How to Reorder a Data Tree To Create U and V Polylines for a Pavilion?

[rhino38 (rhino)]

I have been stuck trying to figure out a way to create polylines in the U and V direction for this patch surface due to how the points are ordered.

The goal is to create straight lines based on the patch to create straight purlins/rafters for fabrication. I need the U and V directions separate because one set of purlins is under the other. The image below shows what I have so far, which are curving, not straight purlins, intersecting each other, instead of one under the other.

So my process to make the polylines was getting the intersections, starts, and ends of the isocurves as shown below.

The problem is how they are ordered as visualized below. To get the U direction’s polylines, I had to manually type in their index. On the other hand, the V direction polylines are impossible to get due to the weird order.

1. Is there a way to reorder these points so that I can easily get the U and V direction polylines?
2. Is there a less manual way to get the points instead of what I did, which was manually writing weirdly ordered points? (but fixing number 1 should mean number 2 is no longer a problem)

Here are the .3dm and .gh files.

Pavilion Base for Grasshopper Forum 3.3dm (114.5 KB)

Pavilion for Grasshopper Forum 3 Order List.gh (28.0 KB)

Thank you very much!

[martinsiegrist]

Are you planning on fabricating this structure with straight wooden laths? Like a grid shell?

[rhino38 (rhino)]

Hi again Martin, we’re planning to use straight bamboo split beams and yes kinda like a grid shell, though there a big main and minor arcs just not included in the files

[martinsiegrist]

I’ve only done one grid shell concept with Kangaroo. But based on all the things I saw doing this many years ago, I really don’t think your process is the way it is done. I think you would have to start with a flat quad mesh and sort of simulate the bending force of the wood in a Kangaroo definition. The flat quad mesh is bent / pushed into a doubly curved shell. This is form finding.

Looking at your topics, it seems you’re doing the opposite. You defined the shape with a patch over five manually modeled edge curves. I don’t think you’ll end up with the right shape the way you approached it.

Here’s a link to a simple gridshell example by @DanielPiker: Bending-active structure using kangaroo2 - #4 by DanielPiker

And again. Put links to previous topics about the same overall project.

[rhino38 (rhino)]

Thanks a lot for pointing me in a better direction @martinsiegrist I’ll try your suggestion out and play with Kangaroo and yes sorry I forgot to include the list to my previous posts.

Also yes, the patch surface was defined by the edge curves because in reality they are big arcs attached to existing concrete pedestals. I guess if I’m going to use Kangaroo, I would still need to have those edge curves as some sort of constraint.

For anyone else reading this, here they are

1. https://discourse.mcneel.com/t/creating-a-single-smooth-roof-surface-for-an-organically-shaped-pavilion/217943/12
2. Retrimming the Surface Division of an Untrimmed Surface - #13 by rhino38

[martinsiegrist]

@rhino38

Also yes, the patch surface was defined by the edge curves because in reality they are big arcs attached to existing concrete pedestals

The foundations are existing, that’s clear.

The arcs too?

[vincent1 (Vincent)]

You might also have a look at shapemap

check at about 27 min in

[martinsiegrist]

The beams of Centre Pompidou Metz are not flat pieces bent into a grid shell. The beams are massive cnc machined laminations. So not the same situation…

[Volker_Rakow]

Reorder a Data Tree To Create U and V Polylines VR 1.gh (28.9 KB)

Regardless if a patch or a found Kangaroo form is the correct surface to be using here, this is how to get polylines from U, V and perimeter curves.

---

Mistake in the above. The foundation corners are skipped.

[martinsiegrist]

@Volker_Rakow

found Kangaroo form

If the form is found with Kangaroo, you don’t need to rebuild any isocurves. In that case the mesh edges are the axis of the laths

[Volker_Rakow]

@martinsiegrist

If the form is found with Kangaroo, you don’t need to rebuild any isocurves.

Then show, because I have no idea how any of that works.

[martinsiegrist]

I already shared a link to a simple @DanielPiker example above

Bending-active structure using kangaroo2

Yes, though the short edges around where it is trimmed aren’t ideal. It keeps things cleaner if you can cut the grid along its main directions or along its diagonals. There’s an example file similar to this here

It’s a simple four sided grid shell. The mesh is transformed from a flat orientation to double curved shell.

The grasshopper definition of my project from 10 years ago unfortunately does not run anymore due to various plugins.

If in the OP’s project modeled surface the isocurves are used to extract the axis of the laths and the laths are not bent laminations, how is this going to be assembled? A grid shell is usually assembled in flat state. A shell like Centre Pompidou Metz is not a true grid shell à la Frei Otto.

[rhino38 (rhino)]

@vincent1 Thank you very much for the suggestion

[rhino38 (rhino)]

Thank you very much for this @Volker_Rakow

[rhino38 (rhino)]

That is a very cool model. In our case, we’re not making a true grid shell.

[Baris]

I think the Frei Otto priciple that @martinsiegrist mentions, should be essential for every person in the aec sector designing (gridshells): Form follows force. The beauty from this structures comes from the self organisation of the material and not from the designer.
Ignoring this principle in real life often leads to unbuildable, ineffective and ugly structures.

[Volker_Rakow]

A few messy changes to get the foundation points included and the start/endpoints of the U/V polylines aligned on the original curves:

Reorder a Data Tree To Create U and V Polylines VR 3.gh (45.1 KB)

I have to back up the other comments in this thread though: without some bend-active form finding, most of this structure is wishful thinking.

[rhino38 (rhino)]

Thanks a lot @Baris and @Volker_Rakow . I totally agree with the form follows force thinking and have shifted my approach by using kangaroo to generate the surface instead. Just having trouble generating the form with the perimeter arcs as the constraints, but that would be for another thread. It’s just another learning curve to get through as it is very unfamiliar to me. Thank you so much for pointing me in the right direction and thanks again @martinsiegrist for the same.