Canvas based neural networks better predict airflow in varying rooms
Comparative Study of Canvas- and Coordinate-Based Geometry-Conditioned Neural Networks for Real-Time Prediction of Indoor Environments with Varying Room Dimensions
Artificial IntelligenceMachine Learning
Summary
Predicting airflow and temperature in rooms is important but slow with traditional physics simulations. The authors compare two types of AI models that learn from these simulations to make predictions faster, even when room sizes change. One model uses a fixed canvas to represent room size, while the other uses coordinate information. They found the canvas-based model gives more accurate results, which can help architects quickly test air flow in different room designs.
What this means in practice
- •For architectural designers: Provide near-instant feedback on air flow and cooling efficiency in rooms with different dimensions during building design iterations.
- •For hvac engineers: Rapidly estimate temperature and airflow distribution for customized indoor environments without running full fluid dynamic simulations.
Authors
Shaofan Wang, Philipp Geyer
Abstract
Computational fluid dynamics (CFD) is computationally expensive for iterative building design. Deep learning surrogates offer faster predictions, but generalization across varying room dimensions remains challenging. This study compares two geometry-conditioned neural network surrogate models for rapid indoor airflow and temperature field prediction under varying room sizes. During the architectural design stage, this method is relevant for real-time assessment of whether conditioned supply air will effectively provide fresh and cool air throughout rooms with varying heights and lengths. The canvas-based neural network (CaNN) explicitly encodes room size by preserving different occupied regions within a fixed canvas, whereas the coordinate-based neural network (CoNN) uses normalized coordinate channels. Both employ U-Net autoencoders to separately predict airflow and temperature fields. Evaluation uses simplified rooms with varying heights and lengths as a controlled benchmark. The results show that CaNN achieves higher agreement with CFD results than CoNN, demonstrating its potential to support performance-driven building design.