Abstract

Abstract

We develop a wavelet-based three-dimensional convolutional neural network (WCNN3d) for superresolution of coarse-grained data of homogeneous isotropic turbulence. The turbulent flow data are computed by high resolution direct numerical simulation (DNS), while the coarse-grained data are obtained by applying a Gaussian filter to the DNS data. The CNNs are trained with the DNS data and the coarse-grained data. We compare vorticity- and velocity-based approaches and assess the proposed WCNN3d method in terms of flow visualization, enstrophy spectra and probability density functions. We show that orthogonal wavelets enhance the efficiency of the learning of CNN.

Abstract

In this work we identify the coherent vortices in the leeward wake of an HQ 17 airfoil with a Gurney flap. This high-lift device consists of a small plate perpendicular to the airfoil, at or near the trailing edge, with a height H of the order from 1% to 4% of the chord. The overall effect is to produce a significant increase in lift, together with a relatively minor increase in airfoil drag. Its geometry generates vortex shedding of spatial scale H, which modulates the wake and thus influence the global configuration of the flow. A numerical study was performed with the commercial software ANSYS Fluent. Simultaneously, experiments were carried out in the wind tunnel of the Aeronautical Departmental Area from the Engineering School of the National University of La Plata. With a Dantec StreamLine hot-wire anemometer, records of the instantaneous horizontal and vertical velocity components in the airfoil wake were obtained. Both the experimental data and those obtained by numerical simulation were analyzed with the continuous wavelet transform technique, in order to identify and characterize the coherent vortices in the wake. Results show the shed vortices structure and the organization degree of this periodic shedding.