Complex shaped roof

[Spork]

Hi all,

I am trying to create some relatively complex roof shapes, and can't seem to figure out the best way to achieve my design in Revit. Here is an elevation of what I am going for:

http://imageshack.us/a/img21/8601/screenshot20130316at336.png
89482

And section:

http://imageshack.us/a/img834/7725/screenshot20130316at337.png
89483

So you can see the roofs are shaped in both plan/elevation and section (basically trapezoidal). I first tried to create a mass and use Roof by Face, but on the faces that are close to vertical I got a weird result (mass highlighted in blue):

http://imageshack.us/a/img405/2598/screenshot20130316at342.png
89484

You can see that they don't appear to be created perpendicular to the face, but rather it looks more related to the Z axis.

I also tried to create the roof by extrusion, but I can't figure out a good way to trim it in plan/elevation. I know there's the Vertical Opening tool, but again that's as if you took a cookie cutter and sliced straight down on the roof (i.e. you can't get the trapezoidal shape in elevation). It seems like I should be able to do this in Revit: I'm in graduate school and really want to push Revit to achieve a complex design. Any help you can give me will be greatly appreciated.

JP

[Mike L Sealander]

It looks like some of these roofs are good candidates for Roof by Extrusion, with some cutting done by roof voids. You might also think about not being wedded to the Roof system family. It looks like a roof in function, but it might be easier to make these in the Family Editor as generic models.

[Spork]

Thanks for the reply.

What do you mean when you say 'roof voids'? I tried creating a Void Mass and using it to cut the roof, but Revit doesn't appear to work that way. Also, using a Vertical opening creates weird results (i.e. it doesn't cut laterally, only vertically).

I will also just try generic Families if this doesn't work.

[dhurtubise]

I would simply model the roof InPlace

[Steve_Stafford]

As Daniel suggests, in-place modelling will give you free reign over solids and voids and it/they can be assigned to the Roof category.

Keep in mind that complex design does not automatically mean good design. Right now the elevation looks kind of like a bad haircut, sorry but you'll be facing a crit sooner or later. :) What about natural light? The roofing curling over the walls like a page boy haircut, covering up nearly all the vertical surfaces is going to create some strange window conditions? Section is interesting but again is the concept relying solely on artificial light? Skylights?

It is sort of stretching and pushing a Gambrel roof style now. The great variety of shapes on the long elevation just seems a bit too chaotic and forced.

[Spork]

It seems like modeling in place is the best solution, so I will attempt that. As far as roof structure goes, can you achieve the different layers (i.e. deck, insulation, membrane, etc.) by modeling in place? That's mainly why I wanted to actually use a roof type.

I also plan to show the structural framing that support each of these. I haven't used Revit structure in a complicated way like this before - can you attach/align structure to an in-place family?

In respect to the design comments, these wrapping roofs are separated by several feet in elevation, creating clerestories at each overlap. I also am beginning to experiment with some of these being solid wall/roof elements and some being more of a perforated screen, to allow views.

Thank you all for your comments.

[mthurnauer]

Question: Have you tried wall by face on these in lieu of Roof by face? You could accomplish the same thing and it may not be so finicky about the vertical vs horizontal orientation.

[dhurtubise]

Each layer will be much work because you would need to model all layers. I would start with a overall one and then, when the design is more "fixed", start adding the layers.

[Spork]

Question: Have you tried wall by face on these in lieu of Roof by face? You could accomplish the same thing and it may not be so finicky about the vertical vs horizontal orientation.

Interesting, so I would basically have almost horizontal 'walls' at some points. I will try that as well.


Each layer will be much work because you would need to model all layers. I would start with a overall one and then, when the design is more "fixed", start adding the layers.

That makes sense. I think I'll either manually add the layers in my wall section view, or try the wall by face as suggested above.

Thanks for everybody's comments, you've been most helpful.

[rosskirby]

Depending on what you're doing, there are a few different options. As others have stated, modeling either by-extrusion or by-face on a mass, or doing an in-place model would all be good solutions if you know that the model won't change. However, if there's any possibility that the roof configuration is going to change, then you should really consider adaptive components. Take a look at Zach Kron's blog (http://buildz.blogspot.com/), where he does walkthroughs of how to leverage adaptive components for design flexibility.

[damon.sidel]

you should really consider adaptive components

Thinking about this method, are all of the roofs six segments like the original section you posted? And does each segment always have four sides? If so, I agree with rosskirby that this looks like a good candidate for an adaptive component. Just a warning, that if you start to go into layers, etc., the adaptive component will get a little tricky to model. I just finished DD a few months back using a pattern-based curtain panel for a high-rise diagrid facade. It was the perfect solution for awhile. Then when the geometry of the building became relatively fixed and we needed more detail about the windows, which were part of the diagrid curtain panel. At that point, I switched out the pattern-based adaptive curtain panel with a generic model. The overall angle of the diagrid was fixed, but the thickness of the "columns" and the configuration of the windows (nested non-wall-hosted window components) and spandrels continued to change. Later in the project, I switched it out one more time: I removed the diagrid columns from the generic model and modeled those as face-based walls and left just the windows.

So to make a long story short, I think the adaptive component method would be great early in a project. Then you may want to develop a mass-hosted solution to use actual roof or floor built-up extrusions (or In-Place models if all else fails) as the geometry changes less often.

[Alfredo Medina]

Another comment on this thread: Modeling the roof is not complicated, but just tedious and time consuming, because all the points that define the roof segments seem to be in random locations, so the user would need to adjust the location of each one of those points, unless these points follow some pattern that you can establish as a repetition. Another issue is: what's going to hold this roof? The section in the original post shows some thick slabs for the floors; however it does not show any indication of the structure of the roof. Considering the randomness of the roof, the structure will be equally random and time consuming to model, too. For building something like this, all those points that define the roof and the structure would need to be documented precisely. Lots of work.

[AP23]

I've done a similar project with roof surface that flows into a wall and floor (something Autodesk doesn't always understands). The trick is to use reference points in the massing tools which you can pulled into space and reference lines can attached to the points which creates surfaces. Wall by face is not an option because it always cuts the wall ends perpendicular to the face. And it doesn't allow you to create variable thicknesses along the the surface of a wall. So you will need to use in place families and use the surfaces created in the mass families as work planes.