Form Follows Evolution: Multi-Objective Evolutionary Optimization with Use of Genetic Algorithms

Home / Form Follows Evolution: Multi-Objective Evolutionary Optimization with Use of Genetic Algorithms

Form Follows Evolution: Multi-Objective Evolutionary Optimization with Use of Genetic Algorithms

Project Details

Student/s: Vangel Kukov

Date: August 29, 2016

Most architectural practices nowadays still follow linear design process, meaning, an architectural form is first conceived and only after that its structural, environmental, constructional, economic aspects are taken into count. Although a certain shift is being witnessed towards a closed design process, in an attempt to integrate multiple design aspects in the early design phase, it is proven to be impossible for the designer to process the associated complexity with analogical means. The present diploma thesis introduces a new design paradigm of integration, analysis and evaluation of a plethora of design criteria, bridging a vast array of scientific fields, at a conceptual stage. One that explorers a multidimensional search space of possible solutions to the architectural problem, leaving us with a set of equally good solutions to choose from and modify. By doing so the thesis describes the methodological milestones that make this exploration possible- from the generative tools that designers have in their disposal and the analytical tools for the performative evaluation of the design to the Genetic Algorithm as a powerful search engine, inspired by the evolutional force of Nature. Furthermore the role of the designer as a conceiver of the architectural form will be challenged by the one of a creator of the design process. The viable applications and perspectives of this exiting branch of the computational design will be examined.
To study built artifacts of this design approach we must look at the work of inventors, architects and engineers of the past century who managed to harvest to great extend the integrated performativity of natural structures. From the organic shapes of Antoni Gaudi- translating a hanging model into the building rational of a conventional brickwork and using “smart” ruled geometry and fractal logic in dealing with space and structure and the invention of “Ferro Cemento” by Pier Luigi Nervi- using prefabrication in the materialization of complex free-form geometric forms to the exploration of shells with the use of three- dimensional catenary models by Heinz Isler and the use of number of generative physical models in the conception of Frei Otto’s lightweight structures, all this examples come to teach us of integration of multiple design criteria, fused to produce objects of great performativity and beauty. A Performativity that seems to lack from modern architectural production struggling between “mono- dimensional”, highly formalistic approaches to design. The Methodology behind Multi-objective Evolutional Optimization consists of three distinct fields of interest that come together, building design complexity. These fields being morphogenesis, analysis- simulation and optimization, allow us to obtain an evolutional design process. Morphogenesis, in the context of this study, is the procedural generation of form guided by the variables, constants and restrictions the design introduces. These are the factors that define the topography of solutions to the architectural problem. A simulation is an imitation of the behavior of a model in a virtual environment. The results of the simulation and analysis of building behavior are the main tools in “feeding” the optimization process. The optimization process itself is cared by a Genetic Algorithm that mimics the evolutional principles of natural selection, mating and mutation. Undoubtedly the major transition from the physical world to the cyberspace that design undertakes comes with a great changes in the way a designer thinks and operates. This by no means decreases his level of control, neither his creativity. On the contrary it reveals a whole new world of diversity, rendering synthesis an ever evolving process and architecture, simply its byproduct. This design methodology is not without its limitations dough. Today it’s practically applicable to deterministic problems of limited complexity. Architectural, on the other hand, are problems of great complexity, often hard to be defined explicitly. Another drawback is the extensive computational power required for the evolutional process to take place. A Big step in this direction is the use of computational resources as a service on the web (computation cloud), taking the capabilities of evolutional design to a whole new level, hoping that one day we will be capable of making use of the same forces of nature that created us in the creation of architecture.

 Previous Next