Structural Member Undergoing Pure Bending: Detached Mesh Regions

[roldy]

I'm working on a structural member that is undergoing pure bending. The load is at the end of the member. I am modeled this as a 3 dimensional solid. I meshed using the bottom up approach and associated the meshed regions to their appropriate geometries. The problem I am having is that when I look at the deformed member, the center region does not deform as would be expected. Instead it remains static and there is an apparent gap at the bottom of the region and it looks like the top of the region intersected the deformed geometry. It looks like the center region is not connected to its surrounding geometry. I am not sure how to go about fixing this. Attached is a picture showing the member being tested and the deformed mesh.

 

[SteamKing]

Sorry, no picture was attached to your post.

 

[roldy]

Sorry about that.

 

[Mech_Engineer]

Are they separate solids, and if so have you defined contact conditions for them? Looks like only one portion of the mesh is seeing the load.

 

[roldy]

This is supposed to be one solid. There are reliefs on each side of the beam. The distributed load is located on the extrusion on the top of the beam.

 

[Mech_Engineer]

What software package is this?

 

[roldy]

Abaqus CAE

 

[Mech_Engineer]

Well all I can say is this looks like a mesh problem. You're going to need to fix the mesh such that its properly attached, whether that's going element by element or re-meshing the geometry. It is up to you.

 

[roldy]

I've remeshed about 3 times now and I get the same problem. I'm wondering if it's because of the relief? I'll probably post the file up here tomorrow.

 

[Mech_Engineer]

What do you mean when you say relief? Are you referring to the cutout in the sides if the beam which forms the thin web center portion, or is there a small gap which separates the beam's flange and web?

 

[roldy]

I meant the cutout in the sides.

 

[Mech_Engineer]

There's definitely something wrong with the finite element mesh, check to make sure it's continuous and well spaced in the region where the problem is happening.

 

[roldy]

I could not figure out the problem so I've included a link to download a zip file containing the necessary files. I have 3 variations of the model but I am only focusing on the model named "default".

Strut

 

[Mech_Engineer]

I'm sorry but I don't have the software so I can't help in that respect. I really think there's a fundamental error in your model where you've split the geometry into multiple solids or something and haven't defined contact regions.

Good luck.

 

[roldy]

I split some faces, but not solids. If it's a problem with splitting the faces, then the problem would of showed up everywhere else I created a split.

 

[Mech_Engineer]

Did you import the solid through a geometry interface (like *.step or *.igs) , or did you construct it from scratch in Abaqus? Maybe there's an error in the geometry import?

The fact that the inner portion of the mesh isn't deforming means that the edges of the web have their own nodes, but there are matching edges from the portion of the mesh that deforms as well. Why did these edges get separate nodes instead of just one line of nodes?

 

[roldy]

I created the geometry within Abaqus. I seeded the edge located in the cutout. I then seeded the edge located directly above it (in the z direction). I think that may be the problem. I will redo the mesh tonight with this in mind.

 

[Mech_Engineer]

Can you show a zoomed-in detail picture of the area you're talking about? From what youre describing it sounds like there's a gap there and you're meshing two separate edges, in which case you do not have monolithic mesh in the area...

 

[roldy]

I used the query tool to find any mesh gaps and it found some. I've attached two pics. One shows the nodes on the edge of the cutout. The other shows the gaps. I'm not sure how to go about closing the gaps.

 

[Mech_Engineer]

Can you merge the geometry into a monolithic piece before meshing? It might make the mesh not quite as pretty, but would guarantee no gaps.

 

[roldy]

Can you merge the geometry into a monolithic piece before meshing? It might make the mesh not quite as pretty, but would guarantee no gaps.

I don't quite understand what you mean. The geometry was modeled in Abaqus. It shouldn't have any gaps, right? If you mean merging the partitions into one, then no I would like to avoid doing this.

 

[Mech_Engineer]

If you don't want to merge the solid, then you have to define bonded contact conditions between the separate pieces right? How is Abaqus supposed to know you want this behavior otherwise?

 

[roldy]

What I don't understand is that I should be getting gaps at all the partition edges. The attached image shows how I partitioned the solid.

 

[Mech_Engineer]

Is there a setting for the meshing process which connects/integrates the adjacent sub-meshes?

 

[roldy]

I'm not sure. That's what I've been trying to figure out.