SRC Forum  Message Replies
Forum: Reliability & Maintainability Questions and AnswersTopic: Reliability & Maintainability Questions and Answers
Topic Posted by: Reliability & Maintainability Forum
(src_forum@alionscience.com
)
Organization: System Reliability Center
Date Posted: Mon Aug 31 12:47:36 US/Eastern 1998
Original Message:
Posted by: John Mathis
(John.Mathis1@West.Boeing.com
)
Organization:Boeing
Date posted: Wed Jul 28 12:18:14 US/Eastern 1999
Subject: Weibull vs Exponential
Message: I have some questions concerning Wiebull vs Exponential failure rate grading. According to MILPRF55365, paragraph 1.1, table III and table IV a Wiebull failure rate level B (0.1% failures / 1000hours at a 90% confidence level) graded part is substitutable for a Exponential 'S' (0.001% failures / 1000hours at a 60% confidence level) part. What is the technical rational behind this? If the reliability of a wiebull 0.1% is equal to or better than an exponential 0.001% does the system level prediction or FMECA reflect the difference? Is the difference due to the confidence level of 90% vs 60%? If a system had all "S" (0.001%) parts and you substituted all of them with "B" (0.1%) parts wouldn't the system reliability be considerably worse? Can anyone explain this!
Reply:
Subject: Weibull Part Grading
Reply Posted by: Bruce Dudley
(bdudley@alionscience.com
)
Organization: Reliability Analysis Center
Date Posted: Thu Aug 12 10:41:21 US/Eastern 1999
Message: I don't have all the details on why the exponential test and the Weibull have different percent per hour failure statements for the same grade of part, but the basic cause is the fact that many passive components have a distinct decreasing failure rate with respect to time. The Weibull distribution describes this event far better than the constant failure rate of the exponential. The second consideration is the Weibull tests are accelerated tests where the old exponential ones were conditioned at rated voltage which does not induce the infant mortality failure that an accelerated test condition can produce. A very interesting paper is included in the 5TH CAPACITOR AND RESISTOR TECHNOLOGY SYMPOSIUM (CARTS) in 1985. The paper is "Weibull Qualification Procedures in MILC39003F" by John Raye and Brian McNicholl. This paper describes the qualification procedures and the statistics for the Weibull testing parameters for tantalum capacitors which are well known for a decreasing failure rate with respect to time. Information/example is shown on the conversion of the two distributions.
