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

Back to message list Show all replies Topics List About this forum
Original Message:

Posted by: June
Date posted: Thu Jan 30 22:06:17 US/Eastern 2003
Subject: Early Failure Rate
For early failure rate. why is number of failures (fpm) corrected using 60% confidence level (using poisson statistics)? a) Why not 90% confidence level, etc b) Why poisson (since for reliability data, it is mostly derived from Weibull, exponential distribution)


Subject: Confidence Level
Reply Posted by: Jorge L. Romeu ( )
Organization: RAC
Date Posted: Mon Feb 3 10:53:30 US/Eastern 2003
Regarding the problem of confidence, the customer is right: there is nothing special about 60% (nor 90% or 95%, for that matter). It is a matter of convenience and usual practice. The confidence level deals with the percent of times, in the long run, that the probabilistic statement is right. Can the customer live with being right six out of ten times? Or 19 out of 20 (for a 95% confidence)? We refer the reader to RAC START sheet: Statistical Confidence. Romeu, J. L. RAC START: Volume 9, Number 4. Regarding the second part of the question: why Poisson, when data is mostly derived from Weibull or Exponential, it is convenient to remember two things. First, that the Exponential is a special case of the Weibull, when the shape parameter is unity. Secondly, if life times are distributed Exponential, then the number of consecutive deaths, (events) in a period of time t, is a counting process distributed Poisson. So, there is indeed a statistical relationship between Exponential and Poisson distributions. We refer the reader to RAC START sheet: Statistical Assumptions of an Exponential Distribution. Romeu, J. L. RAC START: Volume 8, Number 2.

Reply to this message