Data Science for Performance-Driven Design
The modeling of energy-efficient buildings and sustainable urban development is an increasing concern in both the building design and sustainability consulting industries. Early adoption of building performance simulation software for decision-making during the design phase is essential to achieving sustainable design goals. Guiding designers to pursue sustainability in their built environments will bring favorable outcomes and low-cost adaptations. Machine learning (ML), Deep Learning (DL) and data science are promising approaches to shaping the design process and offer instantaneous performance feedback. The active use of data science techniques increases the efficiency and accuracy of building simulation workflow and the optimization of building geometry.
This class will leverage data science and performance simulation as the primary drivers in determining design decisions. In the last decade, the fundamentals of building performance simulation tools for energy, daylighting, airflow, and renewable energies have been translated into performance simulation tools and metrics with relevant measures. There are great advantages for students learning to use such tools, including the ability to calculate metrics and to apply related methodologies in their building designs. However, such utilization requires a high level of understanding of the computations necessary for the geometric modeling process, as well as relevant programming skills. These programming skills and analysis techniques will be explained in this class with practical hands-on workshops to impart environmental information and predict building performance in response to design changes. This course will also introduce data management skills, ML-based surrogate modeling, data analysis and visualization for advanced research. The final deliverables will be an optimized building design option/ design space utilizing data science techniques on the design decision making process.