BIM and Data-Driven Predictive Analysis of Optimum Thermal Comfort for Indoor Environment

Area of Science:

• Building Science
• Computational Fluid Dynamics
• Machine Learning

Background:

• Mechanical ventilation is a major energy consumer in buildings.
• Natural ventilation is key to reducing energy use and maintaining occupant comfort.
• Optimizing natural ventilation requires advanced analytical tools.

Purpose of the Study:

• To present a novel computerized framework for analyzing optimal thermal comfort using natural ventilation.
• To integrate Building Information Modeling (BIM) with machine learning for predictive thermal comfort analysis.
• To evaluate the effectiveness of natural ventilation in reducing energy consumption and ensuring comfort.

Main Methods:

• Utilized BIM for geometrical and semantic building data to define computational domains and boundary conditions.
• Employed Computational Fluid Dynamics (CFD) to simulate airflow and temperature distribution for thermal comfort assessment.
• Developed machine learning models, incorporating graph theory's adjacency matrix, to predict thermal comfort based on BIM-CFD data and spatial features.

Main Results:

• Natural ventilation can reduce cooling energy consumption but may not satisfy all thermal comfort criteria.
• Seasonal performance of natural ventilation is crucial for balancing energy savings and comfort.
• The proposed framework enables efficient examination of various design options and operational strategies.

Conclusions:

• The integrated BIM-CFD-ML framework offers a powerful tool for optimizing building ventilation strategies.
• Balancing natural ventilation with mechanical systems is essential for year-round thermal comfort and energy efficiency.
• This approach supports improved HVAC system design and real-time control for enhanced indoor environments.