Tube Bending and Stretching in Kangaroo

Hello guys,

Can anyone know how to bend a tube/cylinder on Kangaroo? What I'm trying to do is to bend the tube and at the same time tube can be able to stretch. I have a look in many tutorials and example videos but most of them is how to bend line, curvers etc.

Thanks

George

Replies to This Discussion

Permalink Reply by Michael Makris on April 28, 2014 at 1:37pm

For example:

I just ran a quick study of what happens when you vary the number of mesh divisions in a two-way slab. I maintained the strength, gravity, and stiffness factors among all three meshes. However, the final resting slab in each case was very different. Further, the deflection behavior shown does not accurately reflect the deflection patterns for two-way slabs. With increased mesh accuracy, the deflected shape should converge towards a realistic solution. Instead, with increased modeling accuracy, it diverges in this example.

Permalink Reply by Daniel Piker on April 28, 2014 at 3:04pm

There are a number of reasons why I think this will not give accurate results, even aside from the bending behaviour.

You say the "gravity and stiffness factors" were maintained, but do you mean you adjusted the lumping of the gravity load to keep it constant per unit area, or you kept it constant per vertex?

Similarly with the stiffness. As I mention in my other reply, the only proper way to model the membrane stretching is with the 2d elements, which aren't available in the current version, but even so, I think the existing 1d elements can be used to give a somewhat reasonable approximation only if the stiffnesses are appropriately adjusted for area.

Permalink Reply by Michael Makris on April 28, 2014 at 3:23pm

You're right. Since I threw this together fast, I didn't catch the gravity load is per vertex. As I increased discretization, I increased the load.

As to the stiffness, I haven't delved into the parameter. Is the value given an absolute stiffness for the triangle pair, a stiffness per unit area, or stiffness per unit length of common edge?

Permalink Reply by Daniel Piker on April 28, 2014 at 2:56pm

Is it possible to get a look at your thesis? sounds very interesting.

It's true that the bending aspects could certainly use some more work on testing and validation.

For the bending of 1d elements it should be relatively straightforward to link the parameters to some real numbers (and I think Dragos Naicu did some testing of this during the process of the gridshell in Romania we worked on together).

For the bending of shells using the hinge component, the version in the currently available release of Kangaroo isn't rigorously linked to material properties, as it was originally added mainly for origami folding. However, I have recently been working on a properly derived shell bending force.

Also - approximating the stretching and shear of shells or membranes through axial elements alone is always problematic, and dependent on the meshing.

This is why I've now added 2d elements (a constant strain triangle), and that last video you mention shows also the 3d elements (constant strain tetrahedra), for simulating volumetric effects. These 2d and 3d elements are both implementations of work by professor Chris Williams at Bath university, derived from established principles of elasticity, and using standard material parameters such as Young's modulus and Poisson's ratio. While they still need testing and validating to make sure I've implemented everything correctly, I think it should be possible to get some reasonable numerical results with these.

Permalink Reply by Michael Makris on April 28, 2014 at 3:13pm

I think this link to the thesis should work for anybody, as I wasn't logged in when I copied it.

http://digitallibrary.usc.edu/cdm/ref/collection/p15799coll3/id/305175

Unfortunately I can't upload the GH files since I'm not at home, but I will do so ASAP.

I'm really interested in GH and Kangaroo implementations of traditionally engineering-related problems. Could I take a look at your 2d and 3d constant strain examples? I'm working as design engineer at a firm up in Seattle now.

Permalink Reply by George L on April 28, 2014 at 10:29am

Hello Michael,

Thank you for your answer. I've never used bending before, but I'm trying through different examples, so I don't really know the whole idea.

Actually yes, what I'm trying to bend is a series of small parts connected together, that are able to bend forming a spline and also lengthen/stretch under slider movement.

The solid object, eg. the cylinder, focused more on the idea of understanding how a solid can bend through Kangaroo rather than only a curve.

Permalink Reply by Michael Makris on April 28, 2014 at 10:55am

George,

I'm sorry, I still don't understand entirely what you're referring to. When you say

The solid object, eg. the cylinder, focused more on the idea of understanding how a solid can bend through Kangaroo rather than only a curve.

What solid object are you referring to? Do you mean the solid parts within the spine?

As to that spine of smaller parts, what are you hoping to do with it? So far, reading your description, it sounds like you're concerned with geometry more than anything else. You can define a spine of smaller parts that follow a curve and can grow with a slider without using Kangaroo. If you are trying to use Kangaroo to carry a force through that spine, however, I don't think it is capable of what you seek right now, at least using actual bending forces.

Permalink Reply by Matthias Rolf on April 28, 2014 at 10:48pm

Hi George,

I was the one to do the work on the Bionic Handling Assistant that was recently in the news. (The learning and all that stuff wasn't Festo's Engineers as 'designboom' reports, but that work was done during my time at Bielefeld University, Germany).

We HAVE a 3D visualization and kinematics computation that is well published and open source! Check out my page:

http://matthias-rolf.blogspot.com/p/the-bionic-handling-assistant.html

It's not perfect, but for our purposes worked excellently.

Best,

Matthuas

Permalink Reply by ng5 Alex on April 29, 2014 at 1:52pm