GENERATION OF LAYOUT OF MULTI-FAMILY RESIDENTIAL BUILDINGS USING EVOLUTIONARY ALGORITHM
Space planning, evolutionary algorithms, generative architecture.
As the complexity of projects has increased, the computer has been used as a
design assistant with the aim of improving the architectural design process. It was
from 1960 onwards that computational design had its origins with the definition of a
theoretical framework in order to systematize the design process as a methodology.
In these discussions, space planning was disseminated, which studies the process of
arranging rooms in a given space, in which distance, adjacency and other functions
are the main objectives. Just as evolutionary algorithms, which make up generative
systems, contribute to obtaining better solutions. Space planning problems have
variables that are sometimes conflicting and that need to be negotiated so that all
restrictions can be satisfied. In the traditional design method, the chosen layout is
based on personal taste as a criterion, in order to make more assertive decisions
evolutionary algorithms can be applied to this type of design problem. This master’s
dissertation aims to explore the possibilities of automated generation of layouts for
multifamily buildings through the application of evolutionary algorithms. This is a
simulation research with a quantitative approach and exploratory scope, using the
Rhinoceros software and its plugins Grasshopper, Termite Nest and Wallacei. Five
simulations were carried out with the objective of arranging environments in a limit
perimeter applying the evolutionary algorithm. As a result, it was possible to obtain
optimized layouts, informative graphics for decision making and the algorithm ready
to be applied by other researchers. It is therefore concluded that the application of
evolutionary algorithms in space planning generates optimized solutions that actually
meet pre-established restrictions in a more assertive way, being an area with great
potential to be explored.