Breathing Aggregations

AA EmTech Dissertation with Aarathi Muralidharan and Lei Lu, 2012-2013, London
with sincere thanks to Michael Weinstock, George Jeronimidis, Evan Greenberg and Mehran Gharlegi

Can we control the urban microclimate through its form? Can we 'design' temperature, humidity, wind speed, etc. on an urban scale?  This research takes these questions to heart; it investigates how we can design neighbourhoods that avoid the infamous urban heat islands by using agent-based artificial intelligence as a form generator.

Urban heat island is the simple fact that the city tends to be warmer than the surrounding countryside. This is less due to the burning of fossil fuels then it is to the geometry of the city itself. We propose a novel approach not by relying on intuïtion, but by investigating how formal characteristics of a city (such as solar shape factor, surface-to-volume-ratio etc.) influence the way heat is transported and stored in the city fabric.

The proposed method is an artificial intelligence algorithm, capable of optimising for these different formal parameters with a varying relative importance. It is made out of hundreds of designer-agents, and uses a subtly balanced system of indirect communication and independent decision-making while the structure is being build. Different typologies “grow” out of the system, depending on the used settings. The resulting structures are thus an example of true bottom-up growth. In its’ principles, the system is based on the organisation of termite colonies, who build mounds that are environmentally optimised in many ways.

Similar algorithms were used to further enhance the emergent urban features of the generated neighbourhoods, the most significant being a way-finding script for making a pedestrian network (Breathing Networks), and a fluid flow simulation for the detailed design of the towers (Breathing Towers).

Parts of this research were published in AD System City, July 2013 (ed. M. Weinstock).