Designs on the computer.

Annette LeCuyer discusses two radically different approaches to the creative use of computers in design. Frank Gehry uses computers in design development. Peter Eisenman uses computer-generated forms as his starting point.

Computers are now common currency among architects. Because the most widely used systems and software packages are those developed specifically for the profession, they consequently tend to serve architectural preconceptions about form-making, relegating computers to mere tools of production. However, some architects are transcending these limits by borrowing creatively from other disciplines, using widely divergent computer means and ends.

From the early days of studs and chain link -- what Anthony Vidler calls `bricolage populism' -- the work of Frank Gehry has evolved into increasingly complex and monumental assemblages. Despite their plastic complexity, the schemes are still all conceived and developed manually through physical models. For conceptual work, Gehry firmly contends that models and drawings are much faster and more responsive than computers. However, it is the daunting task of building Gehry projects which has led the practice to embrace computers.

Gehry's office uses Catia software on IBM RISC 6000 computers. Originally developed by the French aerospace experts Dassault Systems for the design of fighter planes, this software is significant because it is based on surfaces in lieu of polygons. Since the system was released to the public 10 years ago, it has been adopted by industry with dramatic results. For example, Catia has been instrumental in enabling Chrysler Corporation to compete successfully with Japanese manufacturers by reducing the research and development period for new models from 72 to 36 months. Chrysler is currently working towards 24 months, an overall 67 per cent reduction of the R &D cycle.

Previously, Chrysler developed their designs for cars using artists' sketches, technical drawings, and clay models on plywood armatures laboriously made by teams of technicians. Neither aesthetics nor cost could be effectively evaluated until the physical models were complete, and changes were costly. Now, as designers sketch three-dimensional surfaces on the computer, Catia simultaneously translates all information into a geometric database which governs the milling machines which produce the clay model and controls the final manufacturing process. Chrysler has also developed fullsize three-dimensional colour modelling with high speed anima6On, a system which is making physical models obsolete. Because aesthetic and cost judgments can be made using computer models, modifications can be entered readily in the database and evaluated virtually instantly. The benefits for a competitive product-based industry -- the reduction in design and production time and therefore unit cost -- are obvious.

Like Chrysler Corporation, Gehry's office is using Catia for surface modelling of non-polygonal forms. Their primary aim is not to reduce design time but rather to make their projects more buildable. Materials like the render of the Vitra museum (AR November 1989) and the sheet metal of the University of Toledo Centre for the Visual Arts (AR August 1993) have a built-in tolerance which is forgiving of Gehry's complex and fluid forms. However, producing the same forms in stone presents an enormous technical challenge.

The recently completed American Centre in Paris (AR August 1994) was designed and built without computers. Consequently, it was difficult to produce the complex surfaces to a level of accuracy sufficient to eliminate the `pillow' effect of discontinuous curvature from stone to stone. In contrast, the stone cladding for the Disney Concert Hall in Los Angeles has been developed with computers, using Catia both for design and cost control. Using the thickness of the raw block of stone and cutting time as prime governors of cost, it is easy to understand that cost increases geometrically with the progression from flat surfaces through to single, double and complex curvature. Gehry's design -- made by hand -- was digitised, then rationalised by Catia to achieve repetition without sacrificing form. Using the database of the rationalised design, a physical model was computer milled, compared to the original cardboard model and adjusted where necessary.

The Catia database, which has also been used to generate construction documentation, has subsequently been passed on to the stone subcontractor. The stone cladding and cladding support package was tendered to 14 firms with three- and five-axe milling capability. As part of their submission, four shordisted tenderers were required to build a 3 x 8 metre piece of wall using only numerically controlled computer milling. The selected subcontractor, Harmon Contract from Minnesota, is $500 000 under budget, or four per cent of the $11.5 million subcontract package.

This happy result is not unique to the Disney project. Gehry's Prague office building further explores complex curvature, contrasting precast concrete with steel and glass. Using the Catia database together with Pro-Engineer software on Sun computers, the architects have been able to model the glazed surfaces to ensure that every adjustment remains within agreed parameters. Pro-Engineer also identifies clearances and interferences. All millions are complex curves and, while the glass is flat, no two pieces are the same shape. The initial budget for the glazing package was S200 per square foot. With the help of Catia models to detail fabrication and layout, Gehry has worked closely with the subcontractor, the Italian firm Permasteelisa, to reduce the cost by one-third to $135 per square foot.

While many architects are putting more and more design responsibility in the hands of others, Gehry's use of computers has led the office into closer relationships with contractors and direct relationships with suppliers and subcontractors. By generating the database to which everyone works, the practice increases its responsibility. The hope is that the accuracy of the information and the elimination of middlemen reduces everyone's cost and risk, and makes the schemes more buildable. However, notwithstanding the high degree of accuracy of computer generated construction documentation, the achievable tolerances for both fabrication and erection which are inherent in the materials used remain largely unchanged, and Gehry is conscious that this must be kept in mind continuously when working with computer models.

While Gehry has turned to the computer as a pragmatist, Peter Eisenman is using computers as the conceptual starting point for his most recent projects. Seeking to break free of Cartesian absolutes as a priori beautiful forms, Eisenman is using the computer to explore more dynamic, unpredictable systems of organisation. He is seeking to release architecture from the restriction of being object-oriented to become textual -- that is, to convey the process of being conceived.

In his search for a design inventory which is free of human preconceptions, Eisenman has married Form Z Macintosh software to natural phenomena such as waves, quasi-crystals and slime moulds, all of which, in his terms, have no a priori knowledge or contingencies. The so-called contextual clue for the parti of his Haus Immendorff project in Dusseldorf is the `turbulence' of a river inlet adjacent to the site. The starting point is a cube deformed by soliton waves, a dynamic system bordering on chaos. Solitary waves are physical phenomena that occur in seemingly random ways, but are caused by measurable physical factors such as abrupt changes in depth or subterranean seismic patterns in water. Doubling the solitary wave forms solitons, pulses of energy moving through solids, liquids or gases which form non-linear interactions. These non-linear interactions do not lead to chaos, but instead produce spontaneous self-organising emergent systems. For Eisenman, the soliton analogy for the deformation of the cube has produced outer and inner volumes whose surfaces intersect as they twist vertically, forming a vortex-like cone of space. The form could not have been foreseen but was discovered only through the computer. The recording of an emergent system is the objective of the design process. To build such a scheme -- that is, to crystallise it as an object -- may therefore be totally incompatible with its theoretical premise.

Eisenman's proposed Centre for the Performing Arts for Emory University in Atlanta is his first computer-generated scheme to bridge between theory and execution. The starting point is again Cartesian, made up off our programmatic typological boxes determined by the nature of the different performance spaces and, on a larger scale, the gridded master plan of the university extended into nature like the Jeffersonian grid. These idealised forms are subjected to a series of deformations arising both from the context of the site and from a non-material context that deals with the multiple layers of reality and meaning in the environment of the electronic information age of collapsing time and space.

A steep ravine on the site causes a deflection of the grid which Eisenman identifies as an impulse of energy registered as sine and cosine curves similar to sound waves. The multiple waves are built into a system of harmonics. Each wave is a product of the depth and the width of the ravine and is musically approximated in such a way that depth represents amplitude and width represents frequency.

Using these waves, the boxes are subjected to two deformations. The first Eisenman calls the small-scale fold which is contained entirely in a defined field of energy of harmonic lines; the second is the large-scale fold which spans two sets of harmonic lines. The resultant buildings are the intersection of the deformations, and the resulting forms are a product of always selecting the outermost envelope. The skin of the buildings is therefore composed of parts of both harmonic folds, and each harmonic fold is continuous beneath the skin. The results could not have been predetermined or designed but were entirely dependent on the manipulation of the computer. Using Autocad, Eisenman has translated the Form Z model of harmonic folds into construction documents which include both conventional drawings and a complex three-dimensional computer model and database for the buildings. The project is now ready to start on site. Eisenman is the agent provocateur, using the computer not to save time or money but to challenge authority. He is looking for a means of transforming architecture into what he calls a `condition of open writing' which is independent of brief and typology. It is a process which does not produce a `guaranteed aura' for a building but which uses the computer to loosen the restrictions of architecture, provide absence instead of presence, and produce the unexpected.

Eisenman and Gehry represent extremes of American architectural discourse: cerebral East Coast and pragmatic West Coast. Eisenman's conceptual starting point is the computer; Gehry does not turn to the computer until well into design development. Eisenman's focus is on the inferiority of collapsing time and space; Gehry, like Chrysler Corporation, is primarily concerned with surfaces. For Eisenman, who is engaged with the metaphor of design, buildings may not be the point; Gehry's objective is to transform an idea into physical reality. Through computers, Eisenman is looking for the unauthored process, while Gehry is producing the ultimate designer signature product.

This article is based on a series of conversations with Michael Holmes of Chrysler Corporation, James Glymph of Frank O. Gehry & Associates, and Selim Koder of Eisenman Architects, and a recent symposium on computer generated design at the College of Architecture and Urban Planning at the University of Michigan.