No convergence of nonlinear problem

[tomislav.lalic]

Hi,

I get the error message on my 3D Truss model supported by cables, I have tried the solutions on this forum but nothing works. Please help.

Im attaching my model below.

Edit: new model attached, for some reason Robot made the cables as bars. Error still there.

[Krzysztof_Wasik]

Hi @tomislav.lalic 

Truss structure with such geometry is instable. Change model type to Frame 3D. Define releaes for bar which should be pinned.

[Simau]

@tomislav.lalic 

Some tips about structure:

As said by @Krzysztof_Wasik  switch to 3D Frame

To avoid type 1 instabilities add posts and diagonals 'verticaly and horizontaly.

Increase the gap between cables

For selfweitght case uncheck "opposite to axis sens"

Increase members area

Try other setup cables to reach convergence for case 1

Test model attached

[tomislav.lalic]

I changed the structure type to 3D Frame and I'm still getting the same error. I'll attach the new model with some changes I didnt notice before in case I defined the releases wrong.

Changing the structure itself is not possible since this is an existing structure that is already built. I just got the assigment to make the model.

Thanks for your help.

[tomislav.lalic]

Changing the structure itself is not possible since this is an existing structure that is already built. I just got the assigment to make the model. I'll attach the new 3D Frame model that has some changes I didnt see before in the geometry but the error still persists Maybe I defined the releases wrong?

Thanks for your help.

[Simau]

@tomislav.lalic 

Please check units and SW direction

[tomislav.lalic]

Thank you for noticing my load are in wrong units, I fixed that and unchecked the self weight parameter but the error is still there. Addition error came up: "Matrix is not positive definite - the critical load may have been exceeded."

Corrected model attached.

[Krzysztof_Wasik]

Hi @tomislav.lalic 

• Leave releases for RY only in poinned-pinned label (add other required releases after results verification)
• Set the following parameters for calculations of non auxiliary load cases.

[tomislav.lalic]

Could you elaborate on the following:

"Leave releases for RY only in poinned-pinned label (add other required releases after results verification)"

Im not sure what you mean.

Thanks.

[Krzysztof_Wasik]

Hi @tomislav.lalic 

Change releases definition to the following

When after model calculation you will obtain non-zero moments in nodes where you expect zero then add release in required point 

[tomislav.lalic]

But since its a truss structure, no moments should be present right? Except maybe in the upper and lower chords (I think that is the translation but not sure), so I'm not sure where do I make other necessary releases. Is it on the bars that have any moment value, or just on the bars that have substantial moment value?

I made 2 models with Ry release as you suggested. One with undivided upper and lower chords without any releases applies, and the other with divided upper and lower chords with Ry release applied.

Model 1 had no type 1 instability warnings during the calculations, Model 2 did.

Models and results in attachment.

Thanks again for taking the time to help.

[Krzysztof_Wasik]

Hi @tomislav.lalic 

Truss model is a kind if idealization of the reality (I would say that there are very limited number of structures in reality which works as pure truss).  Since you simulate existing structure you have to decide where moments should be equal to zero (you should define additional releases in that points) and where non zero moments are allowed.

Instabilities type 1 are caused mainly by possibility of nodes rotation is nodes like node 1. In node 1 pinned connection works when release is defined for one member only. node 1 should be attached to second member (without release) to remove type 1 instability.

.

[Krzysztof_Wasik]

Hi @tomislav.lalic 

Instead of modification of releases system in model 2 (to avoid instability type 1), you can also define small rotational elasticity (for instance 0.001 kN*m/rad). It will remove instability and minimize moments obtained in member ends.