Does it matter too much whether to use the centroidal results in FEM model in fatigue analysis?

[Znjmech]

Hi All

a software I am using , doesn't give the centroidal results easily, does it really matter to use centroidal or elemental or nodal to be used in Fatigue analysis ?

 

[SWComposites]

Yes.

 

[Znjmech]

Shall you please elaborate it ?

 

[goeasyon]

Is it solid or engineering element (beam/plate/shell), and element order?
If your fatigue analysis algorithm is based on the stress/strain values, which is originally calculated on the Gauss points in conventional FEM. Centroidal/nodal results are essentially interpolated from Gauss point solutions. Thus, in my opinion, the Gauss point stress/strain can give you the most accurate fatigue results. For your case, if the node-based fatigue algorithm is supported for your element type, use the nodal results.

https://welsim.com

 

[nlgyro]

Quoting:

> Thus, in my opinion, the Gauss point stress/strain can give you the most accurate fatigue results.

Not necessarily. Suppose one is doing a fatigue analysis for a cut-out corner, you cannot use gauss point results as they would not be at the edge of the cut-out. You'll have to use extrapolated nodal stresses at the edge of the cut-out to obtain the "correct" Kt stresses for a fatigue evaluation.

 

[SWComposites]

SFT - you need to post a lot more details for us to give you useful advice.

 

[rb1957]

it depends. If you're dealing with large scale structures, without discontinuities, you can use centroidal stress to characterise the stress of a large structure.

but if you're analyzing a detail, you probably need nodal stresses ...
unless your fatigue algorithm uses centroidal stress.

clear as mud, eh?

another day in paradise, or is paradise one day closer ?