PDA

View Full Version : 2012 How to create complex Roof Sheds with warped slabs



opesch
2012-06-14, 05:22 PM
Hello,

I need your help because I'm really stuck with that!

So if, perchance, the architect who design that or someone else has some advice for me, I promise to drown him with Belgian beer :beer: (or maybe conclude a consultancy contract)! :mrgreen:

If you need some detailed illustrations these are in the attached PDF. Figure 3 showing the completed Roof done with 1st method is also joined as PNG

Project Requirements:

We need to reproduce the waved roof as per the architect’s design (using their own unknown program) within our own BIM model (preferably in REVIT 2012).

Roof Shape Details:

The roof consists of concrete slabs waving South to North and spanning East to West between two radial axes (B & C on the following figures) and four top and bottom sloped surfaces (yellow & cyan triangles on figures 1 & 2 respectively). These rise from South to North and are folded at one of their diagonals.

Slabs form "V-shaped" concrete sheds as shown in figure 3. The bottom level of the slab wave corners are dictated by the intersection of those surfaces and their respective main folds with radial and parallel axes shown on figure 2.
The top of wave folds are reinforced by a 250 or 500mm thick trapezoidal beam (TS on fig.3) and are truncated by a plane at 300mm offset below the upper sloped surface (in yellow on fig.1).

Previously Adopted Methods and Problems Encountered:

1. Creation of standard Revit floors and modification of their Sub Elements to acquire accurate slab border levels : (see figure 4)

Successes :

Slab bottom corner levels and faces are as per the client’s design.
· 2D & 3D aesthetic appearance is good
· Ability to measure, except where there are voids cutting the geometry. (i.e. TS)
Theoretical Problems :
· The slab thickness (Floor type driven parameter) is taken vertically and not perpendicular to the concrete surface.
Unmanageable Consequences :
· The actual concrete slab thickness is thinner than specified due to the above.
· Concrete slabs cannot be divided into individual pouring parts volumes.
· The slope for each single slab must be measured in order to compute the slab thickness deviation. From this, one floor type and thickness per slab must be created (as slopes for each face vary) and its offset parameter from the reference surface (see figure 1 & 2) should be adapted accordingly.
Manageable Consequences (by post treatment) :
· Slab borders join vertically hence slabs will not chamfer as they would in reality. This situation is amplified at joins between slabs with different thicknesses as shown on figure 5.


2. Creation of standard Revit floors and setting their unique mean slope, slope arrow direction & starting offset level : (see fig.6)

Successes :
· Slab thickness (Floor type driven) is taken perpendicular to the concrete surface and is thus in accordance with the client’s design.
· The concrete slabs can be divided into individual pouring parts volumes.
Theoretical Problems :
· The resulting slab corner bottom levels differ from the specified levels (circa. 20mm) perhaps due to the slope direction approximation. (see fig.7)
Unmanageable Consequences :
· The concrete slabs join inaccurately themselves.
· Slabs corner levels and faces are incorrect.
Manageable Consequences (by post treatment) :
· All slabs must be created as per the previous method to enable measurement of the mean slope and the upper corner level of every single slab face. Further to this, the slope arrow and its tail offset must be set within a new floor entity placed at the same location.
· Slab borders join vertically hence slabs will not chamfer as they would in reality. This situation is amplified at joins between slabs with different thicknesses. (see fig.8 )

3. Creation of standard Revit Multi-layered floors (variable air layer + constant concrete layer) and modification of their Sub Element elevations to acquire accurate slab border levels :

Successes :
· Slab thickness (Floor type driven) is taken perpendicular to the concrete surface and is thus in accordance with the client’s design. (see fig.9)
· Volumes Schedules shows correct values but only in materials schedules and not for whole entity. (fig.10)
Theoretical Problems :
· The resulting slab cannot be separated from the whole floor entity.
Unmanageable Consequences :
· Concrete slabs cannot be shown in 3D views. (see fig.11)
· Concrete slabs cannot be divided into individual pouring parts volumes.
Manageable Consequences (by post treatment) :
· Walls cannot be attached to the slab bottom face unless both these entities are structurally well layered.
· The floor offset from the reference level parameter must remain below the lowest corner level and will be the Air Layer bottom face. (fig.11)
· Air infill material must be ignored when scheduling quantities. (fig.10)


4. Creation of Adaptative Revit Family Component (based on adaptative generic model template) and picking its slab border levels : (see fig.12)

Successes :
· Slab thickness (Floor type driven) is taken perpendicular to the concrete surface and is thus in accordance with the client’s design.
Theoretical Problems :
· The resulting slab is incorrectly generated, non-cuttable & non-schedulable
Unmanageable Consequences :
· Concrete slabs randomly extrude in the wrong direction for certain faces and for an unknown reason, whatever the point number order.
· Concrete slabs cannot be divided into pouring parts volumes.
· The void created between two extrusions at slab folds cannot be filled.
· Concrete slabs cannot be cut by voids residing outside the adaptative family even if they shares the same category.

cliff collins
2012-06-14, 06:18 PM
You could create a Mass and apply Floors, Roofs and/or Walls (with custom material/subcomponents) to the faces of the Mass.

opesch
2012-06-15, 04:18 PM
Thank for your proposition Cliff ;)

But, yes I tried

first : to create a mass volume of simpler tetrahedrons & pyramids forming on point sets based on similar intersecting ref planes but I only get an undocumented suggestion "try Reorder Profiles button !" What's this undocumented button ?!? (see attached images)
85889 85888

Second : I rebuilt a single tetrahedron conceptual mass composed only of surfaces joining the points & vertices
-> it can be imported & host entities by face.... but I get a non-adaptative mass accepting only placement of Walls by face (why is it not possible with floor by face ???) not accepting any cutting method other than using Voids

Third : I created a surfaced adaptative component
-> it can be imported & host entities by face.... but now I get an adaptative generic model which cannot be cut by voids (category cannot be changed to mass type) but accepting also only placement of Walls by face and still not accepting any cutting method other than using Voids

Fourth : I created an empty conceptual mass and pasted in it the previous adaptative points & faces
-> it can be imported & host entities by face.... but I also get a non-adaptative mass accepting only placement of Walls by face not accepting any opening method other than using Voids
85887

Whatever the solution used I also got the problem of the gap coming between 2 extrusions at their common fold edge...85890

so my questions are :
_why can't I find an Adaptative MASS template in Revit's Library ?
_why can't I place Floor by face on mass or generic families (and use shafts with them and schedule them as required)?
_why aren't face/wall opening acting on mass hosted walls ?
_is there a way to force walls to join over the folds as they do (trim) when they're vertical ?

Thank a lot for your help... you've pushed me a bit closer to a valuable solution ;)

damon.sidel
2012-06-15, 05:15 PM
Here's another suggestions: Could you create the geometry in another program like Rhino and simply import it in as an SAT? If you import it into an in-place family you can make it walls, floors, roofs, or whatever. You could import different pieces into different in-place families to make them different categories for scheduling purposes. I'd be happy to help if you do not use or don't have access to Rhino. It can be a very good workflow for special geometries.

Alfredo Medina
2012-06-15, 05:25 PM
Have you seen the AugiWorld May 2012? (current issue) http://www.augi.com/augiworld/current-issue/

I wrote an article, entitled "A tribute to Felix Candela's work" in which I demonstrate that some complex forms can be modeled easily in Revit starting from a simple flat projection of the form, using reference lines and reference points, lines for the straight edges, and splines with three points for the curved edges. Then, create the form or the group of surfaces when all the projection is flat at the reference level. Then, you go to an elevation, you set up reference planes, and give them a name. Then, simply lift the points to the correspondent reference plane, and the flat projection will unfold and take shape in 3d. It seems to me that this approach will work in this case. About scheduling and categories, Wall by Face is the most flexible, or powerful of the 3 options (Wall, Roof, Floor), so if it works with wall, at least you get the form, and can move on.

opesch
2012-06-15, 05:29 PM
Yeah !!!

Fifth : I've found that if I cannot put floor by face on the mass, I can put Roof by face !!!

which have a CONSTANT THICKNESS !! can be cut by shafts, by voids, automatically fill the gap at folds over 180°, accept wall and column attachment !

so that's the bestsolution until now.... the only last problems are :

_roof creation was canceled on one face because "some faces (of the Adapative component) have different extrusion direction than the other".... but I don't know how to change this direction as they are no parameter/arrow for that in the adaptative family !?!

_Roof fascias are only settable to plumb cut and it seems to be the same at folds between faces

_The Adaptative component isn't a mass element but a generic model that I cannot categorize afterwards as a Roof or a Floor.... so I've to put it in a guiding geometry workset and hide it !

Thank a lot Cliff ! You pushed me on the right way ! you got it ! :beer::beer::beer::beer::beer:

Thanks also to Damon & Alfredo for their answers....

And I'll look to your article right now Alfredo !
Because now that I've a solution for my roof entity I was ready to pick manualy all slab face corners.... but if you've a technique to do it automatically it would be of a great help ! ;o)

In fact I have 18 totally different set of roof composed of 10 folds set like the one on my last attachments.... so if you've a trick to place those folds set using an array-like method where folds growth being hanged to their radial axes i'll certainly use it !!

opesch
2012-06-22, 10:55 AM
Thank for your suggestion & article Alfredo but I have to follow exactly the architect design and cannot use Candela's smooth forms. I also prefer to use Roofs by face in place of walls because joins between faces were worst with walls and I had less flexibility with openings & attachment to other objects like beams, walls & columns.


So ! Now that I've got my roof module I need to copy it 180x on a radial array waving between 2 sloped refplanes....

My idea is thus to automate a bit the procedure to place the adaptative roof module into an in-place mass component were it will follow a series of guiding lines that I'll pick on my actual project. Then I'll put roof by face on created mass faces in the project itself.

What's strange is that I try that in a test mass family but :
_ I cannot modify the adaptative point positions after module placement by picking the points on the guiding lines.
_ maybe it's link... I cannot copy the module along the guiding lines as I see people do with adaptative shapes on Youtube !! =o/

Could someone help me ? Or tell me at least where I'm wrong ?
I join the test mass family including the adaptative component

Alfredo Medina
2012-06-22, 12:42 PM
Some questions go beyond the scope of public, free, help. This is one of them.

opesch
2012-06-25, 09:43 AM
Mail Sent ;o)

So... my Boss prefer to not use billed consultancy unless our joint venture officially ask for it !

So I've to model it as best that I can for one gallery (10modules) and then I've to tell them how much time it takes to do it (ie. to manually place all adapatative components point by point on the model) and let them decide if we need to pay for expertise because it takes too long or if they will let it done by a bunch of indians or let it do in an other program...

But I realy dislike that as it implies that Revit cannot do it and is slow at doing it when it's just that I'm not as experienced as I would like to ! :cry:

So here is where I am :
_I can nest the adaptative GM component into a Conceptual Mass hosting some guiding lines
_I can use those lines to constrain the adaptative component
_I can only copy it by ctrl-draging it aside.... no copy or array allowed so no fixed distance allowed also !
_copy the component aside deforms it completely.... I know that we can draw adaptative points on other points refplanes... but how to set those point ref planes in a non orthogonal specific direction ?
--> I would like to be able to constrain the adaptative points to stay complanar as shown by the green hidden lines on below PNG
_when I put a roof by face on the adaptative component, face normals directions are varrying.... and roof creation fails when normals are in unconcistent directions !
--> I can show them with zach's trick : http://buildz.blogspot.be/2011/03/that-aint-normal-diagnosing-point.html but haven't find yet a way to change them inside the component !

Well.. I think that's all for now.... If you've some hints for me to do it a bit less "monkey work"...

Thanks a lot

[edit] I found how to modify adaptative points placement when they were picked at a wrong position... dragging them wasn't accurate as they were no "snap" to geometry !
_do not select the component but do hover one of its adaptative point and select it (without anything else)
_click on "pick new host" button on the ribbon and now you can snap in 3D to the existing geometry !

So I've done the monkey work and got now a new problem.... in the valley of my wavy roof there's a concrete infill... not horizontal but sloped as everything else !

The infill has a little slope at its top face which follows the roof folds lowest point evolution & catwalk slab (infill top face &catwalk slab are at +600mm above this lowest point) so I need to do an adaptative component which will hang on Roof Fold Face borders at desired z and will fill the valley with a concrete volume which has to be joined to Roof folds afterwards (same pour -> no lines !)

as you can see in the attachmant its bit like doing a paramatric pyramid on its head.... I only found tutos to make pyramids composed of surfaces but not yet of volumes !

I tried to extrude all faces inward and then cut everything above the top plane but I got errors as Revit doesn't seems to accept the join of the 6 extrusions !
I tried to extrude all faces as voids outward and then cut the top plane extrusion but they were gaps between the 6 extrusions as they do not tapers !

have you already seen somebody doing a pyramidal concrete infill with adaptative components somewhere ?

opesch
2012-06-26, 01:30 PM
AAAAALLLLLLL Right !

Eureka I found a solution ! (well... I hope so ! glups !)

After a lot of unfructuous thinking I was wandering how to avoid a max the use of cuts and joins ....

Then I realize that as my top face is driven by a multiple of 2 points, I can create a blend between 3 vertical triangles and then cut only the last part of the extrusion.

So I placed a 9th point (not adaptative) on the bottom ridge's near-horizontal-worplane and trace the third triangle

Good News ! As expected it works and extrude a blobby blend going trough the 3 triangles !

As I need flat faces, I redo the same with only 2 triangle at once so Revit will extrude straightforward between the 2 triangles.

Then I realize that the last one was still a bit curvy on its below faces and that even cut with an ending void it will never have the right bottom faces.

So I decide to place the 9th point as close to the previous colinear point as Revit will let me do.... so 0,2mm

And \o/ youhou ! Its done !

I think I cannot go closer to what I need unless somebody tells me that he has a better solution (even without giving it for copyright reasons ;o) )

I which I could have a tool like Autocad's "Sclupt" in Revit which can convert a watertight surface bundle into a solid volume

Thank a lot for everybody who answer to my questions ;)

opesch
2012-06-28, 12:03 PM
Damn !

When I resolve a problem I get a bunch of others !

SO

making concrete roof waves of constant thickness --> resolved with adaptative component hosting roofs by faces
--> problem : roof top faces show a stepped join because of roof "plumb cut" raft locked parameter.
--> No solution, should probably be corrected in 2D.
roof waves evolve between multiple sloped planes and their intersections --> resolved by using shafts and wraped floors to acquire & pick waves hosting coordinates
--> problem : when I rise the shafts to exclude the roofs by faces created at point 1 from the shaft range, adaptative hosting roof folds erases themselves and roofs by faces become un-updatable !!
--> No solution, roof by face slopes stays where were picked and cannot be updated anymore without placing a new adaptative module and attaching new roofs by face entities.
creating sloped roof valley infills --> resolved by cutting Roof by face from point 1 with a sloped reference floor extrusion and then place an adaptative infill by picking the resulting points
--> problem : when I uncut the sloped reference floor extrusion to recover the original roof by face geometry done at point 1, all adaptative Infills erases themselves !
--> Alternative : replace the whole bottom geometry by the infill entity by acquiring the outermost (thus bottom face) roof intersections with the sloped ref floor to host the infill points an then try to join roof slopes and the infill so I haven't to unjoin the cutting sloped ref floor
--> problem : I cannot join one of the roof by face slope (the only triangular one) with the infill ! This certainely because of the previous post's trick used to create the infill (2 bottom points separated by 0.2mm to simulate a pyramial end)
--> No solution until now unless placing model lines where all adaptative infills should be hosted..... There are still 180 modules -> 180 infills needing 6 lines per infills that I will have to place manually and will not be able to delete afterward !!:roll:


SO QUESTIONS ARE :

Is there a solution to model in an adaptative family the volume of the extrusion from any polygon to ONE apex point other than a square based pyramid (we should be able to extrude volumes off ALL those forms : http://en.wikipedia.org/wiki/Pyramid_(geometry)#Pyramids_with_regular_polygon_faces )... at least a tetrahedron !
Is there a way to "free" an already hosted adaptative component from its hosting entities without changing its own geometry. or loosing it...


Have you got any thought over those last problems ? :(