Alternating pattern with checkerboard pattern based curtain wall?

[graphite]

I have been struggling to understand how this type of pattern could be created with revit's pattern based curtain walls. I think the biggest hurdle I have come across is that of revits ability to introduce two different panels into the pattern applied to the surface. Is it possible that each panel in this facade is actually composed of multiple panels giving the illusion of being tiled individually over the surface? Also how did they get the panels to alternate interlocking with each other?

Any thoughts on how Morphosis may have done this would be appreciated. I also believe they use generative components by bentley so i am wondering if maybe they generated the pattern in bentley and then brought it over to revit? 89267

[damon.sidel]

First off, a question: where did you read that this was done in Revit? Not saying that it wasn't.
Second, another question: I don't think it is a purely repeating pattern of two components (a parallelogram and a trapezoid). It looks a little more compositional than that.

I don't see an immediate solution using a purely pattern-based curtain panel family.

Then a first-thought solution: Create a conceptual mass cube with a triangular or rhomboid pattern. Create a simple four-point adaptive component. Manually place the component where you want it to achieve the composition that you are looking for.

Second-thought solution: Create a trapezoid+parallelogram component, then use divide and repeat in 2013.

Pictures of the design show a lot of variation in the surface of the panels. If they modelled this surface texture, it wasn't in revit... just sayin'.

[Alfredo Medina]

This is no problem for Revit. It could be done as a divided surface with curtain panels, nested, or as a curtain wall with custom curtain panel families. Can you post images of what you have tried so far?

[graphite]

First off, a question: where did you read that this was done in Revit? Not saying that it wasn't.
Second, another question: I don't think it is a purely repeating pattern of two components (a parallelogram and a trapezoid). It looks a little more compositional than that.

I don't see an immediate solution using a purely pattern-based curtain panel family.

Then a first-thought solution: Create a conceptual mass cube with a triangular or rhomboid pattern. Create a simple four-point adaptive component. Manually place the component where you want it to achieve the composition that you are looking for.

Second-thought solution: Create a trapezoid+parallelogram component, then use divide and repeat in 2013.

Pictures of the design show a lot of variation in the surface of the panels. If they modelled this surface texture, it wasn't in revit... just sayin'.


This is the article I found where they mentioned the use of revit for the precast panels. http://blog.archpaper.com/wordpress/archives/24261. However after looking further I found a blog by Morphosis and their IT department http://synosis.org/ reviewing this project and from one of the pictures it looks as though it was started in GT digital.

Thanks for the alternative approaches to this project in revit. But I think you are correct when you said the pattern and the panels are modeled in something else as the sweeps on the panels are continuous across different panel geometry. I have a feeling the bulk of the modeling was done in gt digital all the way to fabrication for the precast guys and the rest of the project was documented in revit.

On another note, would those two suggestion you made be the only way to model a pattern with two or more units composing the pattern? I am familiar with adaptive components and that seems like it would be quite a manual process to place each panel. I am not familiar with divide and repeat though.

Also what do the checkerboard patterns provide that the regular patterns do not? I have looked around and I haven't found good examples of the checkerboard being used.

Thanks for your reply really appreciate it. Trying to figure out the modeling for this project has been bugging me for awhile since I thought the entire thing was done in revit...

[Alfredo Medina]

It is possible in Revit. A panel can contain another panel, then the composed panel is applied to the divided surface. This image shows just a flat sketch. With some time, a panel could be modeled exactly as shown in the photographs. Another option is through custom panels and curtain walls. Probably the latter option might be less problematic in regards to custom edges and corners that turn around. But, in my opinion, this facade is possible in Revit.

[damon.sidel]

Alfredo, you have definitely come up with a nice way to model a repeating pattern that resembles the Cube. However, the Morphosis project is much more complex than that and isn't easily reduced to a simple pattern. I think the minute you go beyond the pattern, it becomes much more difficult to get the level of control in Revit that this design requires.


This is the article I found where they mentioned the use of revit for the precast panels. http://blog.archpaper.com/wordpress/archives/24261. However after looking further I found a blog by Morphosis and their IT department http://synosis.org/ reviewing this project and from one of the pictures it looks as though it was started in GT digital.

Thanks for the alternative approaches to this project in revit. But I think you are correct when you said the pattern and the panels are modeled in something else as the sweeps on the panels are continuous across different panel geometry. I have a feeling the bulk of the modeling was done in gt digital all the way to fabrication for the precast guys and the rest of the project was documented in revit.

That is definitely GT Digital Project or even just CATIA.


On another note, would those two suggestion you made be the only way to model a pattern with two or more units composing the pattern? I am familiar with adaptive components and that seems like it would be quite a manual process to place each panel. I am not familiar with divide and repeat though.

There may be other ways to do it, but without really digging in, they are the only two ways I could think of. They would be quite manual. Divide and repeat would help, but even if you ignore the texture on the panels and just want to the combination of trapezoid and parallelogram panels, it doesn't look like a repeating pattern. If it doesn't repeat, by definition you couldn't use a pattern-based solution or a divide and repeat solution. The next non-manual option would be a script.

Interestingly enough, in the colorful image you post, there are vertical and diagonal joints. If the panels are actually rectangular and just have "fake" diagonal joints, that might suggest a solution with a pattern-based panel that you then swap out the oddballs from the pattern to make it more irregular.


Also what do the checkerboard patterns provide that the regular patterns do not? I have looked around and I haven't found good examples of the checkerboard being used.

The checked patterns leave holes and do not place a panel in the hole.


Thanks for your reply really appreciate it. Trying to figure out the modeling for this project has been bugging me for awhile since I thought the entire thing was done in revit...

Always good to be curious about these kinds of things. Most of the time with something this complicated: irregular composition, overlayed textured, etc, I always start thinking about scripts and more sophisticated methods than just OOTB Revit conceptual tools.

[Alfredo Medina]

Alfredo, you have definitely come up with a nice way to model a repeating pattern that resembles the Cube. However, the Morphosis project is much more complex than that and isn't easily reduced to a simple pattern. I think the minute you go beyond the pattern, it becomes much more difficult to get the level of control in Revit that this design requires....

I was referring to what I saw in the first image in the original post, and then in the photographs in the links. What I see now in the second image with the colors, does not have a clear relation to the pattern shown in the first image or in the photographs. What I see in the photographs is a pre-cast concrete panel. Again, given enough time, this is possible in Revit. What we see in the color image is different and I don't understand the relation between that and the first image or the photos.

[graphite]

Alfredo, the colored image displays the different extrusion types which "texture" the precast panels. Each panel is actually composed of a collection of these extrusions which were cast individually and then assembled to form the trapezoidal and parallelogram panels.

This weekend I spent some more time dissecting this problem and the color image provided some further clues that moved me much further than I thought I would get. So far I have figured out the patterning issues using 4 different adaptive components hosted on a larger adaptive component the width of the building. The devil is in the details though...if you will notice in the colored image the textural extrusions run seamlessly from one panel to the next. I figured I would simply create adaptive components for the majority of the extrusions and host them on the 4 different panels after subdividing their respective surfaces. However creating a seamless transition between two panels Is proving the difficult part as there are so many "fractional" texture pieces as the result of the panel ending and a new panel beginning.

To get the seamless transfer between two panels would either require lots of time spent placing fractional adaptive components. Or building in the fractional pieces to the main panel and ensuring they would mate with other panels (coordination). Any thoughts on this part of the problem?

It's too bad revit does not have a project command similar to rhino. Then the trapezoidal panel grid could be projected to the cube and used to create the main pattern.

Thanks for your thoughts. So far this has been a fairly good exercise for me in terms of getting to know adaptive components better. The goal isn't to recreate what they have done but rather consider how it might be modeled or approached within the domain of revit.

Picture attached with circled fractional texture pieces.

[damon.sidel]

graphite, I wonder about the trapezoid-parallelogram pattern... is that really what the panels are? From this colorful image, there the extrusions seem to be based on a staggered, rectangular grid pattern. Two thoughts on this: (a) Could you model the whole thing based on a staggered, rectangular grid (and therefore not have fractional extrusions) and then add the trapezoidal joint in another manner? or (b) could you create TWO facades, one with the trap-paral pattern of large panels and a second applied facade of a staggered rectangular grid of the extrusions?

Also, to Alfredo's ideas, the one thing that keeps getting me is that it's a solution that creates a regular pattern. From the first black and white image you linked and from other images I've found on the web, it looks like it is an IRRegular pattern. Therefore, you have to either manually place all the panels or create a pattern then replace a few panels that are exceptions. graphite, in your exploration of this design, have you figured out if it is a regular pattern with exceptions or completely compositional?

[Alfredo Medina]

I still don't have any idea what Graphite is talking about in regards to those colors. All I can say is, if I look at the photographs of the construction, (which you can see in those links) it's a precast concrete panel, that follows a regular pattern, repetitive, and that has special pieces to turn the corners, that's all. The ins and outs of the volume of the panel could be done in Revit, too. It would take some time to do one of them carefully, but I insist, all this is doable in Revit.

[damon.sidel]

it's a precast concrete panel, that follows a regular pattern, repetitive.

If you look closely, it is not a regular pattern. Look at the direction of the angles. If it were regular, you'd always have a diagonal up, then down, then up, then down, etc. In fact, you'll see a number of places where you get two or more diagonal ups or downs in a row. Nothing too complicated, but adds to the complexity and cannot be a purely pattern-based curtain panel.


The ins and outs of the volume of the panel could be done in Revit, too. It would take some time to do one of them carefully, but I insist, all this is doable in Revit.

I don't doubt, nor do I think graphite doubts, that this could be done in Revit. The "ins and outs" on the panels that are represented by the colors in the image graphite shows, are are quite varied. They look like relatively straight-forward forms that could be modeled in Revit, yes. The composition of them is quite complex and could be done manually with quite a bit of time. My question is not just how you can do it in Revit, but how did they do it in another program and was it more automated?

I think this is a fascinating case to explore in Revit on a number of fronts. I think we can all agree it would be possible. My thoughts go to efficiency of designing this: can we explain when and why the exceptions in the large pattern occur? What is the best method for addressing these exceptions? Can we explain when and why the smaller extrusions/"ins and outs" occur, what is their logic, and what is the best and most efficient method for adding them, modifying them, etc.

Graphite, what do you know about the exceptions to the large pattern?
What do you know about the generation of the smaller pattern? Do any of the articles mention how they decide where to play what extrusion? Did they generate it based on an algorithm? Or something else?

[Alfredo Medina]

If you look closely, it is not a regular pattern. Look at the direction of the angles. If it were regular, you'd always have a diagonal up, then down, then up, then down, etc. In fact, you'll see a number of places where you get two or more diagonal ups or downs in a row. Nothing too complicated, but adds to the complexity and cannot be a purely pattern-based curtain panel....

yes, there are different rows with different patterns, but the number of different patterns is not too many, and each type of pattern could be a type of curtain pattern based panel or curtain wall panel.

[graphite]

graphite, I wonder about the trapezoid-parallelogram pattern... is that really what the panels are? From this colorful image, there the extrusions seem to be based on a staggered, rectangular grid pattern. Two thoughts on this: (a) Could you model the whole thing based on a staggered, rectangular grid (and therefore not have fractional extrusions) and then add the trapezoidal joint in another manner? or (b) could you create TWO facades, one with the trap-paral pattern of large panels and a second applied facade of a staggered rectangular grid of the extrusions?

From what I can tell each trap or Para panel has the same application of textured pieces within a standard module width with varying heights. The repetition of the texture between similar type panels can be seen here circled in red. This being said I would imagine there are probably three types for each panel1)no texture 2) large amounts of texture 3) less texture. I think this "family" approach is probably what took place as it would be much easier in the fabrication process to itemize and account for all extrusions needed for panel X. Also based on this second image it would seem that these panels were documented in such a way to note possible custom extrusions at the ends needed to create continuity between panels. The synosis blog also speaks about a total of only 20 individual textural pieces for the whole project. "Those 20 types would literally be constructed into 20 unique plywood and masonite molds, into which could be cast multiple flexible reusable formliners. These formliners were then placed into rectangular, flat forms, and arranged in groups to match the patterning logic clearly laid out in Morphosis construction drawings."

A) Would voids within a family be a possibility?
B) Since the extrusions are based upon the panel I dont think this would be as effective.

[graphite]

If you look closely, it is not a regular pattern.

Definitely not a regular pattern. It seems to be repeating bands on the one face but the other face there are less bands to work with. I have attempted a number of ways to model this so far:

1)textural pieces (adaptive component family- multiple types to create perceived sweep) hosted to rectangular face (adaptive component) hosted to larger adaptive componenet multiple times in both parrallelogram and trapezoidal shapes to create banding pattern A. Repeat process and vary pattern to create banding pattern B.= Shaping the adpative panel with hosted extrusions did not like it when I made the sides angled in the case of both the parra and trap shapes.
2) curtain panel pattern= did not work
3)Adaptive panel with hosted adaptive extrusions hosted to mass with subdivided surface. worked for planar surfaces but organic geometry proves problematic if adaptive panel does not have enough adaptive points. I would imagine the hosted textural pieces would inherit the same problems the panel encountered.
4) Curtain panel subdivided surface with adaptive modules.
So far option 4 has been the closest to providing trapezoidal and parralellogram geometry with textural pieces hosted. However pushing this geometry into a non-repetitive pattern still seems to be the issue.
89334
89335

[graphite]

On another note, how would simply paneling an organic form such as the interior work? Would adaptive component with 10+ points placed as repeater be needed to adapt to all the curvature?

[damon.sidel]

graphite, that's quite the adaptive component panel you have there. It adapts correctly to a non-rectangular grid?!

I don't see an option that you tried that creates two separate divided surfaces, one for the large trapezoid/parallelogram pattern and a second for the textural pieces. Based on the colorful image you posted earlier, it looks like the two patterns are quite independent and it could be helpful to keep them that way.

As for the organic form you are showing here, I'm stumped on this one. I'd definitely like to see what you come up with. The 10-point adaptive component may very well work, but you'll definitely be doing this manually.