A Vision for Parametric Design
Architecture's Parametric Tool
Some researchers believe that parametric architectural software would ideally capture conceptual, three-dimensional design intent, enable the automatic parametric generation of iterative solutions throughout design development and documentation, and accommodate the accumulation of data as the computer model ultimately becomes a complete "kit of parts" representation of a building.
Furthermore, aesthetic, technical, and economic implications of proposed modifications could ideally be evaluated and accommodated at any point during the design process, and the geometry of all components and subassemblies affected would update automatically in association with changes to the overall configuration.
In pursuit of this vision, Robert Aish, the director of research of Bentley Systems Inc., has been creating and testing CustomObjects, a prototype associative and parametric computer modeling program specifically designed to be used at the building scale.
CustomObjects is a unique architectural program in that it allows the user to define and completely control the behavioral rules of the software by graphical or code-based interaction and direction. And it does so with a full B-spline palette of geometry that accommodates more complex geometries than existing architectural modelers.
In his development process, Aish has been collaborating with leading architects and engineers, including Foster and Partners, Gehry Partners, KPF London, Arup, and Buro Happold. He has also been working with researchers in digital design and construction from around the world.
In May 2002, 150 attendees at the fourth annual research seminar at the Bentley International User Conference witnessed 12 presentations demonstrating the potential of CustomObjects. The presenters included architects working both with innovative geometry and with more conventional geometry but responding to context-specific constraints.
One way to test new software is to demonstrate how it might have improved the design process for a recently completed building. At the seminar, David Kirkland, of Nicholas Grimshaw and Partners, presented a study he and Aish conducted on how CustomObjects would have assisted in the design of the Waterloo International Rail Terminal.
Conceptually, Waterloo is an elegant and sophisticated train shed in the tradition of the great Victorian railway stations. Working with complex site and programmatic constraints, Grimshaw had to vary the spacing between trains throughout the length of the terminal. His solution was a complex and beautifully sinuous form framed on a series of asymmetrical, hinged assemblies of "banana trusses." The trusses were conceptually similar, but they varied in size along the terminal's length.
What if, Kirkland asked, one truss could have been digitally modeled and then dynamically modified in a malleable, rule-based, interactive process? And what if this assembly could then be arrayed down the length of Waterloo Station with each truss adapting automatically to the building's varying width?
If CustomObjects had been available to capture the designers' intent through those parameters, it could have responded to changes late in the design process. The architects and engineers could have altered the appropriate parameters and each truss design would automatically respond.
Waterloo Station is a unusual building, but Kirkland's study highlights a principle that is applicable to all building types. As CustomObjects is further developed, it may become possible to model both custom and standard building components and then assemble them into dynamically malleable digital objects like the Waterloo trusses.
Parametric Architectural Surfaces
My own practice, The Freeform Research Studio, collaborated with Aish on the testing of CustomObjects by exploring the software's capacity to parametrically study the enclosing glass and polymer skin of a conservatory.
The result of this collaboration is a computer model composed of 80 surfaces, and six layers of 720 nested objects. Through graphic programming, the overall architectural form can be manipulated so that a theoretically infinite number of iterations can be dynamically evaluated.
The overall form can be manipulated in real time by graphically altering a variety of parameters that automatically adjust heights, footprints, and the "bulge factor" of surface curvatures. Furthermore, the building's form can respond formally to any site-specific topography.
Computer Programming as a Design Tool
One common debate about computational design tools is whether designers want to do computer programming to achieve design results. I believe that all programming decisions, especially for parametric software, are essentially design decisions. The question is whether a designer wants to be passively or actively engaged in programming.
Designers who use off-the-shelf software are passively accepting the limitations embedded within the code and the graphical user interface. However, few architects are interested in becoming skilled programmers.
Nonetheless, a few far-sighted practices, notably Foster's and Gehry's, have computer programmers not only on staff but fully integrated into design teams. To make parametric modeling feasible for more mainstream practices, CustomObjects will allow the designer to chose how much to customize the software by programming.
The intention is that designers using CustomObjects may chose to write code directly or to "program" via graphic manipulations. Such graphically driven programming has the potential to harness additional talents of architects who might otherwise have only a passive relationship with their digital tools.
Kevin Rotheroe owns The Freeform Research Studio in New York City, an architectural practice and research organization investigating applications of advanced digital manufacturing. He is also a professor of digital design and manufacturing at the School of Architecture at the University of Illinois Urbana-Champaign.