Contours of the real part of normalized output pressure fluctuations for a supersonic jet at Mach = 1.5 for a forcing frequency of St =0.33. (a) n=1. (b) n=2. (c) n=3. (d) n= 4, where n is defined as the output mode. According to these images, small disturbances near the jet can trigger the propagation of an instability wave downstream which generates noise. (Image courtesy of Jinah Jeun, et. al, 2016)
Turbulence is often considered a random phenomenon. However, aeroacoustic fields originating from turbulent flow structures often produce highly coherent acoustic fields—so much so that researchers hope that an increased understanding of one of these phenomena will create a greater understanding of the other.
New Research Can Help Engineers Understand Aeroacoustics and Turbulence
A research team of the University of Minnesota has recently carried out a study investigating noise generation associated with jets. The goal of the study was to understand “how forcing the velocity equations inside the jet produces sounds in the far-field.”
Simply put, the study aimed to understand the link between fluid turbulence and noise generation for the engineering community. The study involved the analysis of small perturbations for subsonic and supersonic isothermal jets using a modified k-ε model and large eddy simulation (LES) solver over a range of Mach numbers.
Application of input-output analysis was used to understand the sound generation mechanisms associated with turbulent jets and the conversion of near-field aerodynamics fluctuations into far-field acoustic structures.
Traditionally, most aeroacoustic modeling for jets is carried out using high-fidelity models such as LES, as it is difficult to capture noise data experimentally due to extreme operating environments. This study has used both Reynolds-averaged Navier–Stokes (RANS) and LES techniques to investigate the generation of noise.
Noise is generated from fluid disturbances, but in order to capture these disturbances, high-fidelity solvers are needed, which are computationally expensive. The work carried out by the University of Minnesota has shown that lower-fidelity solvers can be used to generate accurate results and can be more advantageous.
One example is to consider where LES predictions of optimal gain were larger than those predicted by RANS simulations. The difference was attributed to small residual errors and the emergence of laminar shear layers from the nozzle within the LES model.
As the RANS model is less computationally expensive than the LES model, engineers can simulate the fluid dynamics to a higher convergence value. Though the residual errors for the LES model were quite small, they still were large in comparison to the RANS simulation which was one cause for the over predicted optimal gain values.
Additionally, the initially laminar shear layers near the nozzle resulted in large gradients being present in this area which resulted in higher fluctuation levels near the nozzle tip in comparison to experimental data. Large fluctuations resulted in increased wave instabilities and in the higher levels of gain recorded.
The most striking aspect of the study was the analysis of recovered acoustic energy as a function of retained output modes. When using an input-output model, retaining 24 output modes or less seemed to satisfy the simulation of the acoustic field. The study noted that this information could then be used to “obtain reduced-order models of noise generation.”
The above study has shed light on a new methodology for acoustic analysis. It has shown that there is a possibility to construct an accurate yet reduced-order model for noise generation analysis, which would be highly beneficial for the industry as research could move away from the constraint of requiring computationally expensive simulations for accurate simulation data.
To learn more about the discovery, read: Input-output analysis of high-speed axisymmetric isothermal jet noise.