Structural design exploration of high rise design concepts using genetic algorithms

Roland Schagen, Delft University of Technology

CHALLENGE - The adoption of optimization algorithms has been slower in conceptual design compared to the later design phases. The choices made during the conceptual design phase have a large impact on the performance of the result. This research aims to find out why it is so hard to use optimization in conceptual design.

SOLUTION - The Central Chinese Television (CCTV) building provides an interesting case for conceptual design exploration because it exemplifies the tendency of designers to ‘think outside of the box’. This method involves the parameterization of the CCTV building envelope, the calculation of its structural performance using an abstracted finite elements analysis, and the use of NSGA II algorithm provided by modeFRONTIER. The implementation of this method has resulted into insight on how the software can be used. One particular improvement was the implementation of an analysis termination condition in the Grasshopper to modeFRONTIER link. This allows the designer to establish rules for when the model is finished. The reason for this feature was the millipede analysis component which would prematurely provide an empty output causing the
model to shut down.

BENEFITS - The result is a deeper insight into why optimization is difficult in the conceptual phase, a method that is considerate of these difficulties, and an analysis tool for this method.