The fatigue modifying size factor and the real world structures

[bxtsafe]

Hi,

Frequently I have to do fatigue life assessment of parts using the results of an FEA. Well, the matter is that I've been studying fatigue analysis for a while, specially the S-N approach, and there is an issue that is botthering me: It's about the modifying size factor "Csize".

I've took 7 books, among Machine Elements books and speciallized metal fatigue books, and all they report the same thing: in the 50's and 60's some researches concluded that the fatigue limit of CILINDRICAL specimens decreases as its diameter increases. The result of this can be seen in the book "Fatigue testing and analysis: theory and practice" of Yung-Li Lee, pg 137 (this book can be found in the google books). There is considerable scatter in the results, but it's possible to fit a (conservative) curve that help the engineer during the fatigue design.

Theses books also show about an research carried out in 1960 by Kuguel which showed that it's possible to find the Csize for PRISMATIC beams with a non-uniform section (rectangular, I-beam and C-beam section). For calculate the Csize of this section shapes, we have to determine an equivalent diameter based on the cross section area loaded by more than 95% of the maximum stress.

However, most part of the structures we analyse by FE don't have constant cross section. They are complex and sometimes have "organic" shape. And some structures don't even have cross section, as car body floor or roof.

So, my query is How do you usually calculate (or recommend to calculate) the Csize for:

1) the solid structures which don't have constant cross section that we analyse daily by FEA? and

2) the shell structures that cannot be considered as a beam, as carbodies?

P.S.: there are attached some pictures of Finite Element Analyses that can explain my query.


Im sorry for the long question, but I think it's better to explain in detail in order to make the query clear.

Any help will be appreciated.


Best Regards,

bxtguard.

 

[corus]

I've only ever seen this for rotating shafts where there are factors for surface finish and size, but ulitmately these factors are only for satisfying infinite life criteria. In general I tend to follow what ever design standard recommendations there are for the component, ie. in cranes, or pressure vessels. Generally most codes are concerned about welds which dominate fatigue life considerations. For plain (unwelded) steel you'd need specfic SN curves of similar sized material.

Tara

http://tinyurl.com/4ydjg7m

 

[bxtsafe]

Thanks for the answer, corus!

I continued the searching for this answer and I found it in this book:

"Metal Fatigue Analysis Handbook: Practical problem-solving techniques for computer-aided engineering"
Autors: Yung-Li Lee, Mark E. Barkey, Hong-Tae Kang;

Hardcover: 632 pages
Publisher: Butterworth-Heinemann; 1 edition (August 31, 2011)
Language: English
ISBN-10: 0123852048
ISBN-13: 978-0123852045

Among all books, that is what is most practical in the application of fatigue, FEA and real world structures. The book explain what is the Csize to consider when analysing shell structures of thickness "t", like carbodies. I think the book really worth.

regards,

bxtguard.

 

[rb1957]

size effect relates to the fact that a larger volume of material will be more likely to have larger flaws imbedded in the grain structure.

 

[corus]

I believe that in the british pressure veseel design standard there was a factor used if the thickness was greater than 16mm, along the lines of (t/16)^0.25 if I remember correctly. I think this factor was dropped in later 'euro' versions of the standard though. Here's a paper that also considers the effect too

http://publications.lib.chalmers.se/records/fulltext/132254.pdf

Tara

http://tinyurl.com/4ydjg7m

 

[gurmeet2003]

I think that the basis for size factor is rather shaky. If one was dealing with a part with no stress concentrations, the size factor may make some sense, as RB957 pointed out. But all real parts have stress concentrations. The fatigue failure really occurs at these locations. Therefore important consideration would be how much material volume is subject to the near peak stresses. Also how fast do the stresses drop off around the peak.



Gurmeet
Time is an illusion. Now is the only thing there is.