Learning Forum

[Ben_Ben]

I am using the dynain method above to perform sequential simulations with beams. Similar to the topic discussed here: https://forum.ansys.com/forums/topic/error-when-transferring-solution-to-new-model-in-the-project-schematic/

I have a structure that is composed of 60 beams and is slightly deformed in the first simulation and then in the second it is compressed. However, when I apply any BCs in the second simulation I get general errors due to out of range velocities. I am using Workbench LS-Dyna 2022 R2, here is the break down of my problem:

When I use this method for beams it combines 60 beam bodies into 1 beam body:

->

This would be fine as I can still select lines and nodes to apply BCs. 

However, when I apply BCs I get exponential kinetic energies. 

For example, if I apply a displacement to the top edges in the picture below and use a fixed support on the bottom edges. I get error due to runaway velocities:

If I apply those BCs to the geometry in the first simulation (the one that doesn’t take geometry from dynain), I can compress the structure without exponential Kinetic energies. 

Do you have any advice for using beams in sequential simulations with dynain?

Thanks in advance, 

Ben

 

[Ram Gopisetti]

Hi , having out of range velcovits means , the model might have instabilities and check the following link on how to deal with it 

https://www.dynasupport.com/faq/general/have-you-any-tips-on-how-to-to-combat-instability

cheers, Ram 

[Ben_Ben]

Hi Ram, 

I will check my model for these things. 

Is it normal for my 60 line bodies from my first simulation to be converted into 1 line body in my second simulation after using the dynain file? 

Cheers, 

Ben

[Ben_Ben]

 

Hi Ram, 

I have tried versy simplistic BCs (very small displacement to an edge) and still getting the errors. 

The problem is that after my first simulation my geometry consists of 60 line bodies, 60 edges, 120 Verticies. When I transfer this to my second simulation, my geometry consists of one line body, 422 edges, 422 vertices.

I think something is happening during the transfer. 

Also cross sections don’t transfer with dynain. So I’m creating a circular cross section in the second simulation.

Ben

 

[Ben_Ben]

 

Hi Ram, 

Have been doing some debugging and found that in my first simulation I have a spike of kinetic energy in the energy summary from the solution files: 

Could this be the reason I am getting errors in my second simulation?

To investigate some more I assigned a kinetic energy tracker for all the bodies and found that the spike is not from the bodies and also the kinetic energy is an order of 2 magnitudes out: 

 

I checked hourglass energy and is 0. I turned damping off and it made no difference. Also the simulation looks like it behaves as it’s supposed to. Please could you advise me on anything I can check. 

 

Cheers, 

Ben

 

[Andreas Koutras]

Hi Ben,

I suggest you build a small model with just few beam elements and verify the dynain workflow. 

Dynain does not import geometry but only imports the mesh, i.e. the nodal coordinates and elements, the *INITIAL_STRESS keyword, and the *BOUNDARY_SPC keyword. Everything else should be redefined in WB.

If for some reason this approach does not work well with beam element, you may try with the LS-DYNA Restart module.

I hope this helps.

AK

[Ben_Ben]

Hi Andreas and Ram, 

I am still strugling with this. Please can you provide guidance on how to over come it. I simplified the problem down to one beam and I still get the same issue. 

Thanks, 

Ben

[Ben_Ben]

Thanks for your suggestion AK. I simplified my sim down to one beam but the second simulation does not behave as expected. I have outlined my test scenario below, all analysis settings are the same etc. 

In my first sim I applied a displacement inwards towards centre at each end

Here are the results 

In the second simulation I applied the same displacements but towards the outside. 

But I get ridiculous results, here are the results on non-deformed geometry and deformed geometry

Please could advise on how to run this scenario with a restart analysis. Where I can change my boundary conditions. 

Thanks, 

Ben

[Ben_Ben]

I asked the question about the LS-Dyna Restart module because I tried to emulate the exact same test I outlined above. I had some problems in that even after applying a full restart and flipping the boundary conditions my results are exactly the same. Here is my Project Tree:

Here is the First simulation set up and results: 

And here is the restart simulations set up and results: 

As you can see it has just reexecuted the boundary conditions from the first simulation. Please could you advise on performing subsequent simulations with differing boundary conditions? 

Thanks,

Ben