Shell form finding

I have a question related to gridshell design,
Should the base surface geometry first form found before applying the grid on it ?
I am wondering if I can do form finding -or maybe optimization- of Geometry that decreases the internal stresses, bending, avoid inextentional deformation ?
And the second part of the question, is it better to have geometry totally under compression-like funicular resulting from hanging curves or mesh- or it is better to introduce some anticlastic curvature to have some kind of stiffness ?

Hi @anon95022132,

This is kind of a chicken-before-the-egg and vis versa question. Either scenario is possible really!
You can start from a geometrical model and optimise its topology (example 1, example 2), or generate the shell geometry with a forced-based simulation (example 1, example 2).

Sure, I guess Karamba3D can do at least some of these things, if not all! There’s also millipede and don’t forget about Kangaroo (which already comes with Rhino). I guess combining Kangaroo and Karamba would be the way to go here.

Hm, in Ancient Rome, it was surely best to focus on compression, since your building materials for monumental buildings (e.g. Pantheon, aqueducts) were basically limited to stone and opus cementitium (precursor to modern concrete). Thus keeping things “simple” and focussing on funicular surfaces was probably best, since their meridional forces are compressive only and only the hoop forces at the base are tension prone.

However, many modern structures (e.g. Vladimir Shukhov, Frei Otto, etc.), especially tensile structures for shells are interesting because their elements carry only tension and no compression or bending. Synclastic or anticlastic matters less here.
The Romans basically used the reversed concept to generate compressive structures.

Since, you’re writing about applying a grid, I’d maybe focus on tensile structures!
Again the grid can be applied later or be part of the simulation, for instance for evaluating cross sections.

1 Like

Thank you for this very detailed answer! That is more than helpful.
Regarding the second part of my question, I forgot to mention that it is about timber bending active gridshell, so I mean if funicular geometry in pure compression will be a good solution -I will apply a 3 way geodesic gridshell- or it would be better if I added some anticlastic curvature for the sake of stiffness?

I’d recommend buying this book first thing. It pretty much covers everything you need to know. Both in terms of structural principles and computational modelling:


Hi Anders
Yes it is a great book, I did actually but since I am still reading it I am still confused about some concepts.
Thank you so much!

1 Like

I don’t know if the same principles really fully apply here, because of the bending part.

Generally speaking, wood is kind of a very mediocre material, since it’s resistance to compression is rather weak (vs. concrete) and to tension also not super good (vs. steel).
Usually compressive funicular structures have less material on top - to save some weight - and are thicker at the bottom, because here the forces are greater due to the weight of the whole structure. Already this could be a challenge, since you’ll only be able to bend strips with a certain maximum thickness, which means that scaling the strip cross sections towards the bottom is challenging.

Fortunately, wood resists compression best, when the forces run parallel to the wood grain, which kind of would be the case for bent strips, but it’s hard to say how the bending moments play into this.

I guess the 3-way gridshell can do more for the overall stiffness than anticlastic curvature can?

1 Like

Yes , in this case the stripes will be double layer- 6 stripes - on these parts or more like in minna no mori gridshell.

I totally agree to this, it is more clear now to me.
Thank you very much !

1 Like