Hostname: page-component-7b7567c8c-z4zkx Total loading time: 0 Render date: 2026-11-16T11:29:07.063Z Has data issue: false hasContentIssue false

Direct numerical simulations of the interaction of temporally evolving circular jets

Published online by Cambridge University Press:  28 April 2025

Tomoaki Watanabe*
Affiliation:
Department of Mechanical Engineering and Science, Kyoto University, Kyoto 615-8540, Japan
Tatsuya Inagaki
Affiliation:
Department of Aerospace Engineering, Nagoya University, Nagoya 464-8603, Japan
Takahiro Mori
Affiliation:
Department of Aerospace Engineering, Nagoya University, Nagoya 464-8603, Japan
Kirari Ishizawa
Affiliation:
Undergraduate Department of Mechanical and Aerospace Engineering, Nagoya University, Nagoya 464-8603, Japan
Koji Nagata
Affiliation:
Department of Mechanical Engineering and Science, Kyoto University, Kyoto 615-8540, Japan
*
Corresponding author: Tomoaki Watanabe, watanabe.tomoaki.8x@kyoto-u.ac.jp

Abstract

Direct numerical simulations are performed to study turbulence generated by the interaction of multiple temporally evolving circular jets with jet Mach numbers $M_J=0.6$ and $1.6$, and a jet Reynolds number of 3000. The jet interaction produces decaying, nearly homogeneous isotropic turbulence, where the root-mean-squared (r.m.s.) fluctuation ratio between the streamwise and transverse velocities is approximately 1.1, consistent with values observed in grid turbulence. In the supersonic case, shock waves are generated and propagate for a long time, even after the turbulent Mach number decreases. A comparison between the two Mach number cases reveals compressibility effects, such as reductions in the velocity derivative skewness magnitude and the non-dimensional energy dissipation rate. For the r.m.s. velocity fluctuations, $u_{rms}$, and the integral scale of the streamwise velocity, $L_u$, the Batchelor turbulence invariant, $u_{rms}^2 L_u^5$, becomes nearly constant after the turbulence has decayed for a certain time. In contrast, the Saffman turbulence invariant, $u_{rms}^2 L_u^3$, continuously decreases. Furthermore, temporal variations of $u_{rms}^2$ and $L_u$ follow power laws, with exponents closely matching the theoretical values for Batchelor turbulence. The three-dimensional energy spectrum $E(k)$, where $k$ is the wavenumber, exhibits $E(k) \sim k^4$ for small wavenumbers. This behaviour is consistently observed for both Mach number cases, indicating that the modulation of small-scale turbulence by compressibility effects does not affect the decay characteristics of large scales. These results demonstrate that jet interaction generates Batchelor turbulence, providing a new direction for experimental investigations into Batchelor turbulence using jet arrays.

Information

Type
JFM Papers
Copyright
© The Author(s), 2025. Published by Cambridge University Press
Figure 0

Figure 1.A schematic of the initial field for the DNS of temporally evolving jet interactions. The initial jet regions are depicted in grey. The number of jets shown is illustrative and does not match the actual value of 40. The figure also displays the coordinate system $(Y, Z)$, which is used for statistical evaluations.

Figure 1

Figure 2.Local Mach number $M=\sqrt {u^2+v^2+w^2}/a$ on a $y$–$z$ plane for M16 at ($a$) $t/T_J=0$, ($b$) $t/T_J=8$ and ($c$) $t/T_J=13$, where $a$ is the local speed of sound.

Figure 2

Figure 3.Temporal variations of ($a$) mean streamwise velocity $\langle u\rangle$ and ($b$) r.m.s. fluctuations of streamwise velocity $u_{rms}$ at $(Y/S,Z/S)=(0,0)$, $(0.5,0)$ and $(0.5,0.5)$ for M06 and M16.

Figure 3

Figure 4.Temporal variations of the ratio between streamwise and lateral r.m.s. velocity fluctuations $u_{rms}/v_{rms}$ at $(Y,Z)=(0,0)$.

Figure 4

Figure 5.Temporal variations of ($a$) the turbulent Reynolds number $Re_\lambda$ and ($b$) turbulent Mach number $M_T$ at $(Y,Z)=(0,0)$.

Figure 5

Figure 6.($a$) Temporal variations of the r.m.s. fluctuations of density ($ \rho _{rms}$) and pressure ($ P_{rms}$), normalised by the mean density and mean pressure, respectively. ($b$) Dependence of the normalised r.m.s. pressure fluctuation on the turbulent Mach number ($ M_T$). All results are evaluated at $ (Y, Z) = (0, 0)$.

Figure 6

Figure 7.($a$) Distribution of dilatation $\Theta =\partial u_i /\partial x_i$ on a $y$–$z$ plane at $t/T_J=80$ for M16. ($b$) Probability density function of $\Theta$ normalised by its r.m.s. value $\Theta _{rms}$. The black line indicates the Gaussian distribution.

Figure 7

Figure 8.($a$) Dependence of the velocity derivative skewness, $ S_{\partial u / \partial x}$, on the turbulent Reynolds number, $ Re_\lambda$. The DNS results are taken at $(Y, Z) = (0, 0)$ after $ Re_\lambda$ reaches its peak. ($b$) Dependence of the velocity derivative flatness, $ F_{\partial u / \partial x}$, on $ Re_\lambda$. The DNS results are taken at $(Y, Z) = (0, 0)$ after $ Re_\lambda$ reaches its peak. The figure includes data from studies of incompressible turbulence (Kuo & Corrsin 1971; Antonia & Chambers 1980; Van Atta & Antonia 1980; Kerr 1985; Mydlarski & Warhaft 1996; Sreenivasan & Antonia 1997; Mi et al.2013; Watanabe & Nagata 2018; Watanabe et al.2019; Zheng et al.2021).

Figure 8

Figure 9.Temporal variations of $u_{rms}^2L_u^3$ and $u_{rms}^2L_u^5$ in ($a$) M06 and ($b$) M16.

Figure 9

Figure 10.($a$) Dependence of the non-dimensional energy dissipation rate, $ C_\varepsilon$, on the turbulent Reynolds number, $ Re_\lambda$. ($b$) Temporal variations of $ C_\varepsilon$. The data are taken for the period following the peak of $ Re_\lambda$.

Figure 10

Figure 11.Temporal variations of $ u_{rms}^2$, $ L_u$ and $ C_\varepsilon$ over time: ($a$) M06; ($b$) M16. Thin black solid lines indicate power laws with exponents determined using the least squares method. Dashed and dash-dotted line lines represent the power laws for $ u_{rms}^2$ and $ L_u$ with the exponents for Batchelor and Saffman turbulence.

Figure 11

Table 1.Power law exponents for $ u_{rms}^2$, $ L_u$ and $ C_\varepsilon$, denoted as $ n$, $ n_L$ and $ n_C$, respectively, and the virtual origin $ t_0$. The theoretical values of Batchelor and Saffman turbulence for $ n$ and $ n_L$ are denoted by superscripts $ (B)$ and $ (S)$, respectively. The relative differences between the DNS results and the theoretical predictions are calculated as $\Delta n^{(\alpha )} = (n - n^{(\alpha )})/n^{(\alpha )}$ and $ \Delta n_L^{(\alpha )} = (n_L - n_L^{(\alpha )})/n_L^{(\alpha )}$, where $ \alpha = B$ for Batchelor turbulence and $ \alpha = S$ for Saffman turbulence.

Figure 12

Figure 12.Dependence of the decay exponent $ n$ of $ u_{rms}^2$ on the fitting range $ t_S \leqslant t \leqslant t_E$: ($a$) $ t_S$ dependence with $ t_E = 5000T_J$ fixed in M06; ($b$) $ t_E$ dependence with $ t_S = 2000T_J$ fixed in M06; ($c$) $ t_S$ dependence with $ t_E = 5000T_J$ fixed in M16; ($d$) $ t_E$ dependence with $ t_S = 800T_J$ fixed in M16.

Figure 13

Figure 13.Temporal evolutions of the three-dimensional energy spectrum, $ E(k)$, for ($a$) M06 and ($b$) M16. The spectra are premultiplied by $k^{-4}$ or $ k^{-2}$ to facilitate comparison with $ E(k) \sim k^4$ of Batchelor turbulence and $ E(k) \sim k^2$ of Saffman turbulence.

Figure 14

Figure 14.Streamwise variations of velocity variance $v_{rms}^2$, integral scale $L_v$ and non-dimensional dissipation rate $C_\varepsilon$. The error bars indicate possible statistical errors estimated with reduced samples. The velocity variance is normalised by the uniform mean velocity in the decay region, $U_0$.