Service Life Adjustment

[Andrew Robertson]

Folks,

I am calculating an MTBF for a pump based on actual service data for a similar to pump which has been in production for about 7 years. This is an aircraft application so the actual operational use of the oldest pump is only about 4000 hours. The expected life of the pump is around 10,000 hours. Clearly there may be failure modes not yet experienced which would ultimately effect the MTBF.

Is there an excepted statistical method for extrapolating partial life failure data to account for future failure modes? Some sort of service life adjustment?

Is this Beta? I am not a reliability engineer just someone trying to sanity check a reliability report I am reviewing

Thanks

- Andy Robertson

[Tarik]

HI,

In Weibull++ you can enter your failure data for the 4000hrs of operation that you have for all the units in operation. You can then make an estimate of your reliability at 10,000 and can use Confidence interval to estimate your uncertainty based on the variations you have seen on the units that have been in operation.

Note that this would not predict any failures modes that you haven't seen yet ! I guess you can use Weibull++ reliability estimate at t=10,000 as your 'best' or 'optimistic' estimate.

I am not aware of any 'crystal ball' statistical method that will predict unseen failure modes.

If anyone in this forum knows about other methods, please let us know.

However, I can suggest that you perform an accelerated test with the help of ALTA to discover any unseen failure modes.