This repository contains OpenFOAM v2306 case setup files and results for channel flow at $Re_\tau = 1000$.

In the data M1, M2, and M3 correspond to 3 different grids, increasingly fine. The M2_rk and M2_sigma cases correspond to simulations on the M2 grid using the RKSymFOAM solver, and the Sigma LES model.

A typical Allrun script is provided inside M1. Things are quite straightforward. But of course the user should first modify the decomposeParDict file to select the appropriate number of ranks. Also it does not account for the restart after transient ends and the turbulent statistics need to be reset (unless fieldAverage has been disabled during the transient phase).

Our suggestion is to keep fieldAverage always on, first run the case from 0 to 250, look at the statistics and see if they are generally reasonable. Then delete all the 250/*Mean files and the 250/uniform folders and run the case up to time 750. It is always a good idea to look at statistics earlier, even if they are contaminated by transients. They typically provide a good inidcation on whether the case is correctly set up.

All the results were generated using OpenFOAM v2306. Additional software includes:

• swak4Foam A large toolkit for OpenFOAM. In this project only funkySetFields is used.
• RKSymFOAM Low-dissipation solver with Runge-Kutta time steppers.
• runTimeChannelBudgets Tool for computing TKE budget terms.
• ofreaders Tool for handling various OpenFOAM-generated files in Python. Here used for converting probe data to HDF5.
• Standard Python packages for scientific computing: numpy, matplotlib, h5py.

• The folder analysis_scripts contains Python scripts that can be used to generate figures found in the paper, and compute quantities like spectra.
• The folder dns contains a bash script for downloading the DNS data by Lee and Moser (2015). The paper also makes use of another DNS dataset for skewness and flatness, obtained by us through private communication with P. Schlatter and S. Toosi at FAU, Germany. As such, we are not at liberty to share it publically.

For each of the cases the following is provided:

• A dummy 0.org directory with constant initial conditions.
• A constant directory, with no mesh.
• A system directory with all the necessary configuration files, including a blockMeshDict to generate the grid.
• A postProcessing directory with subdirectory collapsedFields containing the statistics at time 750 and (for selected cases) subdirectory averageYPlus containing the development of spatially-averaged $y^+$ in time.
• A graphs/750 directory containing the profiles of the TKE budget terms.
• A plane_stats.txt file containing single-point statistics computed using probe data. These can be compared to those in the postProcessing directory.
• z_spectrum_*.txt files containing the spanwise spectra for each wall-normal value.
• coherence_*.txt files containing the linear coherence spectra with respect to $y^+ = 2$.

Due to the size of the produced probe data files (> 100 Gb for the M3 case) we are unable to easily share them at scale. The data can be considered available upon request. However, we provide a portion of the data for the M1 grid via the following links. In these files only every 4th time-step is retained.

streamwise component

wall-normal component

spanwise component