No convergence of nonlinear problem. (panels, 2D shell design model)

[1234eddie]

When modelling 2 panels above each other in which the upper panel is just resting on the lower one in a 2D shell design model, I used compression bars to model the link between the two panels. When running the model the problem then occurs that there is no convergence in the nonlinear problem. I tried to use different load increment numbers and enlarged the tolerance for forces and displacements a little bit, but all of this didn’t help to let the calculation converge.

 What could be hindering the analysis from converging, and is this the right way to model a resting panel ?

Thanks in advance.

[Rafacascudo]

Is this the correct model ?

There are only 4 compression bars on the model , all in the extremities of the upper plate. The rest of the upper plate is unsupported, no compression bars.

Also , there is a pinned support applied to the upper plate on one of these extremities .

[1234eddie]

@Rafacascudo  my fault, it was a test model.

I have attached 2 new models, both with different warnings.

what i'm trying to achive is that the L-shape only will give pressure on the bottemplate. see this section.(the vertical shape is not modelled yet). this because the L-shape will only be connected with 4 simple "pop nails", so i don't want to give any structural contribution from these to the model.

In model version 2 i have added compression bar with a center to center distance of 0,1m.

when running i get this error

In model version 21 i have added compression bar with a center to center distance of 0,1m. and 2 simple bars at the end of the plate.

when running i get this error

I hope you or someone else can give me some advice how i should model this. 

Many thanks in advance.

gr Edward

[Rafacascudo]

 I created a more refined mesh of 10mm size on both plates and created compression only bars for every node of this mesh .Only on pinned support side,  regular bars were created with moment releases on top node.

RSA struggles and cannot converge the analysis for the self weight case. Probably due to top plate being a lot stiffer than the bottom one . Running the 2 plates separately it is easy to notice the difference by looking at the vertical displacements

So in fact , the top plate would be "carrying" the bottom plate if they were connected by regular bars on the whole area.

So I created a load case with a 1kpa over the top plate ,and it converged with no problem.

Distance between the plates center lines should be 3mm instead of 2. I also modified the compression only bars sections

model attached

[1234eddie]

@Rafacascudo thanks for your information.

1: I tried following youre approach but it doesnt work for me. See Model version V29.

what i have done in the model:

- changed topplate thickness to 2mm.

- added compression bars every 10mm(square), and regular bars at the end with moment release.

- added load cases and 2 combinations wich includes dead weight.

- changed the setting to use DSC algorithm

This is the error. I'm lost how to solve this myself while your model works.

2: I also tried to model further see model V30 where i have added the vertical flange of the ange iron. this is also the part of the structure which will receive te load in reality.

but still the same errors occur.

3: and to continue, i added a second L angle(copy from first) model V31, because in the model i want to create (when i understand how) there will be around 25 L-angles, which mean a total with of 1m and depending on the situation with a maximum lenght of 6 metres. the vertical parts of the L-angels will be supported horizontally by a stifner plate(not welded) with a center to center distance of 0,5m, i'm thinking of modelling this with a simple rigid link with only Ux,Uy,Uz.

I hope i won't bother you too much with all my questions and models(but i wan't to understand what;s going on before making a big model with all these issues). I really appreciate your help and insight youre providing.

Thanks in advance.

Gr Edward

[Rafacascudo]

Can you provide a sketch/drawing showing  the complete picture of your structure ?

and why you want the 2mm bottom plate to be part of the resisting structure?

[1234eddie]

here is a simplified sketch of the structure.

Black is the bottom plate which will have folded sides.

red are the L-shapes

blue represents the horizontal support for the vertical parts of the L-shapes.(just connected by a gap in the plate, and resting to the folded sides.

2. i want to model the bottom plate because it all works together to resist te force acting on it.(snow, live load due walking person)

hope this make is clear. thanks.

[Rafacascudo · Accepted solution]

2. i want to model the bottom plate because it all works together to resist te force acting on it.(snow, live load due walking person)

If you mean as "works together" that the angles will transmit the force to the bottom plate by contact only , I am afraid that this will never happen . The difference in stiffness between a single angle and the 2mm plate is huge.

See the difference in vertical displacement just for the self weight.

Difference is around 400x . This means that the load to make the angle touch the bottom plate displaced only by self weight would have to be huge.

Even that analysis converged for your loads/combinations, results would never show the angles transmitting forces to the bottom plate.

[1234eddie]

@Rafacascudo 

Thanks for your opinion. I have to check it but you may be right.

One other queston: in your model case 1 deadload turned into auxilary. how can i do that?

thanks in advance

[Rafacascudo]

On analysis options ,double clicking the load case and marking "auxilary" on the popup window or select the load case and click on parameters

[1234eddie]

Hi @Rafacascudo 

I've been working and thinking of this model and worked a little bit on it but i'm stil running in to some issues.

I upgraded the model to have 10 L angles next to eachother. without any base plate. and added bars with an rectangular section on the places where the vertical flange of the L-shape is cutted to a depth of approximatily 15mm. in reality the stiffnerplate will have the same cut so they slide in eachoter. that's why i think that modelling this as bars would be correct, when i use the bar-end release where only the z rotation is free.

when running the model in just "static case/combination" one small instability of type 3 occurs. this is not a problem.

but when turning the analysis of the case to buckling and combination of buckling cases it gives me te following errors(see picture. i stopped analysis after 25minutes because it should solve it quicker in my opinion. but i don't know what i'm doiing wrong.)

Can you or someone else have a look at my model?

because the end model will exist out of almost 40 L-angles with different flanges height. but for now i'm just trying to get this running before i go big.

Thanks in advance.

gr Edward

[1234eddie]

@Rafacascudo @Rafal.Gaweda 

I'm not sure why this model i posted in my previous post gives the errors i mentioned.(still curious about it)

I did a rebuild on the model and it works fine now(with almost 40 l angles).

i can't share the rebuild because i'm not allowed to share it.

Many thanks for your help and advice.

gr Edward