Computers Building Buildings

by by By Raillan Brooks & Lucas Morduchowicz

illustration by by Robert Sandler

In November 2011, Michael Hansmeyer’s exhibit “Subdivided Columns – A New Order” at the Gwangju Design Biennale made quite the splash. Hansmeyer, an architect and computer scientist, built 16 structurally sound support columns. Or rather, his computer did. Hansmeyer specified certain parameters, like the proportion of the capital to the shaft, and allowed the rest of the column to be created using a computational generative process. Basically, the computer took the “rules” fed to it by Hansmeyer—what shapes are allowed in the final structure, equations that determine the most efficient load-bearing frame, and so on—and subdivided the surface based on that algorithm. He made no further aesthetic or design decisions. Hansmeyer cut the facets out of plastic using a computer numerical control machine and layered them on top of one another, translating the column from virtual to physical reality. This process created a series of complexly detailed columns that, according to Hansmeyer’s website, “exhibit both highly specific local conditions as well as an overall coherency and continuity.” He let the computer run its course, placing no constraints on the size of the columns. No resulting column was shorter than eight feet. The average diameter of a column was 50 centimeters, but the largest column’s circumference reached eight meters because of its many intricate undulations and edges. Some columns had as many as sixteen million facets in dazzling arrangements of the shapes he initially specified. On the use of computation in architecture, his website says that “a computational approach to architecture enables the generation of the previously unseen. Forms that can no longer be conceived of through traditional methods become possible. New realms open up.”

As a relatively young field of study, not many concrete names have been set forth to define what Hansmeyer did. Different terms abound—computational design, data-driven fabrication, generative architecture—and all challenge the same notion. Designing a structure, be it a building or part of one, is about making decisions about the way one wants the piece to look, feel, and interact with the world around it. With new innovations in the fields of computers and architecture, the traditional process of architectural design is changing. Joy Ko, a professor of architecture at the Rhode Island School of Design, echoes Hansmeyer’s conceptual conceit. She says computational architecture “encompasses many approaches that make use of computation in that architectural design process.” In other words, computational architecture is architecture in which the decision-making process is not simply assisted by calculation and computation. It is architecture radically transformed, in which all of the architect’s decisions are made prior to designing, and those decisions are limited to simple parameters instituted all at once.

Hansmeyer’s columns in 2011 are part of a professional oeuvre exploring the applications of powerful computation to design. And he’s been at it for a while. His project “L-Systems,” released in 2003, took a model developed in the 1960s to describe plant growth and fed it into the same program he later used to create his columns. Other projects include creating forms based on the mechanisms of cellular structure, sculptural designs that use cubes as the only basic unit, and a forthcoming project involving mathematical models of insect nests and their applicability to creating human dwellings.

The use of computation in architecture is not new in itself. Architects have been using 3D modeling programs to assist them in the design process since the early days of software. However, very few architectural firms have historically used computation as an integral part of their design process. In fact, even ten years ago, the number of firms relying on computational architecture could be counted on one hand. It wasn’t until so-called “StarChitects” like Frank Gehry popularized it in the early 1990s that using modeling software was considered a part of architectural practice at all. Gehry’s Dancing House in Prague, built in 1992, with its nonlinear support system and abstract arrangement of shapes and lines, was among the first buildings in the world to have its structural integrity modeled by computer software. Today there are many firms using computation to heavily assist their design process. They often have internal computational design consultants that assist other designers who are less acquainted with computers in using novel computational architecture approaches and methods while maintaining their signature style.

Hansmeyer isn’t the first designer to leave architecture entirely in the hands of machines. Daniel Libeskind, the famed Polish-American architect behind the Jewish Museum in Berlin and a number of other high-profile projects, constructed what he called a “writing machine” in the early 1980’s as a conceptual project. His question was, What if we deconstruct a building into its constituent parts and assemble them according to a static, rule-governed system? By turning a series of cranks, this machine, built from what look like typesetting blocks etched with conventional structural elements, would produce the floorplan for a building, oftentimes creating uninhabitable spaces and absurd renderings of structures with no distinct separation of interior from exterior. But it was the first recorded effort to create an autonomous architecture, letting gears and cranks do the work.

Hansmeyer and Libeskind participated in computational architecture as artists, but there are plenty of examples of the practical applications for the field. One particularly useful movement that has come out of computational architecture is the gradual acceptance of programs that fuse 3D modeling with structural engineering. The development of ArchiCAD is evidence of this shift. ArchiCAD is a BIM (Building Information Modeling) computer-aided design program used by architects. When it was originally released by the Hungarian company Graphisoft in 1987, it was the first program that ran on personal computers to have support for both 2D drawings and parameter creation of 3D shapes. Today, in its fifteenth iteration, ArchiCAD fuses 3D modeling and structural engineering. In a few short years, Graphisoft has added virtual tracing, view rotation functionality, and, most recently, a 3D editing plane and a 3D feedback plane. These additions have improved architects’ ability to model their work in three dimensions while keeping their design structurally sound. While presently this program and others like it don’t replace actual structural engineers, they do allow for a much more streamlined process in the design of architecture.

Computational architecture is no magic bullet, though. Joy Ko views it as a collection of transformative new methods rather than the final frontier of architecture. Traditionally, the introduction of new tools and procedures to the world of architecture has brought about major changes to the field. Computational architecture already permits architecture to rethink the building as a static concept. For example, Marco Verde, one of Hansmeyer’s collaborators, has used ArchiCAD to create floor tiles equipped with sensor technology. The sensors pick up foot traffic in a room and also allow the tiles to communicate with one another, making any room with these tiles automatically customizable (with, say, heat or light) based on use patterns. What design like this means for the field as a whole remains unclear. But one thing is certain: people like Hansmeyer and Verde are thinking hard about what else computers can build.

RAILLAN BROOKS B’13 and LUCAS MORDUCHOWICZ B’12 are rule-based brainchildren.