I’ve been searching for a way to extrude/pipe a triangular grid as structural support for the facade of a project of mine. I have managed to put together something like a “Frankenstein” of a script from some suggestions from here:
The problem is that because the surface that I am applying the grid to is a free-form one, I get always at the ends of the extrusions unfitting and ugly results. Does anybody have an idea how to get all the continuous lines joined together, so I can maybe rebuild them a bit later, to make them smooth, and also how to apply rectangular piping to them so I get some nice-looking joints also for the interior renders later?
Picture from the continuous seams I think, might help the joints to look better
I assume that you know what a Mesh Normal is. If yes … first issue is that torsion free beams little matter. Then … the node join is rather a nighmare to design and implement on site (general case: any blob type of envelope).
So … I would strongly suggest Plan B: stand alone frames on per Face basis. This approach - some years ago - was a bit unthinkable … but after the Parametric delirium/pandemonium AND “some” real life issues here and there … people start to understand the obvious advantages.
That said nothing beats a Mero KK System … but that’s another animal.
thank you very much for your reply! It means a lot that somebody is trying to help me out in this case!
It turns out that this is not working for me, because after I baked the surface it turned out that it is actually a polysurface, after I turned it into a mesh this happened…
I guess I have to re-mesh and divide it into equal counts of mesh faces. Well, I tried to do that and still didn’t work as I was expecting it to work, maybe I am doing something wrong…
I guess you are right about the site implementations of such a structure, but how do these frames get connected? Now I see there is a rather noticeable big gap between them…
Anyway, I was looking more for a single rectangular profile for the structure of the facade and not for divided triangular frames, since the beams are easier to explain and this is a university project and will never get built
Now it is also getting also a bit more complicated, because I wanted to apply more detail to the facade by adding a seam that moves along the volume, which divides the previously single surface into two faces. And I would like to apply this triangular grid structure to both of them. Do you think you can help me with this?
The gap is user defined. Is there for applying various sealing things (from MIGUA type of stuff to other complex things).
The stand alone frames are bolted together. Obviously with no gap … well … in real life that’s impossible (construction tolerances, thermal expansions, seismic activities and the likes).
Any envelope is designed having in mind the final skin. This means a rational harmony between the LBS and the modules that seal the structure from the elements. 1:1 real-life details are requred PLUS a bottom to top design mentality: first you deal with pistons/valves and THEN (and only then) you can think about the engine. Most Architects have absolutely no idea about what I’m talking about.
For instance let’s assume that you are after the finer (and most expensive) solution known to man: carbon fibre MERO W truss (you can barely see it) + planar glasing > this dictates a lot of design constrains and has nothing to do with the stand alone concept provited. Or go after some tensegrity truss (even more expensive).
The “top” skin must be airy and not constrained by “walls” … otherwise the aesthetics would suffer. 99.99% of the AEC things done these days are visual atrocities. This means a skin separated from the walls. So do a proper Mesh and mastermind some policy for the “sides”.
In plain English: AVOID doing a blob (as a whole) and then trying … blah, blah.
Even I.M.Pei (the greatest ever etc etc) couldn’t do a pretty blob (but a blob is ugly by definition anyway).
Consider tensile membranes: this IS the Ducati of the AEC world.
All in all: Forget anything: do some realistic 1:1 details related with the envelope as you were in the Final design phase. Take the construction (and cost) into account as well.
Tip: If you fail to prepare, prepare to fail (as we say in racing).
Bad news: Designers are born, not made.
Ugly news: It’s unlikely that you can anything realistic without code.
ResetNowForEver: avoid going after stupid/naive/ugly function follows form things like this:
PS: Almost(?) any(?) Mesh can be torsion free … but this requires various things far and away from the scope of this thread. See the 2nd test Mesh used:
first, I have to thank you for the broad insides of this! As it turns out after I read your comments I have created something that I wasn’t prepared to finish structurally on my own. Anyway, this is why I am here, mostly to learn. If you never try it, you never get to understand how hard actually stuff like this gets to be made. Because is a university project, I mostly want to get the best result possible, for the time I have already invested into this, which is already a lot. I had to do all kinds of research and writing in order to get to do this envelope. And am also working on the proportions of the facade/envelope, to fit them with the floor plans.
I am not a believer in the form follows function, I think form follows investments.
Thank you very much for this script, I have to look at it more deeply in order to fully understand the functionality. I’ll keep you posted on how the process works.
Some day you’ll pay a very heavy price (I.e. when the Parametric - form related - nonsence would come into an end - nothing lasts for ever, mind).
BTW: in our Times where money is virtual (and everything is fake and/or fabricated accordingly) the term investment has a very relative short term meaning.
Tip: try to leave traces behind you that are worth a second look.
BTW: I have a C# that does what you asked for: I.e. rectangular profile struts using planes defined - per edge - in Y with the edge adjacent Faces unitized normals sum and in X with the CrossProduct of Y and the edge direction.
But … without solving the torsion situation this approach is 100% pointless. Of course you can “address” this by using 2 planes, 2 profiles and a Sweep1 Method … that yield a twisted unrealitstic/stupid strut.
For instance if the Mesh is planar … well even kids can do that:
Obviously the chaos is bigger in naked Edges (where only one Face Normal is around)
Plan C: abandon ship > do instead a proper MERO KK type of truss (unless you can relax the Mesh as a torsion free one). Not convinced? Just think the stand alone frames solution. Imagine beams from, say, 5 * 18mm marine plywood glued sheets (the number should be always odd due to the required connecting L plate). The top/bottom faces are tapered. Can you tell me how to smoothen them without spending a year and half?
For instance using lot’s of C# lines (about 1500) we can do this … that look “cool”. But in fact is a naive/stupid solution that has very little in common with some engineering worth the name.
