High-Fidelity Urban Wind Flow Modeling for Safe and Efficient UAM Operations Supporting Power Infrastructure Inspection

Document Type

Conference Paper

Publication Date

2025

Publication Title

PROCEEDINGS OF ASME 2025 INTERNATIONAL MECHANICAL ENGINEERING CONGRESS AND EXPOSITION, IMECE2025

Abstract

Urban Air Mobility (UAM) operations for power infrastructure inspection require precise characterization of urban wind hazards, as high-fidelity simulations necessary for detailed flow field analysis in complex city environments are computationally expensive and resource-intensive. To overcome this, we evaluate three data-driven reduced-order models: Standard Dynamic Mode Decomposition (DMD), Extended DMD (EDMD), and Convolutional Autoencoder-based DMD (CAE-DMD) using 1980 large-eddy simulation snapshots of flow past an idealized urban cylinder cluster. At a reduced dimension, Standard DMD (epsilon = 0.145) accurately captures the principal vortex-shedding frequency but under-resolves secondary shear pockets; EDMD (epsilon = 0.130), incorporating quadratic observables, recovers mid-frequency wake-merging dynamics; and CAE-DMD (epsilon = 0.120) most faithfully reconstructs both sustained shear-layer cores and transient vorticity bursts. ROM-derived hazard diagnostics classify 26.06 % of spatial locations as persistently hazardous, 26.50 % as frequent surge sites, and 12.47 % as rare exceedance events, with CAE-DMD matching full-simulation statistics within +/- 1 %. These results demonstrate that CAE-DMD can deliver high-fidelity, computationally efficient wind-hazard forecasts, enabling adaptive UAM trajectory planning for infrastructure inspection.

Comments

Paper presented at: 
2025 International Mechanical Engineering Congress and Exposition-IMECE, Memphis, TN, NOV 16-20, 2025

DOI

10.1115/IMECE2025-167174

Volume

9

Link to Full Text

Share

COinS