How to model an unbonded contact?

davefodavefo
I am trying to model a precast concrete floor landing supported by a steel Channel beam section. For the temporary case the precast landing is just sitting on the top flange of the beam. but for the final load case there will be two welded studs tying them together.

1. How do I use the laminate material to model the reinforcement within the precast concrete landing
2. How do I model the contact between the precast landing unit and the PFC top flange so that this is like a compression only contact + friction.

I am trying to show the true deflection/twist of the PFC as a result of the orientation of the PFC.


Comments

  • VictorVictor
    1. I'm not sure about using laminate material to simulate reinforcement bars. One way could be to make a steel layer with the same cross sectional area as the total rebar cross section in that plane. Then the overall tensile stiffness in the direction of the rebar would the equivalent. But it might be incorrect for loading in other directions. It might be more accurate to use beam elements inside a solid mesh, rather than the laminate material.

    2. Use CalculiX with Nonlinear static analysis type and a Contact (CCX) instead of Bonded contact.

  • davefodavefo
    1. I was going to try a steel layer but as you have mentioned decided that it would not be correct for all directions. Do you have any examples of how to used beam elements within concrete. Is it possible to define the concrete material correctly. i.e. strong when under compression but very weak under tension.

    2. With non linear static 3D what values should be provided in the analysis settings? 
    3. When i change the analysis type to this the shells are no longer valid. Do i have to convert everything to solid elements? 
  • VictorVictor
    1. See other post

    2. You can leave all the analysis settings at their defaults of 0 or default. If you need time-dependent loading, then specify time step information.

    3. In Mecway version 5, the error messages are for the internal solver, not CCX. So you can still use shells with CCX but not with the internal solver. Version 6 which is now in beta has the error messages appropriate for the solver.


  • davefodavefo
    tried using ccx 2.10 with contact in mecway version 6 but still cannot get it to solve. it runs for ages but doesnt give any result.

    what have i set wrong. lml file is attached

    Regards

    David

  • VictorVictor
    I think the problem is that the contacts may be initially open so on the first iteration, the force has no constraints to provide any reaction and pushes the plate off to infinity. I checked this by putting a fixed support on a small part of it and it found a solution in 20 iterations.

    This is a common kind of problem with contact and there are various ways around it. One is to put some soft springs in the force direction to provide some reaction as the iterations get started. Another is to use a prescribed displacement rather than force initially until the two parts are in contact, then apply the force and remove the prescribed displacement. For this, you may need two *STEP sections and enter some of the CCX cards manually.

    One of the beams is also lacking any Y-constraints.
  • UGMENTALCASEUGMENTALCASE
    I'm trying to do exactly this. Something which is not initially connected, but say for example a force is applied, moves the object into another and then deformation will appear. I can't find or understand how to use the STEP functions in CCX solver options. Do you know of any basic tutorials or pointers knocking about?
    I've tried a bit of googling but I seem to get more linux based instructions which for use on a windows machine doesn't really work.

    Cheers
  • VictorVictor
    Most of the CCX information will be assuming Linux but defining steps and just about everything else will work just the same on Windows. The difficulty might be doing it through Mecway. For multiple steps, that's not really easy so it might be better to just work on the .inp file directly using a text editor, at least until you know how it should be.

    A simple way to get multiple steps working is to start with a correct .inp file and just copy and paste the existing step section so there are two. Then it should do two steps. After that's working, you can modify them to be more useful. A step section is everything from *STEP to *END STEP inclusive.


  • UGMENTALCASEUGMENTALCASE
    Hi Victor really struggling with this could you help please? I've attached a txt file which i exported out from the solver. I've tried all sorts of copy and pasting and tried following some other instructions from abaqus files, and i can't get it to work.
    I'm basically trying to move a very basic model in an x direction, then a y direction. I've started basic and tried to move it 2 separate directions in separate steps then i'll move on and see what i can do.
  • VictorVictor
    Here you go. The solution contains two steps - one with each direction's displacement.

  • UGMENTALCASEUGMENTALCASE

    Hi Victor, thank you for the file

    That's exactly what I did but when I import the file to Mecway and run it I only get movement in one direction? I've tried different analysis types and all sorts and can only get it moving in the initial direction.


    Cheers

  • VictorVictor
    These more complicated .inp files aren't fully recognized by Mecway and show an error message on Analysis settings after you import them to identify what was missed. To include the extra step via Mecway, paste into CCX -> Custom step contents like in the attachment.

    Mecway thinks you're adding lines inside the STEP section but really you're ending the first step and starting a new one, so it begins with *END STEP and doesn't end with another *END STEP.

    I also changed it too nonlinear so that Mecway v6 will recognize the time steps in the postprocessor. This causes a few unnecessary iterations in the solve. There are ways to work around that too if you want.

    As you can see, Mecway's not very good at using more complicated CCX features like multiple steps, and you sort of have to work around a lot of limitations.
    xy.liml 29.1K
  • UGMENTALCASEUGMENTALCASE

    Ah ha, thank you. Got it moving in 2 directions now. A bit more editing and I think I should be able to get it moving in x 80mm then stay at x 80mm and then move in Y.


    Thanks again


    Cheers

Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In Register