Jump to content

Structural Topology Optimization


Recommended Posts

Hi All


I´m looking for some good sites on Structural Topology Optimisation and virtual evolution af form. I´m not sure these words make sense to everyone but what I mean by Structural Topology Optimisation is the act of analysing a specific structure and optimising the way the material is distributed (Try it here: http://www.topopt.dtu.dk/ )


Virtual evolution covers disciplines where the designer is reduced/elevated to a role where he or she is the one that descibes a set of parameters that will clearly define what is demanded of a given structure for it to be "succesfull". Then these parameters are fed to a computer that will generate random form and compare each iteration to the criteria defined by the parameters mentioned - this process continues until the form reaches a succescriteria.

Parameters could be: Light, wind, traffic, average height of a human, average height of a thigh etc.


Take a look at the work of Karl Simms: http://alife.ccp14.ac.uk/ftp-mirror/alife/zooland/pub/research/ci/Alife/karl-sims/creatures-demo.mpg


9.1 MB MPG movie


I have a little studio here in Copenhagen and I am very interested in working with optimization tools.


Oh - There is of course also Gehry Technologies: http://www.gehrytechnologies.com/


I don´t really go to many other forums but maybe you guys have some good links?!?

Link to comment
Share on other sites

There's been some work, although perhaps not as scientific as your suggesting (at least from my point of view).


Greg Lynn (http://www.glform.com/)


Karl Chu (http://www.sciarc.edu/v5/gallery/student_g_01_vert_chu.php)


Marcos Novak (http://www.centrifuge.org/marcos/)


All were professors while I was at UCLA, although I only had Marcos Novak (by choice, as I don't really buy into these types of experiments, at least as isolated design mechanisms - I did love Novak's class, though).


There's others, like Ben Van Berkel, Koolhaas, Hani Rashid, and probably numerous more by now, although the 'blob' era has thankfully passed us, for the most part.


There has been a lot of debate (and many a thesis using Maya scripting) about the 'author' and if it's really removing the designer's hand from the equation. Personally, I don't think you can and shouldn't.

As generative formal experiments approaching this can lead to new discoveries, and for that purpose it's great stuff, it's just when people claim that they are not part of the equation, or that it's truly 'pure' design (meaning based only on data and certain criteria), that I get skeptical.

In the end, the designer or architect will always control the final output, whether it be by entering numbers or cutting some basswood.


It's fun, though, and makes for pretty pictures ;-)


If you want more educated or specific data, I suggest you go to Archinect and post. You'll be sure to get some heated debates going on both sides!



Oh, I should note, Gehry (didn't have the opportunity to take his class, although I did have a great materails class with Randy Jefferson, his right hand engineering man) taught there, too, but I don't think his design goes anywhere near this methodology.

Link to comment
Share on other sites

Hi Marcus


Thank you very much - I think I will post this in the forum you suggested.


Gehrys own method of giving form is not related to this but Catia and Gehry Technologies might be usefull though...


From your post it seems that you consider this strategy to be abandoned? Can you elaborate a little on that?


I know Gregg Lynn of course but to me it doesn´t seem like he or any of these other guys really do what I´m after. To me it seems like the methods they are using are primarily directed at generating new shapes - not neccesarily appropriate ones.

I´ve seen one project where Lynn used traffic flow to generate shapes but the connection between the parameter and the generated shape was very abstract - as in "one more car = a litlle more blob". This is of course nice and as you say gives you pretty pictures but what I´m looking for is something like what the engineers are working on at TOPOPT. Simply and rationally shaping material into extreme appropriateness.

For this to happen I am thinking that the architect should define a set of criteria that a given randomly generated shape should be held up against. Out of one thousand generated shapes there might be five to consider as a basis for the next iteration. Do this one million times and you end up with somethin interesting.

I am aware that a building is very complex to define a set of succescriteria for but the experiments of Karl Simms shows that the aproach can yield very interesting results.


As you say I am perhaps more interested in the scientific/engineering approach. I would so love for the architect to be more of a scientist than a taste-judge. For me a more humble approach to form and willingness to abandon "reference architecture" (like the blobs or the height of a building in yards being the same as a the year a political document was written) would be a great goal to pursue.


Thanks again for answering Marcus

Link to comment
Share on other sites

I don't think it's been discarded, it's just been done by numerous people for a while now.


I'll have to respectfully disagree with you about the role of the architect. To me, architecture is about the 'art', the form making, the shaping of a building into something both rational and beautiful simultaneously.

I've seen projects that did what you suggest (as I mentioned, there have been several thesis projects based on using scripting to make controlled buildings. In the end, I've never seen or heard of anything that produces something 'optimal'. This is simply, imho, due to the fact that architecture is subjective. There is no way to remove 'taste' from the equation, no matter how many numbers are punched in. In the end, someone has to decide if the building is white or red, green or blue glass, stucco or metal, etc., etc.


There certainly are pragmatic applications that could use what you are suggesting, such as how Gehry uses Catia. This is were it will actually allow the architect to sculpt the form more freely, devoid of the limitations that are prevalent now (again, as Catia has freed Gehry to do what he wants, formally, without any noticeable limitations). But these 'programs' will only provide data, it'll always be up to the architect to decide how to use it to make a building work.

I strongly believe that there is so much more to architecture than simple providing shelter. To me, it is 'art' in it's best form, and that is what I strive towards.


Another program you may want to look at is Bentley's Generative Components. I believe it's similar to Revit, but not sure about the details. Morphosis (Thom Mayne was my thesis advisor at UCLA - he loved playing with form, but always with control) was testing it about a year ago.

Link to comment
Share on other sites

Hi Marcus


I understand your point of view and I think that in a way you are right. The only thing is that I think the truly great piece of architecture is actually done exactly in the way I imagine it - using a great simulation tool - the brain to get there...


We might disagree here to but to me the mind represents the best in computing power and by far the best in optimized data combination - Intuition and such.


I think the most timeless beauty is created in nature for now - our own bodies for example. They are certainly a result of a great and ongoing evolution and the beauty we see is to me a reaction to profound coherence in form and function in the broadest possible sence (including the ability to to feel emotion and react to any stimulation)


I have a little four year old girl and I know that the reason she truts me is because I don´t let her down (well, not a lot). Her trust is a result of countless iterations where my reaction has been incorporated into her database of emotional responses. She knows that there is a 100% probability that I will pick her up from kindergarden for example. If I start not living up to this responsibility she will ajust her mindset to this situation and stop trusting me.


I am not saying that there is only one solution to a certain problem - that would mean that you would be able to include any and all of the parameters affecting a certain situation... meaning everything that happened from day one till now.

But I do think that there is a very high probability that our own technology eventually will foster something that will rival and surpass our own capabilities - in a way I think this is our main purpose.

After all - what are we but a result of the evolution of a green technology that had the opportunity to encompass the entire planet. The earth too was formed once and not at all filled with life. Everything on it is a result of a big realtime evolution... Who is to says that it isn´t a simulation in itself (remember "The Hitchhikers Guide to the Galaxy"?)


The other day I read an article about gamers in online communities trading realestate, characters and items for REAL money... Not all of the game suppliers (if any) approve of this but think of the consequences... Reality was just expanded infinitly - new layers of reality added. Who can say that the events and actions taking place in a "virtual" environment are without effect in "reality" when people are actually trading "virutal" goods for "real" values - and what is a real value? Magnetic coding on you credit card or is it constrained to exchanging eggs and potatoes?


The virtual and real is mixing in a major way right now. The technology that allowed this is only a few years old and who is to say what will happen. Maybe simulated gaming environments will provide new solutions to "real" problems - and create new ones of course (chair shaped buttocks ;)).

It is marvelous to be able to interact with someone in a team even though they are at the end of the "real" world. This might very well be our future working environments - hey I´m hanging around in THIS virtual community communicating with you - maybe not exactly work but certainly something interesting for me.


Did you check out the topology optimisation page? You can try your own compliance design both 2D and 3D. All you have to do is to define a set of criteria for the form to be generated. Besides those criteria the algorithm in itself is of course also a parameter that affects the outcome - but I still got optimised(=beautifull) form out of defining a set of relevant paramters.


Also - reading a book on Islamic Architecture these days - beautiful stuff some of it. Many of the traditional setlings are arranged around a central parameter - keeping the temperature beareable. This results in beautifull wind towes, little fountains and other artisticly arranged elements. With the right computing power this could certainly have been simulated.

It took a long time to cultivate this typology because of the whole trial and error methodology combined with simulations within the minds of the builders. Maybe this is not art but to me it is certainly beautifull and I would very much like to pursue this method. It is "form follows function" expanded and elaborated - and I don´t mind at all if the guys building the windtowers do a little ornamentation...


I am arranging a little seminar on all of this in the spring so I am really interested in all the references you might have and I certainly respect you point of view - to me its just another set of parameters I have to incorporate into our communication ;) - and it is so much easier to define your own views when you have someone with a different standpoint - so thanks a lot!


Oh - by the way - don´t you think that the Virtual Creatures of Karl Sims are cool?- so few parameters and such fascinating unexpected beauty.


Do you know of anyone I could contact who is really on the cutting edge with this stuff? It would be great with someone with a scientific background?

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Create New...