# Fatigue life calculation of a component in multi-axial stress state

I am dealing with a high cycle fatigue life calculation of a component which is under a constant amplitude proportional loading, however this component is in multi-axial stress state. In fatigue life calculation we deal with uniaxial stress. Now which uniaxial stress measure (equivalent, maximum principal, minimum principal or absolute maximum of maximum and minimum principal stress) should be used in fatigue life calculations?

### If you want a simple

If you want a simple uniaxial measure, try equivalent stress/strain measures. For multiaxial stress states a better approach is to use a critical plane model, such as Fatemi-Socie or Brown-Miller. An added benefit is that these methods also predict the dominant fracture plane.

Dear Dhanashri,

If you make do with the estimation of the fatigue limit under multiaxial loading, you could check the FatLim database on http://www.pragtic.com/experiments.php. You can set the expected loading type there and check what quality of results the individual criteria provide. If some of them is acceptable for your purposes, you can download PragTic software for free and let it be computed on FEA-results...

There is one problem anyway. The experiments in FatLim database are collected from experiments run on smooth unnotched specimens. Because you seem to be interested in fatigue prediction on a real component, I assume there is some stress concentrator. The direct use of the criteria described in FatLim on FEA local stresses is likely produce over-conservative results, because of the localized highly-stressed volume. Because of that, the strain-based methods (Julian mentioned two of them, which you can run with PragTic as well) could be a better solution, under the condition, that the target life is in the low-cycle regime. Their use in the high-cycle fatigue can lead to a very large deviation from reality.

So, in order to get some acceptable data, you can try to use the stress-based methods implemented in PragTic on the local stresses, but the results will be too conservative. There are two ways to get closer to reality - application of stress gradient solution or the theory of the critical distance (none of which is smoothly implemented in PragTic, but the latter one can be evaluated with some caution). Well, if you are interested in this way of solution, write me (papuga@pragtic.com) and I'll try to help you.

Good luck and regards,

Jan

### Thanks a lot to both of

Thanks a lot to both of you.

Jan, I am going through PragTic, will write you if I come across any doudt. Thanks once again.