Integration of FEM, NURBS and Genetic Algorithms in Free-Form Grid Shell Design

Abstract

Popularity of free-form grid shells grows every day since they represent a universal structural solution for free-form shaped architecture, enabling the conflation of structure and facade into one element [1]. The infinite number of possibilities of generating a grid structure over some surface calls for an automated method of design and optimization, in contrast to the standard trial-and-error routine. This paper presents some results of the comprehensive research dealing with the optimization of grid shells over some predefined free-form shape. By combining static analysis and design software on a basic C++ level we try to statically optimize a grid shell generated over a given surface. Using Genetic Algorithms for the optimization we are able to significantly reduce stress and displacement in a structure, thus save material and enhance stability. The presented method of structural optimization is constructed as a C++ based plug-in for Rhinoceros 3D, one of the main NURBS (Non Uniform Rational B-Splines) geometry based modeling tools used by architects for free-form design today. The plug-in communicates iteratively with Oasys GSA, a commercial FEM software.