The three-dimensional structure of the flow behind a heaving and pitching finite-span wing is investigated using dye flow visualization at a Reynolds number of 164. Phase-locked image sequences, which are obtained from two orthogonal views, are combined to create a set of composite images that give an overall sense of the three-dimensional structure of the flow. A model of the vortex system behind the wing is constructed from the image sequences. Variations of the Strouhal number, pitch amplitude and heave/pitch phase angle are qualitatively shown to affect the structure of the wake.

A visual study of the flow structures behind a heaving and pitching finite-span wing

Multi-fidelity deep neural network surrogate model for aerodynamic shape optimization

We propose a method using supervised machine learning to estimate velocity fields from particle images having missing regions due to experimental limitations. As a first example, a velocity field around a square cylinder at Reynolds number of ${\rm Re}_D=300$ is considered. To train machine learning models, we utilize artificial particle images (APIs) as the input data, which mimic the images of the particle image velocimetry (PIV). The output data are the velocity fields, and the correct answers for them are given by a direct numerical simulation (DNS). We examine two types of the input data: APIs without missing regions (i.e., full APIs) and APIs with missing regions (lacked APIs). The missing regions in the lacked APIs are assumed following the exact experimental situation in our wind tunnel setup. The velocity fields estimated from both full and lacked APIs are in great agreement with the reference DNS data in terms of various statistical assessments. We further apply these machine learned models trained with the DNS data to experimental particle images so that their applicability to the exact experimental situation can be investigated. The velocity fields estimated by the machine learned models contain approximately 40 folds denser data than that with the conventional cross-correlation method. This finding suggests that we may be able to obtain finer and hidden structures of the flow field which cannot be resolved with the conventional cross-correlation method. We also found that even the complex flow structures are hidden due to the alignment of two square cylinders, the machine learned model is able to estimate the field in the missing region reasonably well. The present results indicate a great potential of the proposed machine learning based method as a new data reconstruction method for PIV.

Experimental velocity data estimation for imperfect particle images using machine learning

From active learning to deep reinforcement learning: Intelligent active flow control in suppressing vortex-induced vibration

In the present study, we use the dynamic mesh method based on the radial basis function interpolation for the two-dimensional simulation of harmonically oscillating NACA0015 airfoil. Under various flapping frequencies, heaving and pitching amplitudes, the observed wake flows can be divided into seven types, including the B\'enard-von K\'arm\'an (BvK) vortex street, the reversed BvK (RBvK) vortex street, the $1P$ wake, the $mP$ wake, the $2P+mS$ wake, the $2S+mS$ wake, and the $mS$ wake, where $m$ is around 4 and $xS+yP$ signifies $x$ single vortices and $y$ vortex pairs shedding per oscillation period. Then we have constructed two phase diagrams of the wake types in terms of the flapping frequency, heaving and pitching amplitudes. Importantly, we have combined the propulsion performance of the flapping airfoil with the wake map and found that $\ensuremath{\alpha}(\frac{T}{4})$, the angle of attack at $t=T/4$, can determine the wake type: negative value corresponding to drag-dominated wakes, while positive value corresponding to thrust dominated flow wakes. With the increase of $\ensuremath{\alpha}(\frac{T}{4})$, the wake transforms from the $mP$ to $2S+mS$ then to RBvK and eventually to $1P$ wake. Furthermore, the coherent structure analysis and spectral analysis are conducted for all the types of wakes by using dynamic mode decomposition. And there is a positive correlation between the strengths of vortices shedding at $i$ times flapping frequency and the modulus of the $i\text{th}$ dynamic mode decomposition mode, which can further reveal the differences among different types of wakes.

Propulsion performance of a two-dimensional flapping airfoil with wake map and dynamic mode decomposition analysis.

Unsteady flow prediction from sparse measurements by compressed sensing reduced order modeling

In the present work, we propose an optimization framework based on the active learning method, which aims to quickly determine the conditions of tandem flapping wings for optimal performance in terms of thrust or efficiency. Especially, multi-fidelity Gaussian process regression is used to establish the surrogate model correlating the kinematic parameters of tandem flapping wings and their aerodynamic performances. Moreover, the Bayesian optimization algorithm is employed to select new candidate points and update the surrogate model. With this framework, the parameter space can be explored and exploited adaptively. Two optimization tasks of tandem wings are carried out using this surrogate-based framework by optimizing thrust and propulsion efficiency. The response surfaces predicted from the updated surrogate model present the influence of the flapping frequency, phase, and separation distance on thrust and efficiency. It is found that the time-average thrust of the hind flapping wing increases with the frequency. However, the increase in frequency may lead to a decrease in propulsive efficiency in some circumstances.

Hongyu Zheng

Papers

From active learning to deep reinforcement learning: Intelligent active flow control in suppressing vortex-induced vibration

Propulsion performance of a two-dimensional flapping airfoil with wake map and dynamic mode decomposition analysis.

Unsteady flow prediction from sparse measurements by compressed sensing reduced order modeling

Active learning of tandem flapping wings at optimizing propulsion performance

Kinematic parameter optimization of a flapping ellipsoid wing based on the data-informed self-adaptive quasi-steady model