WE THINK THIS IS A NEW PARADIGM FOR DESIGNING
Hattie Hartman talks to Patrik Schumacher of Zaha Hadid Architects about Digital Project (DP). It was created by Gehry Technologies in 2004 by licensing the Catia software technology originally developed by Dassault Systemes for the automotive and aerospace industries. Frank Gehry & Partners had been using Catia since 1991 and formed Gehry Technologies in 2001 to market this expertise to the construction industry. Digital Project is distributed in the UK by CenitDesktop. It is used by about 250 major practices worldwide, of which about 30 are in Europe.
What's different about DP? What does it enable you to do that you weren't able to do before?
We started last year with a series of test projects using Digital Project. We are always trying to upgrade our digital processes because our work relies on these processes to fulfil our dream of a uid and organic architecture.
Curvilinear surfaces produce a lot of non-repetitive geometries and we initially sketch-modelled these in programs like Maya and Rhino. Maya is the fastest for pure sketch modeling.
Before DP, we had to rebuild the geometry in Rhino, and we found limitations there. So far we have been using DP for projects which have already been designed and call for the control of complex geometries with non-repetitive elements. Once you build up the complexity with multiple angles and curved surfaces, if you change, even just by 5cm, you have to adjust the geometry and redraw everything. It was our intuition that if you could set up the geometry parametrically from the beginning, you could build in anticipated variations. Then you can adjust the parameter without rebuilding the complex geometry because the elements are networked together.
This is particularly interesting when you have a building which has many individual elements which are similar - a genotype, like the tile on our project for the Basel Concert Hall, which is a space-frame structure with a tessellated surface of cast aluminium segments. You can set up the geometry of the panel and the joint detail. Then it's easy to adjust, for instance the location or angle of the joints, without losing all the work.
You can produce schedules of elements automatically.
The other aspect is the intellectual challenge. We think this is a new paradigm for designing. That means we want to design from the very beginning, not individual artefacts, but genotypes of potential artefacts. What are the parameters? What is the essential morphology and what is the universe of possibilities in which it varies? On the more academic side of our work, we work in Vienna with a Masterclass and with the Design Research Lab at the AA on parametric urbanism.
We are also starting to produce urban studies and we currently have large masterplans in the office, in Bilbao and Istanbul. The idea is that you can treat a segment of a uid grid as a genotype and you can build a diagram and let it proliferate across a field where each cell is different because there are distortions, and then you change the key parameters of the genotype or you change the network and allow it to proliferate again. If you have a similar setup of a building versus a site and 150 or 250 sites, you can design a genotype building model and let this proliferate. This is the kind of thing we are doing under the notion of parametric urbanism.
You can do this not only with a lump of mass, but also by building internal complexity, using voids, cores and key systems such as circulation, floor plates, and envelope. Subsystems can be linked together, allowing for an internal complexity to run through the field.
We have fields of towers and no one would like to contemplate the game of repetition of the '70s. These towers need to be quite malleable, to come down in scale, to flock around a centre and grow, perhaps to stretch at the same time, so that the growth in the vertical dimension is matched with a certain factor of growth in the horizontal dimension. You can script these by writing functions. You write change of dependency, and that is exciting. We have realised enormous productivity gains, designing urban fields using adaptation.
Adaptation means that the genotype is able to be modular with respect to conditions, which then produce a philotype. A condition could be, for instance, context. We usually like to work with gradients. It's a great paradigm shift that you don't want to have only discrete types. You want to work with variations and create morphological gradients. Height gradients are very simple, and also typological gradients.
Some of the tools we were working with previously, like Maya, give you similar techniques, such as lofting and morphing, but they don't give you a proper handle on the components, they just give you surface. If you think of this in terms of components which have a tectonic logic, then if the global envelope changes, the parts have to readjust. This is where we use DP - for the visioning aspect.
More immediately, productivity depends a great deal on time in manufacturing firms, like Permasteelesa, a great facade company which has geared up to pick up these kind of files.
Certain engineers, like Adams Kara Taylor, are currently using DP, and in this case you can build an integrated structural 3D model and plug in surfaces and facade tiles. Ideally you can do M&E systems.
We have to build in quasi-physical resistances. In Maya we were free to design any form we wanted to. In DP you can set constraints for the bending capacity of materials at the initial stage, and then you can play against it. As we work, we don't have to continuously stop to measure and ask ourselves 'have I reached this radius?'.
DP refuses to mould beyond the material capacity. It also refuses to allow components to interpenetrate, so when you model you can build in a 2cm tolerance joint from time to time. This enables us to be sure we can handle large numbers of different components precisely. You then draw manufacturing schedules from the model.
Can you elaborate on the design process?
For form-finding, there is potential through constraint setting, but it is not the same as fast sketching, because you have to think through your system logic before you build it. We are starting to get used to it. The next loop will be when we start a new project and can think ahead about how to generate a project from geometric components.
Nearly all our projects have complex geometry, or lawful differentiation. We're not into random forms of free form. We like to mathematically differentiate form according to formal laws.
This is an aspect of beauty, like physical natural systems which have a very strong coherence such as fluid dynamics or certain landscape formations like glaciers. We use certain organic systems as inspiration and create force fields in which forms are shaped.
This paradigm is not only functionally adapted, but becomes visually appealing by reducing complexity, for instance with the cornerless flow of spaces like at BMW.
Although the architecture is a rule-based discipline, it is extremely complex so it is difficult to stick to the rules using only your design intuition. When you make a drawing using some curves, such as a parabola, or a corner with two large radii or two small radii, it becomes incoherent and lacks elegance. With these new tools, we realised that the scripted setups could achieve a higher level of coherence in design - which means beauty - in the project.
Isn't it still a subjective judgment?
I'm always inquisitive, so I have to go and analyse: why does this look better than that so that I can teach someone to tell the difference. Why is this line not working? Is there a change in the character of the curve? We have done a chair that is not fully satisfactory. You compare and learn why your aesthetic sense is reacting and how you can avoid these disappointments.
We've learned to insist on machine fabrication. With the hotel in Madrid, the manufacturer initially attempted to avoid milling every surface, but we had to reject it.
The otherworldliness had been taken away. We realise that our language is slightly different from Gehry's. We don't want to build complexity if we cannot fully resolve it. We'd rather make it simpler and avoid sharp angles to bring it under control. We only want to make things complex if we can fully resolve the manufacturing aspects.
What constraints do you put into the model to make a building?
We are doing a twisted tower in Milan. It's a split tower with a gap that we shift around with a curtain wall of glass sheets. We've designed it to a degree of twist in the surface so that you can use straight glass and pull the corner to force the glass sheets into a slightly warped frame. The maximum warp of the sheet is used as a constraint to generate the surface.
Is it difficult to train people on DP?
It has a bit of steep learning curve and we are very computer literate. We had a great series of sessions over three or four months when the projects were translated. They became more elegant, more precise, more controlled. The surface quality was increased, which is a whole other discourse because it is super sensitive. You want to heighten your steadiness in all directions.
It's so easy to make mistakes. There's an enormous difference between elegant beauty and something which looks hand-finished. Very sophisticated techniques are required to control this in the computer; otherwise it becomes all blodgy and dented.
DP can help generate so-called class A surfaces. I suspect you need people who are already computer literate and have been trained in the logics of programming, such as what it means to set up functions. We chose people who had an inherent interest and keenness, who conceptually knew what they were looking for.
They had a week-long training seminar in Oxford to do the groundwork and came back here to start project-focused work by rebuilding things from rhino with hands-on tutorials.
Chris Jacata, one of the minds behind DP in Gehry Technologies who had a lot of experience of construction, has been helping us.
There was a good workshop atmosphere.
We used the Glasgow train museum, which has a doublecurved shed roof with complex surfaces. We were able to model the whole surface, which is tessellated with shingles with a set overlap.
How has DP affected the relationship between your productivity and creativity?
Productivity, creativity and elegance. The creative aspect is that you expand your horizon, you feel freer to explore forms which you would have shied away from before.
There is an inherent potential creativity in this kind of software which enables you to set up systems which produce fields of forms. You free up your own thinking because you gain a certain confidence. We are about realising buildings and we have clients we have to convince and we have to convince ourselves that we can get this under control.
For the mind of the architect who loves constructing and problem solving, there is also an interest in new technology in its own right. The first round is not yet more productive. We are investing, gearing up.
On the manufacturing side, you don't necessarily find someone who can take this intelligence and go further.
We are doing some of the work which normally a contractor would have to do.
Sometimes the contractor is not set up with a technical staff able to receive DP and ends up redoing everything. We are making an advance investment which will come to fruition in a few years when all the systems are streamlined down the line.
Different aspects will come to fruition. DP is very good at delivering pure repetitive systems, but it also has a radical capacity.
Experience in the aircraft industry shows that this is the future.
For large complex buildings, all the intelligences will feed into one digital model as in aircraft design.
You've spoken of your dream of a uid organic architecture. Can you elaborate on your urban vision?
The Singapore project was a uid grid, which was about adapting a new urban texture to diverse contexts and absorbing these external differences with an internal smooth differentiation.
We can unify a number of previously different urban textures.
In our masterplan for part of Beijing, which had about 25 buildings, we came up with three phenotypes: a point tower, a boomerang slab and a perimeter block. These blocks shared certain features, such as corner treatments and directionalities. We had more than 20 towers and we didn't want them to be all the same, yet they have the same logic, the same phenotype. We can construct a building diagram with a number of elements, and let this species populate the field. This is a grand vision.
We are now working on a very ambitious large territory, which is a new subcentre in outer Istanbul. We have underlaid a soft grid and used a number of types, which have quite a range of modulation. Courtyards can migrate to the interior of a tower as an atrium. Another transformation sits on the crossing point as a tower where the angle changes, the height changes, the bleeding out of the four sides changes. We want to build these as genotypes and we have the idea of a number of species populating a field.
These are biological metaphors. Different species can be each other's conditional context.
This means you have an ecology of adaptation because you can set up relationships, so that if a tower is close, you can reduce the height. Any variable, such as transparency, direction, colour, or angle, can be networked to any other variable. For example, the transparency value of a glass surface can be related to the proximity or height of another building. Or you can set all your proportions in relationship with the height. We are also doing this in a tower. The envelope is differentiated so this element is intelligent and it always fits itself. If you change the envelope, the oor fits itself. You can also have central control objects which all other objects look towards to receive their value but in a different way. If you reduce the height of this tower, each of the surrounding buildings has a prearranged way of responding. You give yourself hands for manipulating the field. The exciting thing is that this is kind of like cross-eyed laws of nature. You are writing the lawful correlations of element to element. That becomes a grander vision, a new ontology of the design world.
Before, you were restricted to build a shape, and with each new shape you would have to make a new decision. You cannot handle it. You run out of time. Talent is the ability to have an intuitive grasp of these kind of laws. You are giving yourself a rigour, an eye hand control, an intuitive judgment of where there is coherence with simple things like proportion. We learn from dynamic equilibrium composition or pseudo- physics which proportions have to do with an intuitive grasp of what feels right structurally. You expand the range when you lose symmetry, but you still have a range of proportionalities. This is talent. You can reproduce what talent was delivering on a higher level with more complexity and more precision.
What does Zaha think of all of this?
What is satisfying for Zaha is that a lot of the new paradigms and stylistic agendas vested in this are within the desire which existed pre-computer: distortion, the fast-moving hand, the idea of a lot of fragments cohered by a force field. The sensibilities are the same.
Has this new tool enhanced the understanding that informed your original designs and extended your creative range?
Yes, definitely. It's been accelerated. When I look back at our designs of five years ago I see the relative immaturity.
Are your buildings going to look dated?
These tendencies are not like the fashion world. You can't invent a new architecture every Monday, as Mies said. Our generation is so much more sophisticated. I believe these advances are also productive with respect to the life process.
We have two projects which are the apotheosis of all our agendas, both in Dubai. One is a large project with three towers, retail, a financial centre, a cultural building and a bridge, which are all interconnected. We can now handle this scale without it being jarring. That goes all the way into the detail where things key into each other seamlessly and you don't pollute with corners.
We can also see the progress from deconstructivism where we allowed ourselves to see interpenetrating forms and juxtapositions which very quickly became a collage which was chaotic and unmanageable. Now we can handle interpenetration and multiple direction so that you can flow through it with ease and understand the composition even though each side is different.
On an urban level, you know when you are drawing closer to the centre, there is a logic to the way things unfold.
That is why we are winning these competitions even though the work is so extreme because we can make an argument why this is productive, why this is meaningful. That even goes to the detail, minimalist detail, because the non-minimal detail would distracts from understanding the space as intersection of wall units or a series of planes which follow through. I don't believe that these aesthetic desires have no meaning. There is an underlying functionality which compels me to this arrangement in space.
Digital Hadid by Patrik Schumacher, Birkhäuser, 2004, £9.