The computational design process

Algorithms and pure data, rather than textbook examples and experience of similar projects, today increasingly inform urban design choices. Terms such as ‘access to direct sunlight’, ‘shelter from wind and rain’, ‘optimal views’ and ‘connectivity to important points of interest’ are now encoded terms in computational models specifically designed to support and enhance the process of urban design.
The simulation software and even the hardware in supercomputers are now something practitioners build themselves. The models are constructed to resemble (impending) reality and are based on thousands of people-hours in development time and are built on the shoulders of countless brilliant people, thanks to an open source philosophy that provides easy access to knowhow. As a result, we can now perform complex CFD analysis of wind and rain, use space syntax metrics combined with agent-based model statistics and run dynamic thermal models in parallel on clusters in-house, all of which have never been seen on this scale before. Some computational models need increasing amounts of computing power (e.g. CFD an example shown in Figure 11 and Network analysis), and some just need to be refactored and repurposed from existing analytical functions, such as seen in the example use of Manning equations [1] in Figure 5.