Prediction of Turbulent Temperature Fluctuations in Hot Jets

• Published in: AIAA journal 2018-08, Vol.56 (8), p.3097-3111
• Main Author: DeBonis, James R
• Format: Article
• Language: English
• Subjects: Computational fluid dynamics; Computer simulation; Finite difference method; Fluctuations; Fluid flow; Heat; Heat flux; Heat transfer; Induction heating; Large eddy simulation; Mathematical models; Navier-Stokes equations; Nozzles; Particle image velocimetry; Prandtl number; Predictions; Raman spectroscopy; Rayleigh scattering; Simulation; Temperature; Temperature distribution; Velocity distribution; Velocity measurement
• ISSN: 0001-1452; 1533-385X
• DOI: 10.2514/1.J056596

Large-eddy simulations were used to investigate turbulent temperature fluctuations and turbulent heat flux in hot jets. A high-resolution finite difference Navier–Stokes solver was used to compute the flow from a 2 in. round nozzle. Three different flow conditions of varying jet Mach numbers and temperature ratios were examined. The large-eddy simulation results showed that the temperature field behaved similarly to the velocity field but with a more rapidly spreading mixing layer. Predictions of the mean u¯i and fluctuating ui′ velocities were compared to particle image velocimetry data. Predictions of the mean T¯ and fluctuating T′ temperatures were compared to data obtained using Rayleigh scattering and Raman spectroscopy. Very good agreement with experimental data was demonstrated for the mean and fluctuating velocities. The large-eddy simulation correctly predicted the behavior of the turbulent temperature field but overpredicted the levels of the fluctuations. The turbulent heat flux was examined and compared to the Reynolds-averaged Navier–Stokes results. The large-eddy simulation and Reynolds-averaged Navier–Stokes simulations produced very similar results for the radial heat flux. However, the axial heat flux obtained from the large-eddy simulation differed significantly from the Reynolds-averaged Navier–Stokes result in both structure and magnitude, indicating that the Reynolds-averaged Navier–Stokes model was inadequate. Finally, the large-eddy simulation data were used to compute the turbulent Prandtl number and verify that a constant value of 0.7, which is typically used in the Reynolds-averaged Navier–Stokes models, was a reasonable assumption.