Analysis On PIV Data Of Turbulence Generated By Oscillating Grid In Wave Tank

Ocean turbulent mixing is an important factor regulating the ocean circulation.Surface waves play an important role in the upper ocean turbulent mixing.Wave-turbulent interaction is an important mechanism for turbulent mixing generated by non-breaking waves.Using ADV（Acoustic Doppler Velocimeter）and PIV（Particle Image Velocimetry）measuring the water velocity,the research team carried out three types of experiments in the wind-wave-flow multifunctional experimental water tank:experiments of turbulence generated by oscillating grid,experiments of mechanical wave and experiments of both mechanical wave and grid-produced turbulence.By the comparison of these three types of experiments,it is purposed to clarify the intensity and mechanism of the wave-turbulence interaction.This article mainly analyzes a portion of PIV data obtained in experiments of turbulence generated by oscillating grid,focusing on the accuracy of PIV image sub-pixel matching,on the effect of particle concentration,diameter and time interval of each pair of photos on the measurement results,and on the basic characteristics of turbulence generated by oscillating grid.Computer simulated particle images are used to precisely control the simulated particle sub-pixel displacement;The matching accuracy of the quadratic polynomial surface fitting method and the gradient-based sub-pixel displacement method under different windows are compared and the effects of particle concentration and size are discussed.For the situation of idiealinzed homogeneous advective flow,it is found that the matching accuracy of the quadratic polynomial surface fitting method increases with the increasing size of the matching window,and the matching accuracy of the gradient-based sub-pixel displacement algorithm almost does not change with the size of the matching window.For matching windows with the size of 16×16 pixels,32×32 pixel,64×64 pixel,the matching accuracy of the two aforementioned sub-pixel algorithms are both within 0.1 pixel.For the particle image of an ideal vortex flow,in the case of a matching window of 16×16,both matching methods can well characterize the vortex flow field,and the average absolute errors are also within 0.1pixel.The particle concentration and size affect the PIV measurement results.The larger the matching window is,the larger the optimal particle concentration range,which can be attributed to the amount of gray information contained in the matching windows.In terms of particle size,the results of the two sub-pixel algorithms show that the errors are smaller when the particle diameter is 3 pixels,indicating that it is appropriate to select particles with diameter of 3 pixels in practice.For the PIV data of the turbulence produced by the oscillating grid in the wave tank,three sets of experiment data with a time interval（Δt）of 5000,8000,and12000us for each pair of photos and sampling frequency of 14.5Hz are selected to analyze.Moreover,a set of longer time series of data with time interval（Δt）of 5000us（sampling frequency 6.0Hz,total sampling 600s）is used to get the long term variations of turbulence.The performance of the two aforementioned sub-pixel displacement matching algorithms in practical PIV image are evaluated,the effect of time intervals（Δt）on PIV results is discussed.Given the matching window is 64×64 pixel,the results from the quadratic polynomial surface fitting method and the gradient-based sub-pixel displacement method are compared.It is found that the probability distribution of turbulence velocity obtained by the former method tends to fit better with normally distribution,while the latter one shows a more obvious beam-like characteristic.Based on that we conclude that for the PIV data of turbulence produced by oscillating grid,the results obtained by the quadratic polynomial surface fitting method are more reasonable.Using the quadratic polynomial surface fitting method,the turbulence fields under different time interval（Δt）conditions are obtained.From the instantaneous flow field,eddies of different scales can be identified,and the time-averaged flow field appears as a counterclockwise vortex over the entire measurement area.Under the same time interval（Δt）,the probability distribution of the horizontal velocity U and the vertical velocity W are slightly different,but they basically conform to the Gaussian distribution.As time interval（Δt）increases,the probability distribution of the flow velocity meets better with the Gaussian distribution,and the beam-like characteristics gradually weakens.The velocity spectrum of the turbulent flow generated by the oscillating grid basically conforms to the-5/3 power law of Kolmogorov theory,except for the frequency spectrum of the horizontal velocity U,which shows a significant peak at 3 Hz.When time interval（Δt）is 8000us and 12000us respectively,the averaged wave number spectrum of vertical and horizontal flow velocities along the vertical and horizontal directions also conform with the power law of-5/3 power of Kolmogorov theory.On the homogeneity and isotropy of generated turbulence,at different time interval（Δt）,the calculated root mean square velocity and the magnitude and spatial distribution of W_rms/U_rms vary with sampling time interval（Δt）.Taking 30%as a measure of the homogeneity and isotropy of turbulence,that is,when the root-mean-square velocity is between 0.7～1.3 of its spatial average,the area is considered to be an area with quasi-homogeneous turbulence and when W_rms/U_rms is between 0.7 and 1.3,this flow is regarded as an quasi-isotropic turbulence.The results show that within the measurement area,the turbulence generated by oscillating grid can be regarded as quasi-homogeneous and quasi-isotropic among over 90%of the area.The results of the long-time data,which lasts for 600 seconds and sampled with a time interval（Δt）of 5000us,show that the turbulence generated by oscillating grid in the wave tank changes evidently with time.During the observation period,the averaged turbulent kinetic energy is 2.28cm²/s²,while the maximum value and minimum reaches 10.16cm²/s² and 0.45cm²/s² respectively.