Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations GregLocock on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Cosmos Struggle - Mesh Control

Status
Not open for further replies.

emalina

Mechanical
Jun 15, 2009
4
Hey guys,
I am trying to mesh a very basic cylinder in cosmos. The problem is that I am simulating the impact of a probe (similar to a Charpy impact test) and I would like to heavily refine the center of the cylinder. I have tried to mesh control a point on the center of the face, but the mesh didn't change, and I have drawn split lines on the face - but the mesh just refines in very odd ways. Does anyone know how I could mesh an area really finely without the edges?

I know that in other softwares (Abaqus) that you can create a mesh gradient.. This is exactly what I would like to do with a dense mesh at the center, and a coarse mesh on the exteriors. I do not know how to use Abaqus, so I would really like to just use COSMOSWorks.

Also, can you simulate 2D simplifications in Cosmosworks? I would like to simplify the impact and simulate in an axis-symmetric form.
 
Replies continue below

Recommended for you

Hello,

I'm not too clear on what you are describing - is it possible to upload some screenshots?

You can create a mesh gradient in COSMOSWorks by adjusting the two mesh control parameters below the element size one in the mesh control menu. The first one is element growth rate and defaults to 1.5. I usually use 1.2 to make the change in element size more gradual. The second one is the number of layers of elements it should take to transition from the smaller size to the larger one. It is usually set to 3. Change this to 5 and again it makes the transition more gradual.

As for 2D simplifications I don't think COSMOSWorks is geared up for that. You could approximate 2D by making a really thin slice from your cylinder and applying symmetry restraints on the cut sides.

Abaqus is a much better software in my opinion. I've been campaigning to get my company to buy it! Definately worth working up the learning curve.

Hope this helps.

James
 
Piece of cake.

Create a spherical surface on the end of your cylinder in the area of interest.

Create a split feature with the surface and keep both bodies.

Put a mesh control on the little "scoop of ice cream" that is very fine and put the coarsest mesh you can get away with on the global mesh. Use a 1.1 growth factor and 4 or more levels so the mesh doesn't change to radically and make sure the default is set to bond touching bodies.

KTOP
 
James - Thank you very much for your response!


The 2D stuff is as I expected. It sounds like that might work, especially if I restrain the faces. But enough of that.

Attached is an image of the gradient I have now produced, along with the mesh control I used.

Your idea helped. I have now produced a gradient into the cylinder, and in the correct direction. But the problem is that I cannot get the gradient all the way into the cylinder. The mesh element size cannot be smaller than the mesh control size or the mesh control becomes irrelevant. I also cannot think of a way to mesh control from the inside-out.

The center of the cylinder is where the deformation will occur, and I cannot have the rings associated with split lines. Do you have any ideas?

Basically, I need tiny elements in the center, and coarse elements on the exterior. I am progressing, but have not been able to use split lines, and the gradient from layers and ratio is not working as I'd like... Am I just approaching it incorrectly??

Thanks!
 
 http://www.uvm.edu/~emalina/Files/images/reverse_gradient.bmp
Thanks for the pic - I get it now!

I can think of one thing you can try to get the gradient right into the middle of the cylinder but you are going to have to split your part up into multiple parts.

When you are assigning a mesh control you can select an entire part for the control from the assembly tree. So, if you could split your part up into a cylindrical core surrounded by hollow cylinders. You could assign mesh controls to the entire volume of each part (not just a split line).

Make sure that you dimension all the parts to interface exactly and that contact is set to 'global bonded' so it behaves as a continuum. Also, in the mesh control menu make sure you tick the 'Use same element size' checkbox.

I had a go and have attached a pic of the result. Is this what you mean?

Let me know how it goes

James
 
 http://files.engineering.com/getfile.aspx?folder=bb6409c4-382b-4cbc-9ce8-4ff8057d66cd&file=Cylinder_mesh_gradient.JPG
Just read ktop's idea - looks like it would be more efficient for this application as you only mesh refine in the 'crater' area where the probe hits.
 
Thank you both!

I wasn't able to make the sphere work. I don't quite understand what to do because I can't have the sphere there... Oh well.

James - your mesh looks awesome. How did you get inside the circle to mesh so perfectly? I think mine will work (attached).
I guess I just didn't quite understand the layer thing, because I was using a similar method (split lines rather than splitting the feature) and I could never get it right.

Thanks a lot for your help,

Evan
 
 http://files.engineering.com/getfile.aspx?folder=221ecbdb-cd78-4ad0-bbc1-11437f7c7942&file=mesh.bmp
James - It is now more important for my mesh to be consistent.

What is it that created such a consistent mesh on the inside of your "split" circle?

Also - it is to my knowledge that COSMOSWorks only supports 4 (Draft) and 10 (High) nodal elements. Can you use 8 noded cubical (type - not necessarily same length sides) elements?

Thanks again
 
Hello, I'm not at work at the minute (5 hours ahead) so I don't have COSMOS in front of me. To get the consistant element size you have to have separate parts and you have to apply the mesh control to the part (selecting 'Use same element size') not just the boundary surface.

In my assembly I used an inner cylinder of radius 2.5mm - mesh control element size 0.25, growth rate 1.2, layers 5. Then another pipe around that (inner radius 2.5 outer radius 12.5) - element size 2, growth rate 1.2, layers 5 then another pipe around that (inner radius 12.5 outer radius 20) element size global.

I'm afraid you are right and you only have the draft and quadratic element types to pick from.

I'm back at the office on Friday and if you like I can upload the assembly and part files so you can see exactly what I did.

Take it easy

James
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor