Sample Size Calculation for PPM (Parts Per Million) Prediction

I'm asked to calculate the appropriate sample size to inspect for a process with a process average in the range of 200-500 ppm. Is there any particular formula to calculate the sample size so that a defect can be catched?
Anyone pls advise.
Thanks.

Would like to help

Tohbin -
I'm sure others like myself would like to help, but you provide insufficient information.

Is it attribute or variable data?

What do you mean by inspect? Inprocess indicator? Accceptance sample? What is your technique?

What risk are you amiable to? Both Alpha and Beta.

Do you have any knowledge of the extant variability in the process?

There are probably other factors, but we can't even begin until you give more information.

Tohbin,

As Dave mentioned this is a very complex statistical question, even though it doesn't seem so at first. I'm no statistics guru, (I'm knowledgable enough to be dangerous), but I think it would help if you got a copy of, or did a Google search for information on, MIL-STD-105E and/or ANSI Z1.4-1993. These standards involve acceptance sampling and I am sure there are internet sites that can give you a decent amount of background knowledge on them and the basic statistical principals involved so that you can answer your question.

Hope this helps -- let us know.

Re: Sample size calculation

Originally posted by tohbin
I'm asked to calculate the appropriate sample size to inspect for a process with a process average in the range of 200-500 ppm. Is there any particular formula to calculate the sample size so that a defect can be catched?
Anyone pls advise.
Thanks.

Click to expand...

Tohbin,

A Handbook is attached in the thread "Number of responses".

http://elsmar.com/Forums/showthread.php?s=&threadid=6022

Chapter 4 of the handbook might resolve your issue.

use this link for mil-std-105

go here for a copy of mil-std-105

This may help.

Regards,
CarolX

PPM

I'm using attribute data. Inspection here is carried out on a testing process where the product is inspected for any kind of defects induced after the testing process. I only have data that shows the process average is around 200 - 500 ppm. I just want to know what should be the number of samples to be inspected in order to make sure that my testing process is performing well.

Hope this info is sufficient to answer my question.

THanks.

tohbin,

Thanks for reposting with more information. Unfortunately I am still confuse because of the terms you are using, test and inspection. Is this an inspection after a test to determine if the test is operating correctly or is this simply the number of units to test to assure the process average is 200 - 500 ppm?

If you want to determine how many units to sample inspect to assure the ppm is 200 to 500, you need to know what risk level you are willing to accept that you might make a bad decision based on the sample results. In other words, if you have a million parts and there are 200 defectives in that universe (200 ppm) how many do you have to look at to be sure there are no more than 200 ppm. You would have to look at many more to be a 99% sure than you would have to look at to be 50% sure. Another complication is the sample size to discriminate between 200 ppm and 100,000 ppm will be much smaller than the sample size to discriminate between 200 ppm and 1,000 ppm.

As Mike and Dave indicated, this is not a simple question, it involves determining both supplier and customer risk factors. My suggestion is you need to learn more about the basics of sampling. There are a number of good texts available. Ed Schilling has an excellent advanced text that is available through ASQ. Dale besterfield's book, Quality Control, has good sound basic information on setting up a sampling scheme.