sean hanna
research and design

  research | design | speculation | performance/interaction | publications | about me

 

BODY / SPACE / FRAME
by Antony Gormley

date: 2005-2007
Lelystad, Holland

Antony Gormley's Body / Space / Frame is a 25 metre high open steel lattice in the shape of a crouching figure, sited on the end of an 800 metre polder and facing outward from the coast of the Zuiderzee. My role in the project has been in the creation of methods for generating a form which both describes the body form and is geometrically and structurally appropriate.

 

   

The complexity of the structure has called for a series of new technologies to aid both its design and fabrication. From the start, the design has progressed simultaneously in the form of physical and digital models, with techniques for working between the two. Software recently developed by Roberto Cipolla of Cambridge University has been used to digitise the three-dimensional form of physical models by analysing the pixel data from series of two-dimensional images. The reverse process of realising the virtual working model has also been provided by rapid prototyping technologies such as laser sintering and stereolithography to produce scale models with thin structural members barely two milimeters in width. The resulting ability to work in both modes has allowed details to be visualised at crucial stages in the design and the form of the piece to be developed in parallel with analysis of structure site and buildability. The same level of integrated technology will continue to assist throughout the construction of the project. The resulting design is comprised of nearly two thousand steel members, each unique in size and shape, and these are to be scribed and cut automatically using computer numerically controlled devices.

Within this framework, algorithms have been devised specifically for the project to assist in placing the structural members and their connections in a manner that both describes the form of the body in space and provides structural integrity. Natural processes like plant growth and soap bubble formation have been simulated to guide the geometry of the interconnected elements during the initial design phases, and finite element analysis incorporating wind loads and structural capacity of the steel has been used to optimise and fine tune the final design. Several of these are described below.

   

 

Phyllotaxis and close packing
The processes that lead to efficient close packing of buds and branches in plants were simulated on the irregular surface geometry of the body scan to generate a set of node points that most effectively describe the form. In doing so it is necessary to very the density of points such that areas of high curvature have more frequent nodes than flatter areas. In most plants, an efficient spiral packing is produced by buds occuring at angles of 137.5 degrees around the single dimension of the circumference (the 'golden section' or 'golden mean') as the plant grows over time, and the same effect has been demonstrated in experiments with inanimate magnetic fields (Duady and Cauder). The same principle was applied here to the two dimensions of the surface. The algorithm was run within a cellular automaton model constructed on the topoligically irregular mesh of the body scan, in which the sensitivity of each cell was weighted in proportion to its local degree of curvature. As a result, nodes are clustered at a higher resolution around the knees and elbows, and spaced farther apart across the back.

 

 

   

Surface triangulation, Voronoi diagrams and sea foam
The geometric concept that evolved in design was that of an external skin of open polygons, a sort of tensile net resembling bubbles or sea foam, surrounding a triangulated interior of 'starbursts' of structural members. In addition to the differing structural implictions of rigid and pin jointed nodes, the coupling of the two systems required several approaches to member placement.

The open polygons of the skin were based on a Voronoi tesselation of the surface, but calculated neither in plane nor in three dimensions. The surface is topolically irregular, and so cellular automata were again used to develop a mesh deriving from the bud points described above.

 

   

Branching and progressive growth
Incremental methods of 'growing' structure over and through the body form were also used, where members would extend and branch into new ones in response to local conditions of curvature and desired density. The processes were in all cases parametric, so that experiments could be made in adjucting member lengths and angles to produce varying polygonal and polyhedral geometries. The images below show results for the surface only, none of which were used, but a similar technique was combined with the meshes above to generate triangulated meshes thoughout the interior of the piece.