Nonlinear axial compression of a tube - fatal Error E5076: Maximum number of bisections reached

[daniel.poole]

Please help

I'm relatively new to Nastran so I apologise if there is a simple solution to this.

I have a thin walled tube (75mm OD x 2mm thick  x 150mm long) and I want to apply a load to the end. (nonlinear buckling)

I have set up a custom nonlinear material (though have also done a sanity check with a standard nonlinear one) and no matter how I set up the analysis I always get a failed solution.

The obvious way to set it up would be to fix the bottom and a apply a load to the top.

I have tried variuis constraints: fix the bottom edge, fix the bottom in all directions accept along its surface plane. split the tube along its length and put a symmetry constraint there.

I have also tried adjusting the mesh, changing the steps, basically anything I could but I always get the same error.

I must be doing something wrong as I cannot think of a simpler analysis.

[John_Holtz]

Hi @daniel.poole 

It is too hard to guess what might be wrong from a written description. Please attach the model to the forum post. (It also depends on when the E5076 error occurs. If it is at the beginning of the analysis, the model may not have enough stability. If it is near the end of the analysis, maybe the part is buckling!)

• If the model is a part file, attach the .ipt.
• If the model is an assembly, following the instructions in this post: What files to provide when the model is needed - Autodesk Community - Inventor Nastran

John

[daniel.poole]

Thanks for the help.

I'm attaching two parts. The first (2) has had lots of things altered to try and get a result. The second (3) I quickly created to show how in my mind the set up should have been done.

The aim of my study is to determine actual buckling value. If it is buckling how would I set it up to continue so i can visualise and view values.

[daniel.poole]

Sorry, reply is in the main thread. My mistake.

[John_Holtz]

Hi @daniel.poole 

"Test 3" is not statically stable; it is completely free to move in X and Y. The analysis should end with an E5000 error to let you know the model is unstable. The "E5076 no convergence" error that occurs at the beginning of the analysis is most likely related to the instability. 

I tried restraining the bottom face in all directions. The nonlinear static analysis gets as far as the material yielding and ends with an E5076 error. It looks like it is having problems transitioning from elastic to plastic throughout the entire volume all at once. After you decide on the constraints, we can revisit the analysis if there are any convergence issues.

This article may be helpful, at least to understand what the analysis is going to do: What is a nonlinear buckling analysis in Nastran

"Test 2" will not give you answers you want because it is a 1/4 symmetric model. The pipe is not likely to buckle in a symmetric fashion. However, when you were trying to analyze a symmetric model, these are things that are incorrect:

•  Constraint "Constraint 2" is a Y symmetry constraint (Ty, Rx, Rz). It is applied to 2 faces. The face that is normal to the Y axis is the correct face for Y symmetry. The face that is normal to the X axis is not the correct face for a Y symmetry constraint. (Maybe you have a reason to apply the Y symmetry constraint to the face normal to X.)
• Constraint "No Rotation" (Rx, Ry, Rz) is not doing anything in the analysis. The nodes of solid do not calculate a rotation, so the constraint is not having any effect on the analysis. You should delete this constraint to avoid confusion.

John

[daniel.poole]

Thank you for your assistance it is very much apprechiated.

I have attached a new model which I have been running this afternoon. I think I have the symmetrical constraints correct on this. As per the attached diagram.

As this is a stocky tube i would like to see some sort of local buckling (rather than global - Slender tube), maybe a fattening of the middle. This is why I used a 1/4 model. I think the symmetrical constraints also constrain it in X and Y which solves the problems you describe on test 3.

If I were to use a full tube how is it possible to constrain it in X and Y?

[John_Holtz]

Hi @daniel.poole 

I did not look at the new model yet because I was working on version 3. I split the bottom face so that I could put X and Y constraints to provide static stability. I was having problems getting the analysis to converge after 45% load. (By the way, 100% load would cause a stress of 490 MPa which is twice the yield stress. Maybe 490 MPa is too much load?) I decided to try 100 increments in the analysis. It converged at 48% load, and looking down on the top of the tube, it is still circular.

Figure 1: Nonlinear static analysis with nonlinear materials. No buckling at 48% load.

The next increment at 49% load failed to converge. Nastran bisected the step to try to converge but eventually failed at 48.2% load. The top goes oval which is very similar to the results of linear buckling. (You did run a linear buckling analysis, right?) The nonlinear analysis cannot get beyond the start of buckling because buckling is unstable. A static analysis cannot handle instability (in most cases).

Figure 2: Nonlinear static analysis. The last unconverged step at 48.2% load. The top is buckling in an oval shape.

In summary, the pipe will buckle at 48% * 225000 N load when nonlinear plasticity is included. The linear analysis predicts buckling at 148% load. 

Figure 3: Linear buckling analysis. The top buckles in an oval shape (very similar to the nonlinear static analysis).

John

[daniel.poole]

This is fantastic, and again I apprechiate your efforts.

Having gone through my calculations I realise that I think this particular tube should buckle at around 110KN. Not far from what you are getting. I'd given up looking at my calculations and just wanted something to run to completion. I hadn't realised that a failed run could also be a sign that the part has buckled.