PhD, Genetic Algorithms as Architectural Design Strategy

The PhD aim research was develop a series of genetic algorithms (GAs) applied to different architectural scales. The theory behind Genetic Algorithms is Darwin's natural selection, which suggests that the environment can encourage or discourage the organism reproduction depending on their characteristics. The offspring inherit from the parents some of their characteristics and properties that determine the fitness adaptation to the environment. Therefore, the organisms with less aptitude have fewer options to survive, because their DNA will not spread faster in the population than the most fit organisms. This process is known as evolution of species.

Between 1950 and 1960 a number of scientists began to study independently the evolution of species, with the idea to used as an optimization and search tool. The experiments consisted in create randomly population of solutions for a specific problem. Later on, evaluate the solutions to find the highest ranked solution in the population. The Genetic Algorithms were invented bt John Holland between the years 1960-1970. He was the first one to apply selection, crossover and mutation to the top ranked solutions and apply the algorithms to optimization problems. John Holland as his team at Michigan University studied the natural phenomenon of adaptation and develop ways in which these natural adaptive mechanisms can be imported into computer systems: Holland's schema theorem.

On January 27, 2011 I did my PhD dissertation defence at ETSAB-UPC with the title: Algoritmos Genéticos AGs aplicados a la arquitectura (Genetic Alrithms applied to Architecture).

The thesis was divided in 4 blocks:

  1. What are Genetic Algorithms and how they work.
  2. State of the art in Genetic Algorithms applied to architecture and examples in engineering design.
  3. Technology available, software and techniques.
  4. Genetic Algorithms applied to 4 different architectural design scales:
    • Urban design
    • Building design
    • Space layout design
    • Energy consumption optimization

Generative Urban Models

A Genetic Algorithm is used to find the most efficient route between a series of buildings to adapt in proportions and orientation. For each point where the route is, a rectangle represent a building which is scaled and oriented according the surrounding buildings in the route. The result is a series of volumes spatially related to each other. The algorithm was able to create spaces which can be recognized as squares, roads, open areas as parcs, etc.

Generative Building Models

A Genetic Algorithm must satisfy a series of spatial and volumetric restrictions that oppose each other. The buildings are represented by a list of 1s and 0s (1010), were 1 represents a cube and 0 a void. The combination is a result of parallelepiped geometries. The restrictions to satisfy are: Maximize the volume, minimize the area of ​​land occupation, look for information within the building DNA and penalize the distance to attractor and repeller points. The final form is the result of all satisfied restrictions, a subtle balance between all parameters.

Generative Spatial Layout Plans

A Genetic Algorithm is used to spatially organize and distribute plan layouts, according to objective areas. The process is based on Voronoi diagrams, to subdivide the space and the GA as engine to adjust the layout plans.

Energy Consumption Optimization

What is the shape and windows distribution that consumes the minimum energy (heating and air conditioning) during the hottest and coldest day of the year? The proposal of this work consists of optimize a building by modifying its openings and geometry to reduce the consumption of heating and air conditioning through genetic algorithms. The objective is to maintain the temperature at 20ºC with the lowest possible energy use.

  • Language: Spanish
  • Author: Carlos Ignacio de la Barrera Poblete
  • University: Universitat Politécnica de Catalunya
  • School: Escola Tècnica Superior d’Arquitectura de Barcelona
  • Department: Expresión Gráfica Arquitectónica
  • PhD programme: Comunicación Visual en Arquitectura y Diseño
  • Director: Dr. Javier Monedero
  • Thesis submission: Barcelona, Septiembre 2010

Es posible descargar la tesis completa desde la pagina de TDX (Tesis Doctorals en Xarxa) en este link: PhD thesis


Add a comment