Simulate Early, Simulate Often... In Rhino
Firstly, I am amazed at how simple Scan and Solve is to use.
Well done on that front!
I would really love to see the ability to dynamically change the geometry over several iterations based on the FEA results I am seeing.
Some people are calling this "evolutionary topology". Others call it "structural optimisation". I first saw an example of this back in about 1999 and it blew me away...
Here are a few examples:
Yes, of course. A proper technical term is "topology optimization". It is not that difficult to implement. In fact, it can be done using SnSScript. However, using such technics is not all that simple. One needs to know how to control the process, when and how it fails, etc. Simplicity (of use) is anything but simple! We do have something related in the works. So stay tuned ...
Well I just tried out BESO3D which comes from RMIT university in Melbourne. I had high hopes because it is compatible with Rhino. Unfortunately it (BESO3D) doesn't seem to work very well for me at all. I was expecting to see the new structure evolve before my eyes. But it seems to create a separate txt file for each iteration and you must go into each one and check them individually. I started off with a simple structure (a solid cone) and it justs wouldn't work. There are limited commands, and only point forces can be applied.
I don't have a background in numerical analysis, but rather materials science. A lot of which is physically testing materials. And in real life, all loads are applied through faces. So that is a great feature of SnS.
I really hope you see that there is great opportunity for your company to bring something along these lines to the market. Personal computers are powerful enough now to be able to do the computery stuff.
In the mean time I am going to try out Optistruct, but if it is anything like the other FEA programs that I have tried, I won't be that impressed with it.
The nice thing about SnS is the speed and simplicity of it. I would love to see a topology optimisation feature, with each iteration going to a new layer within Rhino. That would be cool – and yes I would pay extra for that (or even if it means buying two separate programs).
Could you at least include an alpha (transparency) as part of a new legend/gradient so that instead of low stress areas appearing as blue (for example), they disappear completely from view? That should be very simple to implement I would imagine...
I would like to add my name to the list of those eagerly awaiting this as an option. I had to choose between Scan and Solve or Frustum. I went with SnS because of the ease of use and integration with Rhino, but I was really looking forward to finding weight savings structural designs.
I am wondering how you are getting along with this a year down the track?
I found solidThinking Inspire software that can already do this, but quite frankly their support isn't great.