Reliability Estimation: 0-field failure:

[vnigam]

I am trying to estimate the reliability for a particular failure mode. We have over the past 10-15 years, seen no failure in the field. I have the suspension times of about 20 samples which were physically checked but found OK for the failure mode. I have several thousands in the field with no failures.

Would it be fair to run a exponential distribution to estimate the reliability at a given time? I was thinking of using MLE,FM analysis methods and when I do that I do get the reliability plot Vs time.

Please let me know if I am thinking right?

Thanks

Vandana

[Pantelis]

An exponential distribution is a reasonable assumption in this case.

[vnigam]

Assuming an exponential distribution, is there a method available which would help me complete my estimation with a confidence level?

Thanks

[Pantelis]

Use the chi square distribution... if I recall correctly the conf on the mean is (2*Total Accumulated Time)/Chi^2(1-CL, 2 degrees of freedom).

[vnigam]

Thanks a lot!