"Revit can't do that....." series part 1

[AP23]

I'll be posting in the coming weeks a few samples of geometry that is easy to create in other 3d software, but extremely difficult, if not impossible to do in Revit. These posts are not intended to bash Revit, but to discuss workarounds and discover new possibilities.

First example: the glazed roof of Britisch Museum in London done by Foster and Partners.

In Autocad (see attachment) or rhino, you can create this roof by lofting 3 profiles. Doing the same procedure in Revit will end up in an error. So you will need to move the inner profile above the second profile before moving the inner profile back to its original position. However, even so, you get an undesirable shape (see attachment).

Anyone knows how to do this in 2 minutes?

here is a pdf link to the roof structure http://opus.bath.ac.uk/14111/1/ChrisDeakin2001.pdf

[tomnewsom]

I know a guy who helped design that roof. The initial model was very basic, full of straight lines. The guy I know then wrote custom software to 'relax' the shape whilst maintaining an even structural load. Revit can't do that, but neither can anything else.

[AP23]

It would at least help if Revit can make the basic Autocad shape.

[cliff collins]

Tom,

Any idea what software WAS used on that roof? Doesn't Foster use Bentley?

Could that roof shape be done in Revit by creating two or more separate pieces
and then joining geometry, and mirrorring-- instead of trying to force it to do a multiple-profile loft, which it apparently can't do?

Just a thought......

cheers

[twiceroadsfool]

How to do it in two minutes? Import the CAD Model in to a Mass Family.

Dont get me wrong, i LOVE AutoCAD and all of its modeling tools, but frankly.... Were talking about OTHER workflows that mean more than one platform. Modeled in Rhino? Getting detailed elsewhere, i presume. Modeled in AutoCAD? I GUESS you can detail it there, if you want to.

Id just assume use the CAD Model in a mass family, and then use Revit for the rest. But yeah, those modeling tools sure would be nice with the power of Revit behind them. :)

[AP23]

Since RAC2010, the option to import 3d geometry from Autocad or other software has been a lot smoother. You can divide the surface, you can snap to intersection points on the imported geometry. You can even extrude the surfaces from the imported geometry. These things couldn't be done in previous releases, what makes the workflow with other modelersa now a viable option.

But if you decide to do it in Revit, then the question is how. Creating separate pieces as Cliff suggested is an option.

[jeffh]

Could that roof shape be done in Revit by creating two or more separate pieces and then joining geometry, and mirrorring

I have not taken a lot of time to look at the issue but this was going to be my approach.

[Alfredo Medina]

Well, maybe it is possible to model this in Revit. Not in two minutes, though.

[tomnewsom]

Well, maybe it is possible to model this in Revit. Not in two minutes, though.
That's exactly the sort of thing that got fed into my aquaintance's software :)

I don't know what the source and destination software was - this guy did all his work in C on his mac.

[domglam]

say it like it is...

http://buildz.blogspot.com/2010/06/geometry-is-trivial-what-are-you-trying.html

[vgonzales]

I'll be posting in the coming weeks a few samples of geometry that is easy to create in other 3d software, but extremely difficult, if not impossible to do in Revit. These posts are not intended to bash Revit, but to discuss workarounds and discover new possibilities.

First example: the glazed roof of Britisch Museum in London done by Foster and Partners.

In Autocad (see attachment) or rhino, you can create this roof by lofting 3 profiles. Doing the same procedure in Revit will end up in an error. So you will need to move the inner profile above the second profile before moving the inner profile back to its original position. However, even so, you get an undesirable shape (see attachment).

Anyone knows how to do this in 2 minutes?

here is a pdf link to the roof structure http://opus.bath.ac.uk/14111/1/ChrisDeakin2001.pdf

AP23,

Here are my sample solution to your form problem.

[bregnier]

I think there are 2 true statements coming out of this discussion:

1. Revit's modeling tools have room for improvement (although they are getting better with each release). As the open API improves, plugins and third party applications might give some serious help to this end.

2. Complex form generation is almost certainly going to involve more than one program - this is true whether you're using Revit or not. Of equal importance to improving native modeling tools is improving interactivity with other modeling software.

Computer modeling always has hidden constraints, whether it's workplane based extrusions or SubDs. The trick is to know these constraints and use a suite of programs to either leverage or ignore them. Or, as was said above, write your own code.

PS-- one of the most important things about the form in questions is the lack of hard seams at the corners. This makes it a lot harder to model...

[AP23]

Using different software packages isn't the problem, especially when your doing competitions where you only have a few days or weeks to come up with a design. You're likely to use a tool like Maya or Rhino to knock out hundreds of different variations within minutes, before choosing one.

The questions is, what do you do when you have to document them? What are the steps you need to take to rationalize the geometry and make it workable in Revit? What added value can Revit bring to such a process? You can for instance model the Watercube in Beijing, but it will take you ages. So knowing when to make Revit part of the process is crucial, because Revit has it's limitations.

As for roof form, I think I would design it in another software package and try recreate it in Revit for documentation. It will be a challenge to get the exact dimension from the geometry that was created in another package. And while having a split surface isn't favorable, you need to make a chose where the surface breaks. The attachment is the same file as what Leo posted, but lofted differently. As you can see, you get a different shape from the same splines.

[bregnier]

As for roof form, I think I would design it in another software package and try recreate it in Revit for documentation. It will be a challenge to get the exact dimension from the geometry that was created in another package.

This sort of approach makes sense for certain kinds of construction - such as curved interior walls- , but for things as complex as the British Museum roof, or some sort of complex metal panel system, dimensional control in construction is likely going to be ceded to an entirely separate model, perhaps under the control of a consultant. In that case often the consultant takes a drawing set and some exported geometry and builds their own model that is used to generate g-code or cut lists. Even if you're doing it yourself it's usually easier to generate shops in another program.

Revit in my mind acts as a coordinator, kind of a master database for the entire project. Really complex parts can be followed through in some other software, with the assembly as a whole represented in the Revit model (with approximate geometry). The Revit model is then responsible for locating the assembly and referencing any additional information.

I'm not saying that at some point we won't be working out of a single model for every last bit of information. But as it stands right now, doing so with _any_ single software would not be easy or efficient.

[AP23]

What I meant by getting the exact dimension from the geometry is to recreate it exactly the way you made in in Rhino or Maya. I've tried recreating geometry like that in Revit, but you'll never get it to look exactly the same way.

At the end, you will indeed use other software for shop drawings. However, it will then be hard to justify the use of Revit, since you go form an all purpose modeler or parametric modeler (Catia) straight to fabrication.

[bregnier]

Catia would technically work for soup-to-nuts digital design, except for the enormous learning curve and inefficiencies that result from using a very powerful tool to do simple things. If you look at the workflows of offices using Digital Project oftentimes you'll see DP used on the most complicated part of the building and Revit (or even Autocad) used for the rest.

[cliff collins]

I wonder, in Leo's example, if adaptive components could be used to "smooth" out
the hard edges at the "corners" ?

I think Revit is getting very close to having the same modeling abilities as Rhino,
FormZ, etc--but with all the BIM as well--not too bad for an all-in-one package.

I agree that ultimately other packages will be used downstream to feed file geometry
to a CNC cutter or automated fab tool, so whether or not Revit "can do exactly what
XYZ software does" is not critical. The notion of Revit as a Master Coordination tool as stated in the previous post is very valid.

That said, I'd like to see Conceptual Massing tools have even more fluid workflow, and perhaps NURBS ability in future releases.

I would also like to see the option to "release" walls, roofs, floors, curtain systems, etc
which were created by picking Mass faces, so they are not prescriptively and permanently
tied to the Mass--but I know this idea flies in the face of the way the program was designed.

cheers

[narlee]

What a great idea for a thread, AP23! It deserves its own forum spot.

[zachary.kron]

The British Museum roof form is largely a result of calculations done to reduce the variations in the panels that create it and it is, I think, entirely without seams. Here is another method to approximate it in Revit (video in the zip file, 40 second journal replay).

[brenehan]

Thanks Zach.
That's cool.
I've been having difficulties over the past few weeks in trying to get a 3 point ellipse sweep to work. Anyway, eventually I got it working. See attached:
It's a generic model nested into an Adaptive Component. I found the Ellipse tool worked a bit different in an adaptive component family in comparision to a standard Revit family.

Brian

[hugotavares6]

So good :D
very nice! Congrats

[AP23]

The example of Zach is quite interesting, cause it shows a bottom up approach typically used in manufacturing software like Inventor. This means that such a roof is probably more efficient to recreate it in a latter phase when the final geometry is known.

[Gigmahabir]

Not to pile onto what Revit can't do, but it is so frustrating to create sweeps in revit that are based on the picked path of complex curves. It would be great if Revit's error message would give a more detailed explanation of why the element cannot be created. After some hypothesizing, I can only figure that the sweep might barely intersect itself.

The other thing that is frustrating about picked path sweeps is that the profile planes are always tangent to the point on the curve. It's not like the Conceptual massing environment where you can switch between local and global coordinates.

If anyone can offer some insight into whether it's possible to change the profile plane coordinate system that would be mighty helpful.

[brenehan]

I have been having some fun with a canopy for a project we are doing.
The Elevaiton is driven by one spline and the canopy span by another spline.

I'm starting to become quite amazed how powerfull the adaptive components / Massing famalies are.

Brian

[ree.espinoza]

Hi All
I have read thru all the above posts
very impressive Little about why i am here
I have a very complex form to bring into Revit
Was scanned and brought into Rhino My colleage does the Rhino portion
We have acad 2011 and revit structures 2011
I have not touched acad since 2008 (architecture) I have not made much headway past Revit architecture 2009 after I got laid off
In a crunch to learn software and be productive to create structural support for free form in Revit to do clash detection with primary structure and building structure
I love the fact that 3d acad comes into Revit now!
The exports from Rhino are polylines (sim to contour lines with elevations)
So either from acad and or Revit I need to have true 3d form or surface in order to cut a lot of sections in Revit and detail the structural support system. What would be the best approach? Thought I could learn on my own but todays day and age collaboration is so important for just in time delivery. Please help me. Perhaps a starting point. Thanks so so much in advance.!!!!!!!

[hugotavares6]

I have been having some fun with a canopy for a project we are doing.
The Elevaiton is driven by one spline and the canopy span by another spline.

I'm starting to become quite amazed how powerfull the adaptive components / Massing famalies are.

Brian

Nice Brian.. can you give me your e-mail? If you do a tutorial about this, I'd be very grateful. If you want, I'd like to buy it - how to do spacial trusses.
If someone can do that, please e-mail me: hugotavares6@gmail.com

Did you did tha with adaptative componentes, line by line?

Thanks

[AP23]

Hi All
I have read thru all the above posts
very impressive Little about why i am here
I have a very complex form to bring into Revit
Was scanned and brought into Rhino My colleage does the Rhino portion
We have acad 2011 and revit structures 2011
I have not touched acad since 2008 (architecture) I have not made much headway past Revit architecture 2009 after I got laid off
In a crunch to learn software and be productive to create structural support for free form in Revit to do clash detection with primary structure and building structure
I love the fact that 3d acad comes into Revit now!
The exports from Rhino are polylines (sim to contour lines with elevations)
So either from acad and or Revit I need to have true 3d form or surface in order to cut a lot of sections in Revit and detail the structural support system. What would be the best approach? Thought I could learn on my own but todays day and age collaboration is so important for just in time delivery. Please help me. Perhaps a starting point. Thanks so so much in advance.!!!!!!!

This tutorial might help. http://designreform.net/2008/07/rhino-autocad-revit-linking-complex-form-to-drive-massing/

[brenehan]

Nice Brian.. can you give me your e-mail? If you do a tutorial about this, I'd be very grateful. If you want, I'd like to buy it - how to do spacial trusses.
If someone can do that, please e-mail me: hugotavares6@gmail.com

Did you did tha with adaptative componentes, line by line?

Thanks

Hi hugotavares6

Here is a simplified version of the canopy. Mass Family

[hugotavares6]

Thank you. i'm studying everyday the mass tool to be able to do things like that.
Now I see, I thougt that you made a complex family that behave like a spatial truss. In this case, a very complex spatial truss.
I saw an image in buildz that gave a light about how to do a spatial truss. There is. Design Reform and buildz are helping a lot :D

[hugotavares6]

Theres the image.

[william.lcampo]

I once said "Give me a math formula and I will draw the shape in Revit"...
I am now biting the bullet and building an adaptive component that follows the formula in the initial post....

A bit unfare to challenge people to build it in two minutes though... it took me longer than that just to get my head around the variables!

Anyway, I'm on the case now, and hope to have it ready in a couple of hours... any idea of where to find the x,y coodinates of the nodes after relaxation??? it doesn't seem evident from the quoted paper... I suppose I will have a simplified version of the shape instead!

[william.lcampo]

Here's my approach:
An Adaptive Component with 3 points: 2 origin points (to be able to deal with negative coordinates) and one for the projection of the desired node on the floor.

A second Mass Family hosts several instances of the Adaptive Component, and generates the surface using splines through points. If you need more precision in the surface, just need to add more control points, as wherever you land a point it will follow the surface formula.

Not too different to Zach's approach in this post:
http://buildz.blogspot.com/2010/03/look-ma-no-api-rule-based-form-creation.html

just using the freaking complicated set of formulas in the beginning of this trail.

More interestingly, if you know what the "constants" that define the roof mean, you could start investigating variations and refine the shape, as this approach records the design intent as established in the geometric analysis.

I hope it clarifies one way in which Revit CAN do it! (I feel like Locke in Lost!)
Now I'm interested to know how others can follow the precision established in the geometry analysis in other software packages, beyond being able to draw a shape that somehow resembles the shape, but ignores the complexity of it... in two minutes?

Kind regards,
William.

[jonmirtschin]

Interesting discussion, and I'll throw in my 2 cents. I agree with opinions expressed, there is not one single ultimate software tool for projects of this nature and utilizing (and exchanging model data) a suite is likely necessary.

For studying the development of a surface like this in an alternative software such as Rhino or Grasshopper , you can find some relevant blog posts on my blog, http://geometrygym.blogspot.com

In the case of the Great Court, I'd read the same paper and generated a "mathematical" surface using the formula provided. Whilst this might take more than 2 minutes, it is a quick easy process to parametrically calculate this using NURBS. I was more interested in the process of smoother, continuous grid lying on the surface that was achieved by force-density mesh relaxation, and I've enabled tools within Rhino/Grasshopper to do this (demonstrated on the sample model of the Great Court Roof).

But this discussion is more focussed on how to define the original surface. Commonly project inputs will perhaps be less regular and (even more) difficult to define a mathematically formula to define the shape. Analytical form finding is more commonly used (prior to computers was done using physical models like hanging chains and soap film bubbles). Again I've developed (and am actively developing and improving) routines to do this type of calculation in Rhino/Grasshopper, primarily against uniform pressure like a balloon (not against a constant acceleration like gravity). Very useful for form finding domes with irregular perimeters or aspects such as ETFE cushions.

Whilst it's possible to use these tools through to detailed geometry setout, I've tried to focus on tools that are easy to use to test variation on competition and scheme ideas, with the option to pursue the form later in more specialist software (that generally requires technical knowledge and experience).

Other comments I picked up on where difficulties in transferring geometry from "rhino" to more advanced detailing programs such as Revit. Again this is a strong aspect I'm working on developing. I'm not sure if Revit now utilizes NURBS, but a "translation" would certainly explain different appearance or surfaces. At a component level, I've been developing neutral bim exchange using formats such as IFC to ease this exchange process with greater accuracy.

I am pursuing a direct Revit to Rhino/Grasshopper link, I'm interested to find any potential users to help prioritize the developments I'm pursuing. Please get in touch if you wish to learn more. http://www.geometrygym.com

Cheers,

Jon

[Andre Baros]

Very interesting.

[luke.s.johnson]

Here's my approach:
An Adaptive Component with 3 points: 2 origin points (to be able to deal with negative coordinates) and one for the projection of the desired node on the floor.

A second Mass Family hosts several instances of the Adaptive Component, and generates the surface using splines through points.

Hey William, do you want to upload the actual project file for us to see this fancy family in the project environment?