Harding, J. and Derix, C., 2011. Associative spatial networks in architectural design: Artificial cognition of space using neural networks with spectral graph theory. In: Design Computing and Cognition ’10. New York: Springer Science and Business Media, pp. 305-323.
Paper presented at the International Association for Shell and Spatial Structures Symposium, 2011. Abstract:
“This paper describes a new method for the form-finding of funicular structures in two or three dimensions using a zero-length spring system with dynamic nodal masses. The resulting found geometry consists of purely axial forces under self-weight, with zero bending moment at nodes for both shells and tension net forms. A real-time solver using semi-implicit Euler integration with viscous damping is used to achieve system equilibrium. By using a real-time solver, the designer is able to alter the gravitational field or apply new point loads without re-starting the analysis, leading to an interactive experience in generating design options. The advantages of this method over existing approaches are discussed, with its successful application in a recent real case-study project also shown.”
A Preprint of the paper can be found here.