Element Orientation in Solids

kuhlkuhl
Hello all,
I am working on a composite model under thermal stress using solid elements. As the laminate definition does not support thermal expansion and because I need to capture response in thickness direction, I am using solids instaed of shells.
To consider fiber directions that are not parallel to global x coordinate, I have to introduce element orientation for the elements representing one composite layer. That's ok as long as the composite layer is flat and paralle to xy, yz, zx plane.


If, however, the composite layer is angled towards xy, yz, zx and/or - even worse - curved, defining the element orientation is very time consuming.
As an example I defined a quater pipe section (one layer) with an fiber orientation (u-axis) of -45° towards the axis of the pipe and the w-axis normal to the layer.




Is there any clever way to define the element orientation in this context? I'm thinking of someting like defining the w-axis as the normal direction of the selected elements face normal and then rotating the local coordinates system about this w-axis with respect to the projetion of a global axis onto the selected face or with resprect to the projection of an axis defined by selecting two nodes (dotted red line in image).



Alternatively, maybe the cxx solver can be used to capture thermal and mechanical response in thickness direction for laminates as it doesn's use "proper" shell elementes?

I'm thankful for for any advice,
kuhl

Comments

  • disladisla
    edited February 12
    Not sure if you know Element orientation can be coordinate dependent.

  • kuhlkuhl
    Thanks for the hint. That might be helpful in some case, I'll try that.
  • disladisla
    According to your picture, if geometry is centered at zero , you can get a radial distribution of orientations with:




  • cwharpecwharpe
    Nice, disla!
  • kuhlkuhl
    edited February 13
    Hey Disla, thanks for the heads-up!
    Works pretty well for a cylinder, also if it is not centered at zero.




    So in theory, it should work for all shapes that can be described by a function, although it might get a bit complicated / impractical with more complicated shapes.

    However, in my case (mainly prismatic profiles with radii) it is good enough for now, thanks!
  • disladisla
    I think that for prismatic/cylindrical shapes it can be easier to set a general orientation like the one in my previous post and then play with the degrees option to adjust the +-45 or whatever for each layer.


  • kuhlkuhl
    Yes, easier for sure. But thats only possible if the u-axis is normal the the "surface", or am I missing something?

    When using orthotropic material (composites), usually u-axis is fiber direction, v-axis is transverse direction an w-axis is thickness so my aim was to orient the coordinates accoudingly.
    For me and for documentation, it would be a bit confusing to adapt the material defintion so that u becomes w and so on.
  • disladisla
    edited February 13
    Yep, it would be a bit confusing.

    Considering that w is the transverse direction, (also the normal vector for shells), maybe it would be more convenient to change the GUI input to Angle about W instead of U.
    Kind of a "rotation of actual definition about W".

    That could be a feature request.
    What do you think Victor, would that be difficult?. I think your decision comes from the way orientations are coded in ccx right?
  • kuhlkuhl
    Yes, that would be an improvement.
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