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: Leroy Meyers
Date posted: Mon Mar 5 15:51:46 US/Eastern 2007
Subject: Field Data & MTBF
I have field data over a period of many years, and I want to calculate the MTBF with a confidence level. Would it be correct to use the Chi Square distribution? If so, would the Degrees of Freedom be 2n+2 (where n=number of failures)as if it were a time-truncated test?

I'm looking at page 38 of Practical Statistical Tools for the Reliability Engineer.


Subject: Field Data & MTBF
Reply Posted by: Rushabh
Date Posted: Mon Mar 5 17:25:07 US/Eastern 2007
Hi Leroy, I have not read the book you have referred but based on my exp. I would suggest few key points. 1) If the failure data available with you have considerable products with end of life time then it would not be good idea to treat them as time terminated. 2) For example if you have total 1000 product's field data. Out of which 700 products failure time is available and rest of the 300 products are working in field then we can treat these 300 products as suspended items and get the weibull parameter estimates. 3) Best would be to use Automated tools like WinSmith weibull or Reliasoft weibull to calculate weibull parameter estimate/MTBF estimate. It gives flexibility to treat the data as Time terminated or Failure terminated or suspended. 4) It would not be good choice to use Chi-Square distribution blindly because I believe it holds good for exponential distribution only. So if the failure distribution is not exponential then we my get misguided. Regards, Rushabh

Reply to this message