Closed-loop-manipulated wake of a stationary square cylinder

Vortex shedding from a fixed rigid square cylinder in a cross flow was manipulated by perturbing the cylinder surface using piezo-ceramic actuators, which were activated by a feedback hot-wire signal via a proportional–integral–derivative (PID) controller. The manipulated flow was measured at a Reynolds number (Re) of 7,400 using particle image velocimetry (PIV), laser-induced fluorescence (LIF) flow visualisation, two-component laser Doppler anemometry (LDA), hot wires and load cells. It is observed that the vortex circulation, fluctuating streamwise velocity, lift and drag coefficients and mean drag coefficient may decrease by 71%, 40%, 51%, 42% and 20%, respectively, compared with the unperturbed flow, if the perturbation velocity of the cylinder surface is anti-phased with the flow lateral velocity associated with vortex shedding. On the other hand, these quantities may increase by 152%, 90%, 60%, 67% and 37%, respectively, given in-phased cylinder surface perturbation and vortex shedding. Similar effects are obtained at Re=3,200 and 9,500, respectively. The relationship between the perturbation and flow modification is examined, which provides insight into the physics behind the observation.     

Transition phenomena in the wake of a square cylinder

The transition phenomena in the wake of a square cylinder were investigated. The existence of mode A and mode B instabilities in the wake of a square cylinder was demonstrated. The critical Reynolds numbers for the inception of these instability modes were identified through the determination of discontinuities in the St–Re curves, and were found to have mean values of 160 and 204 for the onset of mode A and B instabilities, respectively. The spectra and time traces of the wake streamwise velocity component were found to display three distinct patterns in laminar, mode A and mode B flow regimes. Streamwise vortices with different wavelength at various Reynolds numbers were observed through different measures. The symmetries and evolution of the secondary vortices were observed using laser-induced-fluorescent dye. It was found that, just like the case of a circular cylinder, the secondary vortices from the top and bottom rows were out-of-phase with each other in the mode A regime, but in-phase with each other in the mode B regime. From the flow visualization, it was qualitatively proven that there is stronger interaction between braid regions in the mode B regime. At the same time, analysis of PIV measurements quantitatively demonstrated the presence of the stronger cross flow in mode B regime when compared to the mode A regime. It suggests that the in-phase symmetry of the mode B instability is the result of strong interaction between the top and bottom vortex rows. It was also observed that although the vorticity of the secondary vortices in the mode A regime was smaller, its circulation was more than twice that of mode B instability. Compared to primary vortices, the circulations of both mode A and mode B vortices were much smaller, which indicates that the secondary vortices most likely originate from the primary vortices. The wavelengths of the streamwise vortices in the mode A and B regimes were measured using the auto-correlation method, and were found to be 5.1 (±0.1)D, 1.3 (±0.1)D, and 1.1 (±0.1)D at Re=183 (mode A), 228 and 377 (both mode B), respectively. From the present investigation, mode A instability was likely to be due to the joint-effects of the deformation of primary vortex cores and the stretching of vortex sheets in the braid region. On the other hand, mode B instability was thought to originate from the “imprinting” process.     

The near wake structure of a square cylinder

The near wake structure of a square cross section cylinder in flow perpendicular to its length was investigated experimentally over a Reynolds number (based on cylinder width) range of 6700–43,000. The wake structure and the characteristics of the instability wave, scaling on θ at separation, were strongly dependent on the incidence angle () of the freestream velocity. The nondimensional frequency (St_θ) of the instability wave varied within the range predicted for laminar instability frequencies for flat plate wakes, jets and shear layers. For = 22.5°, the freestream velocity was accelerated over the side walls and the deflection of the streamlines (from both sides of the cylinder) towards the center line was higher compared to the streamlines for = 0°. This caused the vortices from both sides of the cylinder to merge by x/d 2, giving the mean velocity distribution typical of a wake profile. For = 0°, the vortices shed from both sides of the cylinder did not merge until x/d 4.5. The separation boundary layer for all cases was either transitional or turbulent, yet the results showed good qualitative, and for some cases even quantitative, agreement with linearized stability results for small amplitude disturbances waves in laminar separation layers.     

Experimental investigation on wake characteristics behind a yawed square cylinder

The wake vortical structures of a square cylinder at different yaw angles to the incoming flow ($\alpha =$0°, 15°, 30° and 45°) are studied using a one-dimensional (1D) hot-wire vorticity probe at a Reynolds number (Re) of about 3600. The results are compared with those obtained in a yawed circular cylinder wake. The Strouhal number (St_N) as well as the mean drag coefficient ($C_{D_N}$), normalized by the velocity component normal to the cylinder axis, follow the independent principle (IP) satisfactorily up to $\alpha =$40°. Using the phase-averaging analysis, both the coherent and the remaining contributions of velocity and vorticity are quantified. The flow patterns of the coherent spanwise vorticity (${\omega }_z$) display obvious Kármán vortex streets and their maximum concentrations decrease as $\alpha$ increases. Similar phenomena are also shown in the coherent contours of the streamwise ($u$) and transverse ($v$) velocities as well as the Reynolds shear stress ($uv$). The contours of the spanwise velocity ($w$) and Reynolds shear stress ($uw$), however, experience an increasing trend for the maximum concentrations with increasing yaw angle. These results indicate an enhancement of the three-dimensionality of the wake and the reduction of vortex shedding strength as $\alpha$ increases. While general similarities to the wake behind a yawed circular cylinder are found in terms of flow features, some differences between the two wakes at different yaw angles are highlighted.     

Direct numerical simulation of turbulent wake behind a surface-mounted square cylinder

In this research, direct numerical simulation has been performed to study the turbulent wake behind a wall-mounted square cylinder with aspect ratio 4 and Reynolds number 12 000 (based on the free-stream velocity and obstacle side length) in a developing boundary layer. Owing to the relatively high Reynolds number and high aspect ratio of the cylinder tested, the wake is wide spread behind the cylinder and exhibits complex and energetic vortex motions. The lateral and tip vortex shedding patterns at different frequencies, coherent structures downstream of the obstacle, the production rate and distribution of turbulent kinetic energy, and the instantaneous pressure distribution in the wake region have been thoroughly investigated. In order to validate the numerical results, the first- and second-order flow statistics obtained from the simulations have been carefully compared against available wind-tunnel measurement data.     

Transition phenomena in the wake of an inclined square cylinder

The division of flow regimes in a square cylinder wake at various angles of attack (α) is studied. This study provides evidence of the existence of modes A and B instabilities in the wake of an inclined square cylinder. The critical Reynolds numbers for the inception of these instability modes were identified through the determination of discontinuities in the Strouhal number versus Reynolds number curves. The spectra and time traces of wake streamwise velocity were observed to display three distinct patterns in different flow regimes. Streamwise vortices with different wavelengths at various Reynolds numbers were visualized. A PIV technique was employed to quantitatively measure the parameters of wake vortices. The wavelengths of the streamwise vortices in the modes A and B regimes were measured by using the auto-correlation method. From the present investigation, the square cylinder wake at various angles of attack undergoes a similar transition path to that of a circular cylinder, although various quantitative parameters measured which include the critical Reynolds numbers, spanwise wavelength of secondary vortices, and the circulation and vorticity of wake vortices all show an α dependence.     

On the vortex dynamics in the wake of a finite surface-mounted square cylinder

The shedding process in the near wake of a surface-mounted, square cross-section cylinder of height-to-width aspect ratio 4 at a Reynolds number of 12,000 based on free-stream velocity and the obstacle width was investigated. The boundary layer thickness was 0.18 obstacle heights based on 99% free-stream velocity. The study is performed using planar high frame-rate particle image velocimetry synchronized with pressure measurements and hot-wire anemometry. Spatial cross-correlation, instantaneous phase relationships, and phase-averaged velocity data are reported. Two dominant vortex-shedding regimes are observed. During intervals of high-amplitude pressure fluctuations on the obstacle side faces, alternate formation and shedding of vortices is observed (regime A) similar to the von Kármán process. Regime B is characterized by two co-existing vortices in the obstacle lee throughout the shedding cycle and is observed within low-amplitude pressure fluctuation intervals. Despite the coexisting vortices in the base region, opposite sign vorticity is still shed out-of-phase downstream of this vortex pair giving rise to a staggered arrangement of counter-rotating vortices downstream. While the probability of occurrence of Regime B increases toward the free end, the amplitude modulation remains coherent along the obstacle height. Conditionally phase-averaged reconstructions of the flow field are consistent with the spatial distribution of the phase relationships and their probability density function. Earlier observations are reconciled showing that the symmetric shedding of vortices is a rare occurrence.     

Influence of wall proximity on characteristics of wake behind a square cylinder: PIV measurements and POD analysis

Influence of wall proximity on characteristics of the wake behind a two-dimensional square cylinder was experimentally studied in the present work. A low-speed recirculation water channel was established for the experiment; the Reynolds number based on the free-stream velocity and cylinder width (D) was kept at Re_D = 2250. Four cases with different gap width, e.g., G/D = 0.1, 0.2, 0.4 and 0.8, were chosen for comparison. Two experimental techniques, e.g., the standard PIV with high image-density CCD camera and TR-PIV with a high-speed camera were employed in measuring the wake field, enabling a comprehensive view of the time-averaged wake pattern at high spatial resolution and the instantaneous flow field at high temporal resolution, respectively. For the four cases, the difference in spatial characteristics of the wake in the vicinity of the plane wall was analyzed in terms of the time-averaged quantities measured by the standard PIV, e.g., the streamline pattern, the vector field, the streamwise velocity fluctuation intensity and the reverse-flow intermittency. The proper orthogonal decomposition (POD) method was extensively used to decompose the TR-PIV measurements, giving a close-up view of the energetic POD modes buried in the wake. The low-order flow model of the wake at G/D = 0.8 and 0.4 was constructed by using the linear combination of the first two POD modes and the time-mean flow field, which reflected well the vortex shedding process in the sense of the phase-dependent patterns. The intermittent appearance of the weakly separated region near the wall was found at G/D = 0.4. On going from G/D = 0.8 to 0.4, the remarkable variation of the instantaneous wake in the longitudinal direction confirmed that the wall constraint stretches the vortices in the plane of the wall and transfers the energy to the longitudinal component at the expense of the lateral one.     

Proper orthogonal decomposition of wall-pressure fluctuations under the constrained wake of a square cylinder

The proper orthogonal decomposition (POD) analysis of the wall-pressure fluctuations below the constrained wake of a two-dimensional square cylinder in proximity to a plane wall was made on two systems, i.e., G/D = 0.25 and 0.5, which corresponds to the wakes with and without suppression of the vortex shedding, respectively. Here, G is the gap distance and D is the width of the square cylinder. Synchronized measurements of wall-pressure fluctuations were made using a microphone array. For the system G/D = 0.5, the first two energetic modes contribute 34.7% and 23.4% to the total fluctuation energy, respectively; however, the fluctuation energy corresponding to the third mode are relatively small and less than 10%. This sharp variation in eigenvalue is due to the presence and dominance of the Karman-like vortex shedding. However, for the system G/D = 0.25, the considerable reduction in the eigenvalues of the first several modes is due to the suppression of the Karman-like vortex shedding. The spatial wavy pattern of the first several energetic eigenmodes was shown to be a good reflection of convective vortices superimposed in the wakes. The spectra of the POD coefficients determined the frequency of the dominant structures. Based on the coherence of the POD coefficients, an effective method of determining the number of POD modes for reconstruction of the low-order wall-pressure field was proposed. Accordingly, the low-order wall-pressure fluctuations in the systems G/D = 0.5 and 0.25 were reconstructed by using the first four and five POD modes, respectively. The coherence and cross-correlation analysis of the reconstructed wall-pressure fluctuations, which excluded the influence of the small-scale structures and background ‘noise’, gave an insight view of the footprints of the dominant flow structures, which otherwise could not be effectively captured by using the original wall-pressure fluctuations.     

Coherent vortical and straining structures in the finite wall-mounted square cylinder wake

Topological aspects of the turbulent wake of a finite, surface-mounted, square-cross-section cylinder of h/d = 4 are addressed by decomposing the velocity field into a quasi-periodic coherent part and the unresolved incoherent fluctuations. The three-dimensional large scale structure is educed through a reconstruction of planar phase-averaged PIV measurements using the simultaneously sampled surface pressure difference on opposing sides of the obstacle as a phase reference. A topological model for the vortex structure is educed and mean streamwise wake vorticity is explained in terms of the connections between initially vertical structures shed alternately from either side of the obstacle, rather than previously proposed ‘tip’ vortex structures generated at the obstacle free-end. The coherent structure educed accounts for a significant portion of the fluctuating energy in the wake. The turbulent field is further analyzed by finding Lagrangian straining structures that form by induction of the coherent vorticity field, and these structures are related to the energy transfer from the base phase-averaged flow since they act to stretch incoherent vorticity fluctuations in their neighbourhood.     

Investigation of the three-dimensional intermediate wake topology for a square cylinder at high Reynolds number

Simultaneous multi-point hotwire measurements are used to investigate the three-dimensional wake topology of a square cylinder at high Reynolds numbers. Wavelet techniques are applied to detect the flow structures and to inquire on the validity or extension of previously proposed low Reynolds number topological models to turbulent wakes. Our results suggest that a flow topological model similar to the horizontal perturbation model proposed by Meiburg and Lasheras (J Fluid Mech 190:1–37, 1988) but with alternate rib cuts in the horizontal plane is plausible for the intermediate wake topology.

---

Influence of wall proximity on vortex shedding from a square cylinder

Mean and fluctuating surface pressure data are presented for a square cylinder of side length D placed near a solid wall at Re _D=18,900. One oncoming boundary layer thickness, d=0.5 D was used. Measurements were made for cylinder to wall gap heights, S, from S/ D=0.07 to 1.6. Four gap-dependent flow regimes were found. For S/ D>0.9, the flow and the vortex shedding strength are similar to the no-wall case. Below the critical gap height of 0.3 D, periodic activity is fully suppressed in the near wake region. In between, for 0.3< S/ D<0.9, the wall exerts a greater influence on the flow. For 0.6< S/ D<0.9, the mean drag and the strength of the shed vortices decrease as the gap is reduced, while the mean lift towards the wall increases. Evidence is presented that for S/ D>0.6 the influence of the viscous wall flow in the gap is not dominant and that, consequently, inviscid flow theory can describe changes in the mean lift as S/ D decreases. For 0.3< S/ D<0.6, the flow reattaches intermittently on the bottom face of the cylinder and viscous effects become important. Below the gap height of 0.4 D, periodic activity cannot be observed on the cylinder.     

Influence of rounding corners on the wake of a finite-length cylinder: An experimental study

This study aims to investigate experimentally the influence of rounding corners (r) as well as aspect ratio (AR) on the flow structures of a surface-mounted finite cylinder. The cylinders with sharp (r* = r/D = 0) and rounded corners (r*=0.167, 0.25 and 0.5) and aspect ratio or height-to-width/diameter ratio (AR = H/D) between 2 and 7 are utilized. The experiments are based on the five-hole probe and hot-wire measurements as well as the oil flow visualization. Wake measurements are made in an open return wind tunnel at the Reynolds number, Re = 1.6 × 10⁴, where Re is defined based on the side width/diameter (D) of the cylinder cross-section and the freestream velocity. It is found that r* and AR have significant effects on the flow structure from the perspective of wake topology, strength of streamwise vortices, and vortex shedding frequency. For all r* considered, the wake is characterized by a quadrupole type (both the tip and base vortices are present) at AR = 7, while a dipole type occurs for AR = 2 and 4 (the base vortices are absent). The strength (circulation) of the streamwise vortex structures is affected by r*. For all AR examined in the present study, the strengths of tip and base vortex structures decrease with increasing r*. The oil flow visualization demonstrates that the features of the horseshoe vortex are sensitive to r* and AR. With increasing r*, the location of the separation line moves downstream and the distance between horseshoe vortex legs decreases. Velocity measurements reveal that the downwash flow enhances with increasing r*. It is also found that the Strouhal number increases progressively by 60% as r* increases from 0 to 0.5, regardless of AR.     

Turbulence properties in the cylinder wake at high Reynolds numbers

The present contribution analyses the turbulence properties in unsteady flows around bluff body wakes and provides a database for improvement and validation of turbulence models, concerning the present class of nonequilibrium flows. The flow around a circular cylinder with a low aspect ratio and a high blockage coefficient is investigated. This confined environment is used in order to allow direct comparisons with realisable 3-D Navier–Stokes computations avoiding ‘infinite’ conditions. The flow is investigated in the beginning of the critical regime at Reynolds number 140 000. The analysis is carried out by means of 2-D PIV, of 3-C PIV and of high-frequency 2-D PIV. The experimental analysis contributes to confirm the validity of advanced statistical turbulence modelling for unsteady flows around bodies.     

Dependence of the wake on inclination of a stationary cylinder

Three-dimensional vorticity in the wake of an inclined stationary circular cylinder was measured simultaneously using a multi-hot wire vorticity probe over a streamwise range of x/d = 10–40. The study aimed to examine the dependence of the wake characteristics on cylinder inclination angle α (=0°–45°). The validity of the independence principle (IP) for vortex shedding was also examined. It was found that the spanwise mean velocity which represents the three-dimensionality of the wake flow, increases monotonically with α. The root-mean-square (rms) values of the streamwise (u) and spanwise (w) velocities and the three vorticity components decrease significantly with the increase of α, whereas the transverse velocity (v) does not follow the same trend. The vortex shedding frequency decreases with the increase of α. The Strouhal number (St _N), obtained by using the velocity component normal to the cylinder axis, remains approximately a constant within the experimental uncertainty (±8%) when α is smaller than about 40°. The autocorrelation coefficients ρ _u and ρ _v of the u and v velocity signals show apparent periodicity for all inclination angles. With increasing α, ρ _u and ρ _v decrease and approach zero quickly. In contrast, the autocorrelation coefficient ρ _w of w increases with α in the near wake, implying an enhanced three-dimensionality of the wake.     

Three-dimensional phase dynamics in a cylinder wake

Recently there has been a new surge of interest in three-dimensional wake patterns, from both an experimental and analytical standpoint. One of the central discoveries is that the patterns of vortex shedding are dependent on the specific end conditions of a long cylinder span. However, a number of outstanding questions have remained unanswered, in part because techniques had not existed to control such patterns in a continuous fashion and from outside a test facility. In the present work, we have devised a method to control the end conditions of a cylinder span by non-mechanical and continuously-variable means, namely by the use of end suction. The technique allows a continuous variation of end conditions and admits transient or impulsive control. With the method, the classical steady-state patterns, such as parallel or oblique shedding or the chevron patterns are simply induced. These experiments demonstrate that the wake, at a given Reynolds number, is receptive to a continuous (but limited) range of oblique shedding angles (), rather than to discrete angles. There is excellent agreement in these results with the cos formula for collapsing oblique-shedding frequencies onto a single universal frequency curve. The use of suction has avoided the grossly unsteady motions at the ends of the cylinder span brought about by the wakes of mechanical end manipulators, and we show that the laminar shedding regime exists up to Reynolds numbers (Re) of 205. The surprisingly large disparity among reported measurements of criticalRe for wake transition (Re=140–200), over the last forty years, can now be explained in terms of spanwise end contamination.The control technique has also allowed experiments to be performed, which have resulted in the discovery of new phenomena such as phase shocks and phase expansions. A major difference between these phenomena is that phase shocks (involving regions of straight vortices) translate spanwise at constant speed, crossing the complete span in a finite time, whereas a phase expansion (involving curved vortices) requires an infinite time to complete its development across the span. These transient wake patterns are well illustrated using a simple model, based on experimental measurements, that thenormal wavelength for oblique or parallel vortices remains constant. However, a detailed and close comparison between our experimental results and those results from analytical modelling of the wake using Ginzburg-Landau modelling (in collaboration with Peter Monkewitz at Lausanne) is presently underway. These equations yield a Burger's equation for the spanwise wavenumber (or phase gradient), from which both (phase) shocks and expansions are well-known solutions.

---

Sommario Recentemente è rinato l'interesse per i modelli tridimensionali di scie, sia da un punto di vista sperimentale che da uno analitico. Una delle scoperte centrali è che le modalità di produzione dei vortici sono dipendenti dalle condizioni agli estremi di un lungo cilindro. Nel presente lavoro, è stato formulato un metodo per controllare le condizioni finali agli estremi di un cilindro per mezzo di variabili continue di significato non-meccanico, cioè per mezzo dell'aspirazione agli estremi. La tecnica permette una variazione continua delle condizioni agli estremi ed ammette controllo transitorio od impulsivo. Con questo metodo sono semplicemente indotti i classici modelli a stato fissato, come anche quelli che prevedono produzioni di vortici parallele od oblique o quellichevron. Questi esperimenti dimostrano che la scia, ad un dato numero di Reynolds, ammette una distribuzione continua (ma limitata) di angoli per distribuzioni oblique, piuttosto che una discreta. In questi risultati c'è un eccellente accordo con la formula del coseno per frequenze di produzioni oblique e collassanti su di una curva di frequenzauniversale. L'uso dell'aspirazione ha evitato i moti largamente instabili alle estremità del cilindro, provocati dalle scie delle parti terminali dei manipolatori, e si osserva che il regime laminare diffondente esiste oltre un numero di Reynolds pari a 205. La sorprendentemente larga disparità tra le misure di numeri di Reynolds critici, riportati per transizioni di scia (Re=140÷200) durante gli ultimi quattro anni, può essere ora spiegata in termini di contaminazione della estensione della lunghezza del cilindro.

     

Three-dimensional vorticity in a turbulent cylinder wake

The three components of the vorticity vector in the intermediate region of a turbulent cylinder wake were measured simultaneously using a multi-hot-wire probe. This probe has an improved spatial resolution compared with those reported in the literature. The behavior of the instantaneous velocity and vorticity signals is examined. Both coherent and incoherent vorticity fields are investigated using a phase-averaged technique. The iso-contours of the phase-averaged longitudinal and lateral vorticity variances, and , wrap around the spanwise structures of opposite sign and run through the saddle point along the diverging separatrix. The observation conforms to the previous reports of the occurrence of the longitudinal structures based on flow visualizations and numerical simulations. The magnitude of these contours is about the same as that of the maximum coherent spanwise vorticity at the vortex center, indicating that the strength of the longitudinal structures is comparable to that of the spanwise vortices. Furthermore, and exhibit maximum concentration away from the vortex center, probably because of a combined effect of the large-scale spanwise vortices and the intermediate-scale longitudinal structures. Coherent structures contribute about 36% to the spanwise vorticity variance at x/d=10. The contribution decreases rapidly to about 5% at x/d=40. The present results suggest that vorticity largely reside in relatively small-scale structures.     

Convection velocity measurements in a cylinder wake

The convection velocity of vortices in the wake of a circular cylinder has been obtained by two different approaches. The first, implemented in a wind tunnel using an array of X-wires, consists in determining the velocity at the location of maximum spanwise vorticity. Four variants of the second method, which estimates the transit time of vortices tagged by heat or dye, were used in wind and water tunnels over a relatively large Reynolds number range. Results from the two methods are in good agreement with each other. Along the most probable vortex trajectory, there is only a small streamwise increase in the convection velocity for laminar conditions and a more substantial variation when the wake is turbulent. The convection velocity is generally greater than the local mean velocity and does not depend significantly on the Reynolds number.Nomenclature d diameter of circular cylinder - f frequency in spectrum analysis - f _v average vortex frequency - r _v vortex radius - Re Reynolds number U _o d/v - t time - Th , Th , Th _r thresholds for _zp, , and r _v respectively - U _o free stream velocity - U ₁ maximum value of (U _o – U) - U _c convection velocity of the vortex, as obtained either by Eq. (1) or Eq. (2) - U _co convection velocity used in Eq. (3) U _cd, U _cu average convection velocities of downstream and up-stream regions respectively of the vortex - U _cv the value of U _c at y = 0.5 - u, v the velocity fluctuations in x and y directions respectively - U, V mean velocity components in x and y directions respectively - U,V U = U + u, V = V + v - x, y, z co-ordinate axes, defined in Fig. 1 Greek Symbols circulation - mean velocity half-width - x spacing between two cold wires or grid spacing - ₁, ₂ temperature signals from upstream and downstream cold wires respectively - v kinematic viscosity - _c transit time for a vortex to travel a distance x - phase in the cross-spectrum of ₁ and ₂ - _z instantaneous spanwise vorticity - _zc cut-off vorticity used in determining the vortex size - _zp peak value of _z - a denotes conditional average, defined in Eq. (12) - a prime denoting rms value