Abstract
In response to climate change, building design is increasingly shifting its focus toward thermal performance and internal energy flows. Regarding the space as a dynamic system, this paper approaches the characteristics of its thermal behavior through dominant modes. An automated analysis framework is introduced, applying Dynamic Mode Decomposition to the temperature field of open spaces based on multi-zone simulation, followed by a Generative Adversarial Network modeling to speed up the feedback. The predicted modes reveal the organization of energy flows, offering insights to early design considerations such as thermal zoning and partition layout. The workflow is developed on the Rhino-Grasshopper platform, utilizing EnergyPlus as the simulation engine and pix2pix as the network model.
Keywords Thermal response, Dynamic Mode Decomposition, Thermal zoning, Passive house, Thermodynamic Architecture
Copyright ©
Energy Proceedings