fatigue properties of a welded joint 3

[aber1982]

I am looking into the fatigue behaviour of welded joints of MIG welded C-Mn/HSLA structural steel. It is common knowledge that the S-n curves of welded joints exhibit a distinct slope ~1/3 (and i have results of load controlled tests supporting this, nothing new). I am on the understanding that due to the weld fatigue phenomenon, the crack produced in a weld fatigue failure is dominated by crack propagation, not initiation.
Now, I have seen some fracture mechanics based work considering crack growth rates with curves plotted delta K vs. da/dN. Looking at the Paris law regime for a long crack, the slope m, is assumed to be 3 for steels. Someone told me previous that a fatigue s-n curve and crack rate growth curve are related. Am I right in assuming that the slope of a s-n fatigue curve is the inverse of the paris crack growth law slope? does anybody know of evidence/papers/books to support it? Is it common knowledge or just theory, or incorrect??

any help much appreciated
thanks

 

[metengr]

aber1982;
I would suggest you visit the web site below and specifically review the seminar notes under each topic. This is one of the best web sites regarding fatigue and fatigue crack propagation of metals.

http://www.fatiguecalculator.com/index.htm

 

[davefitz]

aber,
Yes, your observation is correct. In the 1970's, some major catastrophic failures of welded joints on british ocean oil rigs led to many deaths, and a quick updating of british standards BS that apply to such joints occurred in the 1980's. All such welds have pre-existing micro-cracks, formed when an arc is struck and discontinued.

The european norms produced in response to the PED pressure equipment directives include fatigue analyses that discriminate between welded vs unwelded failure points, and fracture mechanics dominates the fatigue analysis of welded joints due to the preexisting crack. Weld geometry, inspection methods, and final finishing detail affects the selection of the failure curve in each case. The EU PED incorporates the british standards BS refernced above , plus other basic differrences that better allow direct use of the results of modern finite element programs( as compared to the older ASME style analysis method).

 

[aber1982]

metengr,
thank you for your reply, I have made alot of use of this site previously and find it very useful. However I cant find the answer to my question.

davefitz,
just to confirm, you are saying 'Yes, your observation is correct' to my question of the slope of an s-n fatigue curve is = to the inverse of the paris crack growth law slope?

I couldnt work out if you were saying yes the slope statement or my understanding of the weld fatigue crack propagation statement.

manys thanks both
aber

 

[metengr]

Someone told me previous that a fatigue s-n curve and crack rate growth curve are related. Am I right in assuming that the slope of a s-n fatigue curve is the inverse of the paris crack growth law slope? does anybody know of evidence/papers/books to support it? Is it common knowledge or just theory, or incorrect??

For a stress amplitude versus cycle to failure (S-N) curve, the slope (b) is in the form

delta S/2 = fatigue strength coefficient * (cycles to failure)^b

b is normally called the fatigue strength exponent
and for steels ranges from -0.12 to -0.05

for fatigue crack growth you provided the equation for the slope, where m is an empirical constant and hovers around 3 for steel.

By the way for a plastic strain life curve, a similar equation exists except the exponent is called the fatigue ductility exponent and ranges from -0.5 to -0.7.

End result, no correlation. You can look up fatigue strength exponent values and compare yourself.

 

[aber1982]

thanks for that metengr.
The only reason I asked is that the fatigue of welded specimens in only propagation, and region 2 of fatigue crack growth da/dn curve is propagation??

 

[metengr]

Yes.