Tensile Pavillon

Hi guys,

I’m a new user and I would definitely need some help my project :slight_smile:

I am working to define a Pavillon in my school project with Tensile but at that point my 3D model looks ok but need a real look inside by a pro to help me to make my definition smoother and clean because it’s my first time using grasshopper and rhino6.

(Also I work on a MacBook Pro)

I did this with grasshopper and Kangaroo but in the definition of the top is just a mirror of the bottom with some cursors changed. My question is how can I define the entirely system of my pavillon without doing two iteration (one for the top, one for the bottom and them mirror)

The second point is my mesh look very weird so I cannot unroll and flatten the entire system into simple elements due to double curved surface I try to make something like this.

(my model process using a simple cone to understand, pattern and how its works)

there is my definition : TensilePavillon.gh (35.6 KB)

Thanks guys ! <3


Welcome Valentin,

First of all, very nice documentation of what you need help with, especially for a newcomer.

Here’s an approach using Kangaroo 2:


It’s a little different from your solution, however it uses only one Kangaroo Solver.
You may notice how the bottom mesh kind of crinkles a little. If you want to avoid that you could try to relax the mesh differently, using the soapfilm method to get a minimal surface.

Btw, this was also done on a MacBook Pro. :wink:

TensilePavillon2.gh (33.4 KB)


For the patterning you need to find the geodesic curves and then unroll them. This is necessary if your plan is to actually build it beacause of the tensile stress on the surface.

Have a look on this:



Damn It looks dope ! Thanks P1r4t3b0y

I understand how the definition works, definitely smoother than mine. You build 3 hexagons curve and you moved into a vector Z positive and negative to have the top and the bottom, you faced them for the mesh I’am just trying to understand the weld mesh and weaverbird catmull process.

Thanks Arqmorelos 1 for the information. You’re right I will try to build it for my end of the semester. ^^

OK, so remember how everything started with three hexagons, and in the following process the data was kept structured in three main tree branches with separate data for the top and bottom parts.
After the individual mesh quads are constructed, you first want to join and weld them together for each of the three, individual lofts, the bottom and top ones. I don’t know if this is absolutely necessary for this case, but I like to structure things clearly.
After that you join and weld the three bottom lofts, as well as the three top lofts to form the top and bottom meshes.
I think it’s best to keep the top and bottom separate. Your process above shows that you have some sort of circular metal profile that you attach the textiles to between both top and bottom.

The Catmull-Clark subdivisions are used to smooth the top and bottom mesh and turn their polygonal, low poly boundary into a rounded one, to match your design from above better.

Here’s another example with the above mentioned soapfilm method, which should give you a minimal surface:

The approach is largely the same. The only thing that changes are the simulation parameters for Kangaroo. See how this is less crinkly.

TensilePavillon3.gh (31.1 KB)


For avoiding crinkling, another simple way is to set the rest length of the edges to zero.
(Leaving the Length input empty means the rest length defaults to the starting length of the edge, which can mean some edges end up in compression, which leads to the fabric buckling into these wrinkles. Providing an explicit input of zero here ensures that each edge is always in tension.)

Because you have a nicely structured quad mesh here, you can also sort the edges into warp and weft directions for more control over the shape and how cone/funnel like it is:


P1r4t3b0y and Daniel this is awesome. Thank you for your help I learned a lot