Hi,
I've been asked to determine what sampling plan should be used to prove that the lot is bad and fails. The known input I've been supplied is that the lot has 0.5% defective product. The lot size is variable.

Any information on this would be greatly appreciated.

thanks

Hi,
I've been asked to determine what sampling plan should be used to prove that the lot is bad and fails. The known input I've been supplied is that the lot has 0.5% defective product. The lot size is variable.

Any information on this would be greatly appreciated.

thanks

Here are a couple of Sampling Plans:

1. MIL-STD-1916, DoD Preferred Methods for Acceptance of Product

2. ANSI/ASQ Z1.4, Sampling Procedures and Tables for Inspection by Attributes

3. MIL-STD-105 (which has been replaced by #2), but still available to download at: Assist (http://assist.daps.dla.mil/quicksearch/)

Added in: One other option is to use no sampling and perform 100% inspection.


Just curious here: Why does your Management want you to reject (not accept) the product philosophy? Why not look to accept the product?

When making the calcualtions, you have to be careful jsut what you mean to find.


Do you want to be fairly certain to reject a lot that is at least 0.5% defective (and perhaps also reject some lots that are less defective)?
Do you want to be fairly certain to accept a lot that is at most 0.5% defective (and perhaps also accept some lots that are more defective)?
Do you want to know the actual defect rate within some range (for example "I am 95% certain the defect rate is 0.5% +/- 0.1%


Each of these would require a different approach.



A while ago a posted a spreadsheet that helps you determine an appropriate sampling plan based on specified defect rates. It is described in this thread: http://elsmar.com/Forums/showthread.php?t=12836

The spread sheet is set up to enter 4 numbers. For example 5% chance of accepting a lot with 0.5% defective and 5% chance of rejecting a lot with 0.1% defective. Then you play with the sample size until you find something that works. For such small defect rates, you typically need very large numbers -- for the example I just gave, you would draw about 1300 parts - accepting 2 defects and rejecting 3 defects.\


Tim F

Hi,
I've been asked to determine what sampling plan should be used to prove that the lot is bad and fails. The known input I've been supplied is that the lot has 0.5% defective product. The lot size is variable.

Any information on this would be greatly appreciated.

thanks

If you already know that "the lot" has .5% defects, why do you need to sample?