tensile membrane facade

Replies to This Discussion

Permalink Reply by Hannes Löschke on March 26, 2012 at 3:19am

I don't get what you are pointing out with "Not tense"

The meshes relaxed form depends on stiffness and resting length of the springs. Where resting length acts as the relaxed length of your nylon model mesh.

Stiffness of the nylon mesh is constant over the unstreched fabric. If you build your Kangaroo mesh, you need to take this into account. The Kangaroo mesh normally starts a a convex hull and shrinks in place, while your nylon model starts at a relaxed state and is stretched outward.

So all in all, where your arrow is pointing, the springs need to locally be stiffer/shorter than the rest in order to mimic your nylon model.

Permalink Reply by Jon Mirtschin on March 26, 2012 at 4:48am

Hi Megan,

If by the term "Not Tense" you mean you want to influence the movement at this location to make it "smaller" or closer to the undeformed line, you might have to add an "edge" element with stiffer properties (and some sort of lack of fit or prestress) to influence the degree of movement at this location.

This is pretty commonly done on structures of this nature, in reality there is most likely an edge cable at this location that this simulates.  

I don't know how to do this with Kangaroo, but Daniel or an experienced user might advise.  With GeometryGym this can be defined into the structural analysis data, or the force density solver in StructDrawRhino (that has only primitive capability) can also model this effect, http://geometrygym.blogspot.com/2010/03/minimal-surfaces-for-roofs.... has a simple example of this for a ridge effect.

Cheers,

Jon

Permalink Reply by megan sadler on March 26, 2012 at 5:14am

Hey Jon,

Here are the files, I've striped them down as they were to big to upload, hope they still function ok.

Thanks for your help.

Attachments:

• GH script_full skin.gh, 13 KB
• TEMP.3dm, 1 MB

Permalink Reply by Hannes Löschke on March 26, 2012 at 5:56am

I guess the first thing, you need to understand is that your GH model and the reference model are totally different configurations.

In the GH model, the outer edges of the fabric are pulled towards the center. Because of the large area without fix points, this is where everything is pulled towards. This strong force bends the free sections of the membrane towards the mass center.

This behaviour is absolutely correct. Here is an image of a similar point in Frei Otto's Munich Roof. This shape was generated only by studying soap skin.

Your reference models doesn't show this for two reasons:

a) the proportions of fix points vs open areas are evenly distributed

b) there are threads present that might act as ridge lines, preventing the membrane from curving in more.

Permalink Reply by Jon Mirtschin on March 26, 2012 at 7:06am

Sure, Hannes is absolutely correct.

I've attached how to emulate these ridge elements in StructDrawRhino.

Note you probably want to apply different ridge strengths to different ridges, at the moment SDR doesn't enable this.  But I'm happy to update the component so this can be nominated if you will use it.

I only applied ridges to the roof, a similar effect must be possible with Kangaroo.

Attachments:

• GHscript_fullskin sdr ridge strengthen.gh, 12 KB

Permalink Reply by Arthur Mamou-Mani on March 31, 2012 at 6:01pm

Hi Megan,

I am on Kangaroo right now. First of all, you are working with an older version (download 0.80). Also what you need is to add some "slack" to your rest length. You are probably using 0 right now as your value? If you use a bit more it will release the mesh area a bit more (0 would mean minimum area possible like soap bubble). Have a look below I am varying the rest length from 0 to 1. I hope this helps? Different cloth properties can be changed using diagonals, shear springs, stiffness...etc... see page 7 of the Kangaroo manual, it is really good!