Generative Design & Topological Optimization/ new form finding techniques


(Huskenazy) #1

Hi all!
I’m working with the topological optimization to create some forms for a structure.

I have already introduced the topic here

Then I developed this definition, which give some results, but after many attempts I’m still not really satisfied with the shapes that come out.
3d topology (76.8 KB)

Does anyone have experience with Millipede?
I would like to know how to set design spaces, loads and constraints; I state that I am a student of architecture and I do not fully know the subject, which is really very specific and sectorial, but basically I think the approach I used should be quite correct.
I have tried a lot and I have decided to address myself in this community, maybe someone is interested or better yet has experience with all this and can give me some indication or suggestion.


(Parabolic Design) #2

I think is more complex subject.
You need a tool to mix all the parameters of your project (stress analysis, structural analysis, topology optimization etc) to produce accurate results.
Take a look at Altair site.




I was also looking for topology optimization a whil ago. I ended up using Fusion 360, you can get the full version free and legal(since you are a student).
With this tutorial you have a good start point I think. Results are quite decent in a little time.

If you mean you are not satisfied with the shape that came out, what exactly do you mean? If you refer to the gnawed edges, I think that is normal. You need to remodel it clean. Like in this video:


(Huskenazy) #4

Hi @Baris

I mean, the point is not the gnawed edges. I’ve already found a solution to remodel in the final smooth shape, with the Tspline plugin for Rhino. (I thought about this because I wanted to stay in the rhinoceros environment)

The point is that I would like to have more control over the creation of the form. I would like the end result to be a middle ground between a form that I thought of and a form that the machine created. Look back a little at the workflow used to design this chair.

The objective is to do something similar for a structure.

Let’s imagine having for example 2 floors at different heights and a flight of stairs connecting them; I want to create the support structures that support them.
The typical structure would be a beam-pillar system.
The point is here: I want to create a structure that has the same function, but with an alternative shape and better performance. To do this we’ll use topological optimization.

Going further, that is where I have the first doubt in particular…
In your opinion, Is it more correct to think of a single volume as design space (thought of as a huge pillar-beam) or to think of different detached elements?
In the first way, the program is left to create the form, which is completely arbitrary and unexpected.

While in the second one has more control but the program has little to do as the shape remains within the detached elements manually modeled, so the workflow becomes more strictly ‘engineering’ and less design.

Final question: how to get a middle way??

Improve definition| panelized curved facade