A convolutional autoencoder is trained using a database of airfoil aerodynamic simulations and assessed in terms of overall accuracy and interpretability. The goal is to predict the stall and to investigate the ability of the autoencoder to distinguish between the linear and non-linear response of the airfoil pressure distribution to changes in the angle of attack. After a sensitivity analysis on the learning infrastructure, we investigate the latent space identified by the autoencoder targeting extreme compression rates, i.e. very low-dimensional reconstructions. We also propose a strategy to use the decoder to generate new synthetic airfoil geometries and aerodynamic solutions by interpolation and extrapolation in the latent representation learned by the autoencoder.
Rotor blades of wind turbines in the atmospheric boundary layer regularly experience the aerodynamic phenomenon of dynamic stall consisting of a temporary overshoot of lift and detrimental fatigue loads. Particularly the formation of dynamic stall under three-dimensional inflow conditions raises open questions. Aerodynamic behavior of a DU 91-W2-250 wind profile undergoing light dynamic stall is thus analyzed in a wind tunnel. Effects of a gust with streamwise and spanwise periodic variation are investigated by comparing total and local lift generation with flow formation above the airfoil. Th…
The aerodynamic performance of the high-lift configuration greatly influences the safety and economy of commercial aircraft. Accurately predicting the aerodynamic performance of the high-lift configuration, especially the stall behavior, is important for aircraft design. However, the complex flow phenomena of high-lift configurations pose substantial difficulties to current turbulence models. In this paper, a three-equation k-(v^2)-ω turbulence model for the Reynolds-averaged Navier-Stokes equations is used to compute the stall behavior of high-lift configurations. A separated shear layer fixe…
The present paper numerically investigates flow control over NACA0021 (National Advisory Committee for Aeronautics) airfoil by applying a leading-edge tubercle as a passive flow control device at transitional Reynolds number (Re) regime. The study includes simulations over the unmodified airfoil and modified airfoil operating at Re=120000 using kT-kL-ω based Delayed Detached Eddy Simulation (DDES) model. The calculations include both the pre-stall and post-stall regions, while the performance of the mean flow aerodynamic properties like pressure, lift, skin-friction, and drag coefficients are …
Dynamic stall on airfoils negatively impacts their aerodynamic performance and can lead to structural damage. Accurate prediction and modelling of the dynamic stall loads are crucial for a more robust design of wings and blades that operate under unsteady conditions susceptible to dynamic stall and for widening the range of operation of these lifting surfaces. Many dynamic stall models rely on empirical parameters that need to be obtained from experimental or numerical data which limits their generalisability. Here, we introduce physically derived times scales to replace the empirical paramete…
The unsteady characteristics of the flow over thick flatback airfoils have been investigated by means of CFD calculations. Sandia airfoils which have 35% maximum thickness with three different trailing edge thicknesses were selected. The calculations provided good results compared with available experimental data with regard to the lift curve and the impact of trailing edge thickness. Unsteady CFD simulations revealed that the Strouhal number is found to be independent of the lift coefficient before stall and increases with the trailing edge. The present work shows the dependency of the Strouh…
We investigate a typical aerofoil section under dynamic stall conditions, the structural model is linear and the aerodynamic loading is represented by the Leishman-Beddoes semi-empirical dynamic stall model. The loads given by this model are non-linear and non-smooth, therefore we have integrated the equation of motion using a Runge-Kutta-Fehlberg algorithm equipped with event detection. The main focus of the paper is on the interaction between the Hopf bifurcation typical of aero-elastic systems, which causes flutter oscillations, and the discontinuous definition of the stall model. The paper…
High-fidelity simulations are performed to study active flow control techniques for alleviating deep dynamic stall of a SD7003 airfoil in plunging motion. The flow Reynolds number is $Re=60{,}000$ and the freestream Mach number is $M=0.1$. Numerical simulations are performed with a finite difference based solver that incorporates high-order compact schemes for differentiation, interpolation and filtering on a staggered grid. A mesh convergence study is conducted and results show good agreement with available data in terms of aerodynamic coefficients. Different spanwise arrangements of actuator…
Information-theoretic machine learning for time-varying mode decomposition of separated aerodynamic flows
Department of Aerospace Engineering, Tohoku University（东京东京大学航空工程系）；Department of Mechanical and Aerospace Engineering, Tokyo University of Science（东京科学大学机械与航空航天工程系）
Airfoils pitching in the stalled regime have been of keen interest in recent years due to their desirable aerodynamic force characteristics. In this study, we are numerically investigating the unsteady flow past a NACA 0012 airfoil under sinusoidally pitching motion, using a finite volume based sharp interface immersed boundary solver. The flow is investigated in the low Reynolds number regime (Re=3000) for reduced frequencies of 0.1 and 0.5, at three different pivot locations (c/3, c/2 and 2c/3 from the leading edge). The airfoil is subjected to sinusoidal oscillations with its incidence angl…
A Synergistic Framework Leveraging Autoencoders and Generative Adversarial Networks for the Synthesis of Computational Fluid Dynamics Results in Aerofoil Aerodynamics
Multi-fidelity aerodynamic data fusion by autoencoder transfer learning
Department of Aerospace Engineering, UC3M（航空航天工程系，UC3M）；Theoretical and Computational Aerodynamics Group, Flight Physics Department, INTA（理论与计算空气动力学组，飞行物理部门，INTA）
The influence that the kinematics of pitching and heaving 2D airfoils have on the aerodynamic forces is investigated using Direct Numerical Simulations and a force decomposition algorithm. Large amplitude motions are considered (of the order of one chord), with moderate Reynolds numbers and reduced frequencies of order 1, varying the mean pitch angle and the phase shift between the pitching and heaving motions. Our results show that the surface vorticity contribution (viscous effects) to the aerodynamic force is negligible compared to the contributions from the body motion (fluid inertia) and …
Effect of airfoil thickness on onset of dynamic stall is investigated using large eddy simulations at chord-based Reynolds number of 200,000. Four symmetric NACA airfoils of thickness-to-chord ratios of 9%, 12%, 15%, and 18% are studied. The 3-D Navier Stokes solver, FDL3DI is used with a sixth-order compact finite difference scheme for spatial discretization, second-order implicit time integration, and discriminating filters to remove unresolved wavenumbers. A constant-rate pitch-up maneuver is studied with the pitching axis located at the airfoil quarter chord point. Simulations are performe…
An approximation model based on convolutional neural networks (CNNs) is proposed for flow field predictions. The CNN is used to predict the velocity and pressure field in unseen flow conditions and geometries given the pixelated shape of the object. In particular, we consider Reynolds Averaged Navier-Stokes (RANS) flow solutions over airfoil shapes. The CNN can automatically detect essential features with minimal human supervision and shown to effectively estimate the velocity and pressure field orders of magnitude faster than the RANS solver, making it possible to study the impact of the airf…
Figuring out the right airfoil is a crucial step in the preliminary stage of any aerial vehicle design, as its shape directly affects the overall aerodynamic characteristics of the aircraft or rotorcraft. Besides being a measure of performance, the aerodynamic coefficients are used to design additional subsystems such as a flight control system, or predict complex dynamic phenomena such as aeroelastic instability. The coefficients in question can either be obtained experimentally through wind tunnel testing or, depending upon the accuracy requirements, by numerically simulating the underlying …
Adjoint method is widely used in aerodynamic design because only once solution of flow field is required for adjoint method to obtain the gradients of all design variables. However, the calculation cost of adjoint vector is approximately equal to that of flow computation. In order to accelerate the solution of adjoint vector and improve the adjoint-based optimization efficiency, machine learning for adjoint vector modeling is presented. Deep neural network (DNN) is employed to construct the mapping between the adjoint vector and the local flow variables. DNN can efficiently predict adjoint vec…
The prediction of aircraft aerodynamic quantities of interest remains among the most pressing challenges for computational fluid dynamics. The aircraft aerodynamics are inherently turbulent with mean-flow three-dimensionality, often accompanied by laminar-to-turbulent transition, flow separation, secondary flow motions at corners, and shock wave formation, to name a few. However, the most widespread wall models are built upon the assumption of statistically-in-equilibrium wall-bounded turbulence and do not faithfully account for the wide variety of flow conditions described above. This raises …
Effects of a gurney flap were numerically investigated on the supercritical NASA airfoil by solving the two-dimensional Reynolds-averaged Navier-Stokes equations for a range of transonic Mach numbers and angles of attack, using turbulence compressible KW SST model. The height of the gurney flap was selected to be 1.65 percent chord length. A high-resolution mesh was applied to accurately predict the flow field specifically in the vicinity of the airfoil. Below the drag divergence Mach number, the gurney flap has a remarkable influence on the aerodynamic coefficients especially at -1 and 0 degr…
Efficiently predicting the flowfield and load in aerodynamic shape optimisation remains a highly challenging and relevant task. Deep learning methods have been of particular interest for such problems, due to their success for solving inverse problems in other fields. In the present study, U-net based deep neural network (DNN) models are trained with high-fidelity datasets to infer flow fields, and then employed as surrogate models to carry out the shape optimisation problem, i.e. to find a drag minimal profile with a fixed cross-section area subjected to a two-dimensional steady laminar flow.…
This article presents an original methodology for the prediction of steady turbulent aerodynamic fields. Due to the important computational cost of high-fidelity aerodynamic simulations, a surrogate model is employed to cope with the significant variations of several inflow conditions. Specifically, the Local Decomposition Method presented in this paper has been derived to capture nonlinear behaviors resulting from the presence of continuous and discontinuous signals. A combination of unsupervised and supervised learning algorithms is coupled with a physical criterion. It decomposes automatica…
Predicting and simulating aerodynamic fields for civil aircraft over wide flight envelopes represent a real challenge mainly due to significant numerical costs and complex flows. Surrogate models and reduced-order models help to estimate aerodynamic fields from a few well-selected simulations. However, their accuracy dramatically decreases when different physical regimes are involved. Therefore, a method of local non-intrusive reduced-order models using machine learning, called Local Decomposition Method, has been developed to mitigate this issue. This paper introduces several enhancements to …
