An official website of the United States government
Realistic wind data are essential in developing, testing, and ensuring the safety of unmanned aerial systems in operation. Alternatives to Dryden and von Kármán turbulence models are required, aimed explicitly at urban air spaces to generate turbulent wind data. We present a novel method to generate realistic wind data for the safe operation of small unmanned aerial vehicles in urban spaces. We propose a non-intrusive reduced order modeling approach to replicate realistic wind data and predict wind fields. The method uses a well-established large-eddy simulation model, the parallelized large eddy simulation model, to generate high-fidelity data. To create a reduced-order model, we utilize proper orthogonal decomposition to extract modes from the three-dimensional space and use specialized recurrent neural networks and long-term short memory for stepping in time. This paper combines the traditional approach of using computational fluid dynamic simulations to generate wind data with deep learning and reduced-order modeling techniques to devise a methodology for a non-intrusive data-based model for wind field prediction. A simplistic model of an isolated urban subspace with a single building setup in neutral atmospheric conditions is considered a test case for the demonstration of the method.
more »
« less
Deep Learning for Realistic Wind Field Prediction in a typical Urban Morphology for Application to Small Unmanned Aerial Systems
This works aims to generate realistic wind data in urban spaces, which is essential in developing, testing and ensuring the safe operations of Small Unmanned Aerial Systems (sUAS) using Deep Learning (DL). This provides an alternative to existing turbulence models, specifically aimed at urban air spaces. We devise and utilize a Non-Intrusive Reduced Order Model (NIROM) approach to replicate and realistically predict wind fields in urban spaces. The method uses Large Eddy Simulation data from well-established computational fluid dynamics solvers like OpenFOAM to devise the NIROM. High-fidelity data generated from OpenFOAM is decomposed using Proper Orthogonal Decomposition (POD) into its orthogonal modes and basis. These orthogonal modes obtained over time are trained on specialized Recurrent Neural Networks like Long-Term Short Memory (LSTM) to complete the NIROM formulation. This method combined the traditional reduced order modeling approach with deep learning techniques to devise a framework for easy building and application of Machine Learning (ML) based Reduced Order Models (ROMs). A typical urban morphology subject to the wind is chosen and considered a test case for demonstrating the method.
more »
« less
- Award ID(s):
- 1925147
- PAR ID:
- 10657822
- Publisher / Repository:
- American Institute of Aeronautics and Astronautics
- Date Published:
- ISBN:
- 978-1-62410-699-6
- Subject(s) / Keyword(s):
- Deep Learning Realistic Wind Fields Urban Environment Small Unmanned Aerial Systems
- Format(s):
- Medium: X
- Location:
- National Harbor, MD
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Advanced Aerial Mobility (AAM) platforms are poised to begin high-density operations in urban areas nationwide. This new category of aviation platforms spans a broad range of sizes, from small package delivery drones to passenger-carrying vehicles. Unlike traditional aircraft, AAM vehicles operate within the urban boundary layer, where large structures, such as buildings, interrupt the flow. This study examines the response of a package delivery drone, a general aviation aircraft, and a passenger-carrying urban air mobility aircraft through an urban wind field generated using Large Eddy Simulations (LES). Since it is burdensome to simulate flight dynamics in real-time using the full-order solution, reduced-order wind models are created. Comparing trajectories for each aircraft platform using full-order or reduced-order solutions reveals little difference; reduced-order wind representations appear sufficient to replicate trajectories as long as the spatiotemporal wind field is represented. However, examining control usage statistics and time histories creates a stark difference between the wind fields, especially for the lower wing-loading package delivery drone where control saturation was encountered. The control saturation occurrences were inconsistent across the full-order and reduced-order winds, advising caution when using reduced-order models for lightly wing-loaded aircraft. The results presented demonstrate the effectiveness of using a simulation environment to evaluate reduced-order models by directly comparing their trajectories and control activity metrics with the full-order model. This evaluation provides designers valuable insights for making informed decisions for disturbance rejection systems. Additionally, the results indicate that using Reynolds-averaged Navier–Stokes (RANS) solutions to represent urban wind fields is inappropriate. It was observed that the mean wind field trajectories fall outside the 95% confidence intervals, a finding consistent with the authors’ previous research.more » « less
-
We propose a component-based (CB) parametric model order reduction (pMOR) formulation for parameterized nonlinear elliptic partial differential equations. CB-pMOR is designed to deal with large-scale problems for which full-order solves are not affordable in a reasonable time frame or parameters' variations induce topology changes that prevent the application of monolithic pMOR techniques. We rely on the partition-of-unity method to devise global approximation spaces from local reduced spaces, and on Galerkin projection to compute the global state estimate. We propose a randomized data compression algorithm based on oversampling for the construction of the components' reduced spaces: the approach exploits random boundary conditions of controlled smoothness on the oversampling boundary. We further propose an adaptive residual-based enrichment algorithm that exploits global reduced-order solves on representative systems to update the local reduced spaces. We prove exponential convergence of the enrichment procedure for linear coercive problems; we further present numerical results for a two-dimensional nonlinear diffusion problem to illustrate the many features of our methodology and demonstrate its effectiveness.more » « less
-
In our earlier work (https://github.com/rpsuark/ASEE21-OpenFOAM-Introduction), it was reasoned that open-source software OpenFOAM would be a cost-effective and more accessible alternative for teaching Computational Fluid Dynamics (CFD) than commercial software. Commercial software like Ansys Fluent costs more than $10k per year for one user. The above-mentioned work models wind flow around a building for smooth flow, whereas extreme winds, which tend to be irregular, can cause various structural failures of buildings. These kinds of irregular wind flows are called turbulent flows. Thus, in this contribution, an additional three-week class module is provided for the ‘CFD for Wind Engineering’ class which includes hands-on material on modeling turbulent wind flow around a building using open-source software OpenFOAM and ParaView. To model the turbulence, Large Eddy Simulation (LES) is considered with a logarithmic inlet profile. To connect the log profile in a coarse grid, the law of the wall condition is also introduced in the OpenFOAM environment. To illustrate the application, the wind flow around a cubic building is considered. The current study’s case files and the extended report are provided at https://github.com/rpsuark/ASEE21-OpenFOAM-LES.more » « less
-
Abstract As insects fly, their wings generate complex wake structures that play a crucial role in their aerodynamic force production. This study focuses on utilizing reduced order modeling techniques to gain valuable insights into the fluid dynamic principles underlying insect flight. Specifically, we used an immersed-boundary-method-based computational fluid dynamics (CFD) solver to simulate a hovering hawkmoth’s wake, and then identified the most energetic modes of the wake using proper orthogonal decomposition (POD). Furthermore, we employed a sparse identification of nonlinear dynamics (SINDy) approach to find a simple reduced order model that relates the most energetic POD modes. Through this process, we formulated multiple different models incorporating varying numbers of POD modes. To compare the accuracy of these models, we utilized a force survey method to estimate the aerodynamic forces produced by the hawkmoth’s wings. This force survey method uses an impulse-based approach to calculate the aerodynamic lift and drag based solely on the velocity and vorticity information provided by the model. By comparing the estimated aerodynamic force with the actual force production calculated by the CFD solver, we were able to find the simplest model that still provides an accurate representation of the complex wake produced by the hovering hawkmoth wings. We also evaluated the stability and accuracy of this model as the number of flapping cycles increases with time. The reduced order modeling of insect flight has important implications for the design and control of bio-inspired micro-aerial vehicles. In addition, it holds the potential to reduce the computational cost associated with high-fidelity CFD simulations of complex flows.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.2514/6.2023-1757
