Beyond BIM: transitioning from 3D Generative/Parametric design to the industry (2D?)

Replies to This Discussion

Permalink Reply by taz on March 22, 2010 at 9:24am

Whoa, you're opening Pandora's box (again)...

Note, however, that the automobile and airline industries are much more interested in optimization (and optimization software) since they are mass producing a product. Material weights and quantities play a much more significant role in the cost and performance of a design than they do in architecture.

Architecture, for better or for worse, will always be a unique, one-off product.

More specifically, it's really hard (not impossible...) to have a wholistic discussion since the delivery of most 3D projects is highly specific to the project and the capabilities of the contractor. Case studies may be the most beneficial way to collect information.

Jon's examples (and the article I posted) are digital 3D integration projects for structure. Structure is usually a good candidate for 3D project delivery (since structure is built first and everything falls out from the structure) but in addition to a Contractor with a suitable skill level you also need a knowledgeable Engineer.

The DP article you posted for Beekman Tower looks like 3D project delivery for the facade. The facade is usually the most expensive component of a building and can benefit the most (realize the most cost savings) from rationalization/optimization/parametricism which can justify the cost of employing Gehry Tech (they like to sell service contracts).

Then there's BIM in relation to mechanical engineering... I don't know too much about that, but Contractors like it because it saves them money. Getting PAID for the delivery of BIM model is also a touchy subject.

For other examples you can look at the work/writings/lectures of SHoP for many interesting examples. On the scale of a installation/pavillion/canopy/small house/pedestrian bridge where one entity (A/E) can produce all the fabrication files there is much more control over the design (see Camera Obscura). On the scale of a building, you can see that most of their custom work is the building envelope which is, historically, the responsibility of Architect.

As for software, I would caution against using GH exclusively. GH is still WIP software (read: beta) which is continually changing on a frequent basis. Yes, GH is a useful tool, but using it exclusively for a project in an office is very dependent on the scale of the project and the timeline. GH "investment" in a office may not easily translate from project to project (depending on how much things change from GH version to version). Maybe in a year this won't be as much of an issue...

SHoP used GH exclusively for this pavillion (I saw a talk on it, there may be more about it on designalyze), but it's small and quick.

Nathan Miller used GH for (the structure of?) a stadium but I think with a definition of the complexity shown he's stuck with an older GH version (Nate, I think I'm paraphrasing correctly, but please correct me if I'm wrong).

Just some thoughts...

-taz

Permalink Reply by Damien Alomar on March 22, 2010 at 11:59am

Well, this is a complicated issue, and unfortunately, no specific answers. I think there are two fundamental questions here... one's a software question and the other is really a professional one. Both have significant business consequences so that is still something that should be carefully considered.

In the software realm, I think your correct in that the "BIM" products that are available are more archaic then their "groundbreaking" image will lead you to believe. IMHO these products don't have the capability to really create the proper geometry for a direct model to manufacturing type of situation that is evident in other industries. Also, these products are geared towards creating standard architect where a wall is a wall, a window's a window and a door's a door. Generally, once you step out of that fairly restrictive box, you're left completely up a creek without a paddle in sight. Needless to say, your answer most likely does not lie within those packages. GC may be able to help make microstation a bit more flexible, but I'm really not sure to what extent that will actually satisfy your needs.

The other end of the spectrum is CATIA/DP, and although it is immensely capable, there are a whole lot of other considerations that need to be made. Geometry wise, you will be able to create, control, and manage all of your design exactly how you need to. CATIA is used to build cars and aircraft down to the last nut and bolt (threads and all), so you have all the capabilites that you'll need to do the work that your looking to do. Not only that, but there are ample drafting capabilities within CATIA, so the short term goal of needing 2D drawings won't be left on the wayside (I doubt it ever really will). The problems that go along with this are costs, time, and implications up and down stream. CATIA/DP is very expensive, not only in the cost of the product, but also in training, and you WILL need training. If you're disappointed about what's available from Bentley, then realize that you'll be getting even less from Dassault. GT really is more of a consulting firm at the moment in that a large amount of what they do is actually work with firms to have them use DP. That to me points to it being a hard platform to really have a whole office working with efficiently, but that's more speculation than anything else in this response. The other issue is that CATIA take a lot of time to use. The parametric aspect means that there's a lot to setup, and if that setup isn't exactly what's needed, then you may have to go back and rebuild parts of the model. This is somewhat diminished with experience, but its always there. The other thing is that outside of generating 2D drawings from the model, it may be harder to transfer that model towards other parties (ie contractors/engineers). Some may have the capability to take a CATIA model or an IGES model or something, but others may not. Since a lot about the modeling to manufacturing process isn't really standardized within the AEC industry, there also isn't much in regards to guidance out there.

The other thing to consider here (which you briefly touched upon) is the use of analysis programs on your design. That's a whole other set of issues, and one that you should consider as that will likely become a more important aspect of the design process as the profession progresses.

Your professional questions really seam to center around using a model for manufacturing purposes. At this point, this really only done on smaller aspects of a project as opposed to the whole thing. Also, most firms who are doing a lot of complicated fabrication have people who are familiar enough with the fabrication process to understand how models need to be constructed in order to be manufactured. Modeling for manufacturing has more requirements than just modeling for drawing generation or visualization purposes, so its important to understand those requirements. Working with a fabricator can help make sure that you get them what they need, but getting the model right is an additional requirement of the process. The other side of the model to manufacturing process is having the manufacturers, and in architecture's case, the contractors, on board with this and be willing to completely shift their typical way of working. Some contractors (generally the larger ones) have already started to integrate a "drawingless" process, but as I understand it, its not 100% of the way there. Moving towards this generally involves a lot of risk that the architects need to take, and in a profession where we're told from the top down to minimize or avert risk, this is a significant aspect of taking on this process. Personally, I think that taking responsibility for risk also gives you the opportunity to control it, so if you can properly control it, the end result can be positive. How you feel about risk will dictate what your comfortable with taking on.

I could continue, but I think that I've covered enough for one post.

Permalink Reply by Alpha222 on March 22, 2010 at 1:35pm

Demien, you must be a great teacher! Glad to hear from you.Thanks.

Permalink Reply by Stan Carroll on March 22, 2010 at 10:29pm

For anyone to provide you with the most relevant feed back it would be helpful to understand the type of work you are wanting to do. If you are doing large stadiums or Hospital work I would recommend one approach. If you are doing residential or small retail I would recommend something completely different. Having been in the architectural business for 25 years I offer the following illustration of how I am using Rhino/Grasshopper, its strengths, and its weaknesses as I see them.

My current work hovers somewhere between sculpture and small commercial architecture and involves in-house design and Digital/analogue fabrication, usually metal. In the past I have also worked for large traditional A/E firms of 100 people, and smaller 10 person Architectural Design Firms also so I understand those types of practices as well. I find the Digital Design/Fab process to be a welcome change in a profession that has been due for an evolutionary leap forward.

Rhino/Grasshopper is a very intuitive (especially for users with autocad history) design tool primarily with the added feature of providing a seamless transition to digital fabrication whether in-house or not. Up until I discovered Rhino, about 7 years ago, I used AutoCad for the 10 years prior, which offered virtual 2D hand drafting with a few added CAD features. It was really still just traditional 2D drafting with a mouse instead of a pencil. During my autocad period, design did not really change much and remained a more traditional process and CDs were done in AutoCad. Once I found Rhino, my design world immediately became one anchored in 3D form.

In a nut shell, Rhino and Grasshopper are a design tool and are best utilized to design 3D form well suited for CNC fabrication. It is particularly strong when the forms start to stray very far from Euclidean geometries. That being said, it is not my tool of choice for traditional architectural Construction Documents (CDs) nor does Rhino claim to be a significant CD tool.

In my early years of Rhino, the ability to parametrically study a design solution did not exist. Each significant design iteration required the designer to pretty much start over with a new model unless the change was fairly simple. With the advent of Grasshopper, parametric variant studies are now one of its greatest strengths. Grasshopper provides the ability of any number of extremely complicated relationships to be established then instantaneously varied and studied without writing a single line of computer code. Hundreds of combinations can be studied in an extremely short period of time. This is the power of Rhino/grasshopper I value most. A great many Grasshopper users also choose to create incredibly complicated geometric forms. This is another of Grasshoppers great strengths however, many of these "over the top" theoretical forms, though very beautiful and stimulating, seem to remain theoretical or at least prohibitively expensive to actually construct at an architectural scale and with our current state of the art of construction technology. Since my work revolves around built form, I remain tethered to build-ability. Grasshopper paired with CNC fabrication re-catagorizes many complicated forms from "unbuildable" to "very buildable". But even with reality limiting my outcome, Rhino and Grasshopper are incredibly powerful tools able to manage challenging 3D forms while providing ease of virtual infinite variant study.

The bottom line of today's architectural process analysis is that there is no silver bullet in design software. The current state of the art of the architectural process utilizes many types software even within a single project. The work flow of a particular project is as highly sensitive to project constraints and opportunities as the design solution its self. If your work varies, so will your process.

And to your second point regarding industry guidelines. BIM, digital fabrication, sustainability, are all such new complexities in a profession that is changing at an unimaginable rate; never before seen in the profession. I am not aware of any guidelines to this issue but would love to hear from someone that might know of some guidelines particularly about risk and responsibilities regarding sharing digital models with the contractor.

Stan

Permalink Reply by Philipp on March 23, 2010 at 2:03am

A propas, a little question about that:
there is a pissibility in Revit to link revit components (means -all details) to "mastermodel" made in other softwhare like sketch up or rhino. But it will be wonderful to invent something to connect it to GH model directly.
I thinck this can open a new stage of intagrated design process.. and i thinck this cant be a big problem...
And for me it a bit good guidline for parametric model -> 2d plans goal

Permalink Reply by David Merlin on March 23, 2010 at 3:46pm

Boyd -

Check out this conversation on DesignReForm between David Fano, Federico Negro (both were working at SHoP at the time and now have their own technology firm called CASE), and a guy from StructureCraft named Santiago. The impetuous of the conversation was the relationship between Inventor and Revit, but it quickly starts to address larger issues:

http://designreform.net/forum/index.php/topic,705.0.html

AIA document E202 establishes BIM protocols that could help you develop firm standards. Here is a link to a sample document:

http://www.pat.ca/files/pdfs/AIA_091708_E202-2008_eSample_Blank.pdf

I also think Damien brought up a great point about liability that many firms are grappling with at the moment. Joshua Prince-Ramus of REX speaks to this issue in a recent TED post:

http://www.ted.com/talks/joshua_prince_ramus_building_a_theater_tha...

Hope this helps,

david

Permalink Reply by sivam krish on May 22, 2010 at 9:01am

One way to to ensure independence and interoperability it to maintain the generative key design data separately. One way of doing this is in www.opengenerativedesign.com

Permalink Reply by Santiago Diaz on February 22, 2011 at 11:25pm

Maybe it's not too late to add to this discussion.

I posted a new topic which compares an Inventor model with a GH model, here:

http://www.grasshopper3d.com/forum/topics/refactoring-parametricism

Permalink Reply by dominic on April 8, 2011 at 3:50am

 

Has anyone checked out RhinoBIM?

 

It has a GH plugin for analysis, but does not seem to have much parametrics. It also seems to be bundling VisualARQ in as part of the package. I don't think there are very well integrated at the moment, but I am sure that will change.

 

On the face of it, the mixing and matching of plugins is something that can be super powerful. In fact, I would have thought that it essential long term. I hope RhinoCommon and GH will really provide the underlying framework for more of this. I think Revit now has a Dynamic Update API that addresses a lot of the requirements for managing multiple plugins.

 

Autodesk has had a bit of success pushing Revit and BIM in the UK, lately. I suspect that a lot of firms are looking at whether to 'revit' or not.... again.

 

GH's interaction with Revit or even Rhino needs to change a bit if it wants to make headway in the mainstream BIM environment. I don't think its can continue to throw geometry over the wall, downstream to the nonscripting world, without the driving intelligence reusable. Would be interesting to see how DesignScript handles this. Aish mentioned at AU 2010 that the intention is for DS is to able to transfer design intent by exchanging scripts, that would re-execute the model on the receiving platform. I guess Inventor or Max/Maya/XSI would have a DS plugin? A sort of dxf for scripts?

 

I suppose it would only be a matter of time before we get script to script translators as well. GH <> GC should be relatively easy. Both are pretty imperative and comparatively conventional language wise. Not sure about DS, which seems to have a somewhat more complex language.

 

There are probably people out there already testing out other language paradigms with 3d modeling, like functional reactive languages like Yampa etc. http://www.comlab.ox.ac.uk/ralf.hinze/WG2.8/25/slides/john.pdf And there is always rules based like CityEngine... split grammar stuff. Hopefully, with RC/GH now being open source, we will see more interesting stuff.