolerance? Is your process stable enough to even do sampling?

Often your customer will specify a desired AQL.

I might suggest spending a few bucks on Squeglia's "Zero Acceptance Number Sampling Plans" 5th Ed published by ASQ. Look over the OC curves and probability of acceptance tables. Consider how much sampling you can afford. Talk to your customer. All of this should give you a starting point.

How to pick the AQL: it's a simple process but it's NOT necessarrily easy to execute.

remember that the AQL is the defect rate that is ACCEPTABLE. inother words if the lot is at the AQL defect rate, you will ACCEPT it XX% of the time. XX% is traditionally 95% adn most published tables use 95%.

if your Custoemr doesnt' give you the AQL, then YOU must decide what defect rate you can tolerate in manufacturing or downstream processes if you are plannign incoming inspection or at the Customer or end user if you are planning product release inspection. you must think about teh severity of the effect fo the defect in terms of time delays, availability, cost and customer experience (loyalty).

given the AQL, you can then use various tables or formulas to determine the sample size for a c=0 plan.