GA-based Form Exploration of Truss Bridges
2013-2014
Research group: Anahita Khodadadi, Peter von Buelow
The main objective of this research is to demonstrate a form exploration technique based on parametric form generation and evolutionary optimization. Accordingly, the design process of a truss bridge is considered, and two types of form-finding methods are presented, numerical modeling with computational analyses on one hand versus physical modeling on the other hand. The focus of this research is on form determination using the first method to indicate its ability to explore and evaluate forms based on multiple goals and offering arrays of comparable, good solutions instead of a single “optimal” form.
Additionally, the physical modeling of the truss bridges is considered in order to draw a comparison between the form finding techniques that designers use to comprehend, evaluate or describe the performance of a typical structure. The geometric configuration processing of the truss bridges is based on a mathematical system known as Formex algebra and the respective parametric formulations of the forms are provided by the algebra’s associated programming software, Formian.
The next step of form generation is implemented using the ParaGen method, based on genetic algorithm (GA) concepts. This method provides arrays of solutions that are structurally evaluated and mutated or recombined through an iterating cycle. Para-Gen provides visual representation of the solutions and allows the designer to have a suitable interaction and make decisions regarding personal preferences. The ParaGen framework uses a non-destructive, dynamic population to fill a database with solutions linked to a variety of performance characteristics.