Learning Forum

[kpetrunik]

 Hello,

I have a question about thickness and mass of bodies in 2D explicit dynamic simulations. From what I have read so far 2D explicit dynamic simulation can only be solved with the plain strain option which then assumes the thickness is infinite. 

The simulation I am running is a cylindrical object impacting a flat plate.

I ran a 3D simulation with a body thickness of 10mm and then a 2D simulation of the same body for comparison. The force reaction of the 3D simulation is approximately 10 times greater than the force reaction of the 2D simulation. Each body deforms the same amount over the same time period.

My question is, is there a way to set a mass of a body or thickness for a 2D explicit dynamic simulation?

I am interested in the impact force of the object. I can run my simulations in 3D but obviously running them in 2D is much faster and the cross section of the body is constant throughout its depth, however I need more information on how the force reaction is calculated assuming an infinite depth. Any help is appreciated!

[peteroznewman]

Just a quick comment on Explicit Dynamics 2D vs 3D compute time.  The 2D solver is limited to running on one core, while the 3D solver had code written to solve on multiple cores.  So if your mesh limits the element count in the third direction to be just 2 elements thick, then you are doubling the elements and tripling the node count, but if you have a high core count processor, 6 to 12 cores or more, then the 3D model might actually solve in less time than the 2D model.

[kpetrunik]

Hmm, Ok. So maybe it would be better to have my model only 2 elements thick and run 3D and increase my core count. 

 

I just checked the mesh statistics and I have ~29,000 nodes for the 3D mesh and ~4,000 nodes for the 2D mesh. Currently the 3D model is 5 elements thick, if I decrease to 2 elements thick then the number of nodes drops to ~19,000 but the element quality decreases significantly, is that a concern? My understanding is that elements closer to a cube are preferred, but if I constrain the model from moving normal to the Z plane and only have a velocity in the -Y direction will the elongated elements causing the decrease in quality matter?

[peteroznewman]

The only way to know for sure is to solve both meshes and compare the results. You can even try 1 element in Z, since you are constraining the Z motion anyway, the free nodes in the center plane of the 2 element thick might misbehave while at 1 element in Z, there is no node to go wrong, but of course the element shape has an even higher aspect ratio.  Mechanical meshing has internal limits on what is allowed, so see if it will mesh.