Hostname: page-component-7b7567c8c-7pwfn Total loading time: 0 Render date: 2026-11-16T08:56:50.229Z Has data issue: false hasContentIssue false

Investigation of supersonic twin-jet coupling using spatial linear stability analysis

Published online by Cambridge University Press:  14 May 2021

Petrônio A.S. Nogueira*
Affiliation:
Department of Mechanical and Aerospace Engineering, Laboratory for Turbulence Research in Aerospace and Combustion, Monash University, Clayton 3800, Australia
Daniel M. Edgington-Mitchell
Affiliation:
Department of Mechanical and Aerospace Engineering, Laboratory for Turbulence Research in Aerospace and Combustion, Monash University, Clayton 3800, Australia
*
†Email address for correspondence: petronio.augusto@gmail.com

Abstract

The present work focuses on the study of the resonance and coupling of an underexpanded circular twin-jet system operating at a nozzle pressure ratio of $5.0$. Particle image velocimetry data from previous work were revisited, and a symmetry-imposed proper orthogonal decomposition (POD) was performed. It is shown that the system is dominated by a single POD mode pair symmetric about the internozzle plane, and the resonance loop is modulated by a third POD mode related to shear thickness modulation. A spatial Fourier transform of the leading POD mode pair leads to the identification of the peak wavenumbers and radial shapes of the different waves at play in the screech phenomenon. Locally parallel linear stability analysis around the experimental mean flow is also performed, in order to provide clarification of the mode ‘locking’ mechanism, i.e. the selection of the global mode associated with screech. It is shown that the characteristics of the Kelvin–Helmholtz wavepackets alone are not sufficient to explain the coupling observed in the experimental data. A consideration of the upstream-travelling guided jet mode offers an explanation; only specific symmetries of upstream modes can be supported in the frequency range at which resonance occurs. Results from stability analysis point to structures at frequencies and wavenumbers close to those found experimentally, and their spatial structures show excellent agreement with the POD modes. The present results suggest that the resonance loop is closed by an upstream-travelling guided jet mode for the twin-jet system at high nozzle pressure ratio.

Information

Type
JFM Papers
Copyright
© The Author(s), 2021. Published by Cambridge University Press
Figure 0

Figure 1.Sketch of the coordinate system used in the present analysis. Polar coordinates $(r,\theta )$ are associated with the centre of the twin-jet system, and $(r_j,\theta _j)$ are associated with the centre of a single jet. Both cross-plane (a) and full view (b) are shown.

Figure 1

Figure 2.Power spectral density from the acoustic data of the $\textrm {NPR}=5.0$ screeching jet. The screech tone is clearly identified around $St=0.19$.

Figure 2

Table 1.Symmetry of the different modes in the twin-jet system.

Figure 3

Figure 3.Streamwise (a) and lateral (b) mean flows from PIV experiments. The magnitudes of both velocity components at the system centreline ($y=0$) are also shown in (c) and the mean streamwise velocities at several axial stations analysed in this work are shown in (d). All fields are normalised by the ambient velocity of sound.

Figure 4

Figure 4.Sample mean flow interpolated into the computational mesh for stability analysis (a) and its respective radial $\partial U / \partial r$ (b) and azimuthal $\partial U / \partial \theta$ (c) derivatives ($x/D=4$). Modes are normalised by the ambient speed of sound.

Figure 5

Figure 5.The POD spectra for both SS (a) and SA (b) modes. Blue symbols (and left-hand $y$ axis) depict the fraction of the total energy held by each mode and orange lines (right-hand $y$ axis) show the cumulative energy as a function of the number of modes. All percentages are related to the total energy associated with each symmetry (SS or SA). The total energy of SS modes is $10.4\,\%$ greater than the total energy of SA modes.

Figure 6

Figure 6.Three dominant POD modes for the SS field. Both streamwise (a,c,e) and lateral (b,d,f) velocities are shown. Modes 1 and 2 (a–d) seem to form a pair, and mode 3 (e,f) has no associated pair. All fields are normalised by their maximum. (a) POD-SS mode 1 (u), (b) POD-SS mode 1 (v), (c) POD-SS mode 2 (u), (d) POD-SS mode 2 (v), (e) POD-SS mode 3 (u) and (f) POD-SS mode 3 (v).

Figure 7

Figure 7.Three first POD SS coefficients forming an ellipsoid (a). Amplitude of the $a_1$–$a_2$ cycle ($r_{12}=\sqrt {a_1^2+a_2^2}$) as a function of $a_3$ (b). Phase portrait of modes $a_1$ and $a_2$ for $|a_3-0.02|<0.01$ (c).

Figure 8

Figure 8.The POD SS wavenumber spectra for both streamwise (a) and lateral (b) velocity components. Blue line depicts the wavenumber energised by the interaction between the KH mode (peak of both spectra) and the shock-cell structure.

Figure 9

Figure 9.Real part of the pressure field of KH modes related to azimuthal wavenumber $m=1$, $St=0.19$ and $x/D=4$. The eigenmodes in the cross-plane can be directly associated with the different symmetries allowed in the problem. Modes (a) 1SS ($\mu =0$), (b) 1AA ($\mu =0$), (c) 1SA ($\mu =1$) and (d) 1AS ($\mu =1$).

Figure 10

Figure 10.Growth rates of KH modes related to $m=1$ for $x/D=4$ as a function of Strouhal number (a) and their respective streamwise wavenumbers (b). (c,d) Same quantities as a function of streamwise station for $St=0.19$. (a,c) Growth rates and (b,d) Wavenumbers.

Figure 11

Figure 11.Real part of the pressure field of upstream modes for $x/D=4$ and $St=0.19$ (1SS, 1AS) and $0.20$ (1AA, 1SA). The stable upstream-travelling mode is shown for the 1AA case. Modes (a) 1SS ($\mu =0$, $St=0.19$), (b) 1AA ($\mu =0$, $St=0.20$), (c) 1SA ($\mu =1$, $St=0.20$) and (d) 1AS ($\mu =1$, $St=0.19$).

Figure 12

Figure 12.Wavenumber of the upstream-travelling modes related to $m=1$ for $x/D=3,4,5$ and several Strouhal numbers. The dashed line indicates the speed of sound, which is the point where the guided jet mode becomes propagative. (a) $\mu =0$; $x/D=3$, (b) $\mu =0$; $x/D=4$, (c) $\mu =0$; $x/D=5$, (d) $\mu =1$; $x/D=3$, (e) $\mu =1$; $x/D=4$ and (f) $\mu =1$; $x/D=5$.

Figure 13

Table 2.Peak wavenumbers of the different waves extracted from the POD wavenumber spectrum compared with the wavenumbers of the waves from LSA for $x/D=5$ and $St=0.2075$.

Figure 14

Figure 13.Comparison between mode 1SS from LSA (both (a,b) KH and (c,d) upstream waves) and the dominant POD mode (extracted from the energy peaks at positive and negative wavenumbers of the spectra shown in figure 8). The LSA was computed using $x/D=5$ and $St=0.2075$. All modes are normalised by their maximum amplitude.