Fatigue - varying principal stress directions 2

[hpon]

Hi,

I'm looking for ways to calculate the fatigue life of a cast iron component which is subjected to several non-synchronized loads. The thing I find tricky is the fact that the directions of the principal axis varies significantly depending on the combination of active loads (I'm not even sure along which direction the cracks will form). How do I estimate the fatigue life in this and similar cases? Could you name any relevant methods or other references?

Thanks

/hpon

Regards,
hpon

 

[corus]

I usually take the difference between stress components and calculate the principal stresses from that.

Tata

 

[hpon]

Hi corus,

I don't understand what you mean. Could you expand your explanation a bit?

The problem is not calculating the principal stresses. Instead, the problem is to estimate the cumulative fatigue damage caused by stress states which differ both in amplitude and orientation.

hpon

Regards,
hpon

 

[corus]

If you have Sxx, Syy etc. from case 1 and Sxx, Syy etc. from case 2 then take the difference (Sxx1-Sxx2),(Syy1-Syy2) etc. and calculate the principal stresses from those differences. Use that value for your fatigue assessment.

Tata

 

[johnhors]

Corus, have you got a reference for this method?

http://www.Roshaz.com

 

[corus]

Nope, it was the best way I could think of overcoming the problem which also works for the simplest case. In Abaqus you can do this easily by creating a new step as the difference between two others, and then calculate the principal stress difference from that new step. I'm not sure about other codes.

Tata

 

[crisb]

Found this from a google of the thread title. Don't know if it makes sense to any of you guys?

(excerpt from mechanical design handbook)

http://books.google.co.uk/books?id=...- varying principal stress directions&f=false

 

[hpon]

Corus,

How do you know that your method is accurate/conservative?

Do any of you have experience using Dang-Van's approach? Can you refer to any good texts about it?

/hpon

Regards,
hpon

 

[corus]

The text in crisbs posted reference seems to suggest the same method. As far as accuracy goes, then fatigue assessment is never accurate. Most design codes give a probabality of survival rather than predict accurately how many cycles something will last.

Tata

 

[TERIO]

The method corus describes sounds similar to what is used in the ASME B&PV Code Section III. I can't say that I have dug very deeply into the background of it.

In chapter 8 of Multiaxial Fatigue by Socie there is a review of some methods for nonproportional loading.

 

[hpon]

Corus,

I am trying to understand you method. Do you proceed by calculating some equivalent stress measure based on DeltaSxx, DeltaSyy etc.?

What if you have three or more load cases?

(I cannot access crisb's link.)

/hpon (I hope that the "double-signature" is gone now)

 

[corus]

Yes, calculate the change in each of the stress components and from that calculate what I would call the principal stress range. For additional load cases you'd have to calculate the differences between each load case to use in a rainflow method of assessment.

Tata

 

[hpon]

I see. And the result should be conservative since all the stress changes are assumed to have full impact on the generation of the potential future crack.. A less conservative but possibly more accurate method would (according to my loosely founded understanding) scale the impact of stress changes differently depending on the orientation of the principal axis.

How do you represent the mean stress?

/hpon

 

[corus]

Refer to crisb's link which gives a description, and aanwers your question on mean stress, I believe.

http://books.google.co.uk/books?id=...- varying principal stress directions&f=false

Tata

 

[rb1957]

i'd look into fatigue analysis software, like "n-code"