The velocities measured by particle image velocimetry (PIV) and particle tracking velocimetry (PTV) commonly provide sparse information on flow motions. A dense velocity field with high resolution is indispensable for data visualization and analysis. In the present work, a physics-informed neural network (PINN) is proposed to reconstruct the dense velocity field from sparse experimental data. A PINN is a network-based data assimilation method. Within the PINN, both the velocity and pressure are approximated by minimizing a loss function consisting of the residuals of the data and the Navier–Stokes equations. Therefore, the PINN can not only improve the velocity resolution but also predict the pressure field. The performance of the PINN is investigated using two-dimensional (2D) Taylor's decaying vortices and turbulent channel flow with and without measurement noise. For the case of 2D Taylor's decaying vortices, the activation functions, optimization algorithms, and some parameters of the proposed method are assessed. For the case of turbulent channel flow, the ability of the PINN to reconstruct wall-bounded turbulence is explored. Finally, the PINN is applied to reconstruct dense velocity fields from the experimental tomographic PIV (Tomo-PIV) velocity in the three-dimensional wake flow of a hemisphere. The results indicate that the proposed PINN has great potential for extending the capabilities of PIV/PTV.

Coronavirus disease 2019 has become a global pandemic infectious respiratory disease with high mortality and infectiousness. This paper investigates respiratory droplet transmission, which is critical to understanding, modeling, and controlling epidemics. In the present work, we implemented flow visualization, particle image velocimetry, and particle shadow tracking velocimetry to measure the velocity of the airflow and droplets involved in coughing and then constructed a physical model considering the evaporation effect to predict the motion of droplets under different weather conditions. The experimental results indicate that the convection velocity of cough airflow presents the relationship t^−0.7 with time; hence, the distance from the cougher increases by t^0.3 in the range of our measurement domain. Substituting these experimental results into the physical model reveals that small droplets (initial diameter D ≤ 100 μm) evaporate to droplet nuclei and that large droplets with D ≥ 500 μm and an initial velocity u₀ ≥ 5 m/s travel more than 2 m. Winter conditions of low temperature and high relative humidity can cause more droplets to settle to the ground, which may be a possible driver of a second pandemic wave in the autumn and winter seasons.

Near-wall velocity prediction for wall-bounded turbulence is useful for constructing a wall model and estimating dissipation and wall shear stress. A convolutional neural network is developed to improve the near-wall velocity prediction and spatial resolution for wall-bounded turbulent velocity fields obtained using particle image velocimetry (PIV). To establish the relationship between the low-resolution and high-resolution fields, this machine learning model is trained on a synthetic PIV dataset generated based on velocity fields obtained from the direct numerical simulation of turbulent channel flows at Re_τ = 1000. Using a test dataset with a higher Reynolds number of Re_τ = 5200, the performance of this model is assessed in terms of instantaneous fields, error analysis, velocity statistics, and energy spectra. The influences of the interrogation window, image resolution, and particle concentration on the performance of this network are also considered. We further apply this network to practical PIV data from a turbulent boundary layer at Re_τ = 2200 to assess the network performance under real experimental conditions. The results indicate that the proposed machine-learning-based model can predict missing near-wall velocity fields and enhance the spatial resolution of PIV fields, but the accuracy for Reynolds shear stress prediction needs to be further improved. The presented approach shows the potential ability to predict the near-wall instantaneous velocity of high-Reynolds-number turbulence from low-Reynolds-number flow fields.

Flows around an axisymmetric body of revolution at a zero yaw angle were studied using a hybrid Reynolds-averaged Navier–Stokes (RANS)/large eddy simulation (LES) approach, which employed a full Reynolds stress model (RSM) in the RANS branch with the aim of accounting for the Reynolds stress anisotropy, streamline curvature, and flow separations in the boundary layer. The SUBOFF model without appendages was applied to conduct the simulations, and the Reynolds number based on the free-stream velocity and the length of the body is $R{e}_L=1.2\times {10}^6$ . The results, including time-averaged C_p, C_f, and velocity statistics, were compared with the experimental data and wall-resolved LES results available in the literature, and the overall agreement of the comparisons was satisfactory. To assess the performance of the RSM-based hybrid RANS/LES approach, we carried out shear-stress transport-based hybrid RANS/LES approach simulations under identical free-stream conditions for comparison. The sensitivity of the hybrid RANS/LES approach to the RANS models was observed for separated flow with surface curvature and adverse pressure gradient-induced separation. The RSM-based hybrid RANS/LES approach was found to provide a better prediction for the unsteady flows near the stern. That is because the effects of the streamline curvature and the strong interactions between individual stresses can be captured by the exact production terms in the RSM-based hybrid RANS/LES approach. These effects are important for predicting the development of turbulent boundary layers along the stern.

A novel method is proposed to combine the wall-modeled large-eddy sim-ulation (LES) with the diffuse-interface direct-forcing immersed boundary (IB) method. The new developments in this method include: (i) the momentum equation is integrated along the wall-normal direction to link the tangential component of the effective body force for the IB method to the wall shear stress predicted by the wall model; (ii) a set of Lagrangian points near the wall are introduced to compute the normal component of the effective body force for the IB method by reconstructing the normal component of the velocity. This novel method will be a classical direct-forcing IB method if the grid is fine enough to resolve the flow near the wall. The method is used to simulate the flows around the DARPA SUBOFF model. The results obtained are well comparable to the measured experimental data and wall-resolved LES results.

A kinematic simulation with an approximate deconvolution (KSAD) hybrid model is proposed to predict the Lagrangian relative dispersion of fluid particles in a large eddy simulation (LES) of isotropic turbulent flows. In the model, a physical connection between the resolved and subgrid scales is established through the energy flux rate at the filter width scale. Due to the lack of subgrid-scale (SGS) turbulent structures and SGS model errors, the LES cannot accurately predict the two- and multi-point Lagrangian statistics of the fluid particles. To improve the predictive capability of the LES, we use an approximate deconvolution model to improve the resolved scales near the filter width and a kinematic simulation to recover the missing velocity fluctuations beneath the subgrid scales. To validate the proposed hybrid model, we compare the Lagrangian statistics of two- and four-particle dispersion with the corresponding results from the direct numerical simulation and the conventional LES. It is found that a significant improvement in the prediction of the Lagrangian statistics of fluid particles is achieved through the KSAD hybrid model. Furthermore, a parametric study regarding the wavenumbers and orientation wavevectors is conducted to reduce the computational cost. Good results can be obtained using a small number of wavenumber modes and orientation wavevectors. Thus, we can improve the prediction of the Lagrangian dispersion of fluid particles in the LES by applying the KSAD hybrid model at an acceptable computational cost.