Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-16T11:19:25.016Z Has data issue: false hasContentIssue false

Triglobal resolvent analysis of swept-wing wakes

Published online by Cambridge University Press:  09 January 2023

J.H. Marques Ribeiro*
Affiliation:
Department of Mechanical and Aerospace Engineering, University of California, Los Angeles, CA 90095, USA
Chi-An Yeh
Affiliation:
Department of Mechanical and Aerospace Engineering, University of California, Los Angeles, CA 90095, USADepartment of Mechanical and Aerospace Engineering, North Carolina State University, Raleigh, NC 27695, USA
Kunihiko Taira
Affiliation:
Department of Mechanical and Aerospace Engineering, University of California, Los Angeles, CA 90095, USA
*
†Email address for correspondence: jeanmarques@g.ucla.edu

Abstract

Through triglobal resolvent analysis, we reveal the effects of wing tip and sweep angle on laminar separated wakes over swept wings. For the present study, we consider wings with semi-aspect ratios from $1$ to $4$, sweep angles from $0^\circ$ to $45^\circ$ and angles of attack of $20^\circ$ and $30^\circ$ at a chord-based Reynolds number of $400$ and a Mach number of $0.1$. Using direct numerical simulations, we observe that unswept wings develop vortex shedding near the wing root with a quasi-steady tip vortex. For swept wings, vortex shedding is seen near the wing tip for low sweep angles, while the wakes are steady for wings with high sweep angles. To gain further insights into the mechanisms of flow unsteadiness, triglobal resolvent analysis is used to identify the optimal spatial input–output mode pairs and the associated gains over a range of frequencies. The three-dimensional forcing and response modes reveal that harmonic fluctuations are directed towards the root for unswept wings and towards the wing tip for swept wings. The overlapping region of the forcing–response mode pairs uncovers triglobal resolvent wavemakers associated with self-sustained unsteady wakes of swept wings. Furthermore, we show that for low-aspect-ratio wings optimal perturbations develop globally over the entire wingspan. The present study uncovers physical insights on the effects of tip and sweep on the growth of optimal harmonic perturbations and the wake dynamics of separated flows over swept wings.

Information

Type
JFM Papers
Copyright
© The Author(s), 2023. Published by Cambridge University Press
Figure 0

Figure 1.Set-up for finite swept wing simulation. In the grey boxes, the instantaneous flow field for $\alpha = 20^\circ$, $\varLambda = 15^\circ$ and $sAR = b/c = 4$, with $Q = 2$ isosurfaces coloured by instantaneous $u_x$. Mesh coloured by time-averaged $\bar {u}_x$. In the blue box, isosurfaces of the primary response mode with mesh in light grey.

Figure 1

Table 1.Time-averaged lift and drag coefficients ($\overline {C_L}$ and $\overline {C_D}$) compared with those of Zhang et al. (2020a) for laminar separated flow over NACA 0015 wings with $sAR = 4$, $\alpha = 20^\circ$ and $\varLambda = 0^\circ$, $15^\circ$, $30^\circ$ and $45^\circ$.

Figure 2

Figure 2.Instantaneous isosurfaces of $Q = 2$ coloured by $u_x$ for (a) $\alpha = 20^\circ$ and (b) $\alpha = 30^\circ$. Unsteady shedding near wing root ($\boldsymbol {\diamond }$, red), unsteady shedding near wing tip ($\boldsymbol {\square }$, yellow), steady flow with root structures ($\boldsymbol {\blacktriangleright }$, green), steady flow with streamwise vortices ($\boldsymbol {\blacktriangledown }$, blue).

Figure 3

Figure 3.(a) Resolvent gains and (b) forcing–response mode pairs for $(sAR,\alpha,\varLambda ) = (4,20^\circ,0^\circ )$. For each mode, forcing ($\hat {\boldsymbol {f}}$) is the top half while response ($\hat {\boldsymbol {q}}$) it the bottom half with isosurfaces of velocity $\hat {u}_y \in [-0.2,0.2]$, with free stream directed to the right. Bottom panel of (a) shows power spectrum density of lift coefficient $\hat {C}_L$.

Figure 4

Figure 4.Resolvent gain distribution for the top three mode pairs and forcing–response mode pairs for selected frequencies for $(sAR,\alpha ) = (4,20^\circ )$ and $\varLambda =$ (a) $15^\circ$, (b) $30^\circ$ and (c) $45^\circ$. Primary forcing ($\hat {f}$) and response ($\hat {q}$) modes shown with isosurfaces of velocity $\hat {u}_y \in [-0.2,0.2]$.

Figure 5

Figure 5.Wingspan location of primary resolvent wavemakers with isocontours of $\xi$ for $0.05 \le St \le 0.25$ for $(sAR,\alpha ) = (4,20^\circ )$ wings with (a–d) $0^\circ \le \varLambda \le 45^\circ$. Resolvent wavemaker modes at $St = 0.14$ are shown with isosurfaces of $\hat {u}_y / \| \hat {u}_y \|_\infty = \pm 0.1$ and instantaneous $Q=1$ are grey-coloured isosurfaces.

Figure 6

Figure 6.Wingspan locations of dominant forcing (red) and response (blue) with contours of $\boldsymbol {\varOmega }_{\hat {\boldsymbol {f}}}$ and $\boldsymbol {\varOmega }_{\hat {\boldsymbol {q}}} \in [0.4,1.0]$ for $0.05 \le St \le 0.35$ for $sAR = 4$ wings with $\alpha = 30^\circ$ with (a–d) $0^\circ \le \varLambda \le 45^\circ$. Dot-dashed lines are polynomial fit of maximum $z/c$ of forcing and response at each $St$. Arrows show direction of optimal forcing to response at $St = 0.15$.

Figure 7

Figure 7.(a–d) Wingspan locations of dominant forcing (red) and response (blue) with contours of $\boldsymbol {\varOmega }_{\hat {\boldsymbol {f}}}$ and $\boldsymbol {\varOmega }_{\hat {\boldsymbol {q}}} \in [0.4,1.0]$ for $0.05 \le St \le 0.35$ for wings at $\alpha = 30^\circ$, $sAR = 2$ and $1$ and $\varLambda = 15^\circ$ and $30^\circ$.

Figure 8

Figure 8.Resolvent gain distribution and forcing–response mode pairs over frequency for $(\alpha,\varLambda ) = (30^\circ,15^\circ )$ and (a–c) $1 \le sAR \le 4$. Forcing ($\hat {f}$) and response ($\hat {q}$) modes shown with isosurfaces of velocity $\hat {u}_y \in [-0.2,0.2]$. Mode switching is absent for $sAR = 1$ due to merging of root and wing tip perturbations on the wake.

Figure 9

Figure 9.Eigenvalues of $\boldsymbol{\mathsf{L}}_{\bar {\boldsymbol {q}}}$ for $sAR= 4$ wings with $\alpha =$ (a) $20^\circ$ and (b) $30^\circ$, and $0 \le \varLambda \le 45^\circ$. Here $St_r$ and $St_i$ represent the growth rate and temporal frequency, respectively. Green-dashed line shows $St_r = 0$. Cyan-dashed line shows where discounting parameter is set.

Figure 10

Table 2.Convergence of the randomized resolvent analysis using $k = 5$, $10$ and $20$ test vectors shown for the five leading singular values with $(sAR,\alpha,\varLambda ) = (2,30^\circ,15^\circ )$ at $St = 0.14$.