betaFeedback

WorksPhysics and Astronomy

Energy Transfer Mechanisms in Wake-Modulated Transonic Flutter

Abstract

Transonic flutter is a detrimental aeroelastic instability that can generate large-amplitude structural oscillations, leading to severe vibration, fatigue damage, reduced operational limits, and potentially catastrophic structural failure. Incoming wake disturbances can further amplify this instability, making it critical to identify the underlying aerodynamic mechanisms responsible for predicting and controlling flutter onset. The underlying flow physics is complex with nonlinear interactions between the wake and the wing, shock motion, shock-induced flow separation, vortex shedding and the wing motion. In this study, we perform high-fidelity direct numerical simulations of a sinusoidally pitching NACA0012 airfoil with an underwing cylinder at various transonic Mach numbers and a Reynolds number of 10,000. Through energy maps, we identify that the addition of the cylinder significantly expands flutter boundaries compared to an airfoil-only system. We extend the force partitioning method to partition the power transferred between the flow and the airfoil for compressible flows. Application of this approach to distinct regions of the flow domain indicates that the gap flow between the wing and the cylinder is the dominant contributor to the energy transfer from flow to the wing. The blockage effects due to the cylinder cause flow acceleration on the wing which further enhances the tendency for flutter. We investigate cylinder placement relative to the airfoil to reveal that flutter is enhanced only when the cylinder is placed upstream of the pivot point on the airfoil. The current study highlights how such partitioning methods can parse force and energy transfer mechanisms in complex, unsteady high-speed flows.

Introduction

Transonic flutter is a challenging aeroelastic problem due to the nonlinear interaction between the structural dynamics and the compressibility effects in transonic regimes. Several studies have identified a transonic dip phenomenon wherein the flutter onset speed is reduced in the transonic flow regime [1, 2, 3, 4, 5] thereby limiting the operational regime of commercial and military aircraft. The flutter instability causes growth of oscillations but often saturates into detrimental limit-cycle oscillations due to the nonlinearities rather than diverging monotonically [2, 6, 4]. These limit cycle oscillations can cause structural damage of the aircraft through fatigue loading and can also cause pilot discomfort in high-performance aircraft. The dominant nonlinear behaviour is attributed to the coupling between unsteady shocks and separated flow, including λ\lambda-shock formation, shock-induced stall and viscous effects in the boundary layer [7, 8, 9, 10, 11, 12]. Hence, determining flutter boundaries and understanding the relevant flow mechanisms is essential for expanding the operational regimes of aircraft.

While transonic flutter already involves highly complex flow physics in nominally undisturbed freestream conditions, practical flight environments often include incident flows modified by upstream wakes. Such wake–wing interactions are particularly relevant in formation flight, where the aerodynamic environment experienced by the follower aircraft depends strongly on its position within the wake of the upstream wing [13]. In transport aircraft, wakes generated by engine nacelles have also been shown to alter shock–boundary-layer interactions on the lower wing surface, leading to buffet-like low-frequency shock oscillations and associated unsteadiness [14, 15]. Furthermore, wakes shed from upstream wings undergoing buffet instability can induce significant unsteady aerodynamic loading on downstream tail-plane surfaces [16, 17, 18].

The effects of wing-wake interactions on flutter instability are extensively observed in high performance aircraft mounted with external underwing payloads such as fuel tank, engines or stores. There have been several studies to understand the flutter instability mechanism through flight tests [19, 20] and computational studies [21, 22, 23, 24]. These studies have shown that underwing attachments can also increase susceptibility to flutter and can lower the flutter onset speed. Changes in configuration of these external payloads can substantially modify the aeroelastic response, the onset flutter speed, amplitude of limit cycle oscillations.

The effects of external wakes on wing flutter can be attributed to several mechanisms such as the structural properties of the wing-store configuration, mass distribution, attachment between the store and the wing, transfer of store aerodynamic loads onto the wings and the aerodynamic interaction between the underwing payloads and the wing. The responsible structural mechanisms have been analyzed using complex structural models coupled with inviscid or parametrized aerodynamic models [25, 26, 27]. However, the aerodynamic interaction mechanisms by which stores modify transonic flutter are yet to be understood. Flight tests have revealed that the addition of the underwing attachments modify the pressure distribution on the lower surface of the wing and can change the local shock pattern in the transonic regime [28, 29]. Panel-based methods coupled with linearized transonic aerodynamic model suggest that shock interactions associated with the store can strongly influence the response [28, 30]. However, viscous effects can significantly modify shock location, shock strength and the resulting flutter boundary [9, 10].

Previous studies have incorporated viscous effects through Reynolds averaged Navier–Stokes equations (RANS) and unsteady RANS [31, 23, 24, 32] and shown that viscous effects can substantially modify transonic flutter boundaries relative to Euler-based predictions. However, results from such models are sensitive to the choice of the turbulence model. Further, the fidelity of RANS-based approaches remains limited for isolating the details of the nonlinear aerodynamic interaction mechanisms such as shock-induced separation, vortex shedding. Beran et al. [2004b], Parker et al. [2007] conducted Navier–Stokes studies of Goland-type wing configurations. They indicated that the aerodynamic loads of underwing attachments can increase LCO amplitude, while aerodynamic interference beneath the wing can reduce the LCO amplitude. All these studies highlight the sensitivity of aerodynamics to the fidelity of the underlying fluid dynamics solver.

For such coupled wing-wake problems, high-fidelity simulations are necessary to fully resolve the vortex-dominated perturbations, wake interactions, gap-flow effects, shock motion and shock-induced separation that may arise between the store and wing in transonic flow. However, a full aircraft configuration with underwing store is extremely complex and challenging. We consider a simplified wake-airfoil system to bridge the gap between a full aircraft configuration and a canonical aeroelastic model in order to obtain a fundamental understanding of the transonic aerodynamics. In transonic flows, rod-airfoil and tandem bluff-body/airfoil configurations have served as canonical models for upstream-wake interaction with a downstream lifting surface [34, 35, 36, 37]. While many of these studies were motivated by aeroacoustics or buffet, they establish a computationally tractable configuration in which the fundamental wake-wing interaction can be examined. Motivated by these, we consider a circular cylinder underneath a NACA0012 airfoil in transonic flow conditions as a canonical configuration to enable a fundamental analysis of wake-induced effects on flutter.

Previously, Turner et al. [2024a] used prescribed-motion simulations and energy maps to identify the flutter boundary for a sinusoidally pitching NACA0012 airfoil in transonic flow at a chord Reynolds number of 10,000. They showed that a subcritical instability occurs near M∞≈0.7M_{\infty} \approx0.7 and identified shock-induced stall associated with λ\lambda-shock formation as the primary flutter mechanism. The present study extends that framework by introducing a circular cylinder beneath the airfoil as a prototypical model of a wake-generating underwing store. The cylinder–airfoil system is intended to capture, in a minimal setting, the dominant aspects of underwing-attachment interference, namely blockage, wake impingement on the wing, modification of the lower-surface shear layer, and the formation of locally supersonic gap flow with multiple compression/shock structures (see Figure 1). We prescribe sinusoidal pitching motion for the cylinder-airfoil system and use energy maps [38, 39] to characterize the flutter boundaries and compare them with those of an airfoil-only configuration. Since the dynamics is observed to be very rich with the presence of shock wave motion, flow separation, vortex shedding and gap flow, we need a quantitative framework to discern the effect of cylinder on wing aerodynamics.

Various flow mechanisms observed in a sinusoidally pitching cylinder-airfoil system in transonic Mach numbers

Figure 1. Various flow mechanisms observed in a sinusoidally pitching cylinder-airfoil system in transonic Mach numbers

Force partitioning methods have been widely used to identify the flow structures responsible for aerodynamic forces and moments in unsteady incompressible aerodynamic and aeroelastic systems [40 Menon and Mittal,] 2021b,a, Prakhar et al., 2025]. These methods originate from the work of Quartapelle and Napolitano [ [43] and are closely related to the force-element analysis of Chang [ [44]. The underlying framework employs influence potentials to transform surface integrals of pressure forces and moments into distributed volumetric and surface contributions that can be directly associated with specific flow structures and mechanisms. Extensions of these approaches to compressible flows have been relatively limited. Chang and Lei [ [45], Chang et al. [ [46] extended the formulation to steady compressible flow by introducing a correction term to the Lamb vector and demonstrated the method on RANS simulations of compressible flow over a delta wing. Related Lamb-vector-based decompositions have also been developed using derivative-moment-transformation integral methods [47, 48, 49], with additional correction terms introduced to account for transonic compressibility effects.

In the present study, we extend the influence-potential-based force partitioning framework to a moving-body transonic wake-interaction problem in order to diagnose the mechanisms responsible for wake-modified flutter. Since the central focus of this work is the exchange of energy between the flow and the oscillating wing, we further introduce a corresponding “power partitioning” framework that decomposes the aerodynamic power transferred through the pressure-induced loads into physically interpretable volumetric and surface contributions. This compressible force/power partitioning methodology enables direct identification of the flow structures and regions responsible for destabilizing or stabilizing energy transfer. Although developed here for the transonic cylinder–airfoil configuration, the framework is broadly applicable to high-speed compressible flows involving stationary, moving, and deforming bodies.

The outline of the paper is as follows: Section 2 outlines the numerical setup and computational methodology. Section 3 describes the flow physics and energy maps. Section 4 outlines the compressible force and power partitioning method. We discuss the wake-induced effects in Section 4 and conclusions in Section 5. The grid convergence details are presented in Appendix A.

Problem setup

We perform high-fidelity, time-accurate, two-dimensional simulations of compressible flow over a NACA0012 airfoil and an airfoil-cylinder system undergoing prescribed sinusoidal motion using high-order sharp-interfaced immersed boundary method (ViCAS3D, Viscous Cartesian All Speed Three-Dimensional [3D] Solver) [50]. We conduct several simulations at various Mach numbers and pitching amplitudes at a airfoil chord-based (cc) Reynolds number of Re∞=U∞c/ν∞=10,000Re_{\infty} = U_{\infty}c/\nu_{\infty} = 10{,}000, where U∞U_{\infty}, ν∞\nu_{\infty} are the free-stream velocity and kinematic viscosity, respectively. While this Reynolds number is lower than operational Reynolds numbers for transonic wings, it is sufficiently large so as to generate complex vortex and shock dynamics that are representative of realistic flows, while allowing for a comprehensive simulation-based examination of the large parameter space with a moving boundary. A brief description of the solver and the numerical setup are described below. Further details and a rigorous validation of the solver can be found in Turner et al. [ [50].

Flow Solver

We consider two-dimensional compressible Navier–Stokes equations in conservative form:

∂Q∂t+∇⋅E−M∞Re∞∇⋅F=0.(1)\frac{\partial\mathbf{Q}}{\partial t} + \nabla\cdot\mathbf{E} - \frac{M_{\infty}}{Re_{\infty}}\nabla\cdot\mathbf{F} = 0. \tag*{(1)}

Here, free-stream Mach number is M∞=U∞/a∞M_{\infty} = U_{\infty}/a_{\infty} where U∞U_{\infty}, a∞a_{\infty} are the free-stream velocity and the speed of sound, respectively. The free-stream Reynolds number is Re∞=ρ∞U∞Lc/μ∞Re_{\infty} = \rho_{\infty}U_{\infty}L_c/\mu_{\infty}, where ρ∞\rho_{\infty}, μ∞\mu_{\infty} are the free-stream density and viscosity and LcL_c is the characteristic length scale. The vectors Q\mathbf{Q}, E\mathbf{E}, F\mathbf{F} are given by,

Q=[ρ, ρu, ρet]T,(2)\mathbf{Q} = [\rho,\ \rho\mathbf{u},\ \rho e_t]^{T}, \tag*{(2)}
E=[ρu, ρu⊗u+pI, (ρet+p)u]T,(3)\mathbf{E} = [\rho\mathbf{u},\ \rho\mathbf{u}\otimes\mathbf{u} + p\mathbf{I},\ (\rho e_t + p)\mathbf{u}]^{T}, \tag*{(3)}
F=[0, τ, τ⋅u+q]T.(4)\mathbf{F} = [0,\ \boldsymbol{\tau},\ \boldsymbol{\tau}\cdot\mathbf{u} + \mathbf{q}]^{T}. \tag*{(4)}

Here, u=[u,v]T\mathbf{u} = [u,v]^{T} is the velocity vector in two-dimensions, pp is pressure, ρ\rho is density, ee is the total energy defined as

et=p(γ−1)ρ+u⋅u2e_t = \frac{p}{(\gamma-1)\rho} + \frac{\mathbf{u}\cdot\mathbf{u}}{2}

with γ=1.4\gamma= 1.4 for air and II is the identity matrix. The stress tensor τ\boldsymbol{\tau} and the heat flux vector q\mathbf{q} are defined as

τ=∇u+∇uT−23(∇⋅u)I,(5)\boldsymbol{\tau} = \nabla\mathbf{u} + \nabla\mathbf{u}^{T} - \frac{2}{3}(\nabla\cdot\mathbf{u})\mathbf{I}, \tag*{(5)}
q=−μ(γ−1)Pr∇T(6)\mathbf{q} = -\frac{\mu}{(\gamma-1)Pr}\nabla T \tag*{(6)}

where Prandtl number Pr=0.72Pr = 0.72 and μ\mu is the local dynamic viscosity determined using Sutherland’s law [51]. The results presented in this work are non-dimensionalized by the respective characteristic length, velocity, time-scale and density as airfoil chord cc, free-stream velocity U∞U_{\infty}, c/U∞c/U_{\infty} and ρ∞\rho_{\infty}.

The governing equations are discretized using sixth-order cell-centered compact finite difference schemes, with third and fourth-order compact schemes applied at the boundaries [52]. In the current study, high-order compact filtering is implemented with a filter cut-off of αf=0.45\alpha_f = 0.45 for numerical stability [53]. These compact filtering and differencing schemes are implemented using the Thomas algorithm [54]. The governing equations are discretized on a non-uniform Cartesian grid which requires the following transformation for the evaluation of the derivatives. A fourth stage Runge-Kutta scheme is used for time integration [55]. Additionally, a hyperviscosity method is used for handling the discontinuities caused by shock waves Kawai and Lele [ [56]. This method introduces artificial diffusivity biased towards high-wavenumbers that are unresolved. The coefficients for artificial diffusion are added to the dynamic viscosity, thermal conductivity and bulk viscosity based on the fourth-order derivative of the local strain rate and the grid size, following Kawai and Lele [ [56], Turner et al. [ [50].

Since we are simulating flow over an airfoil and airfoil-cylinder system, the internal bodies are represented by an unstructured surface mesh immersed in a Cartesian fluid mesh. A no-slip boundary condition is implemented on the internal bodies using a sharp-interface immersed boundary method [57] using the ghost-cell method. The cells that have at least one neighboring fluid cell are identified as ghost cells and are used to implement the no-slip boundary condition. We use a Taylor polynomial interpolant of nthn^{\mathrm{th}} order through weighted least-squares minimization. For the current simulations, we use a third-order polynomial interpolant which considers 25 nearest fluid cell solutions to compute the values at the ghost cells. This formulation is the same as the one used in earlier studies [57, 50]. For farfield boundaries, an energy transfer and annihilating (ETA) boundary condition is prescribed to avoid spurious reflections [58].

Problem Setup and Mesh

We consider a NACA0012 airfoil profile with a cylinder placed underneath and ahead of the leading-edge as our geometry as shown in Figure 2. The trailing-edge of the airfoil is slightly rounded (rTE=0.00128cr_{TE} = 0.00128c) so that there are finite number of Cartesian cells that cover the airfoil trailing edge. The truncated chord length of the airfoil is c0=0.924cc_0 = 0.924c, where cc is unit chord. For the airfoil-cylinder system, the cylinder of diameter D=0.12cD = 0.12c (same as the thickness of the airfoil) is placed at (xc0,yc0)(x_{c0}, y_{c0}) at vertical gap of dd from the airfoil chord. For most of the current simulations, unless specified, yc0=−1.5Dy_{c0} = -1.5D and xc0=xLE−0.5Dx_{c0} = x_{LE} - 0.5D. For D=0.12cD = 0.12c, the cylinder center is placed at (xc0,yc0)=(−0.522,−0.18)(x_{c0}, y_{c0}) = (-0.522, -0.18). This cylinder placement is considered for a majority of the subsequent analysis unless specified. The computational domain spans 10 chord lengths in all directions with the airfoil as its center. For the earlier mentioned cylinder placement, a uniform mesh with a grid spacing of Δ=0.00125c\Delta= 0.00125c is used for −0.6≤x/c≤0.5-0.6 \le x/c \le0.5 and −0.6≤y/c≤0.5-0.6 \le y/c \le0.5 to ensure that the cylinder-airfoil system and the dominant physics is contained within the uniform, highly resolved region. The grid is stretched outside this uniform region as shown in Figure 2(a). The upstream and vertical directions are stretched at the same rate, while the downstream region is stretched more gradually to resolve the wake. Overall, we employ a mesh size of 1240×9601240 \times960 resulting in approximately 1.192 million Cartesian cells. We use the same mesh for the airfoil-only simulations. The grid verification details are discussed in Appendix A.

Computational domain and mesh setup, and cylinder placed underneath the airfoil

Figure 2. (a) Computational domain and mesh setup for the current study. (b) Setup of cylinder placed underneath the airfoil.

Both the airfoil only and the airfoil-cylinder systems are pitched about the airfoil mid-chord sinusoidally as per the kinematics given by,

θ(t)=θ0+Aθsin⁡(2πfst+π/2)(7)\theta(t) = \theta_0 + A_{\theta}\sin(2\pi f_s t + \pi/2) \tag*{(7)}

where θ0\theta_0 is set to be 0∘0^\circ, AθA_{\theta} is the pitching oscillation amplitude, frequency is fsf_s and T=1/fsT = 1/f_s is the oscillation period. We set fsf_s to be fixed value for an appropriate comparison with prior study [12]. For all the current simulations the reduced frequency is set to f∗=fsc0/U∞=0.1f^* = f_sc_0/U_\infty= 0.1.

Transonic flutter characteristics: Prescribed motion simulations

We investigate the transonic flutter characteristics of a two-dimensional sinusoidally pitching NACA0012 airfoil and cylinder-airfoil system under at various transonic Mach numbers and pitching amplitudes at a chord-based Reynolds number of 10,000 with a pitching frequency of f∗=0.1f^* = 0.1.

Vortex and Shock Dynamics

We consider the flow physics of the airfoil-only and the cylinder-airfoil systems pitching with Aθ=10∘A_{\theta} = 10^\circ at various Mach numbers of M∞=0.6M_\infty= 0.6, 0.7, and 0.8. The corresponding vorticity-fields at four instances during the pitch down motion are visualized in Figure 3. Here, t/T∈[4,4.5]t/T \in[4, 4.5] captures the pitch-down motion and t/T∈[4.5,5]t/T \in[4.5, 5] captures the pitch-up motion. The instance corresponding to the maximum oscillation amplitude is t/T=4t/T = 4, where TT is the respective time period of oscillation at each inflow Mach number.

Instantaneous vorticity-fields of sinusoidally pitching airfoil-only and cylinder-airfoil systems

Figure 3. Instantaneous vorticity-fields of a sinusoidally pitching airfoil-only (a)-(l) and cylinder-airfoil systems (m)-(x) with a pitching amplitude of Aθ=10∘A_\theta= 10^\circ and reduced frequency of f∗=0.1f^* = 0.1. Vorticity-fields are visualized at four instances during the pitching down motion corresponding to t/T=4,4.125,4.25,4.375t/T = 4, 4.125, 4.25, 4.375 at Mach numbers M∞=0.6,0.7,0.8M_\infty= 0.6, 0.7, 0.8 using ωz\omega_z and ∥∇ρ∥\lVert\nabla\rho\rVert in grey. The shock waves are characterized using the arrows indicate the shock motion direction and S indicates a stationary shock.

For the pitching airfoil-only system (Figure 3(a)-(Figure 3l)), as Mach number is increased, the flow becomes locally supersonic as can be seen as the shock waves. At M∞=0.6M_{\infty} = 0.6, while we see weak compression waves at the leading edge at t/T=4t/T = 4 (Figure 3(a)), the flow remains subsonic throughout the domain. As the airfoil pitches down, we see flow separation and shedding on the upper surface of the airfoil. At M∞=0.7M_{\infty} = 0.7 (Figure 3(e)-(Figure 3h)), we see multiple upstream propagating shock waves as the airfoil pitches down, indicated by the arrows. These upstream propagating shock waves later dissipate into the freestream and are classified as Tijdeman type C [59]. At the maximum pitch-up position at t/T=0t/T = 0 (Figure 3(e)), we see the presence of λ\lambda−shock wave system, where AA and BB denote the branches of the λ\lambda−shock. Downstream of branch AA, we see a separated shear layer. This λ\lambda− shock causes strong flow separation and shedding downstream of branch BB as the airfoil pitches down at t/T=4.125t/T = 4.125 (Figure 3(f)). Turner et al. [ [12] investigated the flow physics and flutter characteristics for this airfoil-only system with Aθ=10∘A_{\theta} = 10^{\circ}. They identified the presence of flutter instability at M∞=0.7M_{\infty} = 0.7 and attributed the flutter mechanism to this shock-induced separation. The flow separation near the trailing edge causes a low pressure region on the upper surface, creating a negative aerodynamic moment during the pitch-down moment resulting in net positive energy transfer as will be discussed in subsequent sections. At this time instant, a normal shock wave originates at the trailing edge due to the vortex roll-up. The trailing edge vortex undergoes shedding and the normal shock wave propagates upstream on the upper surface as the airfoil pitches down. At t/T=4.375t/T = 4.375 (Figure 3(h)), the upstream propagating shock waves on the lower surface later result in the λ\lambda−shock-induced separation during the pitch-up motion. Since the airfoil has zero mean angle of attack, the same physics of the lambda-shock-wave formation and shock-induced stall is observed on the lower surface during the pitch-up motion.

As the Mach number is further increased to M∞=0.8M_{\infty} = 0.8 (Figure 3(i)-(Figure 3l)), the width λ\lambda−shock system expands (Figure 3(i)) when compared to M∞=0.7M_{\infty} = 0.7 (Figure 3(i)) during the maximum pitch-up position. While the branch AA propagates upstream into the freestream, branch BB propagates downstream toward the trailing edge of the airfoil and is relatively stationary. As the airfoil starts to pitch down, the shock-induced separation predominantly occurs near the trailing edge and not on the airfoil surface. Past t/T=4.125t/T = 4.125, (Figure 3(k) and (Figure 3l)) both branch BB and the normal shock originated due to TE shedding propagate downstream on the upper surface. We also observe the presence of stationary shock waves on the bottom surface of the airfoil.

We now examine the vortex and shock dynamics of the sinusoidally pitching cylinder-airfoil system (Figure 3(m)-(Figure 3x)). With the addition of cylinder, vortex shedding and shear layers now interact with the lower-wing surface aerodynamics. A larger density gradient near the airfoil leading edge now is seen encompassing both the airfoil leading edge, the cylinder and the gap region. At M∞=0.6M_{\infty} = 0.6 (Figure 3(m)-(Figure 3p)), unlike the airfoil-only scenario (Figure 3(a)-(Figure 3d)), the flow becomes locally supersonic due to the presence of the cylinder. The presence of cylinder causes flow accelerating in the gap region, where the shear layers on the cylinder and the lower airfoil surface act as a channel, this results in shock waves in the gap region (Figure 3(p)). The blockage effect from the cylinder also results in a more pronounced flow separation and shedding on the airfoil upper surface compared to airfoil-only case.

As Mach number is increased to M∞=0.7M_{\infty} = 0.7, a λ\lambda−shock is formed at t/T=4t/T = 4 (Figure 3(q)). As compared to the airfoil-only system, we notice the advancement of branch AA of the λ\lambda−shock near the leading edge due to the presence of the cylinder. Similar to the airfoil-only system, flow separation on the airfoil surface is seen downstream of branch BB. As seen in M∞=0.6M_{\infty} = 0.6, even for M∞=0.7M_{\infty} = 0.7, we observe a pronounced vortex shedding on the upper surface that is also prevalent after the vortex shedding at the trailing edge. Further, the cylinder adds complexity to the flow physics on the airfoil lower surface, resulting in vortex shedding and shock waves in the gap region. As the Mach number is increased to M∞=0.8M_{\infty} = 0.8, the vortex shedding from the cylinder is suppressed [60, 61]. The shear layer on the airfoil lower surface and shear layer emanating from the cylinder act as a channel with moving walls. The incoming flow accelerates and becomes supersonic as it enters the gap flow and a shock-train is formed [62, 63]. Xiong et al. [ [64] identified that the shock-train undergoes oscillations due to periodic downstream disturbances in a stationary channel with constant area. In the present case, the gap region continuously is moving and deforming and we also see the flow separation and shedding downstream of the gap region. Hence, the structure of the shock-train continuously changes as the cylinder-airfoil system undergoes the pitching motion. Similar to the airfoil-only case, we observe formation of λ\lambda− shock waves on the upper airfoil surface in the cylinder-airfoil system. Overall, the addition of cylinder increases the complexity of the flow physics due to the presence of vortex shedding from the cylinder, interaction between the cylinder vortex shedding with the lower airfoil shear layer, flow acceleration due to the blockage effects from the cylinder.

Energy Maps

The flow physics for both the airfoil-only and cylinder-airfoil systems varies based on the Mach number, pitching amplitude and frequency. In order to systematically identify the flutter boundaries in the Mach number and the pitching amplitude space (M∞−AθM_{\infty} - A_{\theta}), we consider energy maps. Energy maps have been used to quantify the energy imparted to the structure by the fluid and vice versa for moving bodies in both compressible and incompressible aeroelastic systems [38, 65, 50]. Since we are interested in the effect of underwing cylinder on the airfoil, we consider energy maps solely for the airfoil surface in both the airfoil-only and the cylinder-airfoil systems. For a single degree-of-freedom pitching motion, the energy transfer between the structure and the fluid is quantified as

E⋆=∫0TCW dt,(8)E^{\star}=\int_{0}^{T} C_{W}\,\mathrm{d}t, \tag*{(8)}

where CWC_{W} is the power coefficient. The power coefficient is computed as

CW=CMθ˙,(9)C_{W}=C_{M}\dot{\theta}, \tag*{(9)}

where CMC_{M} is the pitching moment coefficient and θ˙\dot{\theta} is the pitching velocity. In the current scenario, θ˙\dot{\theta} is given by the sinusoidal pitching motion prescribed in equation (7). The pitching moment coefficient is computed from the surface pressure on the airfoil surface BB as

CM=∫Bp[(xp−x)×n]⋅ez dx1/2ρ∞U∞2c02,(10)C_{M}=\frac{\int_{B}p[(\mathbf{x}_{p}-\mathbf{x})\times\mathbf{n}]\cdot\mathbf{e}_{z}\,\mathrm{d}x}{1/2\rho_{\infty}U_{\infty}^{2}c_{0}^{2}}, \tag*{(10)}

where xp\mathbf{x}_{p} is pivot point (origin in our case), ez\mathbf{e}_{z} is the unit vector in zz-direction. The moment arm is defined such that the pitch-up moment is positive and the pitch-down moment is negative. We note that the contribution of shear stress to CMC_{M} is negligible and we compute the moment coefficient based only on the pressure loading on the airfoil. We also note that for the cylinder attached to the wing, the surface pressure from both the surfaces will contribute to the moment but the moment would then depend on the precise shape of the underwing structure and the attachment location. In this study, we are focused on how the aerodynamic interactions of the shear layers, wake, gap flow and the shock dynamics between the airfoil and the underwing cylinder effect the flutter characteristics on the airfoil. We therefore ignore the direct aerodynamic forces on store and their impact on the net loading on wing.

The energy maps for the airfoil-only and the cylinder-airfoil systems in the M∞−AθM_{\infty}-A_{\theta} space are shown in Figure 4. The flutter boundaries are characterized by the E⋆=0E^{\star}=0 contour. We can classify the flutter boundary as stable or unstable based on the conditions dE⋆/dAθ>0\mathrm{d}E^{\star}/\mathrm{d}A_{\theta}>0 and dE⋆/dAθ<0\mathrm{d}E^{\star}/\mathrm{d}A_{\theta}<0, respectively. Menon and Mittal [ [38]. The energy map for the airfoil-only system is reproduced from Turner et al. [ [12] (Figure 4(a)). The energy map for the cylinder-airfoil system is obtained using 70 new simulations sampled in the M∞−AθM_{\infty}-A_{\theta} space. For each of these 70 simulations, the energy is computed by averaging over five pitching cycles based on equation (8) after the passage of the initial transients. The energy map is obtained by interpolation using Gaussian process regression (GPR) using the Matern kernel [66].

Energy maps for (a) a pitching airfoil-only system and (b) a pitching cylinder-airfoil system

Figure 4. Energy Maps for (a) a pitching airfoil-only system (reproduced from [12]) and (b) pitching cylinder-airfoil system for Reynolds number of 10,000 and reduced frequency f⋆=0.1f^{\star}=0.1. The flutter boundary of the airfoil-only system are overlayed in (b) in gray. The black dots denote the flow configurations considered for subsequent analysis.

The flutter boundary for the airfoil-only system is overlayed on the energy map of the cylinder-airfoil system using the gray line (Figure 4(b)) and we note that compared to the airfoil-only system, the cylinder-airfoil system is much more prone to flutter instability as the flutter boundaries expand significantly. The maximum energy transfer for the

cylinder-airfoil system occurs at (M∞,Aθ)≈(0.675,8∘)(M_{\infty}, A_{\theta}) \approx(0.675, 8^\circ). Further, the subcritical instability now occurs at significantly lower Mach number of M∞≈0.56M_{\infty} \approx0.56 for the cylinder-airfoil system compared to the M∞≈0.68M_{\infty} \approx0.68 for the airfoil-only system. These initial observations are inline with several previous studies that indicate that the addition of store can enhance transonic flutter [22, 23, 29, 24].

To further examine the effect of the cylinder wake on flutter, we consider several distinct cases corresponding to pitching amplitudes of Aθ=5∘A_{\theta} = 5^\circ and 10∘10^\circ and Mach numbers M∞=0.6M_{\infty} = 0.6, 0.7 and 0.8. From the energy maps (highlighted as black dots in Figure 4(b)), among these flow conditions, the airfoil-only system exhibits flutter instability at (M∞,Aθ)=(0.7,10∘)(M_{\infty}, A_{\theta}) = (0.7, 10^\circ) and (M∞,Aθ)=(0.7,5∘)(M_{\infty}, A_{\theta}) = (0.7, 5^\circ) is right at the flutter boundary. The pitching cylinder-airfoil system is prone to flutter at all these conditions. Hence, these conditions will give us insights into the effect of cylinder flow on the responsible energy transfer mechanisms for flutter instability.

The sinusoidal pitching motion and the temporal variation of CM,CWC_M, C_W for these cases are shown in Figure 5 for three pitching cycles. For all these cases, the reduced frequency is f⋆=0.1f^\star= 0.1. At (M∞,Aθ)=(0.7,10∘)(M_{\infty}, A_{\theta}) = (0.7, 10^\circ), the airfoil-only system has significant positive peaks in CW(t)C_W(t) variation (Figure 5(f)), responsible for positive energy transfer to the wing. The corresponding negative and positive peaks in CMC_M during the pitch-down (i) and pitch-up (ii) motion are highlighted using the arrows (Figure 5(d)). These peaks are associated with the λ\lambda-shock induced vortex shedding at the trailing edge on the upper surface during the pitch-down motion (Figure 3(e)-(Figure 3g)), and on the lower surface on the pitch-up motion respectively. As the pitch amplitude is reduced to Aθ=5∘A_{\theta} = 5^\circ, the fluctuation magnitude of CWC_W also reduces (Figure 5(e)). Similar positive peaks of reduced magnitude appear in CWC_W. The associated peaks in CMC_M are highlighted in Figure 5(c). As the pitching amplitude reduces at M∞=0.7M_{\infty} = 0.7, the shock-induced flow separation is suppressed placing Aθ=5∘A_{\theta} = 5^\circ at the flutter boundary. These peaks are absent in M∞=0.6,0.8M_{\infty} = 0.6, 0.8 as they fall outside of flutter boundaries.

Time histories of pitching motion, moment coefficient, and power coefficient for airfoil and cylinder-airfoil configurations

Figure 5. Time history of (a)-(b) the pitching motion θ(t)\theta(t) and the moment coefficient CMC_M and the power coefficient CWC_W of NACA0012 airfoil (c)-(f) and NACA0012 airfoil-cylinder system (g)-(j), respectively at pitching amplitudes of Aθ=5∘A_\theta= 5^\circ and 10∘10^\circ and M∞=0.6M_\infty= 0.6, 0.70.7 and 0.80.8 for three pitching cycles. The dotted lines in (e),(f), (i) and (j) denote the mean power coefficients for each case.

For the pitching cylinder-airfoil system (Figure 5(g)-(Figure 5j)), we observe several high frequency fluctuations in CM(t)C_M(t) unlike the airfoil-only case due to the presence of cylinder wake. To obtain further insights into the variation of CWC_W and responsible physical mechanisms, we leverage compressible force and power partitioning method.

Compressible Force and Power Partitioning Method

The force partitioning method (FPM) has been widely applied to understand the flow structures and mechanisms responsible for generating aerodynamic forces and moments in several unsteady aeroelastic and aerodynamic systems [40, 39, 67, 42, 68]. This method is based on the work by Quartapelle and Napolitano [1983b] and and is also related to the formulation of Chang [1992]. FPM partitions the aerodynamic forces and moments into volume and surface integrals on an immersed body by projecting the momentum conservation equation on an influence field. The details of the FPM method for incompressible flows can be found in Zhang et al. [2015], Menon and Mittal [2021a].

In this study, we extend this force partitioning framework to compressible flows. The related influence-potential based force element method using the lamb-vector has been proposed in the past Chang and Lei [1996], Chang et al. [1998] for compressible flows past stationary bodies to examine the volumetric components for the aerodynamic forces. However, extension to moving bodies in compressible flows has not been shown before. Furthermore, since energy-transfer between the flow and a oscillating airfoil is not just driven by the pressure loading but also the phase between the pressure loading and the airfoil motion, application of partitioning directly to the power associated with flutter is needed. Using these extensions, we can investigate the flow physics responsible for transonic wing flutter and the effect of wake interaction in the presence of a cylinder. Since energy is a quantity of interest for flutter, in the present study, we solely focus on the partitioning of power.

For partitioning of induced power, we start by constructing the “influence” field ϕ(x)\phi(x) defined as follows:

∇2ϕ=0,withn⋅∇ϕ={n⋅vB,on B,0,on Σ,(11)\nabla^2 \phi= 0, \qquad\text{with}\qquad\boldsymbol{n}\cdot\nabla\phi= \begin{cases} \boldsymbol{n}\cdot\boldsymbol{v}_B, & \text{on } B,\\ 0, & \text{on } \Sigma, \end{cases} \tag*{(11)}

where nn is the unit normal vector on the immersed body surface (BB) (the normals for the airfoil and the cylinder can be seen in Figure 6(a)) and the outer domain boundaries (Σ\Sigma), and vBv_B is the velocity of the immersed surface. Note that this is subtly different from the influence potential corresponding to the aerodynamic forces or moments [39] in that the velocity of the body surface vBv_B also appears in the construction of the influence potential. A few instances of the influence fields corresponding to power for a sinusoidally pitching cylinder-airfoil system are visualized in Figure 6(a)-(Figure 6c) and (d)-(f), respectively. Note that the presence of the cylinder changes the influence field in a small but noticeable manner.

Influence fields corresponding to power

Figure 6. Influence fields corresponding to power (ϕ\phi) (a)-(c) for a pitching cylinder-airfoil system with f⋆=0.1f^\star= 0.1 and Aθ=10∘A_\theta= 10^\circ at three different time instances of a pitching cycle. Note that ϕ=0\phi= 0 at t=0t = 0 as the body velocity is zero at the maximum pitch-up phase.

By projecting the compressible momentum equation (equation (1)) on ∇ϕ\nabla\phi and integrating over the fluid volume V\mathcal{V} we obtain,

∫V∇p⋅∇ϕ dV=∫V[∇⋅τ−∂ρu∂t−∇⋅(ρu⊗u)]⋅∇ϕ dV.(12)\int_{\mathcal{V}} \nabla p \cdot\nabla\phi\,dV = \int_{\mathcal{V}} \left[\nabla\cdot\boldsymbol{\tau} - \frac{\partial\rho\mathbf{u}}{\partial t} - \nabla\cdot(\rho\mathbf{u} \otimes\mathbf{u})\right] \cdot\nabla\phi\,dV. \tag*{(12)}

Using Gauss-divergence theorem, conservation of mass and the boundary conditions for influence potential (equation (11)), we can further simplify the equation 9 into equation 10.

∫Bpn⋅vB dS=∫V[∇⋅(ρDuDt)ϕ+(∇⋅τ)⋅∇ϕ] dV+∫B+Σϕ[−ρn⋅DuDt] dS.(13)\int_{\mathcal{B}} p\mathbf{n} \cdot\mathbf{v}_{B}\,dS = \int_{\mathcal{V}} \left[\nabla\cdot\left(\rho\frac{D\mathbf{u}}{Dt}\right)\phi+ (\nabla\cdot\boldsymbol{\tau}) \cdot\nabla\phi\right]\,dV + \int_{\mathcal{B}+\Sigma} \phi\left[-\rho\mathbf{n}\cdot\frac{D\mathbf{u}}{Dt}\right]\,dS. \tag*{(13)}

Since the outer boundaries (Σ\Sigma) are placed far away from the immersed body where the flow is mostly uniform in space and time, we can simplify the surface integral to finally obtain the partitioned power as

∫Bpn⋅vB dS⏟W=∫Vϕ[∇⋅(ρDuDt)] dV⏟W1+∫V(∇⋅τ)⋅∇ϕ dV⏟W2+∫Bϕ[−ρn⋅DuDt] dS⏟W3.(14)\underbrace{\int_{\mathcal{B}} p\mathbf{n}\cdot\mathbf{v}_{B}\,dS}_{\mathbf{W}} = \underbrace{\int_{\mathcal{V}} \phi\left[\nabla\cdot\left(\rho\frac{D\mathbf{u}}{Dt}\right)\right]\,dV}_{\mathbf{W}_{1}} + \underbrace{\int_{\mathcal{V}}(\nabla\cdot\boldsymbol{\tau})\cdot\nabla\phi\,dV}_{\mathbf{W}_{2}} + \underbrace{\int_{\mathcal{B}}\phi\left[-\rho\mathbf{n}\cdot\frac{D\mathbf{u}}{Dt}\right]\,dS}_{\mathbf{W}_{3}}. \tag*{(14)}

The time-integrated power per one time-period of oscillation amounts to the energy exchanged between the flow and the foil. The respective integrands are denoted as Wj=∫wj(x)dx\mathbf{W}_j = \int w_j(\mathbf{x})d\mathbf{x} where the subscript jj corresponds to the specific partition. From equation 11, we have two volume integrals and one surface integral. The integrand of the first term on the right hand side (W1\mathbf{W}_1) corresponds to the volumetric Lagrangian rate of change of momentum that encompasses the volumetric unsteady and advection effects. The term W2\mathbf{W}_2 represents the volumetric effects of viscous stress divergence on surface pressure. The third right hand side integrand (W3\mathbf{W}_3) encodes the added-mass effects which depends only on the surface acceleration of the immersed body.

The power partitioning framework for compressible flows can be directly compared with the corresponding formulation for incompressible flows [40, 39]. The terms W2\mathbf{W}_2 and W3\mathbf{W}_3 appear in both formulations. In the incompressible limit, the viscous stress divergence term, W2\mathbf{W}_2, reduces to the viscous momentum diffusion term because the velocity divergence vanishes [70]. The primary distinction between the incompressible and compressible formulations arises in the W1\mathbf{W}_1 term. For incompressible flows, this term reduces to the so-called “vortex-induced force”, whose integrand contains Q=−∇⋅(u⊗u)Q = -\nabla\cdot(\mathbf{u} \otimes\mathbf{u}). In contrast, the compressible formulation retains the divergence of the full Lagrangian rate of change, thereby incorporating not only the nonlinear convective contribution, but also dilatational and density-gradient effects that are absent in incompressible flows. Consequently, W1\mathbf{W}_1 represents the combined influence of vortical motion, compressibility-induced dilatation, and density gradients on the surface pressure loading.

The application of the power partitioning method to the pitching cylinder-airfoil system at (M∞,Aθ=0.7,10∘)(M_\infty, A_\theta= 0.7, 10^\circ) is shown in Figure 7 as an example. Figure 7(a) shows that CWC_W computed using the power partitioning method (equation (14)) matches with the convention computation using the surface integral of power. The contributions from the individual terms can be seen in Figure 7(b). The Lagrangian rate of change of momentum term, W1\mathbf{W}_1, is the primary energy transfer term responsible for about 85% of total energy, followed by the volumetric viscous term W2\mathbf{W}_2, which accounts for 14%. The added mass term, W3\mathbf{W}_3, contribution to the overall energy transfer is negligible. The spatial regions of the flow which have non-zero contributions to the energy transfer can be directly visualized using w1(x,t)\mathbf{w}_1(\mathbf{x},t) Figure 7(g)-(Figure 7j). The cylinder wake, the gap flow, shear layers on airfoil surface and the vortex shedding near the trailing edge are highlighted. We also note that the shock waves are decomposed as both positive and negative energy transfer due to the pressure gradient across the shock waves. Furthermore, the contribution of structures diminishes with distance from the airfoil due to the rapid decay of the influence field.

Plots and flowfield visualizations for the pitching cylinder-airfoil system

Figure 7. Application of compressible power partitioning method to the pitching cylinder-airfoil system at M∞=0.7M_\infty= 0.7 and Aθ=10∘A_\theta= 10^\circ. (a) Verification of the power partitioning method. (b) Contributions to CWC_W from individual power partitions. (c)–(f) A few instances of flowfields visualized using vorticity contours and density gradient magnitude in gray. (g)–(j) The corresponding power density fields colored using w1\mathbf{w}_1, ∥∇ρ∥\lVert\nabla\rho\rVert in gray and accentuated using ∥∇ρ∥=3\lVert\nabla\rho\rVert= 3 contour line.

Quadrant-wise Regional Power Partitioning

Splitting the integral into different regions of interest can help isolate the mechanism of flutter instability [39]. Here, we split the integrand into four distinct quadrants: Qk\mathcal{Q}_k, k=1,2,3,4k = 1, 2, 3, 4. These are centered around the mid-chord pivot point of the airfoil (see Figure 8(a)) and rotate with the airfoil as the airfoil pitches up and down. Specifically, we consider body-fitted coordinates denoted as xR(t)\mathbf{x}_R(t) centered around the origin. For a sinusoidally pitching system about the origin, these can be obtained as

Application of regional power partitioning method for flow over a pitching NACA0012 airfoil

Figure 8. Application of regional power partitioning method for flow over a pitching NACA0012 airfoil at M∞=0.7M_{\infty}=0.7 and Aθ=10∘A_{\theta}=10^\circ. (a) Illustration of quadrant-based splitting of power. (b) Contributions to instantaneous CWC_W from individual quadrants for one pitching cycle. (c)–(g) A few instances of flowfields visualized using ωz\omega_z and density gradient magnitude in gray. (h)–(l) The corresponding power density fields colored using w1\mathbf{w}_1 and ∥∇ρ∥\lVert\nabla\rho\rVert in the background in gray along with ∥∇ρ∥=3\lVert\nabla\rho\rVert=3 contour line.

[xR(t)yR(t)]=[cos⁡(θ(t))−sin⁡(θ(t))sin⁡(θ(t))cos⁡(θ(t))][xy](15)\begin{bmatrix} x_R(t) \\ y_R(t) \end{bmatrix} = \begin{bmatrix} \cos(\theta(t)) & -\sin(\theta(t)) \\ \sin(\theta(t)) & \cos(\theta(t)) \end{bmatrix} \begin{bmatrix} x \\ y \tag*{(15)} \end{bmatrix}

We can then determine the contribution to the power coefficient from each quadrant as,

WQk(t)=∫xR∈Qkw1(x,t) dx.(16)W_{\mathcal{Q}_k}(t) = \int_{\mathbf{x}_R \in\mathcal{Q}_k} \mathbf{w}_1(\mathbf{x},t)\,\mathrm{d}\mathbf{x}. \tag*{(16)}

This quadrant-wise division is natural since it allows us to separate the contributions from the upper and lower surfaces of the airfoil and also the leading and trailing edge regions. We now apply the regional power partitioning method the case corresponding to (M∞,Aθ)=(0.7,10∘)(M_\infty, A_\theta) = (0.7, 10^\circ) as it is within the flutter boundary (gray line in Figure 4(b)) and has been investigated in the prior study. The instantaneous quadrant-wise contribution to CWC_W is seen in Figure 8(b). The contributions from quadrants 2 and 3 (Figure 8(b) right) vary from positive to negative during the pitch-up and pitch-down alternatively. The power density fields in Q2Q_2 and Q3Q_3 highlights the leading edge shear region and the shock waves (Figure 8(h)–(Figure 8l)). During the pitch-down motion (t/T∈[0,0.5]t/T \in[0,0.5]), the upper surface shear layer (Q2Q_2) contributes negative energy to the foil. The leading-edge shear layer results in a lower pressure on the suction surface creating a positive aerodynamic moment. Since this occurs during the pitch-down motion (θ˙<0\dot{\theta}<0), we see a negative energy transfer in Q2Q_2 and similarly, a positive energy transfer during the pitch-up motion.

During the pitch down motion, WQ1\mathbf{W}_{Q_1} follows the fluctuations in the total W1\mathbf{W}_1 including the significant positive peak at t2t_2. A few instances of the flow fields and the corresponding power density fields during the pitch-down motion can be seen in Figure 8(c)–(Figure 8l). The flow separation induced by λ\lambda-shock system near the upper surface trailing edge results in the positive peak in WQ1\mathbf{W}_{Q_1} at t2t_2. Through w1(x,t)\mathbf{w}_1(\mathbf{x},t), we can also identify which region of flow separation results in positive or negative energy transfer. The trailing-edge vortex shedding process comprises of both positive and negative fluctuations in WQ1\mathbf{W}_{Q_1} due to the presence of multiple co- and counter-rotating vortices. For instance, the vortex cores at t4t_4 are highlighted with positive w1\mathbf{w}_1 but the region between then is highlighted with negative w1\mathbf{w}_1. Similarly, during the pitch-up motion t/T∈[0.5,1]t/T \in[0.5,1], WQ4\mathbf{W}_{Q_4} explains the positive peak in W1\mathbf{W}_1 at t/T≈0.8t/T \approx0.8. Hence, regional power partitioning reconfirms that the shock-induced flow separation is responsible for flutter in the airfoil-only case.

Influence of cylinder wake on airfoil flutter

We apply the regional power partitioning method for airfoil-only and cylinder-airfoil systems to compare the flow physics and the energy transfer mechanisms at the pitching amplitudes of Aθ=5∘A_{\theta}=5^\circ and 10∘10^\circ and Mach numbers M∞=0.6,0.7M_{\infty}=0.6,0.7 and 0.80.8. From the energy maps (Figure 4), among these flow conditions, the airfoil-only system exhibits flutter instability at (M∞,Aθ)=(0.7,10∘)(M_{\infty},A_{\theta})=(0.7,10^\circ) and (M∞,Aθ)=(0.7,5∘)(M_{\infty},A_{\theta})=(0.7,5^\circ) is right at the flutter boundary. Comparatively, the cylinder-airfoil system is more prone to flutter at all these conditions. The overall energy contribution can be quantified using the quadrant-wise energy contribution EQi⋆E^{\star}_{Q_i} in Figure 9. Here, EQi⋆=∫0TWQi dtE^{\star}_{Q_i}=\int_{0}^{T}W_{Q_i}\,\mathrm{d}t and i∈{1,2,3,4}i\in\{1,2,3,4\}. The difference in the energy transfer due to the presence of the cylinder is computed using ΔEQi⋆\Delta E^{\star}_{Q_i} (Figure 9(c), (Figure 9f)) given as,

Quadrant-wise contribution of energy transfer

Figure 9. Quadrant-wise contribution of energy transfer EQi⋆E^{\star}_{Q_i}. (a) Schematic of color-coded quadrant division and (b) Outline of flutter boundaries for airfoil-only (gray) and cylinder-airfoil (black dashed) systems. The cases considered in (c)-(h) are shown as red dots. EQi⋆E^{\star}_{Q_i} for airfoil-only ((c),(f)), cylinder-airfoil systems ((d),(g)) and the difference between them ((e),(h)) for Aθ=5∘A_{\theta}=5^\circ and 10∘10^\circ and M∞=0.6M_{\infty}=0.6, 0.7 and 0.8.

ΔEQi⋆=EQi⋆cylinder−airfoil−EQi⋆airfoil−only,i∈{1,2,3,4}.(17)\Delta E^{\star}_{Q_i}=E^{\star}_{Q_i}{}^{\mathrm{cylinder-airfoil}}-E^{\star}_{Q_i}{}^{\mathrm{airfoil-only}},\quad i\in\{1,2,3,4\}. \tag*{(17)}

For the pitching airfoil-only case (Figure 9(c),(Figure 9f)), M=0.7M=0.7 shows flutter instability where the positive energy transfer is seen in Q1Q_1 and Q4Q_4 for both Aθ=5∘A_{\theta}=5^\circ and 10∘10^\circ. The shock-induced stall and trailing edge shedding is only present at M=0.7M=0.7. The pitching amplitude, Aθ=5∘A_{\theta}=5^\circ lies on the flutter boundary and the net energy reduces compared to Aθ=10∘A_{\theta}=10^\circ. When the underwing cylinder is introduced in Q3Q_3 (Figure 9(d),(Figure 9g)), Q2Q_2 and Q3Q_3 show net energy extraction from the fluid for both pitching amplitudes and at all the Mach numbers. With the cylinder present , similar to the airfoil-only case, Q1Q_1 contributes to positive energy transfer for both Aθ=5∘A_{\theta}=5^\circ and 10∘10^\circ at M∞=0.7M_{\infty}=0.7 (Figure 9(d),(Figure 9g)). This indicates a similar mechanism of λ−\lambda- shock-induced flow separation on the upper airfoil surface. The corresponding vorticity-fields can also be seen in Figure 3(q)-(Figure 3t). However, at Aθ=10∘A_{\theta}=10^\circ (Figure 9(g)), Q4Q_4 contributes to negative energy for M=0.7M=0.7. The presence of cylinder creates vortex shedding on the lower airfoil surface and limits the formation of λ−\lambda-shock system (see Figure 3(q)-(Figure 3t)).

The influence of cylinder flow on the energy transfer can be isolated by considering ΔEQi⋆\Delta E^{\star}_{Q_i} (equation (17)) (Figure 9(e),(Figure 9h)) where the dominant effects are in the leading-edge quadrants, Q2Q_2 and Q3Q_3. At Aθ=5∘A_{\theta}=5^\circ the addition of upstream cylinder results in an increased energy transfer in all quadrants for all Mach numbers except for M∞=0.7M_{\infty}=0.7 in Q4Q_4. This discrepancy can be explained by the suppression of λ\lambda–shock system on the airfoil lower surface.

To further investigate the responsible mechanisms and the dominant effects due to the addition of cylinder, we consider Aθ=5∘A_{\theta}=5^\circ and M∞=0.6,0.7M_{\infty}=0.6, 0.7 and 0.80.8 for the subsequent analysis. The instantaneous vorticity-fields during the pitch-down motion at Aθ=5∘A_{\theta}=5^\circ are visualized in Figure 10 and we note that the flow behavior follow similar trends as Aθ=10∘A_{\theta}=10^\circ in Figure 3, except for M∞=0.6M_{\infty}=0.6. While the airfoil-only system exhibits flow separation on the airfoil surface at Aθ=10∘A_{\theta}=10^\circ, at Aθ=5∘A_{\theta}=5^\circ, flow separation occurs at the trailing edge. In contrast, the addition of the cylinder causes flow separation on the airfoil surface and the gap region results in flow acceleration to supersonic speeds.

Instantaneous vorticity-fields of a sinusoidally pitching airfoil-only and cylinder-airfoil systems

Figure 10. Instantaneous vorticity-fields of a sinusoidally pitching airfoil-only (a)–(f) and cylinder-airfoil (g)–(l) systems with a pitching amplitude of Aθ=5∘A_{\theta}=5^\circ are visualized using ωz\omega_z and ∥∇ρ∥\lVert\nabla\rho\rVert corresponding to t/T=4,4.25t/T=4, 4.25 at M∞=0.6,0.7M_{\infty}=0.6, 0.7 and 0.80.8. The arrows indicate the instantaneous shock motion direction and S indicates a stationary shock.

Since the quadrants Q2Q_2 and Q3Q_3 have a dominant effect on enhancing flutter in the cylinder-airfoil case, (see Figure 9(b) and (e)) we now focus on a detailed analysis of these quadrants. Quadrants Q1Q_1 and Q4Q_4, which provide a much smaller contribution to the flutter enhancement are included in Appendix § 10.

Quadrants Q2Q_2 and Q3Q_3

The contributions from Q2Q_2 and Q3Q_3 to CWC_W for the airfoil and the cylinder-airfoil systems (WQ2W_{Q_2}, WQ3W_{Q_3}), and their difference (ΔWQ2\Delta W_{Q_2}, ΔWQ3\Delta W_{Q_3}) are shown in Figure 11. In Q2Q_2 (Figure 11(a)–(b)), the energy extraction to the structure from fluid is observed during the pitch-up phase for all three Mach numbers for both the systems. As the Mach number increases from 0.60.6 to 0.80.8, we see a suppression of high-frequency fluctuations in WQ2W_{Q_2} due to the suppression of cylinder vortex shedding as can be seen in Figure 10. The difference in WQ2W_{Q_2} from the airfoil-only system is visualized as ΔWQ2\Delta W_{Q_2} in Figure 11(c). The variation of ΔWQ2\Delta W_{Q_2} for all three Mach numbers show a similar trend. While both the cylinder–airfoil and airfoil-only systems exhibit similar energy transfer during the pitch-down phase, the cylinder–airfoil configuration produces substantially greater energy transfer to the foil during the pitch-up phase (t/T∈[0.5,1]t/T\in[0.5,1]).

Dominant effects of underwing cylinder on airfoil flutter

Figure 11. Dominant effects of underwing cylinder on airfoil flutter. Power coefficient contribution from (a)-(c) Q2Q_2, and (d)-(f) Q3Q_3 at Aθ=5∘A_\theta= 5^\circ and M∞=0.6M_\infty= 0.6, 0.7 and 0.8. Phase-averaged WQ2\mathbf{W}_{Q_2} and WQ3\mathbf{W}_{Q_3} for (a),(d) airfoil-only system, (b),(e) cylinder-airfoil system and (c),(f) ΔWQ2\Delta\mathbf{W}_{Q_2} and ΔWQ3\Delta\mathbf{W}_{Q_3} are the respective differences between them.

Figure 11(d)-(Figure 11e), show the phase-averaged WQ3\mathbf{W}_{Q_3} for the airfoil-only and the cylinder-airfoil systems, and a significant influence of cylinder can also be seen in this quadrant, which contains the underwing cylinder. While the pitch-up phase is the energy extraction phase in Q2Q_2, in Q3Q_3, the pitch-down phase is the energy extraction phase for both the systems. Similar to Q2Q_2, for the cylinder-airfoil system, the high frequency fluctuations are reduced with increasing M∞M_\infty due to the suppression of cylinder wake shedding (Figure 11(e)). At M∞=0.6M_\infty= 0.6 and 0.7, the high frequency fluctuations are significant in WQ3\mathbf{W}_{Q_3} (Figure 11(e)) compared to WQ2\mathbf{W}_{Q_2} (Figure 11(b)) as the cylinder wake shedding directly affects the flow physics in Q3Q_3. The difference from the airfoil-only system ΔWQ3\Delta\mathbf{W}_{Q_3} (Figure 11(f)) shows an increased energy transfer to the airfoil due to the cylinder during the pitch-down phase for all M∞M_\infty. Interestingly, at M∞=0.6M_\infty= 0.6, the cylinder-airfoil system exhibits a significant reduction in the negative energy transfer to the foil during during the pitch-up phase (seen as a positive ΔWQ3\Delta\mathbf{W}_{Q_3} in Figure 11(f)), indicating that the presence of the cylinder effects both the pitch-up and pitch-down phases of the foil. Both M∞=0.7M_\infty= 0.7 and 0.8, show a similar trend in ΔWQ3\Delta\mathbf{W}_{Q_3}, wherein the presence of the cylinder significantly increases the energy transfer to the foil during the pitch-down phase (t/T∈[0,0.5]t/T \in[0, 0.5]), with a slightly negative contribution during the pitch-up phase.

Flow Features and Pressure Distribution

We now examine the flow-fields to further analyze the influence of cylinder in Q2Q_2 and Q3Q_3. Figure 12 shows the power density fields, w1(x,t)\mathbf{w}_1(\mathbf{x},t) with numerical Schlieren (∥∇ρ∥\lVert\nabla\rho\rVert) in the background, and the corresponding pressure coefficient fields (Cp=p−p∞1/2ρ∞U∞2C_p = \frac{p-p_\infty}{1/2\rho_\infty U_\infty^2}) for M∞=0.6M_\infty= 0.6, 0.7 and 0.8 for both systems at t/T=4.25t/T = 4.25 (pitch-down) and 4.75 (pitch-up). The pressure-coefficient fields show the direct impact of the flow structures on the airfoil surface and can be attributed to power, since CMC_M is computed based on the airfoil surface pressure.

Power density and pressure coefficient fields for airfoil and cylinder-airfoil configurations at three freestream Mach numbers

Figure 12. Dominant effect of underwing cylinder on airfoil flutter for Aθ=5∘A_{\theta}=5^\circ and M∞=0.6M_{\infty}=0.6 (i),0.70.7 (ii) and 0.80.8 (iii). Power density fields visualized using w1\mathbf{w}_1, ∥∇ρ∥\lVert\nabla\rho\rVert and ∥∇ρ∥=3\lVert\nabla\rho\rVert=3 iso-contour (a),(b),(e),(f) and the corresponding pressure coefficient fields Cp=(p−p∞)/(1/2ρ∞M∞2)C_p=(p-p_{\infty})/(1/2\rho_{\infty}M_{\infty}^{2}) (c),(d),(g),(h) at instances t/T=4.25t/T=4.25 (pitch-down) and 4.754.75 (pitch-up). Zoomed-in insets near the leading edge are shown in (ii)-(a),(b).

For the airfoil-only system, in Q2Q_2 and Q3Q_3, w1(x,t)\mathbf{w}_1(\mathbf{x},t) is concentrated near the leading-edge shear layers (see the insets in Figure 12(ii)-(a),(b)). Similarly, for the cylinder-airfoil system (Figure 12(i–iii)-(e,f)), w1(x,t)\mathbf{w}_1(\mathbf{x},t) in Q2Q_2 again highlights the leading edge shear layer. In contrast, the quadrant Q3Q_3 for the cylinder-airfoil system (Figure 12(i–iii)-(e,f)), highlights

not only the shear layers on the airfoil lower surface, but also the shear layers on the cylinder surface, the features in the cylinder wake and the gap region between the airfoil and the cylinder.

From Figure 11(a)-(Figure 11b), we identified that the pitch-up phase is the energy extraction phase in Q2Q_2 for both airfoil-only and the cylinder-airfoil systems. For both these systems, during the pitch-up phase (t/T=4.75t/T = 4.75) (Figure 12(i–iii)-(d),(h)), CpC_p field shows a low pressure region on the airfoil upper surface. This low pressure on the upper airfoil surface in Q2Q_2 causes a positive pitch-up aerodynamic moment (CM>0C_M > 0). Since this occurs during the pitch-up motion θ˙>0\dot{\theta} > 0, we obtain a positive WQ2W_{Q_2} during the pitch-up phase. During the pitch-down phase (t/T=4.25t/T=4.25) (Figure 12(i–iii)-(c),(g)), a similar low CpC_p-region also appears in Q2Q_2. But since θ˙<0\dot{\theta} < 0 in this phase, this pressure loading now contributes to a negative energy transfer to the foil. Hence, the low CpC_p in Q2Q_2 is responsible for a negative WQ2W_{Q_2} during the pitch-down motion and positive WQ2W_{Q_2} during the pitch-up motion as seen in Figure 11(a),(Figure 11b). This is also reflected as negative w1\mathbf{w}_1 on the upper surface shear layer during the pitch-down phase (see zoomed-in inset in Figure 12(ii)-(a)), and similarly a positive w1\mathbf{w}_1 on the upper surface shear layer during the pitch-up phase (see zoomed-in inset in Figure 12(ii)-(b)).

For the airfoil-only system, CpC_p near the upper surface leading edge is lower during the pitch-down phase (t/T=4.25t/T = 4.25, Figure 12(i–iii)(c)) compared to the pitch-up phase (t/T=4.75t/T = 4.75, Figure 12(i–iii)(d)). This is very evident for M∞=0.8M_\infty= 0.8 as we see a lower CpC_p near the leading edge on the upper airfoil surface in Q2Q_2 in Figure 12(iii)-(c) compared to Figure 12(iii)-(d). Since CpC_p in Q2Q_2 is lower during the pitch-down motion, the negative energy transfer to the foil in the pitch-down phase is greater than the positive energy transfer in the pitch-up phase thereby yielding an overall negative EQ2∗E^{*}_{Q_2} for all the Mach numbers for the airfoil-only system as shown in Figure 9(c),(Figure 9f).

In contrast, for the cylinder-airfoil system, CpC_p near the airfoil leading edge in Q2Q_2 is lower during the pitch-up phase (t/T=4.75t/T = 4.75, Figure 12(i–iii)-(h)) compared to the pitch-down phase (t/T=4.25t/T = 4.25, Figure 12(i–iii)-(d)). As a result, unlike the airfoil-only system, the cylinder-airfoil system has a net positive EQ2∗E^{*}_{Q_2} (Figure 9(d),(Figure 9h)). This change in Q2Q_2 is due to the blockage effects induced by the cylinder which results in flow acceleration on the airfoil upper surface. During the pitch-down motion (t/T=4.25t/T = 4.25), the flow acceleration results in an increased flow separation on the upper surface for the cylinder-airfoil system (Figure 12(i–iii)-(e)) compared to the airfoil-only system (Figure 12(i–iii)-(a)). During the pitch-up motion (t/T=4.25t/T = 4.25), the flow acceleration creates stronger upstream propagation shock/compression waves on the upper airfoil surface as indicated by the lower CpC_p in Q2Q_2, in the cylinder-airfoil system (Figure 12(i–iii)-(h)) compared to the airfoil-only system (Figure 12(i–iii)-(d)). This effect is especially evident for M∞=0.6M_\infty= 0.6. The airfoil-only system (Figure 12(ii)) is entirely subsonic and and vortex formation on the airfoil surface is absent. On the other hand, the cylinder-airfoil system develops weak compression waves on the airfoil upper surface during the pitch-up motion (indicated by an arrow in Figure 12(i)-(Figure 12f,Figure 12h)). Thus, the cylinder-induced blockage effects, causes flow acceleration on the upper airfoil surface, results in a lower the pressure region during the pitch-up motion than pitch-down motion, resulting in EQ2∗>0E^{*}_{Q_2} > 0.

In Q3Q_3, the pitch-down phase is the energy-extraction phase for both systems (see Figure 11(d),(Figure 11e)). For both systems, a low CpC_p region near the airfoil leading edge in Q3Q_3 produces a negative aerodynamic moment, CM<0C_M < 0. This negative CMC_M during the pitch-down motion θ˙<0\dot{\theta} < 0 results in a positive energy transfer to the foil, WQ3>0W_{Q_3} > 0. Conversely, during the pitch-up motion when θ˙>0\dot{\theta} > 0, a similar negative aerodynamic moment results in negative energy transfer to the foil, WQ3<0W_{Q_3} < 0. For the airfoil-only system, CpC_p is lower near the airfoil leading edge in Q3Q_3 during the pitch-up phase (t/T=4.75t/T = 4.75, Figure 12(i–iii)-(d)) than during the corresponding pitch-down phase (t/T=4.25t/T = 4.25, Figure 12(i–iii)-(c)). This lower CpC_p in Q3Q_3 results in a larger negative energy transfer during the pitch-up phase than the positive energy transfer during the pitch-down phase. Hence, the airfoil-only system shows EQ3∗<0E^{*}_{Q_3} < 0 at all M∞M_\infty as shown in Figure 9(c),(Figure 9f).

For the cylinder-airfoil system (Figure 12(i–iii)-(g,h)), the addition of the cylinder results in a lower CpC_p in Q3Q_3 compared to the airfoil-only case (Figure 12(i–iii)-(c,d)). The cylinder shear and the wake introduce additional low-pressure regions as seen in CpC_p fields (Figure 12(i–iii)-(g,h)). Additionally, the gap region between the cylinder and the airfoil lower surface, causes the flow acceleration, also resulting in a lower CpC_p underneath the airfoil.

Additionally, for the cylinder-airfoil system, we observe that CpC_p on the airfoil lower surface in Q3Q_3 is lower during the pitch-down phase (t/T=4.25t/T = 4.25, Figure 12(i–iii)-(g)) than during the pitch-up phase (t/T=4.75t/T = 4.75, Figure 12(i–iii)-(h)). This difference is most evident at M∞=0.8M_\infty= 0.8 (Figure 12(iii)-(g,h)). At M∞=0.8M_\infty= 0.8, the cylinder wake shedding is suppressed and the wake region is elongated. The interaction region between the airfoil lower-surface shear layer and the elongated cylinder wake introduces a shock-train-like region. During the pitch-down phase, at t/T=4.25t/T = 4.25 (Figure 12(iii)-(e,g)), this interaction region is associated with predominantly positive w1\mathbf{w}_1 and a low CpC_p. During the pitch-up phase (t/T=4.75t/T = 4.75, Figure 12(iii)-(h)), the shock-train-like region is still present in Q3Q_3 but the CpC_p on the airfoil lower surface is higher than during the pitch-down phase (Figure 12(iii)-(g)).

The differences in the pressure loading on the airfoil surface in Q3Q_3 arise due to the asymmetry in interaction region for the cylinder-airfoil system. During the pitch-down motion, the incidence angle in the gap region results in a higher flow acceleration than during the pitch-up motion. Hence the CpC_p in the gap region in lower at t/T=4.25t/T = 4.25 is lower then t/T=4.75t/T = 4.75. This asymmetry in Q3Q_3 is also seen for M∞=0.6M_\infty= 0.6 and 0.70.7 (Figure 12(i–ii)-(Figure 12g,Figure 12h)). Since the cylinder recirculation region length is reduced as M∞M_\infty reduces, the gap region length also reduces for M∞=0.6M_\infty= 0.6 and 0.70.7 when compared with M∞=0.8M_\infty= 0.8 (also see Figure 10(g)-(Figure 10l)). At these M∞M_\infty, we also observe flow separation and shedding on the airfoil lower surface in Q3Q_3. For M∞=0.6M_\infty= 0.6 and 0.70.7, this shortened gap region coupled with flow separation also results in a positive CpC_p on the airfoil lower surface during the pitch-up phase (Figure 12(i–ii)-(Figure 12h)). Therefore, the asymmetry in the gap-region results in a lower CpC_p on the airfoil lower surface during the pitch-down motion, generating a net positive EQ3∗E^*_{Q_3} for the cylinder-airfoil system at all Mach numbers (Figure 9(d),(Figure 9g)).

Phenomenology-based partitioning of Q3Q_3 contributions

As highlighted above, the Q3Q_3 quadrant plays a dominant role in the energy transfer mechanisms of the cylinder–airfoil configuration. However, this quadrant encompasses several distinct flow phenomena, including the airfoil leading-edge shear layer, the cylinder shear layer, the wake–airfoil interaction region, and the shock-train structure. To isolate the individual contributions of these mechanisms to EQ3∗E^*_{Q_3}, we perform a phenomenology-based sub-regional power partitioning of the Q3Q_3 quadrant.

Figure 11(e) shows that the cylinder–airfoil system exhibits qualitatively similar trends in WQ3\mathbf{W}_{Q_3} across all M∞M_\infty. In particular, the M∞=0.8M_\infty= 0.8 case is especially useful because the cylinder wake shedding is largely suppressed, allowing the interaction-region dynamics to be isolated without the additional complexity introduced by large-scale wake vortices. We therefore employ this case to define the sub-regional decomposition of Q3Q_3 shown in Figure 13(a).

Sub-regional partitioning in $Q_3$ to isolate energy extraction contribution from distinct flow structures and phenomenologies

Figure 13. Sub-regional partitioning in Q3Q_3 to isolate energy extraction contribution from distinct flow structures and phenomenologies. (a) Instantaneous sub-region division of Q3Q_3 for a pitching cylinder-airfoil system at (M∞,Aθ)=(0.8,5∘)(M_\infty, A_\theta) = (0.8, 5^\circ). (b) Phase-averaged contribution from the shock-train, cylinder and airfoil shear layers to WQ3W_{Q_3}. (c) Cumulative integral of the asymmetric component WQ3W_{Q_3} for the subregions. (d)-(k) Power density fields visualized using w1w_1, ∥∇ρ∥\lVert\nabla\rho\rVert in gray and ∥∇ρ∥=3\lVert\nabla\rho\rVert= 3 contour line, and CpC_p-fields at highlighted time instances tit_i and T/2+tiT/2 + t_i in (c).

The Q3Q_3 quadrant is divided into five sub-regions associated with the dominant local flow features: the leading-edge airfoil shear layer (orange), the cylinder shear region (blue), the shock-train region (purple), the gap-flow region upstream of the shock train (cyan), and the region beneath the cylinder (yellow). The cylinder shear region and leading-edge shear region are identified by extruding the corresponding solid surfaces with an offset distance of 0.025c0.025c along the local surface-normal direction. The upper and lower boundaries of the shock-train region are defined by the airfoil lower-surface shear layer and the upper cylinder shear layer, respectively, where these shear layers are identified using the vorticity contours wz=±10w_z = \pm10. The left boundary of the shock-train region is determined from the intersection of the ∥∇ρ∥=3\lVert\nabla\rho\rVert= 3 contour with the identified cylinder and airfoil shear regions, while the right boundary coincides with the edge of the Q3Q_3 quadrant at x=0x = 0.

The phase-averaged contributions from these sub-regions to WQ3\mathbf{W}_{Q_3} are shown in Figure 13(b). We note that among the five sub-regions, the shock-train, cylinder-shear and the airfoil-shear regions contribute to positively to EQ3∗E^*_{Q_3}. The cylinder-shear region exhibits the largest variation through the cycle followed by the shock-train and airfoil-shear regions. Since we are interested in the net energy transfer to the airfoil for the pitching motion, we consider the cumulative integral of the asymmetric component of WQ3\mathbf{W}_{Q_3} in Figure 12(c). This asymmetric component is obtained by adding WQ3\mathbf{W}_{Q_3} at a pitch-down instant, tt, and the corresponding pitch-up instant, (T/2+t)(T/2 + t). The cumulative integral of WQ3\mathbf{W}_{Q_3} over the total Q3Q_3 region (black line in Figure 13(c)) converges to a lower value than the shock-train region (purple). This occurs because the other sub-regions corresponding to the upstream gap and the region underneath the cylinder contribute negatively to WQ3\mathbf{W}_{Q_3}. The shock-train region is the dominant source of positive energy transfer in Q3Q_3, accounting for nearly 56% of the total positive contribution. Furthermore, despite its greater distance from the airfoil surface, the cylinder shear-layer region provides the second-largest contribution at 43%, while the leading-edge shear region of the airfoil provides a surprisingly small contribution to the positive energy transfer in Q3Q_3. This phenomenology-based partitioning therefore reveals several counterintuitive features of the energy-transfer process: despite being closest to the airfoil surface, the leading-edge shear layer contributes only weakly to the positive power input, whereas the shock-train region and the more remote cylinder shear-layer structures emerge as the dominant contributors, thereby highlighting the unexpectedly strong influence of compressible gap-flow interactions on the flutter dynamics.

We consider a few time instances during the pitch-down and the corresponding pitch-up motion to further analyze the asymmetry. The power density fields w1\mathbf{w}_1, and the corresponding CpC_p fields at four time instances within the time range from t/T∈[4.2,4.35]t/T \in[4.2, 4.35] are shown in Figure 13(d)-(Figure 13g). The main difference between the pitch-down (Figure 13(d)-(Figure 13g)) and the pitch-up (Figure 13(h)-(Figure 13k)) phases in Q3Q_3 is the structure of the shock-train-like region and the flow separation on the airfoil lower surface. During the pitch-down motion (Figure 13(d)-(Figure 13g)), the cylinder shear layer region and the shock-train region predominantly result in a low CpC_p and a low CpC_p region on the airfoil lower-surface. The corresponding pitch-up instances, t/T∈[4.65,4.8]t/T \in[4.65, 4.8] (Figure 13(h)-(Figure 13k)), show a significant flow separation and shedding near the trailing edge during the pitch-up motion. During the pitch-up phase, at t/T=4.72t/T = 4.72 (Figure 13(h)), flow separation on the airfoil lower surface starts from the leading edge and we observe a vortex roll-up near the trailing edge. This flow separation then results in vortex shedding near the trailing edge at t/T=4.77,4.8t/T = 4.77, 4.8 (Figure 13(i,Figure 13j)). During this flow separation and shedding process on the airfoil lower surface (Figure 13(h)-(Figure 13j)), the CpC_p in Q3Q_3 is higher near the airfoil leading edge than during the pitch-down motion (Figure 13(d)-(Figure 13g)). The asymmetry in the airfoil lower-surface flow induced by the

cylinder between the pitch-up and pitch-down motions modifies the shock-train structure and the pressure loading, producing a net positive energy contribution from Q3Q_3. Overall, the sub-regional partitioning identifies the shock-train interaction region as the primary mechanism responsible for the positive net energy transfer (EQ3⋆>0E_{Q_3}^{\star} > 0).

Influence of cylinder placement

So far we consider a cylinder placed upstream of the pivot point. We now consider various horizontal placements of the cylinder relative to the pivot point on the airfoil to examine the effect of that parameter. The instantaneous vorticity-fields at θ=0∘\theta= 0^\circ during the pitch-down motion are visualized in Figure 14 for (M∞,Aθ)=(0.7,5∘)(M_{\infty}, A_{\theta}) = (0.7, 5^\circ). We observe that as the cylinder is moved downstream of the pivot point, the resulting E⋆E^{\star} is significantly reduced. This indicates that the downstream cylinder placement shrinks the flutter boundaries.

Instantaneous vorticity-fields of pitching cylinder-airfoil systems with different horizontal cylinder placements

Figure 14. Effect of underwing cylinder placement on energy extraction from fluid to the airfoil. (a)–(d) Instantaneous vorticity-fields of pitching cylinder-airfoil systems with different horizontal cylinder placements of xc0={−0.522,−0.25,0.25,0.4}x_{c0} = \{-0.522, -0.25, 0.25, 0.4\} and yc0=−0.18y_{c0} = -0.18 corresponding to (M∞,Aθ)=(0.7,5∘)(M_{\infty}, A_{\theta}) = (0.7, 5^\circ) at t/T=0.25t/T = 0.25. Flowfields are visualized using ωz\omega_z and ∥∇ρ∥\lVert\nabla\rho\rVert.

As the cylinder is moved downstream in the streamwise direction, the blockage effects from the cylinder that resulted in flow acceleration on the upper surface are reduced. This is evident as the flow separation region is moved towards the trailing edge as xc0x_{c0} is varied from -0.522 to 0.40 in Figure 14. Further, when the underwing cylinder is placed downstream of the pivot point, the shock-train and the gap region interact with the trailing edge vortex shedding. All these effects result in a significantly smaller flutter boundary as the underwing cylinder is placed downstream of the airfoil mid-chord.

Conclusions

We investigated the aerodynamic effects of wake–wing interaction on transonic flutter using high-fidelity simulations of a sinusoidally pitching NACA0012 airfoil with and without an underwing cylinder at a Reynolds number of 10,000. The simulations were performed using a sharp-interface immersed boundary method over a range of transonic Mach numbers.

Energy-map analysis showed that the addition of the underwing cylinder significantly broadens the flutter boundary relative to the airfoil-only configuration. Whereas flutter in the airfoil-only system is primarily associated with shock-induced trailing-edge separation, the cylinder–airfoil system exhibits substantially more complex flow interactions involving wake-induced acceleration, gap-flow dynamics, and shock-train formation.

A major contribution of this work is the extension of force and power partitioning frameworks to compressible flows with moving boundaries. The compressible power-partitioning framework enabled direct identification of the flow structures and mechanisms responsible for net energy transfer to the wing. The dominant contribution was found to arise from the volumetric Lagrangian rate of change of momentum term in the conservation of momentum equation. Through quadrant-wise and sub-regional partitioning, we showed that the cylinder fundamentally alters the flutter mechanism by introducing two coupled effects. First, blockage induced by the upstream cylinder accelerates the flow over the upper airfoil surface, enhancing energy extraction during the pitch-up phase. Second, the gap between the cylinder shear layer and the lower-surface leading-edge shear layer forms a confined acceleration region that generates a shock train. Among the identified structures, the shock-train region is the dominant source of energy extraction, followed by the cylinder shear layer. In contrast to the airfoil-only system, the cylinder wake also suppresses lower-surface shock-induced separation.

GridNxN_xNyN_yΔxmin⁡\Delta x_{\min}
Coarse8005200.0025
Medium12409600.00125
Fine180814540.0008

Table 1. Grid resolutions considered in the present study.

The study further showed that cylinder placement strongly influences flutter behavior. An upstream cylinder enhances instability and energy transfer to the wing, whereas placing the cylinder downstream of the pivot point substantially reduces the extracted energy. These findings demonstrate that upstream wake disturbances can strongly intensify transonic flutter through coupled vortex, shear-layer, and compressibility effects.

More broadly, the compressible force- and power-partitioning framework developed here provides a new physics-based tool for analyzing unsteady high-speed flows involving fluid–structure interaction. Beyond the present flutter problem, the methodology has potential applications in shock-boundary-layer interactions, turbomachinery flows, aeroelasticity, rotorcraft aerodynamics, propulsion systems, and other compressible flow configurations where coherent structures govern unsteady loading and energy transfer.

Acknowledgments

The authors acknowledge the support from the U.S. Air Force Office of Scientific Research (Grant number: FA9550-23-1-0010). We also acknowledge the U.S. Department of Defense High Performance Computing Modernization Program (HPCMP) for computational resources.

Declaration of interest

The authors report no conflict of interest.

Grid and statistical convergence

The grid verification test is conducted for the pitching cylinder-airfoil system at Re=10,000Re = 10,000, (M∞,Aθ)=(0.7,10∘)(M_{\infty}, A_{\theta}) = (0.7, 10^\circ). The details of the considered grids are presented in table 1. We consider the phase-averaged lift and drag coefficients, CLC_L and CDC_D, of the airfoil over 5 pitching cycles (figure 15(a), (b)). We observe RMS errors of 2.6% and 0.7% for the phase-averaged lift and drag coefficients between the medium and fine grid. We hence use the medium grid for the current study.

To examine the statistical convergence of the simulations, the cycle-to-cycle variation of E⋆E^\star and the five cycle moving average of E⋆E^\star for the pitching cylinder-airfoil system at (M∞,Aθ)=(0.7,10∘)(M_{\infty}, A_{\theta}) = (0.7, 10^\circ) are shown in figure 15(c). We consider the first four pitching cycles as transient and base all the analysis in the paper on the subsequent five cycles.

Effects Associated with Quadrants Q1Q_1 and Q4Q_4

In the main paper, we have focused on Q2Q_2 and Q3Q_3 which dominate the enhancement of energy transfer in the cylinder-airfoil system (see figure 9). For completeness sake, we present the effects associated with Q1Q_1 and Q4Q_4, which provide smaller but non-negligible contributions to the energy transfer.

For the flow over pitching airfoil-only system, shock-induced separation is the primary flutter instability mechanism and is seen in the positive EQ1⋆E^\star_{Q_1}, EQ4⋆E^\star_{Q_4} at M∞=0.7M_{\infty} = 0.7 and both Aθ=5∘A_{\theta} = 5^\circ and 10∘10^\circ (figure 9(c),(f)). The addition of an upstream underwing cylinder significantly reduces this energy transfer to the foil in Q4Q_4 (figure 9(e),(h)) as seen as the negative ΔEQ4⋆\Delta E^\star_{Q_4}. This shows that the upstream underwing cylinder wake suppresses the shock-induced stall mechanism in Q4Q_4. In Q1Q_1, ΔEQ1⋆\Delta E^\star_{Q_1} is negative for Aθ=10∘A_{\theta} = 10^\circ and negligible at Aθ=5∘A_{\theta} = 5^\circ indicating a more subtle effect on the shock-induced stall mechanism in Q1Q_1.

The phase-averaged WQ1W_{Q_1} and WQ4W_{Q_4} for the cylinder-airfoil system and the airfoil are plotted during the pitch-down and pitch-up motions respectively in figure 16(a)-(b). From the vorticity-fields (figure 16(c)-(d)), the flow separation occurs mainly near the trailing edge region for the airfoil-only system during the pitch-down (t1,t2t_1,t_2) and the pitch-up (t3,t4t_3,t_4) motions. For the cylinder-airfoil system, in Q1Q_1 the flow separation occurs more upstream near the leading edge than the

Grid and statistical convergence plots

Figure 15. Grid and statistical convergence. (a)–(b) Grid dependence test for pitching cylinder-airfoil system at (M∞,Aθ)=(0.7,10∘)(M_\infty, A_\theta) = (0.7, 10^\circ) using phase-averaged CLC_L and CDC_D of the airfoil. (c) Variation of E⋆E^\star for each pitching cycle.

airfoil-only system (Figure 16(c)). Additionally, several weaker vortices are shed when the upstream cylinder is present whereas we observe a more coherent roll-up and shedding in the airfoil-only system. This results in a decreased energy contribution from the shock-induced separation on the upper airfoil surface.

Effect of underwing cylinder on shock-induced separation

Figure 16. Effect of underwing cylinder on shock-induced separation in Q1Q_1 and Q4Q_4. (a)–(b) Phase-averaged WQ1W_{Q_1} and WQ4W_{Q_4} for airfoil and cylinder-airfoil system during the pitch-down and pitch-up motions, respectively at (M∞,Aθ)=(0.7,10∘)(M_\infty, A_\theta) = (0.7, 10^\circ) (c)–(d) Instantaneous vorticity-fields visualized using ωz\omega_z and ∥∇ρ∥\lVert\nabla\rho\rVert at highlighted time instances.

A significant difference due to the cylinder presence is observed in Q4Q_{4} (Figure 9(e),(Figure 9h)). Since the shock-induced separation for the airfoil-only system occurs during the pitch-down motion in Q4Q_{4}, the vorticity-fields during the pitch-down motion at t3t_{3} and t4t_{4} are depicted in Figure 16(d). The cylinder presence directly impacts the formation on the λ\lambda−shock wave system on the lower airfoil surface. The cylinder wake suppresses the shock-induced flow separation seen in the airfoil-only case, thereby resulting in a negative net energy transfer in Q4Q_{4} (Figure 9(g)) for M∞=0.7M_{\infty}=0.7, Aθ=10∘A_{\theta}=10^\circ.

References

  1. [1]B. H. K. Lee. Self-sustained shock oscillations on airfoils at transonic speeds. Prog. Aerosp. Sci., 37(2):147–196, 2001. doi:10.1016/S0376-0421(01)00003-3.
  2. [2]G. Schewe, H. Mai, and G. Dietz. Nonlinear effects in transonic flutter with emphasis on manifestations of limit cycle oscillations. J. Fluid. Struct., 18(1):3–22, 2003. doi:10.1016/S0889-9746(03)00085-9.
  3. [3]D. B. Kholodar, E. H. Dowell, J. P. Thomas, and K. C. Hall. Limit-cycle oscillations of a typical airfoil in transonic flow. J. Aircr., 41(5):1067–1072, 2004. doi:10.2514/1.618.
  4. [4]G. Dietz, G. Schewe, and H. Mai. Amplification and amplitude limitation of heave/pitch limit-cycle oscillations close to the transonic dip. J. Fluid. Struct., 22(4):505–527, 2006. doi:10.1016/j.jfluidstructs.2006.01.004.
  5. [5]O. O. Bendiksen. Review of unsteady transonic aerodynamics: Theory and applications. Prog. Aerosp. Sci., 47(2):135–167, 2011. doi:10.1016/j.paerosci.2010.07.001.
  6. [6]N. C. Lambourne. Control-surface buzz. Technical Report R&M 3364, HM Stationery Office, 1964.
  7. [7]P. Ghosh Choudhuri and D. D. Knight. Effects of compressibility, pitch rate, and reynolds number on unsteady incipient leading-edge boundary layer separation over a pitching airfoil. J. Fluid Mech., 308:195–217, 1996. doi:10.1017/S0022112096001450.
  8. [8]M. Yamasaki, K. Isogai, T. Uchida, and I. Yukimura. Shock-stall flutter of a two-dimensional airfoil. AIAA J., 42(2):215–219, 2004. doi:10.2514/1.9088.
  9. [9]J. P. Thomas, E. H. Dowell, and K. C. Hall. Modeling viscous transonic limit-cycle oscillation behavior using a harmonic balance approach. J. Aircr., 41(6):1266–1274, 2004. doi:10.2514/1.9839.
  10. [10]P. T. Karnick and K. Venkatraman. Shock-boundary layer interactions and energetics in transonic flutter. J. Fluid Mech., 832:212–240, 2017. doi:10.1017/jfm.2017.629.
  11. [11]T. C. Corke and F. O. Thomas. Dynamic stall in pitching airfoils: Aerodynamic damping and compressibility effects. Annu. Rev. Fluid Mech., 47:479–505, 2015. doi:10.1146/annurev-fluid-010814-013632.
  12. [12]J. M. Turner, J.-H. Seo, and R. Mittal. Sinusoidally pitching foils in transonic flow: Insights into flutter from time accurate simulations. AIAA J., 62(3):1148–1158, 2024a. doi:10.2514/1.J063142.
  13. [13]A. Ning, I. Kroo, M. J. Aftosmis, M. Nemec, and J. E. Kless. Extended formation flight at transonic speeds. J. Aircr., 51(5):1501–1510, 2014. doi:10.2514/1.C032385.
  14. [14]S. Sebastian and R. Rudnik. Experimental assessment of wing lower surface buffet effects induced by the installation of a UHBR nacelle. CEAS Aeronaut. J., 15:49–59, 2024. doi:10.1007/s13272-022-00632-z.
  15. [15]T. Lürkens, M. Meinke, and W. Schröder. Impact of 2d engine nacelle flow on buffet. CEAS Aeronaut. J., 15:23–35, 2024. doi:10.1007/s13272-024-00728-8.
  16. [16]C. J. Schauerte and A.-M. Schreyer. Experimental investigation on the turbulent wake flow in fully established transonic buffet conditions. CEAS Aeronaut. J., 15:125–147, 2024. doi:10.1007/s13272-023-00690-x.
  17. [17]J. Kleinert, J. Stober, and T. Lutz. Numerical simulation of wake interactions on a tandem wing configuration in high-speed stall conditions. CEAS Aeronaut. J., 14:171–186, 2023. doi:10.1007/s13272-022-00634-x.
  18. [18]J. Kleinert, M. Ehrle, A. Waldmann, and T. Lutz. Wake tail plane interactions for a tandem wing configuration in high-speed stall conditions. CEAS Aeronaut. J., 15:79–103, 2024. doi:10.1007/s13272-023-00670-1.
  19. [19]C. M. Jr. Denegri, J. A. Dubben, and D. L. Maxwell. In-flight wing deformation characteristics during limit cycle oscillations. J. Aircr., 42(2):500–508, 2005. doi:10.2514/1.1345.
  20. [20]C. M. Denegri, V. K. Sharma, and J. S. Northington. F-16 limit-cycle oscillation analysis using nonlinear damping. J. Aircr., 53(1):243–250, January 2016. doi:10.2514/1.C033315.
  21. [21]R. W. Bunton and C. M. Denegri. Limit cycle oscillation characteristics of fighter aircraft. J. Aircr., 37(5):916–918, 2000.DOI
  22. [22]P. S. Beran, N. S. Khot, F. E. Eastep, R. D. Snyder, and J. V. Zweber. Numerical analysis of store-induced limit-cycle oscillations. J. Aircr., 41(6):1315–1326, 2004a. doi:10.2514/1.404.DOI
  23. [23]G. H. Parker, R. C. Maple, and P. S. Beran. Computational aeroelastic analysis of store-induced limit-cycle oscillations. J. Aircr., 44(1):48–59, 2007. doi:10.2514/1.21051.DOI
  24. [24]M. Iovnovich, D. E. Raveh, D. Michaels, and M. Adar. Computational study of transonic limit cycle oscillation phenomenon on F-16 fighter aircraft. J. Aircr., 54(2):783–793, 2017. doi:10.2514/1.C033918.DOI
  25. [25]J. P. Thomas, E. H. Dowell, and K. C. Hall. Nonlinear inviscid aerodynamic effects on transonic divergence, flutter, and limit-cycle oscillations. AIAA J., 40(4):638–646, April 2002. doi:10.2514/2.1720.DOI
  26. [26]M. A. Padmanabhan, C. L. Pasiliao, and E. H. Dowell. Simulation of aeroelastic limit-cycle oscillations of aircraft wings with stores. AIAA J., 52(10):2291–2299, 2014. doi:10.2514/1.J052843.DOI
  27. [27]M. A. Padmanabhan, E. H. Dowell, J. P. Thomas, and C. L. Pasiliao. Store-induced limit-cycle oscillations due to nonlinear wing-store attachment. J. Aircr., 53(3):778–789, 2016. doi:10.2514/1.C033577.DOI
  28. [28]D.-H. Lee and P. C. Chen. Studies of aerodynamic influence of under-wing stores on flutter characteristics of F-16. In 50th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, 2009. doi:10.2514/6.2009-2463. AIAA Paper 2009-2463.DOI
  29. [29]C. M. Jr. Denegri, J. A. Dubben, and S. L. Kernazhhttpsky. Underwing missile aerodynamic effects on flight-measured limit-cycle oscillations. J. Aircr., 50(5):1637–1645, 2013.
  30. [30]M. Iovnovich. Effect of external stores aerodynamics on analytic fighter aircraft flutter predictions: evaluation of a superposition modeling approach. In International Forum on Aeroelasticity and Structural Dynamics, Savannah, Georgia, 2019.
  31. [31]G. Parker, R. Maple, and P. Beran. The role of viscosity in store-induced limit-cycle oscillation. In AIAA Struct. Struct. Dyn. Mater. Conf, Austin, Texas, 2005. doi:10.2514/6.2005-1916.DOI
  32. [32]W. Yuan, S. Rimple, and D. Poirel. Fully coupled aeroelastic analyses of wing flutter towards application to complex aircraft configurations. J. Aerosp. Eng., 34(2):04020117, 2021.DOI
  33. [34]T. Nakagawa, G. E. A. Meier, R. Timm, and H.-M. Lent. Compressible flows in the wakes of a square cylinder and thick symmetrical airfoil arranged in tandem. Proc. R. Soc. Lond. A, 411(1841):379–394, 1987. doi:10.1098/rspa.1987.0072.DOI
  34. [35]M. C. Jacob, J. Boudet, D. Casalino, and M. Michard. A rod-airfoil experiment as a benchmark for broadband noise modeling. Theor. Comput. Fluid Dyn., 19(3):171–196, 2005. doi:10.1007/s00162-004-0108-6.DOI
  35. [36]J. Boudet, N. Grosjean, and M. C. Jacob. Wake-airfoil interaction as broadband noise source: A large-eddy simulation study. Int. J. Aeroacoust., 4(1–2):93–115, 2005. doi:10.1260/1475472053730093.DOI
  36. [37]B. Greschner, F. Thiele, D. Casalino, and M. C. Jacob. Prediction of sound generated by a rod–airfoil configuration using easm des and the generalised lighthill/fw-h analogy. Comput. Fluids, 37(4):402–413, 2008. doi:10.1016/j.compfluid.2007.02.013.DOI
  37. [38]K. Menon and R. Mittal. Flow physics and dynamics of flow-induced pitch oscillations of an airfoil. J. Fluid Mech., 877:582–613, 2019. doi:10.1017/jfm.2019.627.DOI
  38. [39]K. Menon and R. Mittal. On the initiation and sustenance of flow-induced vibration of cylinders: insights from force partitioning. J. Fluid Mech., 907:A37, 2021a.DOI
  39. [40]C. Zhang, T. L. Hedrick, and R. Mittal. Centripetal acceleration reaction: an effective and robust mechanism for flapping flight in insects. PloS one, 10(8):e0132093, 2015.DOI
  40. [42]S. Prakhar, J.-H. Seo, and R. Mittal. Modal force partitioning—a method for determining the aerodynamic loads for decomposed flow modes with application to aeroacoustic noise. J. Fluid Mech., 1013:A7, 2025.arxiv.org/abs/2501.05331
  41. [43]L. Quartapelle and M. Napolitano. Force and moment in incompressible flows. AIAA J., 21(6):911–913, 1983a. doi:10.2514/3.8171.DOI
  42. [44]C.-C. Chang. Potential flow and forces for incompressible viscous flow. Proc. R. Soc. Lond. A, 437(1901):517–525, 1992. doi:10.1098/rspa.1992.0077.DOI
  43. [45]C.-C. Chang and S.-Y. Lei. On the sources of aerodynamic forces: Steady flow around a cylinder or a sphere. Proc. R. Soc. Lond. A, 452(1954):2369–2395, 1996. doi:10.1098/rspa.1996.0128.
  44. [46]C.-C. Chang, J.-Y. Su, and S.-Y. Lei. On aerodynamic forces for viscous compressible flow. Theor. Comput. Fluid Dyn., 10(1–4):71–90, 1998. doi:10.1007/s001620050051.DOI
  45. [47]J. Wu, X.-Y. Lu, and L.-X. Zhuang. Integral force acting on a body due to local flow structures. J. Fluid Mech., 576:265–286, 2007. doi:10.1017/S0022112006004551.DOI
  46. [48]B. Mele and R. Tognaccini. Aerodynamic force by lamb vector integrals in compressible flow. Phys. Fluids, 26(5):056104, 2014. doi:10.1063/1.4875015.DOI
  47. [49]B. Mele, M. Ostieri, and R. Tognaccini. Vorticity based breakdown of the aerodynamic force in three-dimensional compressible flows. AIAA J., 54(4):1198–1208, 2016. doi:10.2514/1.J054363.DOI
  48. [50]J. M. Turner, J.-H. Seo, and R. Mittal. A high-order sharp-interface immersed boundary solver for high-speed flows. J. Comput. Phys., 500:112748, 2024b.DOI
  49. [51]F. M. White and J. Majdalani. Viscous fluid flow, volume 3. McGraw-Hill New York, 2006.
  50. [52]S. K. Lele. Compact finite difference schemes with spectral-like resolution. J. Comput. Phys., 103(1):16–42, 1992.DOI
  51. [53]D. V. Gaitonde, J. S. Shang, and J. L. Young. Practical aspects of higher-order numerical schemes for wave propagation phenomena. Int. J. Num. Methods Eng., 45(12):1849–1869, 1999.doi.org/10.1002/(sici)1097-0207(19990830)45:12<1849::aid-nme657>3.0.co;2-4
  52. [54]B. Carnahan, H. A. Luther, and J. O. Wilkes. Applied Numerical Methods. Wiley, New York, 1969.DOI
  53. [55]A. Jameson. Numerical solution of the euler equations for compressible inviscid fluids. Numerical methods for the Euler equations of Fluid Dynamics, 1, 1985.
  54. [56]S. Kawai and S. K. Lele. Localized artificial diffusivity scheme for discontinuity capturing on curvilinear meshes. J. Comput. Phys., 227(22):9498–9526, 2008.DOI
  55. [57]J.-H. Seo and R. Mittal. A high-order immersed boundary method for acoustic wave scattering and low-mach number flow-induced sound in complex geometries. J. Comput. Phys., 230(4):1000–1019, 2011.pubmed.ncbi.nlm.nih.gov/21318129
  56. [58]N. Edgar and M. Visbal. A general buffer zone-type non-reflecting boundary condition for computational aeroacoustics. In 9th AIAA/CEAS Aeroacoustics Conference and Exhibit, page 3300, 2003.DOI
  57. [59]H Tijdeman and R Seebass. Transonic flow past oscillating airfoils. Annu. Rev. Fluid Mech., 12(1):181–222, 1980.
  58. [60]D. Canuto and K. Taira. Two-dimensional compressible viscous flow around a circular cylinder. J. Fluid Mech., 785:349–371, 2015.DOI
  59. [61]Y. Liu, Z. Ding, Y. Tao, J. Qu, X. Xie, and X. Qiu. Numerical study on compressible flow around a circular cylinder in proximity to the wall. Phys. Fluids, 35(6), 2023.DOI
  60. [62]K. Matsuo. Shock train and pseudo-shock phenomena in supersonic internal flows. J. Thermal Sci., 12(3):204–208, 2003.DOI
  61. [63]F. Gnani, H. Zare-Behtash, and K. Kontis. Pseudo-shock waves and their interactions in high-speed intakes. Prog. Aerosp. Sci., 82:36–56, 2016.DOI
  62. [64]B. Xiong, X. Fan, Z. Wang, and Y. Tao. Analysis and modelling of unsteady shock train motions. J. Fluid Mech., 846:240–262, 2018.DOI
  63. [65]Y. Zhu, Y. Su, and K. Breuer. Nonlinear flow-induced instability of an elastically mounted pitching wing. J. Fluid Mech., 899:A35, 2020.DOI
  64. [66]D. Fan, G. Jodin, T. R. Consi, L. Bonfiglio, Y. Ma, L. R. Keyes, G. E. Karniadakis, and M. S. Triantafyllou. A robotic intelligent towing tank for learning complex fluid-structure dynamics. Sci. Robot., 4(36):eaay5063, 2019.DOI
  65. [67]Y. Zhu and K. Breuer. Flow-induced oscillations of pitching swept wings: stability boundary, vortex dynamics and force partitioning. J. Fluid Mech., 977:A1, 2023.DOI
  66. [68]S. Kumar, J.-H. Seo, and R. Mittal. Computational modelling and analysis of the coupled aero-structural dynamics in bat-inspired wings. J. Fluid Mech., 1010:A53, 2025.
  67. [70]M. Aghaei-Jouybari, J.-H. Seo, J. Yuan, R. Mittal, and C. Meneveau. Contributions to pressure drag in rough-wall turbulent flows: insights from force partitioning. Phys. Rev. Fluids, 7(8):084602, 2022.DOI

Paper details

vlnc:W_je978j7p
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents