General Submodeling Question

[nds88]

I am working on a submodel (in ANSYS 2021) which comes from a global model that has multiple load steps, varying temperatures, dynamic and static loads, and external data as input. After I got results, my submodel didnt seem to line up correctly when evaluating some reaction forces at key areas, about 50% difference. This suggested to me that I imported loads incorrectly (assuming I was far enough away from my boundary) but I also had a thought about theory that someone here might know.

As I understand it, we are always solving for nodal displacements, to which all of our other relations stem from. So, I guess I am wondering, why do I need to import anything other than my cut boundary displacements? Those cut boundary displacements are a result of all of the loading conditions (no matter what they are) at the end of the load step. So by that logic (and maybe im missing something here) I should not need to import temperature, acceleration, etc. in addition to my cut-boundary displacement.

I got in touch with an ANSYS rep and he didn't really end up having an answer to that.

 

[SWComposites]

You likely need to import temperatures, accelerations and loads for the interior of your submodel in order to get the internal effects.

 

[FEA way]

I’m not sure how this works in Ansys but the general rule is that all nodal DOFs (temperatures too if the analysis is thermo-mechanical) on the submodel’s boundary are driven by values from the global model. And apart from this special boundary condition you only have to assign boundary conditions, loads, initial conditions, and predefined fields to submodel if they are applied to this region in the global model.

 

[nds88]

I gave it some thought and I remembered back to the part of equilibrium equations that accounts for body forces, which of course are volumetric. So I'm kind of further in disagreement that acceleration, or EM loading, or etc, should be included because in this "cut-out" submodel because we no longer maintain the same volume, so, the resultant force (like a reaction force that I measured) must not be the same. In my model, I took the global model's acceleration components and applied it directly to the bodies in my submodel. I wonder if maybe I should scale the acceleration to my new volume? That might work for acceleration, but I dont know that my other loads could be manipulated that way.