arXiv:2512.14725v1 Announce Type: new 
Abstract: Urban wind flow modeling and simulation play an important role in air quality assessment and sustainable city planning. A key challenge for modeling and simulation is handling the complex geometries of the urban landscape. Low order models are limited in capturing the effects of geometry, while high-fidelity Computational Fluid Dynamics (CFD) simulations are prohibitively expensive, especially across multiple geometries or wind conditions. Here, we propose a generative diffusion framework for synthesizing steady-state urban wind fields over unstructured meshes that requires only geometry information. The framework combines a hierarchical graph neural network with score-based diffusion modeling to generate accurate and diverse velocity fields without requiring temporal rollouts or dense measurements. Trained across multiple mesh slices and wind angles, the model generalizes to unseen geometries, recovers key flow structures such as wakes and recirculation zones, and offers uncertainty-aware predictions. Ablation studies confirm robustness to mesh variation and performance under different inference regimes. This work develops is the first step towards foundation models for the built environment that can help urban planners rapidly evaluate design decisions under densification and climate uncertainty.

تم اقتراح إطار جديد للانتشار التوليدي لنمذجة تدفق الرياح الحضرية، مما يسمح بتوليد حقول الرياح الثابتة على هندسة حضرية معقدة دون الحاجة إلى قياسات شاملة أو تفريغ زمني. تجمع هذه الطريقة بين شبكة عصبية رسومية هرمية ونمذجة انتشار قائمة على الدرجات لتحسين دقة وتنوع محاكاة تدفق الهواء.

Se ha propuesto un nuevo marco de difusión generativa para la modelización del flujo de viento urbano, permitiendo la síntesis de campos de viento en estado estacionario sobre geometrías urbanas complejas sin necesidad de mediciones extensas o despliegues temporales. Este método combina una red neuronal gráfica jerárquica con modelado de difusión basado en puntuaciones para mejorar la precisión y diversidad de las simulaciones de flujo de aire.

Un nouveau cadre de diffusion générative a été proposé pour la modélisation des flux d'air urbains, permettant la synthèse de champs de vent en régime permanent sur des géométries urbaines complexes sans avoir besoin de mesures extensives ou de déroulements temporels. Cette méthode combine un réseau de neurones graphiques hiérarchiques avec une modélisation de diffusion basée sur des scores pour améliorer l'exactitude et la diversité des simulations de flux d'air.

A new generative diffusion framework has been proposed for urban wind flow modeling, allowing for the synthesis of steady-state wind fields over complex urban geometries without the need for extensive measurements or temporal rollouts. This method combines a hierarchical graph neural network with score-based diffusion modeling to enhance the accuracy and diversity of airflow simulations.

Generative Urban Flow Modeling: From Geometry to Airflow with Graph Diffusion

Was this article worth reading? Share it

PlantFCE Model Builder

LucidQuery AI

Metaflow AI

Fluig AI

Archipeg

Summiflow

Ready to build your own newsroom?