Lagrangian statistics in turbulent channel flow ---- 10/06/2021 ---- Authors: Davide Perrone @Politecnico di Torino [davide.perrone@polito.it] Hans Kuerten @TU Eindhoven Luca Ridolfi @Politecnico di Torino Stefania Scarsoglio @Politecnico di Torino ---- Collection of Lagrangian statistics in a turbulent channel flow obtained via numerical simulation. Simulations were performed by means of a pseudo-spectral direct numerical simulation code at four different friction Reynolds numbers ReT (180, 395, 590, 950). The velocity field resulting from the simulation is numerically integrated to obtain the trajectories. A total of 600000 particle trajectories are integrated at each Reynolds number. The time step of the trajectories in wall units dt+ is: ReT dt+ -- -- 180 0.225 395 0.29625 590 0.36875 950 0.475 For more details on the method and the parameters employed, see: J. G. M. Kuerten and J. J. H. Brouwers , "Lagrangian statistics of turbulent channel flow at Reτ = 950 calculated with direct numerical simulation and Langevin models", Physics of Fluids 25, 105108 (2013) https://doi.org/10.1063/1.4824795 Starting from the trajectory data, the velocities and accelerations of particles are computed using a 4th order finite differencing scheme. Particle velocities and accelerations are computed in wall units. The fluctuating velocity is obtained subtracting the local mean Eulerian velocity. Lagrangian statistics are then computed at Ny distinct wall-normal locations; Ny differs for different Reynolds numbers. The point closest to the wall is usually at y+ ~ 0.5; points are distributed in both halves of the channel. Particles are binned to a specific y+ location if they enter a measurement volume surrounding that location; they are binned again at the same y+ location if they enter the same volume at a later time, but only if a definite amount of time has passed (nonetheless, a bias may be introduced: see https://tel.archives-ouvertes.fr/tel-01739689/document, page 104). The included statistics are: 1) the mean axial velocity U 2) the velocity covariance matrix 3) the acceleration covariance matrix 4) the mixed velocity-acceleration covariance matrix 5) the velocity autocorrelation functions rho_u_i, computed for Nt = 2000 time steps 6) the acceleration autocorrelation functions rho_a_i, computed for Nt = 2000 time steps 7) the velocity and acceleration time scales (= the autocorrelation functions integrated up to the first zero crossing) 8) the second order Lagrangian structure function ---- Included files for each Reynolds: retXXX\ y.txt --> y+ coordinates (in wall units) of the Ny wall-normal locations at which statistics are sampled tau.txt --> delay vector for the autocorrelation function U.txt --> mean axial velocity U as a function of y (size: [Ny, 1]) vcov.txt --> velocity covariances (size: [Ny, 6]) acov.txt --> acceleration covariances (size: [Ny, 6]) mixcov.txt --> mixed covariances (size: [Ny, 9]) rho_u.txt --> streamwise velocity autocorrelation (size: [Ny, Nt]) rho_u.txt --> wall-normal velocity autocorrelation (size: [Ny, Nt]) rho_u.txt --> spanwise velocity autocorrelation (size: [Ny, Nt]) rho_a_x.txt --> streamwise acceleration autocorrelation (size: [Ny, Nt]) rho_a_y.txt --> wall-normal acceleration autocorrelation (size: [Ny, Nt]) rho_a_z.txt --> spanwise acceleration autocorrelation (size: [Ny, Nt]) TV.txt --> Lagrangian velocity timescale (size: [Ny, 3]) TA.txt --> Lagrangian acceleration timescale (size: [Ny, 3]) lsf_u.txt --> streamwise 2nd order Lagrangian structure function (size: [Ny, Nt]) lsf_v.txt --> wall-normal 2nd order Lagrangian structure function (size: [Ny, Nt]) lsf_w.txt --> spanwise 2nd order Lagrangian structure function (size: [Ny, Nt]) retXXX.h5 --> all of the previous data in a HDF5 dataset ---- This work was sponsored by NWO Exacte en Natuurwetenschappen (Physical Sciences) for the use of supercomputer facilities, with financial support from the Netherlands Organization for Scientific Research, NWO. Additional computational resources were provided by HPC@POLITO (https://www.hpc.polito.it)