Evolving Buildings

• Title: Designing Sustainable Building Using Evolutionary Optimization Systems
• Project Period: 1 December 2008 to 30 November 2011
• Funding Source: MOE Academic Research Fund (AcRF) Tier 1
• Principal Investigator: Patrick Janssen (PI)
• Budget: SGD $148,632.95

Research Overview

The research developed an evolutionary developmental design method, called ‘Evo-Devo-Design’, as well as a supporting computational environment for executing population-based algorithms, called ‘Dexen’.

Example of building scale evolutionary optimization.

Research Outcomes

Key research outcomes include the following:

• The development of the Evo-Devo Design method, defining how designers can encode their design schema as a set of computational procedures, thereby avoiding the need to write any code. These schema procedures define key aspects of the project: the type of brief and site, the criteria for optimization, and an adaptable design model.

• The development of encoding schema for the evolutionary procedures, so that they can be uploaded into the evolutionary system allowing a population of optimized design to be evolved and explored. For this research, a Visual Dataflow Modelling software called Sidefx Houdini was used. The Figure above shows such an evolutionary optimization process being conducted for a residential housing development.

• The development of the Dexen system, a Distribute Execution Environment that was developed to accelerate the process of evolving designs. Dexen allows users to run evolutionary jobs in parallel on multiple networks computers. Although Dexen was created to support the Evo-Devo Design method, it is in fact a highly generic environment that can be used to run a wide variety of population-based algorithms.

• A set of case studies showing how the Evo-Devo Design method can be used to evolve designs for buildings with multiple conflicting performance criteria.