Use of reliability equations 1

[multilinnoob]

Hello Everyone,

I work for an oil and gas operating company under reliability engineering department. We have base probabilities of failures defined for fixed equipment such as piping, vessels and other mechanical machines. The focus is mostly using Reliability Centered Maintenance (RCM)

I am an electrical engineer, and I have to make plans for electrical equipment such as VFDs, motors etc for which there is no company provided probability of failure.

I came across this reliability equation: Probability of Failure = 1 - exp(-lambda * t), where lambda is failure rate per year of a particular component.

My senior reliability engineers (non-electrical) are saying I can only use this equation if:
1. The failure mode is random and
2. The failure rate does not change as years go by.

Long story short, I would like to get opinions of electrical/reliability engineers on this forum:

For a component such as capacitor, IEEE 493 Table 10-4 states that the failure rate is 0.17443 failures per year.
Can I use this lambda to predict my failure probability, say 10 years from now?

Sorry I am having hard time putting my problem into words, please do ask for clarifications where needed.

Thanks!

 

[IRstuff]

Yes, and no. Various capacitor materials and classes result in wildly varying failure rates. Assuming you find rates directly applicable to your specific capacitor family, they were calculated based on the number of failures for some aggregate number of device hours; this is where the constant failure rate is assumed. And, that assumes that the constant failure rate region extends past 10 years, which it might not.

The standard reliability model for electrical components makes use of the "bathtub" curve, a high failure rate infant mortality region, followed by a constant, relatively low failure rate region, followed by an end-of-life "wearout" high failure rate region. The reliability equation cited is only valid in the middle region, and does not apply to either ends of the bathtub curve.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[multilinnoob]

Thanks!

So is there a particular equation that I should be using for a bathtub mode? Or will it be piece wise function?

 

[IRstuff]

Vaguely piecewise, since it's a rather smoothly varying curve. At that point, those two other parts of the curve are also widely variable, depending on the specific component, screening, etc.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[multilinnoob]

Thank you sir!

This clears my questions.