Modelling a welded connection between structural steel decking and a U frame rib

[Anonymous]

I am in the process of modelling a pedestrian steel bridge spanning 20m. The bridge is to be constructed of U frame ribs connected by a structural steel decking, stiffeners and top chords. Due to the nature of the decking and the welded connection between it and the frame, I must essentially model the bottom section of the U frame and the decking as one element (the decking is contiguous and welded to each rib) or as a series of beams with very wide flanges that are connected together (I couldn't find any options in Robot to achieve this). Is it possible to model this in Robot in a way which accounts for the continuity of the decking and its connection to the frame ribs? 

[Rafacascudo]

Can we see the general Layout of the bridge ? How many U ribs (2212mm width) are there?

I can Imagine 3 ways of modelling this.

1- "user defined" orthotropic slab

2- Top steel plate + U ribs defined as bars (Section created on section definition module) and connected to the plate through rigid links .

3-Full shell model.

Probably you will have build more than 1 of these models for different needed results

[Anonymous]

I will defiantly attempt the full shell model, thanks for the suggestion. In terms of the rigid links, how would I apply a link between the plate and the bars? To my understanding I can generate nodes on the bar but not along the plate which I model as a floor member (I am unaware of a way in which I can divide an edge into nodes).

[Rafacascudo]

Set the U section on its exact position according to its calculated CG in relation to the top plate.

Divide this bar in a way that the distance between the dividing nodes is aproximately your chosen panel mesh size.

Copy the dividing nodes to the plate level in the position where the U vertical plates intersect the top plate.

Use rigid links to connect the bar nodes to the corresponding plate nodes.

Copy the U bar with its assigned rigid links to the next positions where the U bars are supposed to exist

[Anonymous]

Should the links restrict all three axis? I have attached a test model to confirm I understood your instruction.

[Rafacascudo]

No , I meant this below , with full rigid links

File attached

[Anonymous]

I am not sure I understand how this model works, the deck seems to be above the U frame ? Also I am unclear on how the links work in this case. I have attached a quick 3D model of a segment of the bridge to illustrate how its put together. 

[Rafacascudo]

I am sorry . I completely misunderstood your bridge layout.

But ,anyway , isn´t the vertical web plate missing from the last picture you sent showing the overall layout and also from the RSA model you sent?

Is this 10mm vertical web plate welded to the base plate?

[Anonymous]

There is a nominal weld between the vertical plate and the bottom plate however is considered non structural; this is why I have left it out of the original model. But you are correct, my 3D model is also missing the side plate, I must have thought purely of structural elements when drawing it. In hindsight I should have probably mentioned this before, sorry if it caused confusion. 

[Rafacascudo]

So , I am guessing that the supports are on the upper nodes of the U ribs? Is that correct?

[Anonymous]

No, the supports are on the bottom of the U-Frame. Below are the diagrams of the elevation and plan of the bridge respectively, with supports indicated by red arrows. 

[Rafacascudo · Accepted solution]

   Well , ,what I pointed before as your options for modelling this structure have not changed.

I would go for a full shell model with the CHS chords modelled as bars . U ribs could be modelled as shell or also as bars with offsets or linked to the botton plate with rigid lings as you did in your model.

   My last question to you was if the supports were in ALL top U rib nodes(suspended bridge) as I couldn´t figure out how the shear forces would be transfered to the 20m apart supports, if your vertical plates webs are non structural.

  There is only the bottom plate with 3 small 75x10mm stiffener plates to carry the whole applied load to the transverse support beam.

[Anonymous]

  There is only the bottom plate with 3 small 75x10mm stiffener plates to carry the whole applied load to the transverse support beam.

That's correct.

  I would go for a full shell model with the CHS chords modelled as bars

I'll attempt this now.

I have made a model which utilizes beams for the u frame and the shell element for the deck with rigid links however I get a "No convergence of nonlinear problem" error. 

[Anonymous]

In the case of the full shell model should the stiffeners be shell elements too? I am struggling to match up the meshes. I assume they do as they are welded to the bottom plate

[Rafacascudo]

    Your model is nonlinear because you assigned an elasto-plastict section to  "U-bottom" . I never used it so I cannot help you  on this one . Try searching the forum. There is some information on this thread here  with a manual for this type of analysis attached.

  Keep in mind that for any nonlinear analysis it is not mandatory that you converge all load cases. Only those that have self weight alone or in a combination.

If you want to model the 75mmx10mm stiffeners as shell panels ,you will have to define a small mesh size of at least 25mm for them. Then mesh the stiffeners 1st ,freeze their mesh and only then mesh the bottom plate .

I would keep them as bars as it is easier to read the results .

Testing your model , I turned off the elasto plastic property and for a linear analysis , for your cases ,DL1+Live gives a midspan vertical deformation of 1813mm . In 20m it is just 1/11 Lspan.

[Anonymous]

Thanks a lot, I wouldn't have caught the elasto-plastict section; must have been an accident on my side. I have run the model and get the same results are you, I have also run a version with the side panels which yielded a much better result (9mm deflection). I'll revisit the assumptions made regarding the side panels and reconsider the model, but I now have a general understanding on how it should be modelled, thank you.