Can someone explain in lay man term what is LTPD and AQL. Would appreciate an example. When do we use LTPD and AQL?

Both terms are related to sample inspection using standard sampling plans.

AQL is the Acceptable Quality Level, and is the contractually agreed acceptable amount of defects in a batch. So for example an AQL of 0.1 would mean that the customer accepts 0.1% defects in a batch.

LTPD is the Lot Tolerance Percentage Defective - and I am struggling to recall what that is about at the moment !

Johnny , Martin

AQL is as Martin said the mutually agreed on quality level that the customer finds acceptable. Actually, plans based on AQL will not in general produce lots to that AQL.

AQL is the level which will be accepted most of the time. Most generally for ANSI-ASQ V1.4 plans usually means in the 90 to 95% of the time.

The actual long term average of the lots is another measure AOQL, Average Outgoing Quality Level.

LPTD or Lot Percent Tolerant Defective is the level of quality that will be rejected most of the time. Usually for Dodge Romig type plans at 10% probability.


AQL plans should generally be used as short term stop gap measures to control quality to some tolerable level while improvements aree made. Or so my advisor, Ed Schilling indicates in his book on accpeptance sampling.

LPTD will do more to insure high q2uality to any particular level, but the price paid is increased asampling. Management usually does not like this.

Most managers I have worked with believe that they will not get a lot above the AQL using this kind of a plan. They are usually surprised when a "bad" lot is accepted andd makes it to the floor.

If you study AQL acceptance sampling curves, you might be surprised at how "loose" the common ones are from the customer standpoint.

Does it matter which to use (AQL or LTPD) if I buy-off at in-coming or out-going inspection? I (the QA) buy-off at out-going visual inspection...when do I accept the lot, reject or sample again. Sometimes when there is high reject, Production is requested to performed 200% inspection before the product it brought to QA for buy-off. Question is in what situation it warrant a 200% inspection?

johnnybegood,

In answer to your question of 'what situation warrants 200% inspection'. No situation warrants 200% inspection. You can not inspect quality into a batch of material. All you can do is cull out bad product. Typically to get to around 99% confidence you need to do 350% inspection. It just doesn't make sense.

Now, to the practical side of things. Sampling schemes only work if you apply the scheme not the plan. An AQL plan such as the old military MIL-STD-105 only worked if you used rectification and switching procedures. That is, the specific rejected lot was 100% inspected and the plan went to tightened or loosened inspection based on past results. The sampling scheme was set to control quality levels over time, not for a specific lot of material. As Dave S. said, AQL plans should generally be used as short term stop gap measures to control quality to some tolerable level while improvements aree made. If you need to use sampling plans/schemes, there are many to choose from. As Dave S. recommends, I also suggest you get a copy of Ed Schillings book on acceptance sampling. IMHO it is the bible on sampling. It is a rough read, pretty technical. If you want something a little easier as a starter try Dale Besterfield's book on Quality Control. He dose a pretty good job of giving an overview of sampling plans and the inherent risks associated with them.

Just to muck up the works a bit, you need to evaluate any plan based on its risks. You should review the Operating Characteristic Curve (OC), and the Average Outgoing Quality (AOQ) and Average Outgoing Quality Limit (AOQL) to determine if the risks are accepptable.

Bottom line, you are better off spending resources on fixing the process than improving the sampling scheme. Sampling costs money and does not improve the product.

Hope this helps.

Bump in the hopes someone offers additional help on the subject.