Fatigue Life Estimation

[hammad0252]

I wanted to know if it was possible to estimate fatigue life of a pressure vessel at low stress if you know its failure conditions at a higher pressure. As it is possible for me to test the vessel at higher pressure for lower number of cycles but testing for larger number of cycles will be too much time consuming.

For example if a vessel fails after 20 cycles at 700 MPa, and fails after 35 cycles at 650 MPa, is it possible to estimate the number of cycles it will survive at 400 MPa.

 

[IRstuff]

Hypothetically, yes, so long as you have some function that describes how the higher stress accelerates the life test.

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

 

[drawoh]

hammad0252,

If you do the analysis for high pressure and low cycles, and you build and test it and it works, then your theory is good. You should be able to change the numbers and be confident in the result. Hopefully, you are a pressure vessels person, or you are being closely supervised by one.

--
JHG

 

[metengr]

Yes, keep in mind you are in a regime of low cycle fatigue behavior. There are a large amount of published articles by ASME on this topic.

 

[rb1957]

If you have a good s/n curve for the material you have a better case. Something to plot your test point on so you can justify your extrapolation.

It could be that at a higher pressure the effect of the stress concentrations is conservative (or possibly unconservative).

I suspect that you want to fatigue test at a much higher pressure (10x operating) to get a short fatigue life. It'd be better with a lower factor, maybe 2x (should reduce your life by a factor of 16 to 32).

Something to remember ... don'y "you guys" proof pressure test repeatedly ? This regular over-pressure should have a significant (beneficial) effect on fatigue life. But maybe neglecting this (conservatively) still gives you an adequate life ?

Something practical to think about ... are there seals in this pressure vessel (most likely) ... how will they behave (misbehave?) with the higher pressure ?

another day in paradise, or is paradise one day closer ?

 

[EdStainless]

as with all accelerated testing the greater the difference the less reliable the correlations.
With fatigue as you move from low cycle to high cycle the entire morphology of the event changes.
Reach out to someone who specializes in fatigue.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[gerhardl]

Am I understanding your example figures correctly here? 400MPa= 4000bar?

Anyway. Statistically, you have two points on a curve (with 20 and 35 cycles). Even if this is verified with a number of tests, with vessels of exactly same size, construction, materials, fluid and failure characteristic, there will probably be some variation in result.

Two points of a curve give no indication of how the curve continues, it could as easily be exponential as straight. With more points on the curve you could possibly simplify to an accuracy 'probably more than lowest figure tested, extended somewhat further'.

What is the exact requirements and accuracy needed?

 

[rob768]

theoretically it is possible. I would suggest that a extrapolating a curve based on two points is not very reliable, and considering the amount of spread in fatigue results, i would call it a long shot. Not even an educated guess...