Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-16T08:09:14.584Z Has data issue: false hasContentIssue false

Numerical study of effect of wave phase on Reynolds stresses and turbulent kinetic energy in Langmuir turbulence

Published online by Cambridge University Press:  07 October 2020

Anqing Xuan
Affiliation:
Department of Mechanical Engineering and Saint Anthony Falls Laboratory, University of Minnesota, Minneapolis, MN55455, USA
Bing-Qing Deng
Affiliation:
Department of Mechanical Engineering and Saint Anthony Falls Laboratory, University of Minnesota, Minneapolis, MN55455, USA
Lian Shen*
Affiliation:
Department of Mechanical Engineering and Saint Anthony Falls Laboratory, University of Minnesota, Minneapolis, MN55455, USA
*
†Email address for correspondence: shen@umn.edu

Abstract

The dynamics of Reynolds stresses and turbulent kinetic energy (TKE) in Langmuir turbulence are analysed using data of large-eddy simulations with the wave phase resolved. It is found that the streamwise and spanwise Reynolds normal stresses and the Reynolds shear stress vary appreciably with the wave phase, while the vertical normal stress is only weakly dependent on the wave phase. Budget analyses indicate that the production due to wave straining and the effects associated with turbulence pressure are the dominant mechanisms for the wave-phase variation of Reynolds stresses. The accumulative effect of wave–turbulence interactions on TKE is then investigated using the Lagrangian average. It is discovered that the energy transfer from wave to turbulence is contributed by two mechanisms. The first mechanism is the turbulence production by the Lagrangian mean wave shearing and the mean shear stress, which is consistent with the traditional wave-phase-averaged model. The second mechanism, which is not accounted for in previous studies, is the correlation between the wave-phase variation of the Reynolds shear stress and the wave orbital shearing. A model is proposed for the second mechanism. Comparison of the frequency spectrum with Craik–Leibovich simulation results shows that the correlation effect can affect the turbulence fluctuations at time scales around the wave period, indicating the importance of this effect on Reynolds stresses and TKE.

Information

Type
JFM Papers
Copyright
© The Author(s), 2020. Published by Cambridge University Press
Figure 0

Figure 1.Sketch of the configuration of the wave-phase-resolved simulation of Langmuir turbulence. The grey filled arrow denotes the direction of the surface shear stress $\tau _0$. The angled arrow on the surface indicates the wave phase velocity $c$.

Figure 1

Table 1.Computational parameters of Langmuir turbulence simulation cases.

Figure 2

Figure 2.Sketch of (a) the wave Eulerian orbital velocity $\boldsymbol {u}_w$, (b) the effects of the wave Eulerian orbital straining $\boldsymbol {\nabla } \boldsymbol {u}_w$, and (c) the effects of the Lagrangian straining ${\overline {\boldsymbol {\nabla } \boldsymbol {u}_w}}^L$. In (a) the arrows with solid lines illustrate the wave orbital velocity and the dashed line with arrows indicate the direction of the convection of a fluid particle in a wave-following frame. In both (b) and (c) the distortion effect of the wave straining on fluid elements is illustrated using the change from the dashed rectangles to solid ones.

Figure 3

Figure 3.Contours of phase-averaged Reynolds normal stresses (a,d) $\langle {u'^2}\rangle$, (b,e) $\langle {v'^2}\rangle$ and (c,f) $\langle {w'^2}\rangle$ for (a–c) case 1 (${{La}}_t=0.35$) and (d–f) case 3 (${{La}}_t=0.9$), respectively. The Reynolds stresses are normalized by $u_*^2$. The grey lines in (a) represent the fixed depth $kz=-0.1$ (— $\cdot$ —) and the Lagrangian trajectory with a mean depth $kz=-0.1$ (– – –) that are used in figures 5 and 6, respectively.

Figure 4

Figure 4.Contours of phase-averaged Reynolds shear stress $\langle {-u'w'}\rangle$ for (a) case 1 (${{La}}_t=0.35$) and (b) case 3 (${{La}}_t=0.9$). The stress is normalized by $u_*^2$.

Figure 5

Figure 5.Normalized variation of Reynolds normal stress $\langle {u'^2}\rangle$ at the fixed depth $kz=-0.1$ in case 2 (——). The values of $\langle {u'^2}\rangle$ are normalized by its mean value over the wave period at that depth, $\overline {u'^2}={(L_x)}^{-1}\int _{0}^{L_x} \langle {u'^2}\rangle \,\mathrm {d}x$. The experimental data plotted in the figure for comparison are from: Jiang & Street (1991) with an air speed $U_{air}=2.5\ \mathrm {m}\ \mathrm {s}^{-1}$ and an estimated turbulent Langmuir number ${{La}}_t=0.52$ ($\vartriangle$); Jiang & Street (1991) with $U_{air}=4.1\ \mathrm {m}\ \mathrm {s}^{-1}$ and ${{La}}_t=0.66$ ($\square$); Thais & Magnaudet (1996) with $U_{air}=4.5\ \mathrm {m}\ \mathrm {s}^{-1}$ and ${{La}}_t=0.57$ (case E2) (${\Large \circ }$). The wave surface geometry $\eta$ corresponding to the wave phase $kx$ is illustrated using the solid line above the plot. The dotted line with the arrow above the plot indicates that the variation is measured at a fixed depth (the dash–dotted line in figure 3a) and the phase change is in the ${-}x$-direction in the wave-following frame.

Figure 6

Figure 6.Normalized wave-phase variation of Reynolds normal stresses in the wave-following frame along the Lagrangian trajectory with a mean depth $kz=-0.1$ for case 1 (——), case 2 (– – –), case 3 (— $\cdot$ —) and case 1S (— $\cdot$$\cdot$ —): (a) ${(u'^2)}^l/{\overline {u'^2}}^L$, (b) ${(v'^2)}^l/{\overline {v'^2}}^L$ and (c) ${(w'^2)}^l/{\overline {w'^2}}^L$. For comparison, the theoretical results (3.1) based on the RDT analysis by Teixeira & Belcher (2002) for $ak=0.084$ ($\vartriangle$) and $ak=0.15$ (${\Large \circ }$) are plotted. The wave surface geometry $\eta$ corresponding to the wave phase $kx$ is illustrated using the solid line above the plot. The dotted line with the arrows above the plot indicates that the variation is measured following the Lagrangian trajectory (the dashed line in figure 3a), which points to the ${-}x$-direction in the wave-following frame.

Figure 7

Figure 7.Normalized wave-phase variation of Reynolds shear stress in the wave-following frame for case 1 (——), case 2 (– – –), case 3 (— $\cdot$ —) and case 1S (— $\cdot$$\cdot$ —): (a) ${(u'w')}^l/{\overline {u'w'}}^L$ along the Lagrangian trajectory with a mean depth $kz=-0.1$ and (b) ${(u'w')}^l/{\overline {u'^2}}^L$ at the surface. In (b) the approximation (3.4) is plotted for $ak=0.084$ ($\vartriangle$) and $ak=0.15$ (${\Large \circ }$).

Figure 8

Figure 8.Contours of the dominant budget terms of $\langle {u'^2}\rangle$, including the production due to the wave $P^w_{xx}$, (a) $-2\langle {u'^2}\rangle {\partial } u_w/{\partial } x$ and (b) $-2\langle {u'w'}\rangle {\partial } u_w/{\partial } z$, (c) the pressure transport $T^p_{xx}$ and (d) the pressure–strain correlation $R_{xx}$. The results are normalized by $u_*^2 ak\sigma$.

Figure 9

Figure 9.Contours of the dominant budget term of $\langle {v'^2}\rangle$, pressure–strain correlation $R_{yy}$. The results are normalized by $u_*^2 ak\sigma$.

Figure 10

Figure 10.Contours of the dominant budget terms of $\langle {w'^2}\rangle$, including the production due to the wave $P^w_{zz}$, (a) $-2\langle {w'^2}\rangle {\partial } w_w/{\partial } z$ and (b) $-2\langle {u'w'}\rangle {\partial } w_w/{\partial } x$, (c) pressure transport $T^p_{zz}$ and (d) pressure–strain correlation $R_{zz}$. The results are normalized by $u_*^2 ak\sigma$.

Figure 11

Figure 11.Sketch of the dominant processes contributing to the wave-phase variation of the Reynolds normal stresses. The arrows indicate the direction of the energy flux. The solid arrows represent the wave normal production. The arrows with dash–dotted lines represent the pressure transport. The hollow arrows with dash–dot–dot lines indicate the pressure–strain correlation.

Figure 12

Figure 12.Contours of the dominant terms in the evolution equation of $-\langle {u'w'}\rangle$: (a) $\langle {u'^2}\rangle {\partial } w_w/{\partial } x$, (b) $\langle {w'^2}\rangle {\partial } u_w/{\partial } z$, (c) $\langle {u' {\partial } p'/{\partial } z}\rangle$, (d) $\langle {w' {\partial } p'/{\partial } x}\rangle$ and (e) $\langle {w'^2}\rangle {\partial } u_w/{\partial } z+ \langle {w' {\partial } p'/{\partial } x}\rangle$.

Figure 13

Figure 13.Profiles of the budget terms of Lagrangian-averaged TKE (4.1) for (a) case 1 and (b) case 3: production due to the wave straining ${\overline {P^w}}^L$ (——), production due to the mean current ${\overline {P^c}}^L$ (– – –), diffusion due to the resolved turbulence fluctuations, SGS turbulence and viscous effects ${\overline {T^t}}^L$ (— $\cdot$ —), pressure transport ${\overline {T^p}}^L$ (— $\cdot$$\cdot$ —) and dissipation ${\bar {\epsilon }}^L$ (${{\cdotp \cdotp \cdotp \cdotp \cdotp \cdotp }}$). The results are normalized by $ku_*^3$.

Figure 14

Figure 14.Lagrangian decomposition of TKE production for (a) case 1, (b) case 2, (c) case 3 and (d) case 1S: mean effect ${(P^w)}^{LL}$ (——) and correlation effect ${(P^w)}^{ll}$ (– – –). The results are normalized by $u_*^2 {(ak)}^2 \sigma$.

Figure 15

Table 2.Percentages of TKE production due to the mean effect $\mathcal {P}^{LL}$ (4.15) and the correlation effect $\mathcal {P}^{ll}$ (4.16) in the total production $\mathcal {P}$. The fourth and fifth columns are calculated using the proposed estimation model for the mean effect $\mathcal {P}^{LL}_{estimate}$ (4.20) and correlation effect $\mathcal {P}^{ll}_{estimate}$ (4.28), respectively.

Figure 16

Figure 15.Production of TKE due to the mean effect (4.17a,b) for (a) case 1, (b) case 2, (c) case 3 and (d) case 1S: $-{\overline {u'w'}}^L{\overline {{\partial } u_w/{\partial } z}}^L$ (– – –), $-{\overline {u'w'}}^L{\overline {{\partial } w_w/{\partial } x}}^L$ (${{\cdotp \cdotp \cdotp \cdotp \cdotp \cdotp }}$), and the model estimation (4.19) (${\Large \circ }$). The results are normalized by $u_*^2 {(ak)}^2 \sigma$.

Figure 17

Figure 16.Production of TKE due to the correlation effect (4.18a,b) for (a) case 1, (b) case 2, (c) case 3 and (d) case 1S: ${\overline {-{(u'w')}^l{({\partial } u_w/{\partial } z)}^l}}^L$ (– – –), ${\overline {-{(u'w')}^l{({\partial } w_w/{\partial } x)}^l}}^L$ (${{\cdotp \cdotp \cdotp \cdotp \cdotp \cdotp }}$), and the model estimation (4.27) (${\Large \circ }$). The results are normalized by $u_*^2 {(ak)}^2 \sigma$).

Figure 18

Figure 17.Profiles of (a) the in-phase amplitude of the wave-phase variation of the Reynolds shear stress, ${(u'w')}_{in}$ (4.22), and (b) the scaled in-phase variation, $|{(u'w')}_{in}|/(ak {{La}}_t^{-n} {\overline {u'w'}}^L)$ for case 1 (——), case 2 (– – –), case 3 (— $\cdot$ —) and case 1S (— $\cdot$$\cdot$ —). In (b) the estimation (4.25) (${\Large \circ }$) is plotted for comparison.

Figure 19

Figure 18.Comparisons of (a) ${\overline {u'^2}}^L/u_*^2$, (b) ${\overline {v'^2}}^L/u_*^2$, (c) ${\overline {w'^2}}^L/u_*^2$, (d) ${\bar {E}}^L/u_*^2$, (e) $-{\overline {u'w'}}^L/u_*^2$ and (f) $u_c/u_*$ between the case L1 of the wave-phase-resolved simulation (——) and the CL simulation (– – –).

Figure 20

Figure 19.Comparisons of (a,b) the premultiplied frequency spectrum of the TKE and (c,d) the premultiplied frequency cross-spectrum of $u'$ and $w'$ between the case L1 of the wave-phase-resolved simulation (——) and the CL simulation (— $\cdot$ —). The spectrum is computed along the Lagrangian trajectories with the mean depth (a) $kz=-0.2$ and (b) $kz=-0.8$.