I have modeled a 3D steel bar subjected to a point impulse load. I used FI solid elements for the run. My mesh is also good with mesh size of 0.01in and max aspect ratio of 1.20. But I am getting erratic high stress and strains in the result. What can be the cause of this?
Hi @manoz21,
It could be a multitude of issues. What material model are you using for the steel and are you just using literature values for the parameters? Is the point load just a load on a node of the bar (what keyword are you using)? If you have a mesh of an impactor to represent the point load, then what type of contact are you using?
@negativevolume hey. I am using the material values for A304 steels. The load is defined as a sinusoidal impulse load with an amplitude of 250 lbs and time period of 0.1 ms and 0.2 ms. the load is applied to a single edge node in load_node. the 0.2 ms load shows no such erratic stresses but the 0.1 ms one does.
@manoz21 Okay so it has nothing to do with contacts then. What material model are you using? (ex. *Mat_piecewise_plasticity) And what type of mesh is the bar? Hexahedral or tetrahedral? It could be related to the element formulation. You can also check the energy of your simulation in your ascii or binout files under glstat for any erratic behavior.
@negativevolume I am using elastic material with the hexahedral (brick) element. I am using fully integrated elements. I think it has something to do with shear locking but I don't have the means to verify it.
@manoz21 Fully integrated hexa elements are prone to shear locking. Try ELFORM = 2 and then use a hourglass control of 5
@negativevolume that's the problem, I am studying wave propagation in solid media, fully integrated elements are supposed to give practical results. elform =2 does not need hourglass integration, 1 needs it. I am using 2 for the current simulation in which I am getting high stresses.
@manoz21 Ah sorry I got mixed up and was thinking about shell element ELFORM=2. I meant ELFORM=1 for the solid elements. Have you tried that with that hourglass control?
@negativevolume elform 1 uses just 1 integration point. I am studying wave propagation and fully integrated elements are bound to give practical results. I need to use fully integrated elements. I need ways to verify if my model is undergoing shear/volumetric locking. If you got any idea about that that would be quite helpful. Thanks
@manoz21 Got it. Have you tried ELFORM=-1 or -2? These are both fully integrated but may reduce your shear locking issue. This link will download a .pdf by LSTC highlighting these element formulations starting on page 8.