Line Forces and Line Moments For Battelle's Structural Stress for Fatigue

[Doodler3D]

All,

I'm trying to verify Battelle's structural stress method by Dong for shell elements. For a start, I'm doing a quick calc on a spreadsheet before diving into an APDL script. Is there a way to estimate or hand-calculate a value for structural stress and can membrane stress be negative as in the example provided? Of course, since membrane stress is the average value through the cross section, it should be positive. Also, is this calculation a futile effort as the Battelle Structural stress method is patented and not available outside software addons like FE-Safe/Verity?

Note, the tables are nodal forces/moments scoped using ENFOVECTORS/ENMOVECTORS in the local coordinate system.

Model:

Nodes/Elements:

Line Moments for Element 4970:

Line Forces for Element 4970:

Ref: Selvakumar, P., Hong, J.K.Hide, Robust mesh insensitive structural stress method for fatigue analysis of welded structures, Procedia Engineering, Volume 55, 2013

Thank you.

 

[NRP99]

Not at all familiar with Dong's structural stress method used in assessment of fatigue of welded structure. You may find some information in the book - Metal Fatigue Analysis Handbook by Lee, Barkey and Kang or search any other books or research papers.

But membrane stress can be negative if the load is compressive or producing effect that is overall compressive. For example, axial compressive load on simple bar. The average stress/membrane stress will be compressive. As I see it, the remote force(-x direction as per coordinate system in other image) try to bend the square section at the welds about -y-axis and twist about the +z-axis. So the elements ( 4969 and 4970) may be in compression, if my imagination is correct. Its my guess though. You can check that by plotting min principal stress.

 

[FEA way]

The structural stress method itself and the definition of structural stress are not secret, you can find the description in the research papers (especially those written by Dong) and documentation for fe-safe/Verity, among others. However, it would be difficult to recreate this due to the level of necessary automation. The aforementioned plug-in for fe-safe also offers some additional tricks making calculations easier. For example, it can detect weld lines.

 

[Doodler3D]

NRP99/FEA Way, Thank you.

The elements are in tension compression.

When I repeat the calculation for element 4183, I get a value of 556,619 cycles for a stress ratio R = 0, which is very close to the results Dr. Dong and his team demonstrated in the original 2003 SAE completion. However, for R = -1, I have some errors in my spreadsheet. At this stage, it's just fortuity or complete idiocy.

 

[NRP99]

It seems to me most of the weld is in orange color (0 psi to -3105.6 psi) and hence in compression. But you are better judge than me.

Did you find any information in the book I mentioned? If you have access to/have any contact/friend who can check it in fe-safe, that will be helpful to you to benchmark your results and end your frustration.

 

[Doodler3D]

@NRP99, Thank you, the book is a very good reference and does a great job of summarizing the technique. However, there appears to be no benchmark for this method, apart from a few papers. I will have to look around and see which article offers the best example.

 

[swimfar]

I don't know anything about the method you're talking about. But what do you mean when you say the elements are in compression? Is it based on your minimum principal stress plot? The minimum principal stress is usually going to be negative for most stress states. What does a plot of the pressure look like?

 

[Doodler3D]

@swimfar, the stresses may be inaccurate as the method does not require a thorough convergence check. The idea is to use the internal nodal forces and moments and back-calculate a bending and membrane stress at the weld toe.