surface contact - Shell brick

[Anonymous]

Hello dear community,

I am working on the analysis of a scaffolding connection (only surface contacts, 2 outer tubes, one inner tube, 2 bolts, non linear material, 26 SS contact pairs)

To get a decent result for the v. Mises stress, I need a mesh with 3 elements in the thickness.

With a material thickness of 3.2 mm and a length of the model of 1300mm the mesh gets extremely fine.

At the connection I have a surface contact between the connection pins and the scaffolding tubes (the pin is placed through a hole in the tube).

Now the question:

If I discretise the tube with bricks (3 elements in the height) the calculation takes days (literally)!

My Idea, to reduce the calculation time is to go for shell elements for the outer tubes (midplane discretisation).

I do not get a stable calculation using the shell elements.

Is it possible to use the edge of a midplane shell element as a surface contact for a brick, or is this not possible because the element has only an edge and not a surface in the contact zone?

Regards Boris

[zhuangs · Accepted solution]

It is definitely possible to use the edge of a midplane shell element as a surface contact for a brick.  To use the edge of shell part in a contact pair, you can do like the following procedure:

(1) Select all the lines on the edge which has contact;

(2) Right click and select "Edit Attributes";

(3) Input a surface number (such as 100), note this number should be different from the existing surface numbers in the shell part.

(4) Define the corresponding contact pair in the order:

     (1st part/surface on 1st) with (2nd part/surface on 2nd)  =  (Brick part/surface on Brick part) with (Shell part/edge surface 100 on Shell part)

(5) Define the corresponding contact type as "Point to Surface".

Then deal with it as a common contact pair and it should work.

-Shoubing 

---

bannecke wrote:

Hello dear community,

I am working on the analysis of a scaffolding connection (only surface contacts, 2 outer tubes, one inner tube, 2 bolts, non linear material, 26 SS contact pairs)

To get a decent result for the v. Mises stress, I need a mesh with 3 elements in the thickness.

With a material thickness of 3.2 mm and a length of the model of 1300mm the mesh gets extremely fine.

At the connection I have a surface contact between the connection pins and the scaffolding tubes (the pin is placed through a hole in the tube).

Now the question:

If I discretise the tube with bricks (3 elements in the height) the calculation takes days (literally)!

My Idea, to reduce the calculation time is to go for shell elements for the outer tubes (midplane discretisation).

I do not get a stable calculation using the shell elements.

Is it possible to use the edge of a midplane shell element as a surface contact for a brick, or is this not possible because the element has only an edge and not a surface in the contact zone?

Regards Boris

---

[John_Holtz]

Hi Boris,

In addition to the ideas that Shoubing suggested, here are a couple of other ideas and/or questions.

When you wrote about 3 elements through a 3.2 mm thickness and a 1300 mm long tube and lots of elements, is that because you were making the elements roughly the same size in all 3 directions? In other words, 1300 mm / (3.2 mm / 3 elements) ~ 1218 elements along the length. If you are okay with larger elements along the length, then consider using the layered mesh: "Mesh > Mesh > 3D Mesh Settings > Options > Solid > Bricks and wedges". (see Help > Autodesk Simulation > Mesh Models > Mesh Overview > Meshing CAD Solid Models > Model Mesh Set...)

The other thought is that you may not need to model the entire 1300 mm length of the tube. You may be able to replace a majority of the tube's length with beam elements. See the section "Connect Lines to the Centroid" on the page Help > Autodesk Simulation > Mesh Models > Mesh Overview > Create and Modify Geometry in the FEA Edi....

[Anonymous]

@Shoubing: Thank you, the solution worked well.

@John: I use the "Brick and wedge" mesh option for any thin element as a standard.

But to my understanding a brick element remains a brick, nevertheless if I mesh with „Brick and wedge" or with "Bricks and tetrahedra". The best result for a brick element brings afaik the cube. With a site ratio of 1 to 1. I try to keep the standard site ratio below 1 to 3 (I think I read this aspect ratio somewhere in the literature). So my brick element is supposed to be 3 mm long if it is 1mm thick. Is this to conservative?

The idea to replace part of the tube with a beam element is very good. I did not think about this option. In this case I will not stick to the Idea, because the problem is highly non linear and I would like to use as few simplifications as possible. But for further problems I will keep this in mind.

Thank you and regards

Boris