Issue With Mesh For Three Point Bending Test

[person123]

Hi. I'm attempting to model a three point test for a beam. I'm using a 2D planar model with two vertically restrained boundary conditions on the bottom corners and a concentrated load on the middle of the top side. A crack is also modelled in the beam. As you can see in the image, the vertical position of the nodes on the top side alternates creating this "wrinkly" appearance. I imagine this could lead to inaccurate results when trying to measure the displacement. I'm using the student edition which limits the number of nodes to 1,000 so I would not be able to refine the mesh much more. Does anyone know how I might correct this issue? Thanks.

 

[FEA way]

This looks like hourglassing, try using full integration elements. Also consider changing the way load and boundary conditions are applied - it would be better to distribute them over small parts of the model’s edges.

 

[person123]

Thank you! I tried both of those approaches and it seems to work:

To go into more detail, first I unchecked the reduced integration box in assign element type, which I assume leads to full integration. This removed the hourglassing as you called it, but there was still a large drop in the central node. I then replaced the point load with a pressure load over the length of twice the mesh size, and replaced the point BC with a distributed one. This reduced the drop and it now looks pretty smooth.

There does seem to be a sudden switch between two straight lines on the top side now though. Beam theory wouldn't predict this but I imagine it might be different with a thick beam (as well as maybe because of the crack). Does the result appear reasonable?

 

[SWComposites]

That geometry is way outside the limits of std beam theory.