edit isoCurves, and designing exo skeleton

Dear All,

I have been trying to design an exoskeleton structure to wrap around a complex internal void of my proposed massing (building) - that will have an inner and outer exoskeleton. The resulting mass was a result from a subtraction with a form created using tsplines). With a lack of grasshopper knowledge and limited time, I have been trying to do this manually which is proving to be inefficient (and in addition the iso curves are a mess from the previous tsplines mass).

It would be really good if I could create the exoskeleton in grasshopper (and to have better control of making amendments and variations) but I don't even know where to begin so would appreciate if anyone has any suggestions how I could go about this alternatively.

I have attached a few screen shots of a previous iteration and the current work-in-progress (both created manually).

As mention, any advice would be very much!

Attachments:

Replies to This Discussion

Permalink Reply by KC on May 11, 2017 at 2:48pm

This is the overall mass.

Permalink Reply by KC on May 11, 2017 at 2:49pm

Permalink Reply by peter fotiadis on May 12, 2017 at 3:05am

ExoW is very tricky (reports engulfing issues and stops playing ball quite frequently [even in cases where there's no such apparent issue]).

Other than that:

1. Creating a "skin deep" lattice out of things like the above (say using the Brep edges + some "interconnecting" curves) is OK but only if this is some sort of decorative/artistic/academic stuff.

2. But the problem is that is not ... thus avoid at any cost this catastrophic way of thinking (tres a la mode these days, truth to be said). Why? because this is as far from engineering as a Skoda is from a Bentley. For instance imagine the cost of bending IPE/HEB/HEA/C members in order to approximate a given curvy edge. Then imagine the bespoke nodes etc etc. Then imagine the skin (avoid rain). Then go to Louvre and spend a week studying the Pyramid.

3. Use Mesh Machine and achieve a meaningful "rigid" (kinda) triangulation (planarity by default) and THEN attempt to convince ExoW to work (good luck).

4. If ExoW can cut the mustard ... well then there's only a trivial thingy left: spend zillions and create the "liquid" nodes. Why?

5. Plan B: skip 4 and create a rigid W depth truss with good old triangles.

Permalink Reply by peter fotiadis on May 12, 2017 at 3:24am

BTW: This (MERO KK ball system) costs 100 times less than the above (and the skin costs 1000+ times lass).

Permalink Reply by KC on May 12, 2017 at 10:18am

Dear Peter,

Thank you very much for your response. I forgot to mention that this is for academic study - I am currently an architecture student, but note that your points are valid and important.

I have attached an axo section of the previous iteration, made manually, which shows that there is an inner exoskeleton and outer exoskeleton (one more fluid and other more rigid) . I am in the process of amending these exoskeletons to allow the structure to vary dependent on certain 'attractor' key spaces, but is there a way in grasshopper to achieve this, to allow easier control of making amendments and/or variation, as doing this manually is much less efficient?

Would you suggest the brep edges and interconnecting curves method to be the advised route to go (rather than ExoW with its related issues)? The curved parts would eventually be 3D printed to form a physical model.

I look forward to hearing back from you! Thanks!

Permalink Reply by peter fotiadis on May 12, 2017 at 10:46am

Well ...

1. Spend some time studying the excellent solution that Laurent did (a "bit" expensive to do it mind, he he).

2. Your lines of thought are formulated even when on some Academic environment ... and the bad habits die last as they say. This means that ... well ... the adaptation to more realistic (and meaningful) things later on ...

3. I can easily provide some solution (ultra expensive in real-life) to do what you want but this would be carried over solely via C# code (NOT good for you especially when this would/could be used in some sort of Thesis). To make a very long story short the "curvy" parts is highly recommended being tubes ... and the "liquid" nodes required ... well ...that's another animal UNLESS one could accept an Academic over simplification by using balls of a slightly bigger R than the adjacent tube "struts" (whilst the "iso curves" [per BrepFace] would use an even smaller R and inserting crudely into the Brep Edge "main" curves). But since actually we are talking about a secondary random "lattice" per BrepFace the "iso curves" are actually stuff made via the Surface.ShortPath Method (not sure if this exists as GH component) using random points where their number is proportionally to a given BrepFace area (freaky stuff, trust me). This yields a "uniform" random secondary "lattice" in accordance to the whole "random"/liquid appearance of the T-Splne Brep.

The above a bit naive approach (obviously out of question in real life) can yield a solid thingy if we unite all the parts and bits (Rhino takes ages to do that if we are talking big numbers of Breps) ... thus some 3d printing is doable.

In other words we do a MERO "approximation" by hoping that no German guru reads this thread, he he.

We can provide a Frankenstein type of "pro" connectivity as well: since a Brep is actually kinda a Mesh (with regard connectivity of vertices, edges, faces et all) making the connectivity trees required is not a big deal (GH has the Brep Topology thingy as well).

But the whole solution could be a black box to you: if this what you want?

Permalink Reply by peter fotiadis on May 12, 2017 at 11:55am

BTW: This is what I would use (with A LOT of changes in order to work against BrepFaces [i.e. trimmed stuff] and in order to deal with the fact that the Curve.ExtendOnSurface appears a bit temperamental) with regard the secondary "lattice" per BrepFace.

Since this lattice (supposedly) supports the skin (glass?) it should be quite dense. Code is required in order to "divide" a given BrepFace into curvy pieces.

But this leads us to that (or to some equivalent frame-less system also with double curvature) :

which is the most efficient way to blow zillions for absolutely no reason.

Attachments:

ShortPath_entryLevel_V1.gh, 121 KB

Permalink Reply by peter fotiadis on May 13, 2017 at 12:02am

FYI: Just finished a long Skype session with a friend (Pilkigton UK: masters of glass) with regard a double curvature double layered planar system for some future project that I refuse to do. He said that 2++K (per M2 !!!) appears reasonable to him (it's a matter of perspective I guess).

That said term planar glazing these days doesn't necessarily mean ... er ... planar class. Anyway the above is probably the best for you: primary tubes (edges), secondary tubes (short paths), Karma (a must) plus the cheapo glass > job done > never again.

Permalink Reply by peter fotiadis on May 13, 2017 at 11:25am

BTW: This is what I would do, save millions (maybe zillions) and avoid some nightmare(s) later on (Why God did I that thing? can You forgive me? etc etc).

1. Get a boxy thing + a TSpline blob and do something:

2. Find all the curvy BrepFaces and do a variety or preparation work (for plan A or B):