Hello Guys,
i want to experiment with Netstructures like this:

And i did the following script: (16.0 KB)

The behavior is not like I suspected. Its definetly not physically correct. Has someone a hint how it will workout with specific input properties?

Best regrads

Hi Alex,

Which version of Rhino are you using?
I notice this definition is made with the old version of Kangaroo.
I’d recommend using the new one, which has been included with Rhino since version 6. You can find some example files here.

Also I see you are using these huge List item and Merge components to organize the data - this shouldn’t ever be needed - you can use the Geometry input of Kangaroo to pass lines and keep them in the right order.

For the simulation itself, lines will only be connected if they share a point. So if you have one curve meeting the midpoint of a second curve, that second curve needs to be split into 2 segments so that it has a new vertex at that junction.
Finally, to make sure all the ropes are in tension, you need to set the rest length low enough. One way to guarantee this is to simply set them all to zero, which means they can only ever be in tension.

It ought to be something like this. (41.9 KB) (16.1 KB)

1 Like

Hi Daniel,
thanks for your feedback. I’m working with the Version Rhino 7 since some weeks. I didnt know that i was choosing the old components. I used some old examples which i still had. But thanks for the link, i will try it.

If I understood it right, i’m using the output directly from the kangaroo 2 solver, like in the example from David S. Mavrov? Or is there also another way?

Thanks a lot for all your support. I will dive in it and will share my goals. Thanks

1 Like

Yes - like the example @davidsmavrov shows is good.
Because the inputs are organised into branches with the Entwine component, you can also get the original groupings of the curves from the output by exploding like this: (11.7 KB)

1 Like

Hi David,
thanks a lot. The examples are very helpful! I tested also the second one. I will start to play around with, there are outstanding geometries possible.
Thanks and best regards

1 Like

Fun little side note: I fairly recently designed this children’s climbing net (that wraps around an old water tower). Using K2 for the form finding, GHPython for generating the topology, and Galapagos to dial in the settings (which was mainly to get the cables that attach to the tower right):

The largest challenge was acquiring and fulfilling all the safety standards for playgrounds, lots of this:


Hi Anders,

nice project, that looks very interesting. I’m also starting to think about a structure to climb. Did i get it right, that you used galapagos to optimize the struture with the restrictions of the playground industry? I didnt get the idea of the analysis. Is this the output from galapagos or are you doing a kind of structural FEM analysiys? I would be happy if you want to share some information, if not thanks a lot for showing your project. Best regards Alex

1 Like

Ah yes, sorry that was indeed a bit confusing.

No I used it to dial in the cable lengths/strengths of the 15 cables connecting to the tower. With the fitness criteria of minimising the total distance from their start-points (i.e. at the outer nodes of the outer quadrilateral grid cables) to a plane fitted through these points (i.e. having these nodes sit on a plane):

The analysis I posted is the offending downward facing entrapment angles. Basically, playground equipment can’t have any two downward facing elements meet at an angle smaller 60 degrees (according to the EN1176:2017 standard, if I recall correctly). Another important requirement here was that a “cylinder” the diameter of child isn’t allowed to fall through the network, without being stopped by cables below (see these red cylinders):

I forget exactly what we did for the structural analysis, as I didn’t manage this part. But if I were to do another cable network project, I’d definitely implement K2Engineering for this (@c.brandtolsen works in our inhouse engineering team).


Maybe I have gone nuts, but looking at the spirals and the patterns they made… This resembles a fibonacci pattern I have seen before. Maybe there is a simple algorithm analogue to the galapagos brute force method… In any case, remarcable work as always Anders! :slight_smile:

1 Like

Cheers David. Yes there’s definitely some Fibonacci-action going on in there. For both the original design with the diagrid and the final non-child-killing design, I generated the topology by recursively going from the outer grid and moving inwards, connecting nodes along the way. Here’s an example of a flat diagrid with one of the rules I came up with applied:

And a topology diagram of the resulting inner “fans” in the final design with the regular outer grid:


Hi Anders,

very nice project. Thanks for sharing. I will have a look at K2Engineering.

Best Alex

1 Like

Is the final result achieved only with catenary action or did any of the ropes have to be pre-tensioned?

I’m guessing if pre-tensioning was necessary, cables would be required in lieu of rope?

All the ropes are pre-tensioned (with zero catenary action). With the outer grid being the primary cables, the inner grid the secondary, and the fan grids in between the tertiary. Unfortunately I wasn’t involved in the fabrication and construction, so I’m unsure how accurately the built result matches the form found design. The ropes do vary in tensile strength and diameter as far as I know/recall.