Mesh joints for wood material

Hi,

Soon I’ll want to use Karamba to analyse a project made of planar elements of CLT (cross laminated timber). As I imagine it right now, it will be a tessellate free surface where each planar face will be made of CLT.

But for now I'm learning using Karamba with CLT by analyzing an existing project as a case study, the St-Loup Chapel.

First, I’ll try to use wood of the predefined list material set by Karambe, but the analyze fail every time. When I use other materials (concrete, steel), even if the deformation is weird (probably because of the material selected), it’s seem to globally work well. Wood contrary of the other material is anisotropic. Does Karamba taking this in consideration and is it why it doesn’t work? In this case, I’ll probably do a hypothetic isotropic material for my CLT… If there’s some more hints I’m really open.

Secondly, since it’s wood material, I can’t consider the connexion between my panel as rigid. I know there’s "beam-joints" to add hinges at the ends of beams. But I work with panel, so with mesh, and I would like to know if it’s possible to add some sort of hinges between my meshs? For now, I can’t consider make my model as linear elements (beams) neither for my next experimentation.

Thanks for any help/suggestions!

Zoe

Replies to This Discussion

Permalink Reply by Karamba3D on October 6, 2014 at 11:49pm

Hi Zoe,

thank you for your bug report.

The current version of Karamba (1.0.5) assumes isotropic material behavior. For the relation between Young's Modulus E and Shear Modulus G this results in the conditions that G>E/3 and G<E/2 shall be fulfilled. There is a warning when one tries to define a material with the MaterialProperties-component which does not adhere to these conditions. However for wood, the material properties table has not been updated yet to meet these conditions - which will be done in the next release. As a workaround you could edit 'MaterialProperties.csv' (it is located in the 'Karamba'-folder inside Grasshoppers Components-folder) and set G of wood to say 500[kN/cm2].

The shell elements in Karamba (contrary to the beam elements) do not take account of shear deformation and are based on the Kirchhoff hypothesis. For slender panels this gives correct results.

Regarding the hinges there are currently two possibilities:

You could introduce a thin strip of shell elements of very small cross section height made of a very stiff material.
You could connect neighboring panel edges using springs.

Best,

Clemens

Permalink Reply by Zoe Tolszczuk-Leclerc on November 25, 2014 at 4:56pm

Hi Clemens,

Thanks! The workaround for the wood material works perfectly and it no longer bugs my analyze!

Otherwise, I'll try your suggestions for my hinges problem. For the second one, I don't have significant result for now, maybe it's me but I can't managed to make a functional model. I have better result with your first suggestion, but still have some issues.

I create the thin strip of shell between my wood panel (principal shell elements). If they seem to be taken in consideration at some point (in the deformation the panels don't go all around and stay connected), the cross section have no influence on the analyze, so it could be 1mm or 10cm, it's doesn't change my result. I finally figure that the elements were inactive when entered the assemble. I activeted them by reducing the limit distance for coincident points (in mesh to shell). My elements are now activeted, but now I face an analyze fail - Can not solve static system. I'm pretty out of ideas now.

If you have some or if someone alse have similar issue I'm open to any suggestions!

Thank in advance!

Zoé

Permalink Reply by Karamba3D on November 27, 2014 at 1:17am

Hi Zoé,

did you try to reduce the size of the structure - e.g. to two panels connected with one strip? This could be a good starting point for locating the problem. If you send it in I will have a look at it.

Best,

Clemens

Permalink Reply by Zoe Tolszczuk-Leclerc on November 27, 2014 at 11:45am

Hi Clemens,

The top script model is the full/normal one, I modify the bottom one to work with less panel.

For now when I work with the first set of three panels (as a kind of arch) it's seem to work well, so I suppose it's the laterals thin shell who make the analyze fail. Work on this now.

Here's my grasshopper file

Thanks for looking! :)

Zoé

Attachments:

wood panel origami.gh, 78 KB

Permalink Reply by Karamba3D on November 28, 2014 at 12:22am

Hi Zoé,

in Karamba the shear modulus G needs to be larger than E/2 because an isotropic material is assumed. For wood this assumption is not sufficient since it can have a much smaller G-modulus in the transverse direction of the plate. The problem you experienced is caused by a bug in the material properties table of Karamba - I am sorry for the inconvenience.

I changed the shear modulus in the attached definition so that it works. For slender plates the influence of the shear modulus is negligible.

Best,

Clemens

Attachments:

woodpanelorigami_CP.gh, 79 KB

Permalink Reply by Zoe Tolszczuk-Leclerc on November 28, 2014 at 9:37am

Hi, Clemens,

Actually I had fix the shear modulus problem by editing the 'MaterialProperties.csv' file long ago. I don't think it's that who cause the bug.

If you look on the top one, I draw some indication to show where I think the issue is about. If you play with the boolean toogle tool you'll see what exacly happen. The analyze don't fail at False, but not seem good since it not consider my thin shell element (element inactive), and at True, what I want to have, the analyze fail...

Thanks again!

Zoé

Attachments:

Attachments:

woodpanelorigami_ZTL1.gh, 85 KB

Permalink Reply by Karamba3D on November 30, 2014 at 11:42pm

Hi Zoé,

when you set LDist to 1e-8 the LineToMesh-component starts to produce shell elements. However those are very small as compared to the others in the structure. This can lead to numerical difficulties. Try to increase the width of the gap between your plates.

Best,

Clemens

Permalink Reply by Aryan Shahabian on January 11, 2017 at 3:24am

Dear Clemens,

Could you please explain the possibility number 1 that you've mentioned in your first comment? I guess that a material with low Young's modulus can represent a hinge and not a very stiff one. Please correct me if I'm wrong.

Best,

Aryan

Permalink Reply by Karamba3D on January 11, 2017 at 10:49am

Dear Aryan,

if the material of the joint strip is very soft the two sides would be completely sparated in terms of relative rotations and translations.

The bending stiffness of a plate is proportional to t^3*E, the shear stiffness is proportional to t*E (here t is the shell thickness, E is the shell material's Young's Modulus). If one wants to make the bending stiffness small but keep the shear stiffness at a reasonable value one needs to choose for t a small value, for E a large value.

Best,

Clemens

RSS