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: P Biswas
Organization:Aeronautical Development Agency Bangalore
Date posted: Thu Dec 7 23:52:38 US/Eastern 2000
Subject: System MTBF predction
I am doing prediction of system MTBF that has got large number of components. The predicted value of MTBF of the system depends on (i) number of components in the system (as individual failure rate of each component is accounted in system MTBF) and (ii) their individual failure rate / MTBF value. The system MTBF has been estimated much lower than individual component MTBF since there are large number of components in the system.
[As System Failure Rate = Sum of individual Failure Rate
And MTBF = 1 / (System Failure Rate)
Where as the MTBF estimate based on the data from reliability tests show higher value than what we get from the prediction based on analysis. The mathematical modeling as per the RBD (complex series-parallel configuration) should approximate to the field estimate. The reason for the difference is not clear.
· Whether there is some mistake in the application of this
· Why the system MTBF should be less than the minimum MTBF
of individual components, even after ensuring individual
component MTBF ?
· What alternative approach should be followed to arrive at
correct prediction ?
I need some expert advice who is involved in such prediction work Or can provide answer / solution.
Subject: System MTBF Prediction
Reply Posted by: Jean-Marie CLOAREC
Organization: LIGERON SA
Date Posted: Tue Jan 2 4:46:05 US/Eastern 2001
MTBF prediction can be based on :
(i) reliability data base such as MIL HDBK 217, NPRD, ...
(ii) field experience issue from test or operational observations.
MTBF = 1/sum of all failure rate.
First, if you use data base (i) you will find a difference with real data (ii) because your experience is not the data base experience. With data base you can find better or worse data.
Second, if you used RBD the above formula can be used, you will have different formula and the MTBF with RBD will be better than with the sum of failure rate, because you will take in account redundancy and not take in account non-essential component, i.e. you will calculate a MTBCF (mean time between critical failure).
If you have real data (ii) it will be better because the reflect your own experience.
If you used database, caution because it's not your experience but the "mean" experience among several company.
So it's normal not to find the same figure using one or other hypothesis.
If I am not clear, you can precise your problem.
Happy new year