Question about doing optimisation in Grasshopper with Femap

Hello all the experts,

I am doing my graduation work and want to use grasshopper to do optimisation. At the same time, I want to use Femap to do the FEM analysis. So far I am at the beginning of my graduation work and I know very few about scripting.

Can anyone give me some clue what should I do to connect Grasshopper and Femap.

Plus, for optimisation(to get a light-weight structure) , can I use Galapagos?

Sincerely,

Xue

Replies to This Discussion

Permalink Reply by peter fotiadis on November 10, 2015 at 3:29am

Hmm ... I do this day in/ day out but using the AEC Bentley Systems FEA/FEM stuff (and the ones in Siemens NX for validating 1:1 complex 3d parts). It's a big thing IF you want "interaction" between the model (or more accurately: some "variant" of the model) and the validation.

Additionally IF the "variant" consists from real-life 3d components (say: real nodes, rods, connecting plates, nuts, bolts etc etc) ... well this IS not an easy-busy thing and NOT a task for a beginner in scripting.

Additionally IF the "variant" consists from real-life parametric 3d components (i.e. feature driven as is the norm in real-life AEC cases) ... well ... you are looking at the entirely wrong apps (GH/Rhino).

Hard to recommend where to start: too many topics to cover ... the "least" important of them is been expert in, say, C# (avoid Python on these matters especially having in mind the available resources in C#).

Anyway post some sketch outlining the thing that you have in mind (is it abstract? [lines, points etc]), has some "real-life" 3d geometry AND components? is it structured the proper way? [assembly/component disciplines and the likes]). And given the immediate future (look what happens in UK [BIM becomes obligatory] for example): do you have some BIM umbrella in mind for the whole project?

BTW: Light-weight optimization using Galapagos? In what sense? You mean Kangaroo I do hope.

Permalink Reply by Zhengyu Xue on November 10, 2015 at 3:43am

hello, Sir,

Thank you for your reply.

My idea is that to build the geometry in Grasshopper so that I can change the parameters (shape, section size, height, external load) easily . And by connecting Grasshopper with the FEM software I can do all the analysis in Grasshopper . Then get the FEM result as Fitness for Galapagos to do optimisation.

I have attached a pic for my conceptual design, it would be quite time-consuming to change the parameters manually in FEM software.

If I do not use external FEM software, can I do all the analysis with the existing plugins in Grasshopper? Can Kangaroo and Karamba make this?

Sincerely,

Xue

Permalink Reply by peter fotiadis on November 10, 2015 at 4:22am

Sir: are you kidding me? call me Peter, Friend, Enemy (or use my real name: The Lord of Darkness, he he): we are all friends here since we all live in that yellow submarine.

Jokes apart:

1. This is a truss right?

2. Trusses are NOT made via lines ... they are made by real-life components like balls, rods and other mysterious (and maybe ominous, he he) paraphernalia.Good news for you: lot's of C# stuff around me that do that (but they are not exactly "entry-level").

3. PRIOR talking to ANY FEA/FIM thingy you need to address clash situations: I mean IF a given node is doable or not (because lines they don't rise clash issues ... but rods/struts/tubes/cones do). Good news for you: lot's of C# stuff around me that do that (but they are not exactly "entry-level").

4. Then you have to use some real-life (or at least some "realistic") components like the ones found in, say, a classic MERO "ball" system (and especially the adapter cones between the balls and the tubes). Or at least "some" of them that outline a "realistic" truss.Good news for you: see above.

5. Then you could validate the whole structure AND the parts VS structural loads: I mean there's absolutely no meaning "doing the whole" without taking into account the load bearing capability of the parts. For instance, say, what happens if the geometry (i.e. the topology) is "capable" but a given bolt is weak? That sort of stuff.

6. Now ... this is Academic ... but following the "abstract" way (I don't care about bolts because I'm a student)... this could teach you the entirely wrong way to use FEA/FIM for validating any structural ability of ... anything. And besides FEA/FIM is used for making the damn thing in the real-world ... and that involves (unfortunately) "some" bolts and nuts.

I can arrange a (rather long) Skype session for a demo of all the above ... but first I strongly advise to post here a finished thing (in terms of 3d component geometry) ... and THEN we can examine the whole strategy: what to export, how and especially what could be an "interactive" (both ways) protocol/strategy in order to give the green light for that truss.

BTW: Kangaroo is a physics engine and as such it's used as an abstract "shape" finder. I have no idea what Karamba does ... but always have in mind: BIM things ... are BIM things (meaning that without a serious BIM umbrella ... don't go out when it rains).

Permalink Reply by peter fotiadis on November 10, 2015 at 4:46am

My best proposal:

Start posting "states" of your definition here and I can "insert" gradually C# stuff that prepares things for the big event.For instance a C# that does trusses, another that puts some "realistic" 3d things there, another that checks clash issues etc etc etc.

This MAY become a bit complex from a point and on (especially with regard instance definitions and other mysterious things) but no pain no gain.

BTW: what is your experience with C#? (can you - at least - read code?)

he, he

Permalink Reply by Zhengyu Xue on November 10, 2015 at 5:14am

Dear Friend, Lord of Darkness, Muscular Princess,

This is a tensegrity structure. I managed to make a parametric geometry model in Grasshopper. To do the FEM analysis, I need to define node, element, external load, prestress, material, property, sectional size and make a button to do analysis....These are what I assume to be able to do the parametric design. A lot of stuff and I feel it a mess because I do not have the skill yet...

That's why I am learning C# and other plugins for Grasshopper. About C#, I am a rookie...Almost no experience before, but I am not afraid of the unknowns.

Is the description above clear for you? I am very grateful for your kind help!!

Sincerely,

Xue

Permalink Reply by peter fotiadis on November 10, 2015 at 5:39am

Dear Xue (PS: alternatively you can use my other real-name: Lord of SardineLand).

Well ... where's that definition? BTW: Connectivity data with regard anything is peanuts > forget that for the moment ... the big thing is the "parts".

And what "level" of reality you wish? (assuming - just a hint - that the def or yours makes ... er ... hmm ... lines).

Tensegrity? meaning that "remotely" mimics a tensile membrane ???

Not afraid of the Dark Side you said? If so be prepared for the ultimate PANIC (he, he + he, he) but long is the path (and hilly).

best, Lord of SardineLand, Prince of NoMansLand

Permalink Reply by Zhengyu Xue on November 10, 2015 at 5:56am

Dear Juicy Sardine Prince, Faceless king of the Ash,

Tensegrity is a combination of isolated compression bars inside a net of continuous tension cables.

I've already got a book "Femap scripting language". I think what I need to know is the basic logic sentences in C#, and how to start scripting.

I have attached a pic for the related buttons in Grasshopper. I want to programme something with these functions, to do the FEM thing in Grasshopper.

Best,

King of Pain

Permalink Reply by peter fotiadis on November 10, 2015 at 6:46am

Dear King

Did or did not asked for the definition? What to "program" (or start something, anyway) with thin air on hand? Or maybe you mean: get thin air output from here and mix with thin air output from there > stir > serve chilled.

BTW: these are not buttons they are components from some plug-in (what plug-in? Masaaki (ricecooker) stuff?)

Permalink Reply by peter fotiadis on November 10, 2015 at 7:59am

BTW: Is it possible for the Lord (of SardineLand) NOT knowing what tensegrity is?

Makes Me wonder, he he.

And since I have a strange feeling that you work actually for DARPA (hence the "reasonable" reluctance to post that !$%$! def of yours) ... here's a T thingy straight out of planet Utopia (or Zorg).

But can you tell me why is pure utopia? (*)

PS: this is a world's first: using components instead of C# (I always do the opposite, but I'm a bit disoriented right now ... blame Tequila, what else? he he).

(*) answers to:The Lord, Apartment 666, District 9, North Pole.

Attachments:

tensegrity_from_Planet_Utopia_V1.3dm, 158 KB
tensegrity_from_Planet_Utopia_V1.gh, 33 KB

Permalink Reply by Zhengyu Xue on November 20, 2015 at 6:17am

Dear Lord SardineLand,

Thank you so much for your kind help.

To be honest, sometimes I feel a little bit hard to understand what you mean...I am not reluctant to post my work...It's just I have no work so far..With no skill, no work..Plus, you can see the pic I posted in the beginning of our discussion. The roof structure is different from the traditional Tensegrity, it's not pure tensegrity and there are bending moments in the compression struts..

For form finding I want to first make a flat square net, add nodal force(upward and downward) on points to get the shape of the tension cable net after loading. Then based on this shape I make the geometry of the tensegrity. (Do you feel this a reasonable way?)

In the attachment I sent you my error files...I am also new to Kangaroo..(Holly Jesus! I chose a graduation topic that I need to learn every possible software to reach my goal!!) ..

As much thanks as the sea water to you the Glorious Kangaroo,!!

Zhengyu

Attachments:

error.3dm, 73 KB
error document.gh, 19 KB

Permalink Reply by peter fotiadis on November 20, 2015 at 11:58pm

Dear King of Oblivion, Prince of Despair

...I feel a little bit hard to understand what you mean... that makes 2 of us, he he.

First things first:

http://www.grasshopper3d.com/forum/topics/i-want-to-make-this-form-...

Other than that:

1. Tensegrity is a "static" thingy in the sense that you use some module (let's call it "mode") and repeat. Creating some code that does INVENT new modes for T trusses (Pulitzer/EMMY/Nobel on sight, he he) ... I would strongly suggest to forget that THIS VERY MOMENT.

2. Applying some T "mode" on something (see my examples in the above thread where I use surfaces for the T nodes) is another animal. If you intend to use Kangaroo to "relax" that something (NOT the T itself) well ... you can do it but has nothing to do with T.

3. The Kangaroo def provided is a "way" to test the "rigidity" of the T in use. It's a "post-processing" thing NOT a T solving thing.

4. I have a terrible feeling: are you saying that (a) without knowing a thing (or two) from C#, (b) without knowing K1/K2, (c) with a limited GH experience ... your goal is to write down from scratch a FEA ("Femap") thingy that ALSO does node "relaxation" ? If so ... well ... what about sky diving (without parachute) or that classic Russian roulette "game"?

PS: shown double tetra (classic) and XFrames (classic) T trusses applied in open and closed surfaces.

But of course these are abstract stupid "arrangements" utterly out of question in real-life: read CAREFULLY the discussion in the thread provided above AND also study the 3dPDF attached (with a system out of many available) in order to get the gist about what real-life means (Note: EVEN if no real-parts are used ... the node calculation is different from the abstract "star" connections pictured above - by "star" I mean that cables meet at a single point in space without any "offset" etc etc).

Moral: Seppuku

Plan Z: Skype ASAP

Permalink Reply by peter fotiadis on November 21, 2015 at 1:07am

BTW: your K def has NO anchor points > a black hole absorbs all the things > the so called "event-horizon" point > not bad at all in some parallel universe, he he.

BTW: make a brake; spend some time on these in order to understand what K does (your case COULD BE the Bend_a_3dgrid_... OR forget grids and use K on a surface and then use the relaxed surface division points).

BTW: a XFrame T truss is this thing pictured (BUT if the X frames are NOT planar then you need a complex system that assures rigidity on these, otherwise the system would collapse):