Which stress componenet is relevant to fatigue life

[izax1]

I would appreciate you opinion on which stress component is contributing to the fatigue life. (Max. shear, Max principal, von Mises)

For me this is relevant for Vibration fatigue, that is accumulating cycles across a frequency range. I know Max. principal is often used, but in dynamics the stress direction will change all the time. Thus I am considering uing v Mises. What is your experience on this?

Thank for any input.

 

[rb1957]

i would use max principal, and think i'm being conservative (ignoring the "clocking" of the direction).

btw, von Mises also changes direction (just like max principal). also von Mises is an absolute value (ie +ve for a compression stress).

the really accurate way is to have your orthogonal stresses components for all you load cases and to analyze every direction ... yeah, right !

you could consider using dedicated fatigue analysis software (ncode, for example).

 

[corus]

It's usually the stress range you should look at and as Von Mises is always positive then using this for the range of stress would not be conservative. Better to use the principal stress range.

The manager

 

[CoryPad]

You need to use max principal stress since it is tensile and that is the stress that usually results in fatigue cracking. Shear and von Mises are used to analyze deformation since shear and compression can contribute to yielding, though rarely to fatigue.

 

[hecmo]

Hi, I'm new here but I'll try to help as much as I can. In my opinion, the analysis approach depends on the tipe of problem you are solving and the nature of the stresses you find:

- Is it a uniaxial or multiaxial state of stress? (from your question i guess it is multiaxial, bad luck ).

- Are the loads proportional? This means that during the analysis the principal axes don't rotate and the rate between principal stresses is constant. Usually, this happens when the load spectrum is reduced to a unique frequency (despite the amplitude of the cycles can of course vary). So, find out if your load cycles can be considered to act at the same frequency.

If the stress state is uniaxial and loading proportional, one option is to use the statical equivalent loads approach combined with a static failure criterion. If the stress state is multiaxial, you can obtain the equivalent static stress for each stress component to later combine them with a multiaxial static failure criterion (VM for instance).

If your loading is not proportional, then a more involved approach should be followed, in this case the critical plane approach. You mention that you're solving a vibrations problem. If your have response at many different frequencies, the later should be used.

Good luck with your problem!

 

[flash3780]

I've used a signed von Mises stress in the past when dealing with fatigue in a geometry that undergoes multiaxial stress states. Essentially, you compute the von Mises stress for each loading condition and assign a sign to the stress (positive or negative) based on the principal stress with the largest absolute amplitude.

You may then use your two stress states to compute the mean stress and the stress amplitude, which can be used on a Goodman diagram or a S-N curve.

Be sure that the material data that you have was collected at the same R-ratio (or A-ratio) that your part sees. If you have data that was collected at a different R-ratio, you'll have to use Walker's relation to adjust your results. Here's a paper I found that discusses the Walker and SWT equations:

http://www.fatigue.org/minutes/Spring-2005/DowlingBrasilPaper.pdf

Good luck.

//signed//
Christopher K. Hubley
Mechanical Engineer
Sunpower Incorporated
Athens, Ohio
--

http://engineeringliberty.wordpress.com

 

[izax1]

Thank you for the input, guys. Seems like most of you are looking at Max principal stresses for fatigue, which is problably right given that you are looking at one specific location in a static (not vibration) fatigue case. As I said, this is a case of vibration fatigue with a multiaxial stress state, with stress response levels at several different frequencies. The max stress location and direction of the max principal will of course vary from frequency to frequency, why I tend to agree with flash3780. My question was posted to find out if I was completely alone with my opinions. (It is good to find that someone else is thinking the same