OPTIMIZATION OF SUNSHADES FOR DAYLIGHTING THROUGH EVOLUTIONARY ALGORITHM

Abstract

The dissertation attempts to test the feasibility of Evolutionary Algorithm as a design tool for optimizing sunshades for daylighting through one of Evolutionary Algorithm techniques known as the Genetic Algorithm. Sunshades are an integrated design feature that has a significant impact on the critical evaluation of the appearance and performance of architecture. The conventional approach to sunshade design has been proven to be an effective method until recently. With the emergence of organic buildings and buildings that require unique shading strategies demanding both aesthetics and performance calls for new methods of sunshade design to be explored as the conventional approach simply would not work. Drawing inspiration from the organic evolutionary mechanisms, Evolutionary Algorithm has proven to be potentially helpful to designers to solve complex design problems such as sunshade design that may otherwise be impossible to solve, as well as enhancing their creativity and opening their minds to exciting design possibilities in the future. Through a design experiment, the proposed method involved the construction of a parametrically modeled 3 single cell office spaces in Houdini. Genetic Algorithm, which will be running on Dexen, was used for the optimization, in which different objectives were encompassed by the fitness definition for the Genetic Algorithm population. Through a simulation process, the capacities of the Genetic Algorithm and by that extension Evolutionary Algorithm are confirmed and a series of observations and findings were discussed and concluded. The advantage of starting from a framework of parameters is its repeatability whereby designers can inject different initial ideas into the methodology to produce a variety of design options. Based on the initial set of parameters that the designers select, the resolution of the design variant can vary. The process can be a un-expectancy generator that is capable of producing designs that fulfill both the objectives in aesthetics and performance. The nature of unpredictability of the results from the experiment, satisfying both the aesthetics and the performance, allows for some very interesting design choices. The experiment here is meant to be more than the sequential description of the technique; it is rather the abstraction of the process of what the evolutionary technique entails as a legitimate design tool, enabling designers to expand their arsenal of design methodologies and materialize concepts that have been until now impossible to conceive.