| id |
acadia23_v2_542 |
| authors |
Harrison, Paul Howard; Heinrich, Jason; Cop, Philipp; Veigas, Glenn; Callaghan, Brigid; Fahmy, Janna |
| year |
2023 |
| title |
Relational Fluid Dynamics: Optimizing for Airflow in Urban Form with Comparative AI Predictions |
| source |
ACADIA 2023: Habits of the Anthropocene: Scarcity and Abundance in a Post-Material Economy [Volume 2: Proceedings of the 43rd Annual Conference for the Association for Computer Aided Design in Architecture (ACADIA) ISBN 979-8-9891764-0-3]. Denver. 26-28 October 2023. edited by A. Crawford, N. Diniz, R. Beckett, J. Vanucchi, M. Swackhamer 542-548. |
| summary |
In this paper, we propose a fast approximation of a particularly slow form of simulation: computational fluid dynamics (CFD). These tools are a crucial component in sustainability and comfort analyses, but their considerable computational requirements mean that only a small number of simulations can be undertaken for any given project—if at all. We demonstrate a predictive AI technique for estimating the results of CFD analysis; as this is nearly 300 times faster than traditional CFD methods, it can be used to quickly determine fitness within an optimization routine. In this case, the AI-predicted results have a mean accuracy of 89.0 to 90.8% relative to results derived using a traditional CFD technique. We also demonstrate an optimization technique for urban form using only predictive AI outputs for fitness. When optimizing for minimal wind velocity, randomly-placed urban aggregations eventually form recognizable vernacular solutions like wind breaks and perimeter walls. While a single prediction might only offer a rough approximation of real- world performance, optimization allows us to focus on the relative differences between predictions. This relational approach is inaccurate, of course, but clearly useful. |
| series |
ACADIA |
| type |
paper |
| email |
|
| full text |
 file.pdf (2,009,694 bytes) |
| references |
Content-type: text/plain
|
| last changed |
2024/12/20 09:13 |
|
