SRC Forum - Message Replies
Forum: Reliability & Maintainability Questions and Answers
Topic: Reliability & Maintainability Questions and Answers
Topic Posted by: Reliability & Maintainability Forum
Organization: System Reliability Center
Date Posted: Mon Aug 31 12:47:36 US/Eastern 1998
Posted by: Bob Moore
Date posted: Wed Nov 15 11:48:02 US/Eastern 2000
Subject: Sample confidence
I am manufacturing 3000 components a day, that each have a life requirement of 15 years.
I have a (destructive) screening test that enables me to determine whether samples I test would fail within 15 years. Unfortunately, the test cannot provide failure data indicating failures beyond this point. Therefore the true failure distribution is unknown.
How do I determine the number of samples I need to test in order to provide myself with confidence that say, 95% of the day's output will last at least 15 years?
Subject: Sample Confidence
Reply Posted by: Ed Sherwin
Date Posted: Wed Nov 15 16:00:56 US/Eastern 2000
Dear Mr. Moore:
Your screen test sounds like you are sampling by attributes. The result of the test is "it will last 15 years or it will not last 15 years". Therefore, the reliability analysis is based on ther binomial distribution.
In order to determine the sample size required, I must known the acceptable quality level, AQL. In other words what the maximum percent of the product your customer is will to accept that will not last 15 years. Since it was not provided, I assumed that it was 5%.
Assuming the proportion defective =0.05, the number samples that you must take to be 95% confident that the day's production is no more that 5% defective, will fail before 15 years, is a function of the number of fails allowed in your test plan. The following table shows the sample size base on the failures allowed:
Failures Allowed Sample Size
Contact me if you need further explanation.