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: Julien Nico
Date posted: Wed Oct 20 9:53:56 US/Eastern 1999
Subject: Population sample number to perform endurance test
In order to improve reliability figure got by FMEA analysis on one of our new product, we will perform endurance tests.
We have to define number of product to test in order to have an MTBF with a known uncertainty. More over, due to the component high cost, it is not possible to have a high number of tested part. Statistics define applicable rules but we have to use standard deviation of the MTBF. As it is a new component, it is not possible to use existing data. We are looking for a data base of standard deviation for component used in aeronautical or another means to minimize number of test items.
Many thanks for your help.
Subject: Sample Testing
Reply Posted by: Bruce Dudley
Organization: Reliability Analysis Center
Date Posted: Thu Oct 21 9:10:16 US/Eastern 1999
I do not know of a specific compendium of "new component" standard deviation data bases. What does exist is individual electronic, electro-mechanical and mechanical part failure data bases. All of these data bases are from historical records so they may not be applicable to "new" technology components. Each of the RAC data bases (EPRD-97 and NPRD-95) shows total number of failures, operating hours or cycles and conditions. Distributions and standard deviations are not cited. The best way to determine the distribution of failure for a new component is to perform sample testing and collect the data. Analysis of the data should allow you to determine the mean and a confidence interval. Suggest that you look at books by Wayne Nelson "Accelerated Testing" and Demitri Kececioglu "Reliability and Life Testing Handbook to assist in analyzing the test data.