
Building Performance Simulation
Introduction to Data Science
Course: J-term Introduction to Data Science for Building Performance Simulation and Architectural Design
Instructor: Jung Min Han
Date: Jan., 2021
The modeling of net zero energy buildings is an increasing concern in both the building design and sustainable consulting industries. The objectives that have been raised and recognized will change how buildings are designed, constructed, and maintained. The building design industry will soon be galvanized by regulations and standards designed to encourage net zero energy buildings while still providing comfortable built environments.
Early adoption of performance simulation software in the design decision-making process is imperative to realizing such goals. Passive building design can be achieved in the early design stage. Buiding designers to pursue sustainability in built environments will bring favorable outcomes and require only low-cost changes.
KEYWORDS:
Building Simulation, Data Science, Decision-making, Sustainable Design
OBJECTIVES
Machine learning and data science are promising approaches to shaping the design process, offering instant performance feedback. This class introduces several methods of environmental analysis and a number of building performance simulation tools, including daylighting, airflow, and energy. The required programming skills and analysis techniques are incorporated by importing generic weather information to predict energy use in response to design changes. This course also introduces data management skills including Python scripting, machine learning, and 3D data visualization.

Day 1
Theme: Introduction to building performance simulation and data science
Skillset: Installation of the Python and ML packages and weather data manipulation
Tools: Anaconda, Python, Jupyter notebook
Python Basic - Basic Syntax & Data Structure
View script in Jupyter Notebook
Pandas Data Process - Visualize & Manipulation
View script in Jupyter Notebook




Day 2
Theme: Daylighting simulation and data processing
Skillset: Data processing (imputing missing values, cleaning data)
Tools: Anaconda, Python, Jupyter notebook, DIVA, Rhino, and Grasshopper
Pandas Data Process - Visualize & Sampling
View script in Jupyter Notebook




Day 3
Theme: Energy simulation and parametric study
Skillset: Parametric simulation and optimization using Rhino and Grasshopper
Tools: ArchSim, GH-Python, Rhino, and Grasshopper
Pandas Data Process - Visualize & Multiple Files
View script in Jupyter Notebook




Day 4
Theme: Airflow simulation and visualization
Skillset: Data visualization: 2D (energy) and 3D (airflow)
Tools: Python, Jupyter notebook, Butterfly, Rhino, and Grasshopper
Pandas Data Process - Natural Ventilation & Result Values
View script in Jupyter Notebook






Day 5
Theme: Machine learning and advanced simulation techniques
Skillset: Introduction to ML using simulated data
Tools: Python, Sk-Learn, and Jupyter notebook
Pandas Data Process - Machine Learning Models
View script in Jupyter Notebook
3D CNN - Radiation Intensity Estimation
View full research article



Urban Modeling Interface - Energy Flows for Sustainable Neighborhoods
View MIT Sustainable Design Lab



