Theoretical and Computational Fluid Dynamics最新文献

This study numerically examines the influences of transverse annulation around a cone surface on the characteristics of a flow over an orthocone. This work is inspired by Spyroceras, a fossilized genus of nautiloid cephalopods from the Paleozoic era, whose method of locomotion is understudied. As a baseline case, a flow over a smooth orthoconic model with a blunt cone end is investigated numerically at Reynolds numbers from 500 to 1500. As Reynolds increases, two different shedding mechanisms—hairpin-vortex wake and spiral-vortex wake—are captured. We notice that an introduction of annulation over the cone surface changes the critical Reynolds number for the transition of the shedding mechanism. The dominant shedding frequency increases with the Reynolds number for the smooth and annulated cone flows. Moreover, the annulation reduces the dominant frequency for the same Reynolds number and increases the time-averaged drag coefficient. Modal decompositions—Proper Orthogonal Decomposition (POD) and Spectral Proper Orthogonal Decomposition (SPOD)—are used to capture the coherent structures and their oscillating frequencies. We have captured modes corresponding to the hairpin-vortex wake and spiral-vortex wake shedding mechanisms. Comparing the leading POD modes for the smooth and the annulated cone flows, we find that the annulation can reduce the twisting effects of the coherent structures in the wake. Additionally, we find that the SPOD analysis can identify modes presenting both hairpin-vortex wake and spiral-vortex wake in one flow condition as leading modes, while the POD leading modes only reveal one shedding mechanism in each flow.

This paper presents the morphological evolution characteristics of a droplet impacting a V-shaped wall by using the lattice Boltzmann method (LBM). Four parameters are investigated comprehensively. The parameters vary over wide ranges: surface wettability ((60^circ le theta ^{eq} le 120^circ )), Weber number ((102.27 le text {We} le 3681.82)), bending angle of the V-shaped wall (90(^circ le theta le 180^circ )), and eccentricity ratio (0 (le b le ) 0.5). Two types of collision are observed: deposition and breakage. For breakage, the number of satellite droplets increases against the increment of We. The splashing occurs for a high We. And the lamella ejection is observed on the hydrophilic wall and the neutral wall. The lamella ejection will be slight against the increase of (theta ^{eq}), while it will become obvious against the increment of (theta ). In addition, the nondimensional spreading length, width, and height are measured and analyzed. Regime maps are established based on We, Re, and (theta ).

Vortex asymmetry, dynamics, and breakdown in the wake of an axisymmetric cone have been investigated using direct numerical simulation for a wide range of angles of attack. The immersed boundary method is employed with pseudo-body-conformal grids to ensure the accuracy and resolution requirements near the body while being able to account for topology changes near the cone tip. The separated shear layer originated from the surface of the cone swirls into a strong primary vortex. Beneath the primary vortex on the leeward surface of the cone, a well-coherent counter-rotating secondary vorticity is generated. Beyond a particular threshold of swirl, the attached vortex structure breaks and the flow undergoes a chaotic transformation. Depending on the angle of attack, the flow shows different levels of instabilities and the topology of the vortices changes in the wake. In addition to swirl, spiral vortices that revolve around the primary vortex core often merge with the core and play a role in developing the double-helix mode of instability at the onset of the vortex breakdown. At the angle of attack of 60(^circ ), the time-averaged side force becomes asymmetric at the stage where the drag overcomes the lift. At the angle of attack of 75(^circ ), the primary vortex governs the flow asymmetry and the side force. Flow asymmetry is independent of the vortex breakdown. Finally, the contribution of primary vortices to the total forces is quantified using a force partitioning method.

Flow around a traveling wave-based surface-undulating NACA0012 hydrofoil has been numerically studied. In particular, we determine the effect of the phase speed of the wave, the phase difference between the waves traveling on the top and bottom surfaces, and the wave number on flow dynamics around and behind the hydrofoil and propulsive performance. The flow results in a vortex sheet or a street behind the hydrofoil, where oppositely signed vortices are aligned in either forward or reverse direction. Apart from these, side vortices start forming on either side of the hydrofoil at a higher wave number. The phase difference analysis between the upper and lower surface undulation reveals the configuration better for the hydrofoil’s lateral and longitudinal stability. The hydrofoil can shift from high thrust to high lateral force configuration by changing the phase difference between waves on the top and bottom surfaces. Thrust increases with an increase in the wave number, and a threshold value of phase speed and wave number exists where the drag-to-thrust transition happens. The added mass force-based scaling analysis corroborates with the simulated results.

A novel passive flow control strategy for the mitigation of transient separation and dynamic stall is demonstrated by means of high-fidelity large-eddy simulations. The control technique is based on a properly-sized micro-cavity cut into a wing’s underside near the leading edge, ahead of stagnation. This cavity remains essentially inactive at low incidence. However, as the wing effective angle of attack increases, the stagnation point displaces past the micro-cavity and the accelerating flow grazing the cavity induces a high-frequency resonance phenomenon or so-called Rossiter modes. The self-generated small-scale disturbances are carried around the leading-edge through the boundary layer to the wing’s upper side where the laminar separation bubble (LSB) amplifies these disturbances. This process delays LSB bursting and dynamic stall when the cavity size is selected such that its naturally occurring Rossiter modes are tuned to the receptivity of the LSB. Control effectiveness is explored for a harmonically pitching NACA 0012 wing section with freestream Mach number (M_infty = 0.2), chord Reynolds numbers (textrm{Re}_textrm{c} = 5 times 10^5), and maximum angle of attack of (18^circ ). The flow fields are computed employing a validated overset high-order implicit large-eddy simulation (LES) solver based on sixth-order compact schemes for the spatial derivatives augmented with an eighth-order low-pass filter. Despite its simplicity, the micro-cavity resonance is found to be highly effective in preventing the deep dynamic stall experienced by the baseline airfoil. A significant reduction in the cycle-averaged drag and in the force and moment fluctuations is achieved. In addition, the negative (unstable) net-cycle pitch damping found in the baseline cases is eliminated.

This paper considers the non-isothermal flow of a viscoplastic fluid on a horizontal or an inclined surface with a flat, a step-up and a step-down topography. A particular application of interest is the spread of a fixed mass—a block—of material under its own weight. The rheology of the fluid is described by the Bingham model which includes the effect of yield stress, i.e. a threshold stress which must be exceeded before flow can occur. Both the plastic viscosity and the yield stress are modelled with temperature-dependent parameters. The flow is described by a reduced model with a thin-film equation for the height of the block and a depth-averaged energy conservation equation for the heat transfer. Results show that for large values of the yield stress, only the outer fraction of the fluid spreads outward, the inner fraction remaining unyielded, hence the block only partially slumps. Conversely, for small values of the yield stress, the entire block of fluid becomes yielded and therefore slumps. We present an analysis which predicts the critical value of the yield stress for which partial slump occurs and how it depends on temperature.

Snoring and obstructive sleep apnea (OSA) are often associated with uvula vibrations and pharynx constrictions. However, successful treatment of snoring or accurate diagnosis of OSA has been proven challenging. This study aimed to identify acoustic indexes that were sensitive to underlying airway structural or kinematic variations. Six physiologically realistic models were developed that consisted of three pharynx constriction levels (M1-3) and two uvula-flapping kinematics (K1-2). Direct numerical simulations (DNS) were performed to resolve spatial and temporal flow dynamics, and an immersed boundary method was used to approximate the uvula vibrations. Time-varying acoustic pressures at six points in the pharynx were analyzed using different algorithms in frequency- or frequency–time domains. Signature flow structures formed near the uvula for different uvula motions and in the pharynx for different pharyngeal constriction levels. The fast Fourier transform showed that the acoustic energy was mainly distributed in four peaks (flapping frequency and three harmonics) with descending magnitudes. Their amplitudes and distribution patterns differed among the six models but were not substantial. The continuous wavelet transforms showed clearly separated acoustic cycles (in both frequency and time) in the uvula-induced flows and revealed a cascading bifurcation pattern in the input–output semblance map. Specifically, the multifractal spectrum was sensitive to uvula flapping kinematics but not pharynx constrictions. By contrast, the input–output cross-correlation and Hilbert phase space showed high sensitivity to pharynx constrictions but low sensitivity to uvula kinematics. The frequency–time analyses of DNS-predicted pressures offered insight into the acoustics signals that were not apparent in original signals and could be used individually or in combination in diagnosis or treatment planning for snoring/OSA patients.

Flow past a flexible filament, a two-dimensional splitter plate with negligible thickness, attached behind a circular cylinder is investigated. The Reynolds number based on the free-stream speed of incoming flow and diameter of the cylinder is (textrm{Re}=100). At this (textrm{Re}), the flow for a rigid filament is steady. However, a flexible filament undergoes flow-induced vibration for a range of reduced speed, (U^*), defined as inverse of the first nondimensionalized natural frequency of the filament. Over the wide range of (U^*) considered in this work ((U^*le 240)), it exhibits both flutter and vortex-induced vibration (VIV). Lock-in with various normal modes related to bending of the filament, each in a different regime of reduced speed, is observed during VIV. Interestingly, the fluid–structure system does not lock-in with the first normal mode of bending but with higher modes. The flow is steady for an extended range of reduced speed both before and after the lock-in with second mode. Two patterns of vortex shedding are observed. The (textsf{2P}) mode is associated with high-frequency vibration, while the (mathsf {2,S}) mode is observed during relatively low-frequency oscillation. A symmetry-breaking pitchfork bifurcation leads to static deflection of the filament during the first steady regime. The filament exhibits flutter response, at large reduced speed, with relatively low amplitude and frequency. No vortex shedding is observed during flutter. The fluid forces that cause flutter arise from asymmetry across the two sides of the filament in the zones of recirculation downstream of the cylinder. Comparison of the space-time patterns of energy transfer at the fluid–filament interface for flutter and vortex-induced vibration reveals that the energy transfer is much smaller during flutter compared to VIV. The point of maximum energy transfer is located close to the root of the filament in case of flutter, while it is near the tip during VIV.

Backpropagation of neural network models (NNMs) is applied to control nonlinear dynamical systems using several different approaches. By leveraging open-loop data, we show the feasibility of building surrogate models with control inputs that are able to learn important features such as types of equilibria, limit cycles and chaos. Two novel approaches are presented and compared to gradient-based model predictive control (MPC): the neural network control (NNC), where an additional neural network is trained as a control law in a recurrent fashion using the nonlinear NNMs, and linear control design, enabled through linearization of the obtained NNMs. The latter is compared with dynamic mode decomposition with control (DMDc), which also relies on a data-driven linearized model. It is shown that the linearized NNMs better approximate the systems’ behavior near an equilibrium point than DMDc, particularly in cases where the data display highly nonlinear characteristics. The proposed control approaches are first tested on low-dimensional nonlinear systems presenting dynamical features such as stable and unstable limit cycles, besides chaos. Then, the NNC is applied to the nonlinear Kuramoto–Sivashinsky equation, exemplifying the control of a chaotic system with higher dimensionality. Finally, the proposed methodologies are tested on the compressible Navier–Stokes equations. In this case, the stabilization of a cylinder vortex shedding is sought using different actuation setups by taking measurements of the lift force with delay coordinates.

The three dimensional modal linear stability of the radially stratified q-vortex is investigated. The presence of a radial density gradient in the vortex core biases the vortex stability features over the whole parameter space, i.e. varying the swirl number q, the axial k and azimuthal m wavenumbers and the density-to-vorticity radius ratio (epsilon ). The high swirl vortex, known to be stable in the constant-density situation becomes unstable to the Rayleigh–Taylor instability (RTI) with high amplification rates for vortex cores denser than the ambient. Hence we carry out a comprehensive stability analysis to measure the consequences of the onset of the RTI on the q-vortex linear stability. The damping effect of viscosity saturates beyond a threshold Reynolds number and we mean to address high Reynolds number situations such as those found in aircraft trailing vortices. Hence we place ourselves in the high Reynolds numbers regime for which vortices with a dense core exhibit a significant increase of the global maximum of the amplification rate. The effect of the radius ratio (epsilon ) is twofold. In the high swirl number regime where the homogeneous modes are stable or weakly amplified, the concentration of denser fluid at the vortex core promotes instabilities. In regions of the (k, q)-plane favouring both the homogeneous instability and the RTI mechanism, the amplification rate peaks for a radius ratio around (epsilon approx 2).