New Horizons for Composite Fiber Structures: Technology Transfer From Aeronautics to Long-Span Structures
Architect Michael Silver loves to make things. He is also fascinated by the patterns which computers can generate. And he is intrigued by the extraordinary structural and aesthetic properties of carbon and other composite fibers. He approached Rafael Viñoly Architects in the fall of 2006 with a proposal: to explore the potential for transferring what fiber manufacturers have already achieved in aeronautics and other high-performance engineering fields to the design of buildings. Though composite fibers remain expensive, he believed that their superior strength and lightness offered advantages over more traditional structures. His intuition also told him that the technology of computer-driven fiber placement could reveal new aesthetic solutions which transcend the old opposition between structural frame and infill panel. The research quickly evolved into a close three-way partnership among Silver, designers and engineers at Rafael Viñoly Architects, and Automated Dynamics Corporation, an advanced composite manufacturer in Schenectady, New York.
Understanding Composites
Composites are termed as such because they represent a composite of two very different materials: a fiber made of carbon, glass, or other material, and a matrix made of resin or a high strength epoxy. They were introduced in the late 1950s and have been widely used in contexts in which a premium is placed on the combination of strength and lightness and where high cost can be partially offset by the production of multiple parts from a single design. The construction of airplanes provides an important example. While there are many fabrication processes, Mike’s research focuses on a relatively new one called “computer automated fiber placement,” a technique which Automated Dynamics uses to make oil pipeline fittings and a variety of other high-tech industrial parts, and which aeronautics manufacturers like Raytheon apply to airplane fuselages. In this technique, a computer-controlled machine lays down strips of fiber tape on a mold or mandrel and heat-seals them in epoxy or resin. Computerization allows the machine to execute remarkably complex patterns with great precision, and this in turn allows designers to use the material very efficiently by placing it exactly where and in the quantities required by the structure. While the machines can produce continuous cylinders or even solids built up of numerous overlapping layers, they can also be used to generate complex three-dimensional patterns resembling basketwork. It was these patterns, and their potential for mimicking the lines of force found in an engineering diagram, that fascinated Mike. The question was what they could do for architecture.
Research Questions and Answers
Mike launched his residency at Rafael Viñoly Architects by meeting with senior architects and engineers, led by Jay Bargmann, to develop an approach. Everyone agreed that composites can successfully be used to build furniture, decorative components, small-scale architectural elements, and panels: designing and building an architectural application of composite elements might produce an aesthetically interesting result but would prove little beyond what was already known. The group wanted to push the limits. They agreed on a more daring strategy: to see if computer automated fiber placement could produce a long-span roof structure. The immediate goal of the research project was to design and produce a scale model of such a structure, using the actual machinery and processes of computer automated fiber placement. This required not only developing a successful structural design but also understanding the precise capabilities and limitations of the manufacturing process and programming the movements of the taping head and mandrel. This might well be a year’s work, but the ultimate goal of the experiment was to determine whether composites could equal the structural performance of steel and glass and even out-perform it in terms of cost. To provide a basis for comparison, an actual project was chosen as a benchmark: the atrium of the addition to the Cleveland Museum of Art, which was already in design. At a minimum, therefore, the composite structure would have to meet all of the constraints of the Cleveland design. It would have to span a space of 338 by 114 feet with no internal supports. The depth of the composite structure could not exceed four feet. And the composite roof would have to admit daylight comparable in quantity and quality to the glass skylights specified in the actual design.
Before setting to work, Mike arranged for the Rafael Viñoly Architects team to visit the Automated Dynamics and Raytheon plants and the National Institute for Aviation Research to gain first-hand exposure to the manufacturing process and materials. The Rafael Viñoly Architects team included technical director Charles Blomberg, structural engineer Carlos Soubié, and research director Ned Kaufman, later joined by consulting engineer Phil Khalil. The design proceeded through numerous iterations in which drawings and models were reviewed, tested, and refined. Meanwhile, Mike was also consulting regularly with Rob Langone, Automated Dynamics vice president for composite structures, and with master programmer Chipp Jansen. The basic concept was a 100-foot-long composite member that looked somewhat like a triangular truss. The challenge was to design a three-dimensional pattern of fibers which would efficiently place material where it was structurally needed and not elsewhere, create aesthetically satisfying patterns when replicated across the entire roof, and admit an appropriate amount of daylight. Most of all, the pattern had to be one that the machine could efficiently follow. And it had to be designed for easy and safe removal from the mandrel when it was complete.
After more than a year of work, all were satisfied that they had a workable design. Rafael Viñoly Architects contracted with Milgo/Bufkin Metal Fabrication, a well-known metalworking firm, to build a custom stainless steel mandrel. The drawings, mandrel, and computer algorithms were taken to Automated Dynamics, which contributed materials, expertise, and the use of its machines. A few days later – with the help of some dry ice – two twelve-foot-long models were successfully removed from the mandrel. One is made of carbon fiber, the other of glass fiber. Both are remarkably light and strong.
The experiment to design and build a long-span composite truss has successfully demonstrated several points. First, it is possible, using computer automated fiber placement, to design and manufacture a composite structure over one hundred feet long that performs successfully from an engineering perspective. Second, such a structure is less deep than a steel truss and a fraction of the weight. Third, composite structures used in this way offer aesthetic possibilities, in terms of pattern, light, and color, which are distinctively different from those available with steel and glass.
Next Steps
Other questions remain as yet unanswered. For example, although the engineering analysis carried out by Phil Khalil and Carlos Soubié took wind and snow loads into account, as well as the impacts of maintenance, the structure’s response to ultraviolet light was not considered, nor was its performance in a fire. Also, no cost comparison was attempted. Such a comparison would have to consider not only the cost of fabrication but also of shipping and assembly. Conversely, it would have to account for savings in the supporting structure due to the greatly reduced weight of the roof. These and other questions await further research. Meanwhile, Mike’s work with Rafael Viñoly Architects and Automated Dynamics has opened up promising avenues for exploration.
