Breeding Apartments

During the spring semester of 2015 at the University of Technology Sydney, Connar Blampied, Stewart Cowan and I completed a computational design studio in association with BVN with Paul Wintour as our tutor. We had set out to create a system that would produce a range of possible apartment buildings within any given site and any given zoning regulations each meeting the requirements and rated based on their performance to our predetermined criteria.

This project aimed to interrogate the issues in translation of the abstract spatial requirements of discrete housing and familiar types into a singular building. For this we produced a series of digital prototypes each heavily informed through planning policy and our own criteria through a genetic solver. Galapagos, the genetic solver in question would ideally begin to ‘learn’ how to optimise for the various requirements we have set out to produce the best fitness value. A rating that brings forth a method of categorization and gives some semblance of worth to the various outputs. First was at the urban scale, New York was to be the major influence on the overall organization and urban intelligence of the model working at multiple zonings and scales.

The functionality of our code was set around packing apartment typologies into the Building Envelope. Because some apartment types were incongruent with one another; when packed, their circulation systems could lead to nowhere resulting in their removal. This fissure would be reprogrammed as part of the public space that is accounted for within all yield calculations. Our code gave us the opportunity to produce over 75,000 iterations relating to the chosen subject site. These iterations where tested against each other and rated as to what were deemed the most desirable. We could then interrogate the highest scoring iteration from each generation against our marking criterion to find the architectural merit and discard those that did not meet our ethical goals.

The tool we have created performed as a test case for this method of design allowing us to apply new design techniques to a typology that is often overlooked. This process could be extended to the urban planning scale and personal developments both within the public and private sector. With even the smallest changes in code and regulation we could see the shifts in the design and how that would affect the surroundings with the click of a button. This software is a starting point to a design method where all costs and square meters are accounted for and we can easily justify our architectural and social agendas while still meeting the requirements of the clientele in a fraction of the man hours of traditional techniques.