Data, code and input for "Fine-Scale Jet Mixing Noise: Computing the Source Spectrum with One-Dimensional Turbulence"
Description
This dataset supports the manuscript "Fine-Scale Jet Mixing Noise: Computing the Source Spectrum with One-Dimensional Turbulence" by S. Sharma (German Aerospace Center, DLR, Institute of Aerodynamics and Flow Technology), M. Klein (BTU Cottbus-Senftenberg), L. J. Ayton (University of Cambridge) and C. Bogey (CNRS, LMFA, Ecole Centrale de Lyon); submitted to the AIAA Journal. The paper concerns the fine-scale mixing noise of a round jet at Mach 0.3 and Reynolds number 100,000. Reduced-order predictions usually pair a statistical source model, such as that of Tam and Auriault, with the turbulent kinetic energy and dissipation rate from a RANS simulation, which fixes the shape of the source spectrum by assumption. This work computes that spectrum instead. The One-Dimensional Turbulence (ODT) model, used as a standalone tool, resolves the full range of turbulent scales along a downstream-advected radial line through the jet at a small fraction of the cost of large-eddy simulation (LES). From an ensemble of ODT realisations, a resolved-source-spectrum (RSS) method obtains the fine-scale noise-source spectrum directly from the resolved signal, with no dissipation estimate and no assumed spectral shape. The ODT jet, configured without reference to the acoustic data, reproduces the mean flow of a reference LES, and the RSS prediction carried through the same near-field-to-far-field transformation as the LES source recovers the far-field spectrum the LES source produces. This deposit contains the authors' own material: the ODT simulation data, an axisymmetric RANS field, and the Python and figure code. Data: the processed arrays behind the figures. These are the ODT mean-flow and turbulence ensembles, the ODT resolved source spectra (the radial source-spectrum map and the streamwise-resolved lip-line spectrum), an axisymmetric RANS field used for the Tam and Auriault contrast, one dense-dump ODT realisation with its accepted eddy events, and digitised experimental free-jet profiles. The production ODT solver input file (Appendix B of the paper) and the radial probe locations are included so the simulations can be rerun from scratch. Code: Python post-processing and figure scripts (NumPy, SciPy, Matplotlib) that read the data files and reproduce the figures. A figure-to-script-to-data map is provided. The raw ODT probe time series is not included because of its size; the processed arrays reproduce the ODT figures directly and are available from the corresponding author on reasonable request. The experimental profiles were digitised from Hussein, Capp and George, J. Fluid Mech. 258 (1994) 31-75; credit that source when reused. The ODT simulations use the open-source adaptive ODT code of Stephens and Lignell, SoftwareX 13 (2021) 100641, in the cylindrical adaptive-mesh formulation of Lignell et al. Data and the input file are licensed under CC BY 4.0; code under the MIT licence. Please cite the manuscript and this deposit if you use the data.
Files
Steps to reproduce
Requirements: Python 3 with NumPy, SciPy, Matplotlib and PyYAML (see requirements.txt). No other dependencies. Layout: the deposit is three folders - code/ (scripts), input/ (ODT solver input) and data/ (the arrays). Keep them side by side. The scripts read the arrays in data/ and write their PDF figures into a figures/ folder created at the deposit root; run them from the deposit root. Quick start (from the deposit root): python -m pip install -r code/requirements.txt python code/gen_prod2_figs.py Figures that reproduce fully from the ODT data in this deposit (full mapping in FIGURE_MAP.md): - Development of one ODT realisation with its eddy events: plot_evolution.py. - Centreline decay and turbulence intensity: gen_prod2_figs.py (also compared with the digitised Hussein et al. experiment). - Far-field directivity of the RSS prediction: plot_dirsrc_main_rev2.py. - Potential-core length versus the large-eddy-suppression parameter Z_LES: plot_zlessweep_rev2.py. Figures that compare the ODT with the reference LES (figures 3-7, 10-12, and the LES-matched reference in 13) need reduced LES arrays that are not part of this deposit (the LES authors' data). The scripts plot_fig2_prod2.py, plot_bogey_final_rev2.py, plot_cmp_profiles_rev2.py, plot_pointspec.py, plot_srcmap.py and plot_srcx.py detect the missing arrays and stop with a note. Obtain the arrays from the LES authors and place them in data/ to reproduce these figures. Rerunning the ODT simulations: the configuration runs from scratch on the public ODT release (Stephens and Lignell, SoftwareX 2021) using input/odt_input.yaml; no code modification is required. The 32 radial probe locations are in input/probe_radii.txt. Data conventions: the .npz files are read with numpy.load and store the processed fields (mean profiles, spectra Phi_q(r,St) and convection speeds). The Strouhal number is St = f D / U_j with D = 0.06223 m and U_j = 102.9 m/s. Source-map levels are in dB relative to the maximum. The radial coordinate is signed, since the ODT domain spans the diameter. Not included: the raw ODT probe signals, on account of size; the processed arrays in data/ reproduce the ODT figures without them. The reference-LES arrays (the LES authors' data). The Viswanathan microphone comparison from an earlier draft.