You are on page 1of 14

ELSEVIER

E x p e r i m e n t s on the Flow Past a Square Cylinder Placed near a Wall*


G. Bosch M. Kappler W. Rodi
Institute for Hydromechanics, University of Karlsruhe, Karlsruhe, Germany
Experiments are reported on the flow past a square cylinder placed near a wall at Re = 22,000 (based on the cylinder diameter, D). Visualization studies were carried out for various widths G of the gap between cylinder and wall. Over the range of the dimensionless gap width G / D = 0.35-0.5, the fraction of time with periodic shedding motion increases from zero to one. Below this range, the shedding motion is completely suppressed and, above it, regular shedding occurs at all times. Detailed two-component laser-Doppler velocimeter measurements were carried out for a gap width of G / D = 0.75. Phase-averaged statistics were obtained with a special signal-processing procedure; the phase relation was determined from a pressure signal measured on the cylinder side wall. In addition to timeaveraged mean-velocity and stress components, phase-averaged components are presented. A detailed comparison is made with the results obtained previously for the flow past a free-standing cylinder away from walls. The Strouhal number is roughly the same and so are many flow features, particularly in the base region near the cylinder, but further downstream the flow with a nearby wall develops a distinct asymmetry, with the vortex trajectories moving away from the wall and the wake developing an oblique character. Differences are noted between the inner and the outer vortices, and the greatest differences in the free-standing cylinder case are the considerably larger values of the u fluctuations in the near wake. The flow topology was found to be similar to that in the flow past two side-by-side cylinders. Elsecier Science Inc., 1996

Keywords: vortex shedding, wall-proximity effects, visualization, L D V measurements


INTRODUCTION The flow past long cylinders, usually associated with vortex shedding, can be affected strongly by the presence of a nearby wall, and in fact vortex shedding can be suppressed completely if the wall is close enough. Because vortex shedding causes dynamic loading on bodies and increased mixing in the wake, it is of considerable practical importance and so are the questions of whether shedding occurs or not and what the influence of wall proximity is on the shedding. Examples of engineering applications of this type of flow are tubes placed near heat-exchanger walls, vortex generators to improve the cooling in electronic circuits, wires near walls or chimneys near tall buildings, and pipelines near the ground or a sea or river bed with possible scouring. Most studies of the flow past cylinders placed near walls have so far been carried out experimentally for circular cylinders. G 6 k t u n [1] investigated this flow at cylinder Reynolds numbers in the range Re = 0.9-2.5 10 5 for various wall boundary-layer thicknesses smaller than the cylinder diameter, D. In this range, he found no influence of this quantity on the drag and lift coefficient. He measured a dimensionless frequency (Strouhal n u m b e r ) of St = 0.2 i n d e p e n d e n t of the gap width, G, between cylinder and wall. Bearman and Zdravkovich [2] investigated the influence of G on the vortex shedding and the spectral behavior of the wake as well as the pressure distribution on the wall and on the cylinder. They also visualized the flow in a smoke tunnel and determined the surface streak lines with an oil-flow technique. They carried out their measurements at Re = 5 10 4 and for a constant boundary-layer thickness ~/D = 0.8. They found that, for the dimensionless gap width G / D < 0.3, vortex shedding was suppressed and that, beyond this width, shedding

*This article is dedicated to Professor Peter Bradshaw on the occasion of his 60th birthday.

Address correspondence to W. Rodi, Institut fiir Hydromechanik, Kaiserstr. 12, D-76128 Karlsruhe, Germany.

Experimental Thermal and Fluid Science 1996; 13:292-305


Elsevier Science Inc., 1996 655 Avenue of the Americas, New York, NY 10010 0894-1777/96/$15.00 PI1 S0894-1777(96)00087-8

Flow Past a Square Cylinder near a Wall occurred at St = 0.2 + 0.04. Buresti and Lanciotti [3] observed vortex shedding past a circular cylinder for a gap width G / D > 0.4. The same authors [4] found only small influence of the boundary-layer thickness on the flow behavior in the range 0.11 < 6/D < 1.1. They also report that the mean and fluctuating forces are strongly dependent on G / D but that the Strouhal number is not. According to Grass et al. [5], the critical wall distance beyond which vortex shedding occurs is G / D --- 0.3 as long as the boundary-layer thickness is smaller than 6/D < 2.5. For 6/D > 3.5, they found a larger critical width of G / D ---, 0.5. In contrast with these results, Taniguchi and Miyakoshi [6] found from measurements of drag and lift forces and visualization pictures for the circular cylinder at Re = 9.4 10 4 that there is a clear correlation between the critical gap width and the boundary-layer thickness: the critical gap width is G / D .~ 0.3 for small boundary-layer thicknesses, but, starting from 6/D = 0.4, the critical gap width increases with boundary-layer thickness and has a value of about 0.8 at 6/D --- 1. As to a physical mechanism for the suppression of shedding, these authors suggest that vorticity in the shear layer separated from the lower side of the cylinder is absorbed by the opposite vorticity in the boundary layer of the nearby wall and that highly turbulent eddies from the outer part of the boundary layer penetrate into the separated shear layer and weaken or break the concentration of vorticity there. They also determined that, once shedding occurs, the Strouhal number is St = 0.2, virtually independent of the gap width. Concerning the conflicting findings about the influence of the boundary-layer thickness, it should be added that further parameters such as cylinder aspect ratio and blockage also influence the flow and shedding behavior and were quite different in the various studies. Also, if the Reynolds number is in the range of the drag crisis, the shedding behavior is very sensitive to the various parameters. Only a few studies were carried out for a square cylinder placed near a wall. Taniguchi et al. [7] determined for this geometry a critical gap width of G / D = 0.5 by examining the discontinuity of the Stouhal number. For larger gaps, he found a constant Strouhal number, St = 0.14. Durao et al. [8] performed two-component laser Doppler velocimeter (LDV) measurements for the square-cylinder flow at Re = 1.36 x 10 4. The blockage of the water tunnel by the cylinder was 12.8%, the cylinder aspect ratio 6, the thickness of the boundary layer at the location of the cylinder 6/D = 0.8, and the turbulence level of the approach flow 6%. By spectral analysis of the Doppler signals, these authors found that vortex shedding was absent for G / D < 0.35, whereas a predominant frequency was detected for G / D = 0.375, which for this and larger gap widths had a value of St = 0.133. For G / D = 0.25 and 0.5, detailed LDV measurements of the long-time averaged velocities were carried out, and velocity vector fields as well as contours of the total (including periodic and turbulent components) normal and shear stresses from fluctuations in the plane perpendicular to the cylinder are given. Because of the data-processing method used, no separation into periodic and turbulent fluctuations was possible in this study. Devarakonda and Humphry [9] measured force coefficients and the u-velocity component with a one-component LDV for a square cylinder at Re = 105, a blockage of 13.75%, a cylinder

293

aspect ratio of 7.27, and approach-flow turbulence level of 1.7%. The gap width was varied from G / D = 3.13 (cylinder in middle of channel) to G / D = 0.95. The drag coefficient was found to decrease and the lift coefficient to increase when the gap width was reduced. The Strouhal number determined from the velocity spectra increased from 0.136 for the largest G / D ( = 3.13) to 0.154 for the smallest G / D ( = 0.95). The present study is part of a series of systematic experimental investigations of the flow past square cylinders; the objective is the provision of detailed information for an understanding of the flow mechanisms and of detailed data for a thorough testing of calculation procedures. All measurements were carried out in the same water channel at the same Reynolds number (Re = 22,000), using the same LDV measurement technique. Lyn and Rodi [10] and Lyn et al. [11] examined the flow past a single cylinder placed midway between the tunnel walls (i.e., a virtually free standing cylinder). Kolar et al. [12] and Bosch [13] performed measurements in the flow past two cylinders placed side by side with a distance between them of two diameters, and the present experiments aim to complement these studies by providing similar detailed information for vortex-shedding flow under the influence of a nearby wall. A common feature of all these measurements, novel to previous measurements for similar geometrical configurations, is the use of a special signal evaluation, allowing phase averaging and hence a separation of fluctuations into periodic and stochastic turbulent components. Only this information allows full understanding of the complex flow phenomena and a thorough testing of unsteady calculation procedures simulating the vortex-shedding. In a complementary theoretical study, calculations of the present experimental situation have been carried out by Bosch and Rodi [14], providing further insight into the behavior of the flow. EXPERIMENTAL SETUP AND C O N D I T I O N S The experiments were carried out in a closed water channel with a cross section of 0.39 x 0.56 m, supplied by a constant head tank. The free-standing single-cylinder experiments (Lyn et al., Ref. 11) and the two-cylinder experiments (Kolar et al., Ref. 12) have been carried out in the same channel. To enable optical accessibility for the visualization and LDV measurements, an additional wall was placed inside the channel as shown in Fig. 1, which gives the geometrical details. The wall was equipped with a trailing-edge flap that was adjusted such (about 5 angle) that undisturbed flow around the leading edge and parallel flow along the plate resulted. The boundary layer developing on this wall was measured with the LDV in single-component mode. At the location of the cylinder (but without the cylinder present), the boundary layer was found to be turbulent and to have a thickness of ~ 0.13D. The side length of the square cylinder placed in the water channel was D = 4 cm, resulting in a blockage of 7.8%; the cylinder aspect ratio is 9.8. For visualization studies, the cylinder was placed at various distances G from the wall, whereas detailed LDV measurements were carried out only for the distance G / D = 0.75. The flow situation and the coordinate system used in this study are sketched in Fig. 1. (Note that Lyn et al. and Kolar et al. placed the origin of the coordinate system at the center of the

294

G. Bosch et al. bursts. Owing to slowness in data transfer between transient recorder and computer, the data rates were relatively slow, being about 1 Hz. Periodic vortex shedding was dominant in the case studied in detail in which time-averaged quantities do not provide sufficient information but in which ensemble- or phase-averaged quantities should be determined so that periodic and stochastic turbulent fluctuations can be separated. A measured instantaneous quantity f is therefore decomposed into

t~

pressure transducer

m e a n flow

"~:

""~

....... ........

:1
1

,,,

6D

,i

7.25 D

:i: 3 D ,i

f(t) = (f)(t)

+if(t) =f+

f(t) +if(t).

(1)

Figure 1. Experimental setup and coordinate system. The cylinder spanned the whole height of the channel.

if(t)
Here ( f } is the ensemble or phase average o f f , f is the time average (or average over all phases), f is the periodic component, f ' is the stochastic turbulent component of the fluctuation, and f " is the total fluctuation. For determining the averages at constant phase within a shedding cycle (quantities with ( ) ) , a special data evaluation procedure was used as described in detail in [10, 11]. A phase reference must be provided, and for this a low pass filtered pressure signal was taken that was measured at a tap at the midpoint of the cylinder side away from the wall (see Fig. 1). The time of occurrence of an L D V velocity realization was marked in relation to the pressure signal, permitting an association of the velocity data with a particular phase of the vortex-shedding cycle. All velocity realizations occurring within the same phase bin or interval constituted an ensemble at constant phase, such that statistics at constant phase could be evaluated. Twelve phase bins or intervals were used (cf. 20 bins in Lyn et al. and 8 bins in Kolar et al.), leading to a resolution of 0.044s in time or 30 in phase angle. Each shedding cycle had to satisfy certain criteria to be accepted for the averaging procedure. Based on the pressure signal, the shedding frequency should not deviate more than 20% from the average frequency (1.88 Hz) and the amplitude should have reached at least 70% of the average amplitude. If a cycle had to be rejected for these reasons, the next cycle also was discarded to avoid the influence of any disturbances from upstream. Collecting L D V signals was continued until each phase bin contained at least 400 samples. Because of low data rates (1 Hz), this usually took about 2 h or O(104) shedding cycles. As a consequence, the measurement mesh had to be fairly coarse. Owing to restrictions in optical accessibility, two-component L D V measurements could not be taken closer than 1 cm ( = 0.25D) to the walls. The actual measurement points are given by the origins of the velocity vectors shown in Fig. 7. Lyn et al. [11] performed a number of consistency checks and determined the accuracy of their measurement data. They estimated the uncertainty in the time-mean and phase-averaged velocity components and in the normal stress components to be less than 5% of Uref, whereas the uncertainty of the corresponding shear-stress is in the range 15-25% of the local value. Because virtually the same L D V system was used, the same accuracy can be expected in the present measurements. According to Eq. (1), the total time-mean stresses (e.g., u'tu 't) can be written as the sum of the time averaged

cylinder.) From volume flux measurements and from the velocity profiles measured with the LDV, the average approach-flow velocity was determined to be Uref = 0.54 m / s , so the Reynolds number is Re UrefD/p = 22,000. This is the same as in the previous studies [11, 12]. The turbulence level in the oncoming channel flow was r ~
=

Tu = ]/uT~/ure f = 3%. Measurements were taken at the upstream location x = - 3 . 5 D (see Ref. 15) where the velocity in front of the cylinder was 0.9ur~f and rose to 1.04Ure f further away from the wall and cylinder. The velocity defect at this station was therefore 0.14Ur~ f, which is larger than that for the free-standing cylinder. This is to be expected because the gap also has a blocking effect. The turbulence at this station was also already increased, probably owing to the velocity gradients present already, and it is clear that calculations in which a uniform velocity profile and the approach-flow turbulence as inflow conditions are used should start significantly further upstream [13, 141. In the remaining part of the paper, all distances are nondimensionalized by the diameter D and all velocities by the reference velocity Uref.
MEASUREMENT TECHNIQUES Flow visualization studies have been carried out with the use of two different techniques. In one technique, dye was injected near the front stagnation point from the cylinder; in the other, tracer particles were added at the pump entry and mixed homogeneously in the approach flow. In both cases, a laser sheet positioned at the central plane perpendicular to the cylinder was employed as illuminating light source, and the dye and tracer motion was recorded with a video camera. For G / D = 0.75, detailed measurements were obtained with a two-component LDV; in addition, cylinder-wall pressure and hot-film measurements at the edge of the wake were carried out to obtain spectral information and a phase relation for the L D V velocity measurements. The L D V system employed was an updated version of the one used in the studies of Lyn et al. [10, 11] and Kolar et al. [12]. It consists of a forward-scatter, two-channel system with Bragg cells for frequency shifting. The photomultiplier signals were filtered, amplified, and digitized by a transient recorder; the Doppler shift frequencies were determined by software processing of the digitized Doppler

Flow Past a Square Cylinder near a Wall periodic and turbulent stresses:

295

u"u" = ~ + ~Vh~"5.

(2)

It was found that the sum of the two terms on the right-hand side, d e t e r m i n e d as the mean values of the phase-averaged quantities, generally agreed within 4% with the total stress on the left-hand side, measured by global averaging of all (unselected) total fluctuations. This creates further confidence in the phase-averaging procedure. RESULTS AND DISCUSSION V i s u a l i z a t i o n S t u d i e s for V a r i o u s G a p W i d t h s Both visualization techniques described above were used to investigate the flow for gap widths G / D in the range 0.25-1.5. The observations that can be m a d e from the video recordings are summarized below. Visualization samples obtained with the dye m e t h o d are given in Fig. 2. It should be noted that the white area is not part of the

cylinder but the image of a metal piece for mounting the cylinder on the transparent side wall of the tunnel. A t G / D = 1.5, the wall has little influence and the flow is very similar to that past a free-standing cylinder; sometimes the inner vortices (shed from the cylinder side adjacent to the wall) can be seen to reach the wall where they are reflected. A t G / D = 0.75, there is still regular shedding but also already a fairly strong influence of the wall, causing an asymmetric behavior of the shedding motion. The inner, near-wall vortices a p p e a r to be smaller and get elongated in the streamwise direction so that they are oblique. The separated shear layer on the side of the cylinder is laterally squeezed in the gap and widened on the side away from the wall. A t some instances, the boundary layer appears to be sucked into the region where the counterrotating vortices have a c o m m o n upward motion; thereby it separates from the wall and it seems that at these instances a dead-water region exists briefly in the lee of the separation. F o r this gap width, additional visualization

c/D- 1.5
vortex shedding

( ; / -- 0.75 vortex shedding

G / D = 0.5 vortex shedding (at most times)

(;/D = 0.5
shedding disturbed/ non-periodic flow (at some inst,anccs)

G / D = 0.25 non-periodic flow

Figure 2. Flow visualization for G/D = 1.5, 0.75, 0.5, and 0.25 (dye added, 0.02-s exposure time). Mean flow is from the left-hand side. The additional wall is located at the lower boundary of the pictures.

296

G. Bosch et al. that measured by Durao et al. [8] for a square cylinder placed near a wall and by Lyn et al. [11] for a single free-standing cylinder (St -~ 0.133). However, the differences between all these measurements are not very significant.
Time-Mean Quantities

pictures obtained with the tracer method are shown below in Fig. 7. At G/D = 0.5, the effects of the wall described above are increased, but there is still mostly regular shedding; however, at certain instances, the shedding is already disturbed and breaks down. Visualization pictures of both flow states are shown in Fig. 2 for this gap width. At G/D = 0.375 (not shown), the bistable behavior is even stronger, and here shedding occurs only at a relatively small percentage of the time and the flow is mostly nonperiodic, separated turbulent flow with a fairly long recirculation zone. In the nonperiodic flow times, there is no sucking-in of the boundary layer, and there are large differences in the shear layers on the two sides of the cylinder. At the smallest gap width, G/D = 0.25, vortex shedding does not occur at any time and is hence completely suppressed in this case by the presence of the wall; however, the outer shear layer could be seen to undergo irregular low-frequency oscillations. The flow behaves similarly to that over an obstacle, with a bleed flow of wall-jet behavior near the wall, generated by the small gap. The observations from the present visualization study confirm the previous findings [7, 8] that the critical gap width is in the range G/D = 0.35-0.5. However, there is no sharp transition from purely nonperiodic flow to purely periodic flow at a particular gap width; rather, the time fraction when periodic shedding occurs changes from zero to one when G / D varies over the above-given range. Frequency analyses in the previous studies could not easily detect this transition behavior. The visualization study was taken as basis for the choice of the gap width for which detailed velocity measurements were to be carried out. A width of G/D = 0.75 was chosen because shedding occurred at all instances and there was a relatively strong effect of the wall on the flow behavior. Spectra a n d S h e d d i n g F r e q u e n c y Spectra from the pressure signal measured at the side wall and the hot-film signal measured in the wake (at x/D = 1.75, y / D = 2.0) are shown in Fig. 3. A clear peak can be seen at f = 1.88(+0.05) Hz, which gives a Strouhal number of St = f" D / u r e f = 0.139 + 0.004. This is virtually the same as that measured by Taniguchi et al. [7] for the same type of flow and by Kolar et al. [12] for the flow past two cylinders placed side by side; it is somewhat higher than

Figure 4 displays the measured contours of the components of time-mean velocity and the total stresses (including contributions from periodic and turbulent fluctuations). The corresponding plots from the measurements of Lyn et al. [11] for the free-standing cylinder are included for comparison (they are not taken from the paper of Lyn et al. but were produced from data provided by D. Lyn). Lyn et al. confined their measurements to one half of the flow region to reduce the measurement effort. For their results shown here (Figs. 4-6), contours in the other half were obtained by reflecting the measured flow about the center line, assuming symmetry of the flow. The contours of the longitudinal mean velocity ~ (Fig. 4a) clearly show the loss of symmetry due to the presence of the wall in the present experiments. The velocity minimum is shifted up from the center line of the cylinder, and hence it is not too meaningful to plot the velocity along this center line. In the base region, the flow is still reasonably symmetric and rather similar to that past the free-standing cylinder, but it has a larger reverse-flow velocity and a longer recirculation region (1.1D compared with 0.9D). For the conditions of G/D = 0.5, Re = 1.36 X 10 4, and Tu = 6%, Durao et al. [8] measured a length of the time-mean recirculation region of 1.9D, which, together with the visualization studies, indicates that the recirculation region lengthens as the gap width reduces. Beyond the recirculation region, the present experiments show a faster approach to the free-stream velocity than in the case of the free-standing cylinder. The main differences between the two cases occur, of course, in the wake region close to the wall. In the gap, the velocity rises up to 1.6 (only up to 1.3 in the flow around the free-standing cylinder) and a wall-jetlike flow develops downstream of the gap. Compared with the free-standing cylinder, the wall causes a shift of the ~ profile in the positive y-direction near the wall, which involves larger gradients, especially in the region between x = 2.5-3.5 for -0.05 < y < 0. Near the wall, the velocity has a plateau before it drops to zero at the wall. Unfortunately, two-

EFT hotfilmisignal FFT pressure~ignal --~ ..... 0.1 ,6 E 0.001


o~

0.01

! i.....'.': ........ 0.0001 ........... ..................................................................


- " ~ .......... A ;"L"

Figure 3. Spectra of hot film (at x = 1.75, y = 2.00) and pressure signal, G/D = 0.75. Position of the pressure transducer is shown in Fig. 1.

0.1

0.2

0.5

10 f[Hz]

Flow Past a Square Cylinder near a Wall

297

present
1.5 I

L y n e t al.

1.3-1.2 ....... 1.1 - 0.9 ...... 0.8 . . . . . 0.7 - - 0.6 . . . . 0.5-0.4---0.3 ....... 0.2-0,1-0 .......

t / ~ ~ ~
~
0 0.5 1 1.5 2 2.5 3 3.5 "...... 4
.... ". :.

:.-:-----'~'-'"" ......................... " !~ ~' ~J~') ~-" ':.'f ~ i .................. "~ - "-)~ -.-~ - ....... ........... -"

~o

. . . . . . . . "+"

Y
~)

0.5

-0.5

-0.1 . . . . -0.2--0.4~ 0.35 - ~ 0.3 ........ 0.25 - 0.2 . . . . 0.15 ....... 0.1 . . . . . 0.05---0.05 ~
-0.1 - -

0.5

~..--%.Z?-:o . . . . -:':: . . . . . .................. . ..'<,,~... . . . . . . . . . . . . . . . . . . . . . ,. 1.5 2 2.5 3 35


,

\it~ ::; ~ ~ t/ / ,z

.,,"..

b)
0.5

N
".'."/I~, \ V~,~--::..~"-'~::-:--:":':
""7.//i
0.5 1 1.5 2 2.5 3 3.5 4 0.5
/

0
-0.5

[]

-0.15 ........ -0.2~ -0.25 -0.3 ....... -0.35 ...... -0.4 . . . . . 0.28 - 0.26 ~ - 0.24 ....... 0.22 - 0.2-~ 0.18 ....... 0.16 . . . . 0.14 - - 0.12 ----"
0.1 0.08 - - - 0 . 0 6 ........ 0.04~

I ','C,. 1

'-.. .................

i
3 3.5

1.5

2.5

C) 0.5

-0.5

0.02 . . . . .

~ . ~ . ~ . ~ . ~
0.5 1 1.5 2 2.5 3 3.5 4

/
0.~ ,
t

115

L5

3'.~
I

0.65 0.6 " - ~ 0.55 ....... 0.5-0.45 . . . . . 0.4 ....... 0.35 . . . . . 0.3 ....... 0.25 ...... 0.2~ 0.15 . . . . . 0.1 ........ 0.05

00.5

._~.....:

-0.5 -1

0.5 0.12 0.1 - ~ 0.08 ........ 0.06-0.04---0.02 .......


0 ......

1.5

2.5

3.5

4 0

0.5

1.5

2.5

3.5

4
1

e) 0.5 %";. ...........


r/i///lll/A ~ ~ . "
........

.02 ..... 0 -0.04 .... -0.06 ~ - 0 . 0 8 -----

-0.5 -1

-0.1 ........ -0.12 - -

0.5

1.5

2.5

3.5

0.5

1.5

2.5

3.5

4 37

Figure

4. Contours

of time-mean

velocity

and

stress

components;

comparison

of present

data

with

data

of Lyn et al. for free-standing

cylinder.

298

G. Bosch et al. 14) and the interaction with velocity gradients there contribute to increased turbulence production. The ~,'tc't contours (Fig. 4d) are much more symmetric and not so different from the ones for the free-standing cylinder; the fluctuation levels are similar, but somewhat higher, and the variation with x also is similar. Of course, significant differences can be found in the near-wall region where t,'% ''t levels off (before it goes to zero right at the wall). As in the case of the free-standing cylinder, the 2, fluctuations are considerably higher than the u fluctuations and have their maximum near the center line of the cylinder. This behavior stems mainly from the periodic component, which contributes about 50% of the total fluctuations. The shape of the distribution of the periodic fluctuations (not shown here) is similar to that of the total fluctuations; only near the wall the turbulent fluctuations dominate as the larger-scale periodic fluctuations are suppressed at larger wall distances by the presence of the wall and go to zero at y = - 1 . The total shear stress u'tt ,'~ shown in Fig. 4e also has a clear lack of antisymmetry in the present experiments, with the zero shear-stress line moving upward away from the center line of the cylinder. Similar maximum values can be found, as in the case of the free-standing cylinder, but there is a distinct shift of the profile, especially in the region further downstream where there is a larger area of significant positive shear stress. This is also the area where the ~ gradients are larger than in the free-standing cylinder case, which would lead to increased u' production and could expla!n the larger u' levels. As can be seen from the profiles of tt'tU 't and ~ plotted in Fig. 6 at x = 1.5 and 3.5, the main contribution to the total shear stress stems from the periodic fluctuations. At x = 1.5, the shear stress in the free-standing cylinder case has a double peak behavior that is caused by the periodic fluctuations. In the present experiments, the total shear stress does not have this behavior but follows a rather fiat curve; there is, however, a trend toward this behavior in the fluctuating component. The profiles at x = 3.5 clearly show the larger positive shear stress in a sizeable region of the inner wake (i.e., the part near the wall).

component L D V measurement points could not be placed in the boundary-layer region. Figure 4b shows that the mean lateral velolcity P is not antisymmetric with respect to the cylinder center line, as in the case of the free-standing cylinder. Just behind the cylinder, the behavior is similar for the two cases; but, further downstream, the lateral motion is larger for the near-wall cylinder because of the longer recirculation region and because of a faster "filling" of the wake by the motion toward the wake center. The contours of the streamwise fluctuations u'tu 't (Fig. 4c) are strongly asymmetric except close to the cylinder, with higher values in the near-wall part of the wake. Here, and at larger downstream distances also on the other side of the wake, the u'tu" values are considerably larger than in the free-standing cylinder case. They do not decay by far as strongly as in that case. Figure 5 displays u'tu 't profiles at x = 1.5 and 3.5 and shows not only this total fluctuating component, but also the periodic contribution tiff. According to Eq. (2), the turbulent contribution is then the difference between the two. At x = 1.5, one can see a shift on the inner near-wall side compared with the profile for the free-standing cylinder, but the level of the fluctuations is roughly the same and so is the shape of the profile, which has a minimum near the point of the maximum velocity deficit. The periodic fluctuations can be seen to contribute about 37% in the peak regions, whereas, on the center line, this contribution is nearly zero because there the periodic fluctuations are small (see Fig. 10). At x = 3.5, the u'tu 't profile is strongly asymmetric and the values are considerably higher than in the free-standing cylinder case. The contribution of the periodic fluctuations is now insignificant (i.e., the u-fluctuations are here mainly turbulent). A possible reason for the considerably higher turbulent u fluctuations in the case with wall effects is the presence of sharper velocity gradients, which together with higher shear stress (see Fig. 4e and discussion below) lead to an increased production of turbulent u' fluctuations. Perhaps the engulfment of flow carrying turbulence from the boundary layer into the wake (as can be clearly seen in the calculations of Bosch and Rodi, Ref.

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0

/ /

z=l.5

0.4 0.35 0.3 0.25

0.2 0.15

/ i

"X
|

/ / / / / /
/ | -1 -0.5 0

x=3.5
present
Lyn Lyn
u n It, tt

present fi6.
u ~tu n mirrored

-~....
-g ....

\\ j/
|

0.1 0.05 0 0.5 1 1.5

...= = -..u...,j~_ i . - w - . , * ~ . . m . . m , . m . ,
Im .. Ill" "uw "ll . . . . . . ,I1" m
.

L , i i i ,. -, I " ' " - .

-I

43.5

0 Y

0.5

| 1.5

Figure 5. Profiles of time-mean total normal stress u ' t u " free-standing cylinder case of Lyn et al. [11].

and periodic stress tiff; comparison with

Flow Past a Square Cylinder near a Wall x=l.5

299

0.15

/ / / 7 / / /
I !

0.15

0.I

0.I

/ / ;/
/ /
| !

x=3.5
p r e s e n t ue..~tvet present ~ L y n u et v et Lyn mirrored
--~-" ~ .... x

0.~

0.05

..................................... :,a

-0.05

J'~~

.0.05
-0.1

~.I

-0.15

'

-0.15

-1

-0.5

0 Y

0.5

1.5

-1

-0.5

0 Y

0.5

! .5

Figure 6. Profiles of time-mean total shear stress u'tv't and periodic stress fit3; comparison with free-standing cylinder case of Lyn et al. [11]. The contours of the three total stresses presented by Durao et al. [8] for their measurements at G / D = 0.5 show similar trends concerning the asymmetry and the oblique behavior of the wake, but the values of the stresses are considerably lower. This difference may be caused by partial suppression of vortex shedding by the wall at the smaller gap width, perhaps including periods of nonperiodic flow. better discerned when the flow field is shown with respect to an observer moving with the convection speed of the vortices. Such vector plots have been published already in [14] and were used to determine the location of the saddles. In various figures, the vortex centers are indicated by + and the possible saddles by . The location and importance of the two different kinds of critical points have been discussed at length in Lyn et al. [11] and Kolar et al. [12], and in the latter work a different flow topology with respect to the location of the saddles has been worked out between the single- and two-cylinder cases. In the latter, there are two saddles at the foot of the outer vortex (shedding from the side away from the symmetry plane), located on the symmetry plane. It appears that a similar situation prevails in the present experiment with two saddles on the foot of the upper vortex with negative vorticity. This can be seen best in the picture of 270 phase angle. The close correspondance between the twocylinder case and the near-wall-cylinder case has already been shown in Bearman and Zdravkovich [2]. At phase-angle 0 , a vortex with counterclockwise rotation can be seen to shed on the lower side; it develops a somewhat elongated form with an inclined axis. A counterrotating vortex that has shed earlier from the upper side can be seen further downstream at x = 3, and this is nearly circular in shape and still fairly strong. At the foot of this vortex, two saddle points are likely to be present. At phase-angle 90 , the vortices from the previous picture have moved downstream, the inner one with counterclockwise rotation with its center now at x ~ 2 and more elongated and stretched across the wake; the vortex with clockwise rotation just moves out of the picture. A new vortex with its clockwise rotation is just formed past the upper side of the cylinder. The saddle points have moved with the vortices. At phase-angle 180 , the inner vortex with counterclockwise rotation has moved with its center to x ~ 3.2 and is now much elongated and hence stretching across the whole wake, and a new vortex with clockwise rotation is just shedding from the upper side and is still very strong. At phase-angle 270 , the inner vortex with counterclockwise rotation is leaving the picture and the counterrotating vortex shed previously from the upper

Phase-Averaged Fields and Vortex Motion


Figure 7 shows visualization pictures obtained with the tracer technique, measured velocity vectors, and contours of the vorticity ( t o ) = c ~ ( v ) / O x - O ( u ) / O y at the four phases 0, 3, 6, and 9 (0 , 90 , 180 , and 270 phase angle). The visualization pictures were chosen from the video film such that the motion shown in each picture corresponds as closely as possible to these phases. Digital image processing has been utilized to improve the visual clarity. This was necessary because of the nonhomogeneous light intensity of the laser light sheet. The increments in the vorticity contour lines are 0.2, and negative vorticity (clockwise rotation) is indicated by dashed lines. The vorticity information is somewhat crude because of the coarse measurement grid (the velocity vectors are given for each measurement point), introducing uncertainties in the derivatives and hence the vorticity values. However, the contours still clearly indicate the trends and the location of the vortices as well as their movement downstream. Generally quite good correspondance with regard to the location of the vortices can be seen between the various pictures--closer correspondance should not really be expected. In particular, the vortex centers as seen in the visualization pictures do not need to have the same lateral position as the vorticity peaks; the difference will be discussed further below. The vortex centers (characterized by closed streamlines) and vorticity peaks can be detected, and saddles characterized by the intersection of streamlines also can be discerned in the flow visualization pictures and vector plots, albeit with considerable uncertainty. It should be mentioned that the saddles can be

300

G. Bosch et al.
r-----------

y
1 1

y
0.5 0 -0.5

.~
.,

II1__
+

+ T r.r i

t L/,

0.5

''
o

~ ~ i

I /ill.J....

1 1 l +/I//II...t

.."77.," 71
....,,...~'--..,-,-.~ ~ ~_ __

.t. . . . .
~ ~ .

-0.5 -1

y
1

y
1

0.5 o -0.5 -1

0.5 o -0.5 -1

0.5

1.5

2.5

3.5 X

0.5

1.5

25

3.5 27

a) 0

b) 90

Figure 7. Phase-averaged velocity vectors in reference frame of cylinder; flow visualization (tracer particles, exposure time 0.02 s); contours of ( w ) at various phases. (Continued)

side has now moved with its center to x --- 2.1; it is fairly circular and still very strong. O n the lower cylinder side, a new vortex is already forming, which is soon to shed. The two saddles on the foot of the strong outer vortex with clockwise rotation can be discerned. The location of the vorticity peaks at the various phases is shown in Fig. 8a, which gives an indication of the trajectory of these peaks. Unfortunately, the scatter and hence the uncertainty are rather large, as can be seen from the location of the peaks at x = 2 resulting from two different phases. Similar scatter, particularly in the base region, was also observed by Lyn et al. [11] in the evaluation of their results. As o p p o s e d to the free-standing cylinder, the vorticity peaks for the vortices shedding on the u p p e r and lower side of the cylinder follow different trajectories, as in the case of the flow past two cylinders studied by Kolar et al. [12]. The trajectory of the peaks of the inner vortices is about 0.5D closer to the wall than the trajectory of the peaks of the outer vortices. Initially,

within the experimental scatter, the peaks move horizontally; but, starting at x -- 3, they move away from the wall owing to the interaction with the wall, causing a displacement. As m e n t i o n e d above, there is some difference between the lateral location of vorticity peaks and the centers of vortices, as observed from the visualization pictures in Fig. 7. The trajectories of the vortex centers, as determined from the video film, are also included in Fig. 8a. The trajectories from inner and outer vortices virtually coincide, and the vortices can be seen to move first fairly close along the center line and then upward away from the wall. The streamwise location of the vorticity peaks as a function of time (normalized such that the slope gives the streamwise convection speed directly) is shown in Fig. 8b. Except for the immediate base region, within experimental scatter both inner and outer structures move downstream with the same velocity Uc = 0.8. This is close to Uc found by Lyn et al. [11] for the free-standing cylinder

Flow Past a Square Cylinder near a Wall 3111

iy
0.5 0 , .....--...--"..-""..--"I -0.5 -1

ig
0.5

\ \ ~ ~

\ \\~..,-..,

0 -0.5 -1

y
1
0.5 0 -0.5 -1

Y
1 0.5

V////////A

~:~f.':d'.:('i:!:~i(;~:).i'...;/.).: //
0 -0.5 -1

0.5

1.5

2.5

3.5
X

0.5

1.5

2.5

3.5
X

c) 180

d) 270
Figure 7. (Continued)

(Uc = 0.78). In the immediate base region, the convection velocity can be seen to be somewhat smaller. In Figure 9, the streamwise variation of the peak vorticity (OJ)p at the various phases is plotted. The results from the free-standing cylinder experiments of Lyn et al. [11] are included. The decay behavior, which was discussed at length in Lyn et al., is similar for the two cases. First, the vorticity decays rather strongly, and then further away from the cylinder the decay slows down. Near the cylinder, the vortices shedding from the two sides have the same magnitude of peak vorticity, but then the inner, near-waU vortex has lower vorticity that further downstream again becomes the same as that of the vortices shedding from the upper side. It should be noted that the differences in I( w)pl between inner and outer vortices are no larger than the experimental uncertainty so that not much weight can be given to them. It should also be mentioned, however, that, in the two-cylinder case of Kolar et al., owing to interaction between the inner vortices shedding from the two cylinders, the vorticity of the inner vortices was found

to be consistently lower, apart from the base region. The numerical calculations of Bosch and Rodi [14] have shown that the wall influences both inner and outer vortex structures. Portions of the wall boundary layer with vorticity of clockwise rotation are sucked into the region with strong upward motion between the counterrotating shed vortices, thereby weakening the lower vortices with anticlockwise rotation. These portions are then engulfed into the upper vortices, thereby strengthening them. Unfortunately, these findings cannot be confirmed directly by the present experiments. Contours of the periodic stress components t~fi and 55 are presented in Fig. 10 for the four phases for which visualization pictures and velocity and vorticity results are given in Fig. 7. These stresses characterize the strength of the large-scale periodic motion, and their time-averaged values appear in the time-averaged momentum equations in addition to the turbulent stresses. In many respects, the behavior portrayed in Fig. 10 is similar to that in the free-standing cylinder [11] and even more so to that of the

302

G. Bosch et al.

b)
4

vorticitypeaks: outer "0


inner 4vortex entres: outer inner -~-1

..-

inner and outer peaks

0.5
0

9 10 0 0 ~

II 0

0 + 7

3.5 3 2.5

O Lyn et al. + slopeis 0.8 ........


+....."6

.'~:"'+'7 ...q_

0
1

+ 4- + 3 4
2

-0.5 -1

4. 5

+ + 0 2 0
1

4.

+ ..-"5 .." ..."


4

0 3
i i i

|,5

0.5

1.5

2
X

2.5

3.5

0.5

1.5

2.5 t / St

3.5

4.5

F i g u r e 8. (a) Geometrical position of vorticity peaks (numbers denote phase position, i.e., phase angle/30 ) and vortex centers (from flow visualization). (b) Movement of vorticity peaks in mean-flow direction (peaks at the boundaries of the measurement region have been omitted); comparison with free-standing cylinder results of Lyn et al. [11].

flow around two cylinders [12]. The values of the periodic stresses are generally small in the region of the vortex centers, because in the vortex frame there is no motion at the centers. These centers move along trajectories that are initially close to the center line of the cylinder but then deviate upward from this line further downstream, and the streamwise periodic fluctuations are always small near these trajectories and hence also near the center line. Below the vortex trajectories are inner maxima and above o u t e r maxima of tiff, and the outer ones are generally located near the o u t e r saddles; on the other hand, the inner maxima are found between the two saddles at the foot of the vortices shedding from the u p p e r side. A similar behavior was observed by Kolar et al. [12] for the flow past two cylinders. However, in the present case, the trajectories of the tiff peaks are m o r e inclined; they move away from the wall, which is in line with the general trend in this case of a somewhat oblique wake. Concerning the 6 5
'I~
4 outer peaks O inner peaks + Lyn et al. []

,~

[]

+ []+ [] o

3 2
1 0
I

rn

[]

1.5

2.5
X

3.5

F i g u r e 9. Decay of peak vorticity with x; comparison with

free-standing cylinder results of Lyn et al. [11].

tiff stress, it is further worth noting that there are fairly strong periodic u fluctuations in the separated shear layer on the upper side of the cylinder, whereas in the gap on the lower side there are only very small periodic fluctuations. As mentioned before, here the shear layer is suppressed and the influence of the shedding motion does not extend much into the gap. The tSt5 maxima lie roughly on the trajectories of the vortex centers because the strongest cross-stream motion occurs in this region between the counterrotating vortices. This means that the peaks in ~ occur between the centers of vortices. The peak values, of course, decay in the downstream direction. Finally, it should be mentioned that the frequency of the periodic stresses is twice that of the vortex shedding. Contours of the turbulent stress components (u'u') and (L"t,,') are shown for the same phases in Fig. 11. Much less structure can be found in these contours than in the corresponding ones given by Lyn et al. [11] for the free-standing cylinder and by Kolar et al. [12] for the two-cylinder case. In particular, the sinuous behavior following the shedding motion, which is rather characteristic of these two experiments, is not present in the (u'u')contours at all and only weakly so in the (t.,'u')-contours. O n e reason for the more irregular behavior of the turbulent stress components may be the coarse measurement grid, but there also seems to be more diffusion of turbulence at the edges of the wake and the turbulent flow altogether. Of course, one boundary of the flow is a turbulent boundary layer, and the reflection of vortices at the wall may induce additional turbulence in the near-wall region. Also, as could already be seen in the visualization pictures, the separated shear layer on the outer side of the cylinder is wider, so the turbulent region is wider from the beginning. The strong similarity between the (u'u') and (c,'t~') contours mentioned in Lyn et al. [11] for the free-standing cylinder also is not obvious in the present case; but, in agreement with the observations of Lyn et al., the v fluctutations reach higher levels than do the u fluctuations; however, the latter ones are significantly

Flow Past a Square Cylinder near a Wall 303


UU 0.22 0,2 0.18 0.16 0.14 0.12 0,1 0.09 0.08 0.07 0.06 0,05 0.04 0.03 0.02 0.01 0.22 0.2 0.18 0.16 0.14 0.12 0,1 0.09 O.Og 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.22 O.2 0.18 0.16 0.14 0.12 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.16 0.14 0.12 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01
m 0.6

i373 0.55 . . . . 0.5 ........ 0.45 - 0.4-0.35 ....... 0.3 . . . . . 0.25 ....... 0.2 . . . . . 0.15 - 0.1---0.05 ........

0.5 0 -0.5 -I

!41\",.

i/.<3 3.5 4

0.5
m

1.5

2.5

0
0.6 n

0.5

1.5

2.5

3.5

i m

B m

5"~
..... " . ".'~i

:
//

0.55 . . . . 0.5 ........ 0.45-0.4---0.35 ....... 0.3 ........ 0.25 . . . . . 0.2 . . . . . . 0.15 - 0.1 . . . . . 0.05 ........

... .................. ",..

..J .i ".~ .~ ~ , ~.- #/~." _-~_ 22~71J

0.5

-0.5

"..........."

0.5
m m

1.5

2.5

3.5

4
0.6 m

0 0.55 - 0.5 ........


0.45 1

0.5

1.5

2.5

3.5

m m

,,.,,,

.<.

, .......

0.4-0.35 ....... 0.3 ...... 0.25 . . . . . 0,2 . . . . . . 0.15 0.1--0.05 ........

0.5

-0.5

. ~ . ~ . ~ . ~ . ~ . ~ . ~ . ~
0.5
m

1.5

2.5

3.5

4
0.6 - -

0.5

1.5

2.5

3.5

_m_

m m

0.55 . . . . . 0.5 ........ 0.45-0.4-0.35 ....... 0.3 ...... 0.25 . . . . . . 0.2 ........ 0.15 - 0.1 - - ~ 0.05 ........

" Z 2 ....

---

"~."

"'2 . i ~

"0.5

//II i
-0.5

0.5

1.5

2.5

3.5

0.5

1.5

2.5

3.5

Figure 10. Periodic normal stresses ~t~ and ~ at different phases. higher in the more downstream region x = 3 - 4 than in the free-standing cylinder experiment. Possible reasons for this have been given already. Attempts to produce contours of the production terms of the turbulent stresses were unfortunately not very successful, owing to the relatively coarse measurement grid and the uncertainties in the measurements, particularly in ( u ' v ' ) .
PRACTICAL SIGNIFICANCE

practical situations. Further, in practice, there is a great need to predict the flow past cylinders and related phenomena, and the experimental data presented will allow detailed testing and validation of calculation procedures and, in particular, turbulence models. Because of the presence of stagnation and separation regions and the importance of periodic fluctuations, the flow studied poses a particularly challenging test for any general-purpose turbulence model.
CONCLUSIONS

The practical importance of vortex-shedding flow past cylinders under the influence of wall proximity was stated in the "Introduction." For the square cylinder, the experimental study reported here provides detailed information that may be of help in controlling vortex shedding in

The flow past a square cylinder placed in the vicinity of a wall was studied with visualization techniques for various gap widths, and detailed phase-averaged velocity statistics

304

G. Bosch et al.

ttttt t
0.3

( Ut 73t)
0.55 0.5 .... 0 . 4 5 ........ 0.4-0.35 - 0 . 3 ....... 0.25 ..... 0.2 ...... 0.15 ...... 0.1 - -

0.275 - 0 . 2 5 ........ 0.225-0.2-~ 0 . 1 7 5 ....... 0.15 .... 0.125 --0.1 .... 0.075 0.05~

0.5

T,,

-0.5

0 0.3 m 0.275 ---0 . 2 5 ........ 0 . ~ 0.2-0 . 1 7 5 ....... 0.15 .....


0.125 0.1 0.075 ..... ......

0.5

1.5

2.5

3.5

4 0.55 - 0.5----0 . 4 5 ........ 0.4-0.35 - 0 . 3 ....... 0.25 ..... 0 . 2 ........ 0.15 ...... 0.1 - -

0.5

1.5

2.5

3.5

0.5

-0.5

0.05--

0.5 0.3 m 0.275 - 0.25 ........ 0 . ~ - 0.2--0 . 1 7 5 ....... 0.15 ..... 0.125 ..... 0 , 1 ....... 0.075 0.05 ....

1.5

2.5

3.5

4 0.55 - 0.5 ...... 0 . 4 5 ........ 0.4-0.35 - 0 . 3 ....... 0.25 ..... 0.2 ..... 0 . 1 5 ........ 0.1 - -

0.5

1.5

2.5

3.5

"-".-~.Z-'~,.~
.... ~,~

",, ~
....... z

:,,"

"

1 ]

, : :,
~ "

/ {

..4r,..

".,.

0.5

....

/"

......... ! - ,

V///////,'A

?: ~~ ,
[ ,.~f / :~

,';-Y~
.......... -0.5

,....; ................. ',-.


~..--.--~,.'-~,.-.,
,.,., ~

......... --i12
.

~2"-:-~S.:'~

"-:: . . . . . . . . . . . . . . . 72" - - -1

0 0.3 i 0.275 ..... 0 . 2 5 ........ 0.225-0.2-0 . 1 7 5 ....... 0.15 .... 0.125 ..... 0.1 ..... 0.075 0.05---

0.5

1.5

2.5

3.5

4 0.55 - 0.5 ...... 0 . 4 5 ........ 0.4-0 . 3 5 ........ 0.25 ..... 0.2 ...... 0 . 1 5 ....... 0.1 - -

0.5

1.5

2.5

3.5

0.5

0.3 .......

-0.5

-1

0.5

1.5

2.5

3.5

0.5

1.5

2.5

3.5

Figure 11. Turbulent normal stresses ( u ' u ' ) and ( u ' v ' ) at different phases.

were obtained from t w o - c o m p o n e n t L D V m e a s u r e m e n t s for the gap width G / D = 0.75. The visualization study yielded, in a g r e e m e n t with previous experiments, that the critical gap width below which shedding is suppressed by the presence of the wall is roughly in the range G / D = 0.35-0.5. However, the experiments have also shown that there is not a single critical width but a transition over this range in which the fraction of the time when periodic shedding occurs changes from zero to one. In between, the flow has a very different, nonperiodic character with a large reeirculation zone that also prevails, at all times, when the gap width is below the given range.

The detailed measurements at G / D = 0.75 were comp a r e d directly with those for the free-standing cylinder r e p o r t e d in Lyn et al. [11] to see the influence of the wall. Comparisons were also made with the measurements of Kolar et al. [12] for the flow around two side-by-side cylinders with shedding motion that is symmetrical with respect to the center line midway between the two cylinders. The Strouhal n u m b e r was found to be = 0.14, which is in close agreement with previous square-cylinder measurements, for cases both with and without the influence of a wall. The flow examined in detail has many similarities with the flow past a free-standing cylinder, particularly

Flow Past a Square Cylinder near a Wall in the base region near the cylinder, but the wall causes asymmetric behavior of the flow: the inner vortices shedding from the lower side get elongated and the vortex trajectories move away from the wall, so the whole wake becomes oblique. Also, the separated shear layers on either side of the cylinder behave very differently: the one in the gap gets squeezed, whereas the outer one is considerably wider than in the free-standing-cylinder case. The flow topology was found to be similar to that in the two-cylinder case; in particular, two saddle points were discerned at the foot of the outer vortices stemming from the upper side. O n the other hand, the faster decay of vorticity of the i n n e r vortices compared with that of the outer vortices observed in the two-cylinder case could not be confirmed with certainty. The m e a s u r e m e n t s have shown a longer m e a n recirculation zone and downstream of this a faster reduction of the deficit velocity than in the free-standing cylinder case. They have also yielded considerably higher u fluctuations in the wake, with a strongly asymmetric distribution; the main contribution to these high values comes from the turbulent fluctuations. Possible reasons for the high values were the higher velocity gradients and shear-stress values in the near-wall part of the wake, leading to increased production of u fluctuations and the engulfment of flow carrying turbulence from the b o u n d a r y layer into the wake. In contrast, the v fluctuations show a similar behavior to that in the free-standing cylinder case; they are considerably higher than the u fluctuations, and about half of the contribution comes from the periodic fluctuations. The detailed time-mean and phase-averaged data reported here provide a good basis for a thorough testing of calculation methods, whether large-eddy-simulation or statistical methods, and have been used for this purpose already by Bosch and Rodi [14]. The data can be obtained in machine-readable form from the latter author. The experiments were financially supported by the German Research Foundation through the Sonderforschungsbereich 210. The help of Mr. Bierwirth in constructing the rig and carrying out the visualization experiments is gratefully acknowledged.

305

Vc
U

X, y 8
ok

convection speed, here streamwise velocity of vorticity peaks, m / s velocity c o m p o n e n t in lateral (y) direction, m / s coordinates, m Greek Symbols boundary-layer thickness, m 2-D vorticity, 1 / s vorticity peak value at a certain phase REFERENCES

( ,,,Sp

NOMENCLATURE D f(t) (f) cylinder diameter D = 0.04 m (see Fig. 1), m a measured instantaneous quantity phase or ensemble average long time average periodic c o m p o n e n t of fluctuation stochastic turbulent c o m p o n e n t of fluctuation total fluctuation (i.e., f ' + f ) gap width (see Fig. 1), m Reynolds n u m b e r , Re = UrefD//~' , dimensionless Strouhal number, St = f O / u r e f , dimensionless turbulence level, T u = V / - ~ / u r c f , dimensionless average approach-flow velocity used as reference velocity, m / s velocity c o m p o n e n t in m e a n flow (x) direction, m / s

/ f

f'
f,t G Re St

Tu
L/re f

1. G6ktun, S., The Drag and Lift Characteristics of a Cylinder Placed near a Plane Surface, M.Sc. Thesis, Naval Postgrad. School, Monterey, CA., 1975. 2. Bearman, P. W., and Zdravkovich, M. M., Flow Around a Circular Cylinder near a Plane Boundary. J. Fluid Mech. 89 (1) 33-47, 1978. 3. Buresti, G., and Lanciotti, A., Vortex Shedding from Smooth and Roughened Cylinders in Cross-Flow near a Plane Surface. Aeronaut. Qtly. 30, 305-321, 1979. 4. Buresti, G., and Lanciotti, A., Mean and Fluctuating Forces on a Circular Cylinder in Cross-Flow near a Plane Surface. J. Wind Eng. Industr. Aerod. 41, 639-650, 1992. 5. Grass, A. J., Raven, P. W. J., Stuart, R. J., and Bray, J. A., The Influence of Boundary Layer Velocity Gradients and Bed Proximity on Vortex Shedding from Free Spanning Pipelines. ASME J. Energy Res. Tech. 106, 70-78, 1984. 6. Taniguchi, S., and Miyakoshi, K., Fluctuating Fluid Forces Acting on a Circular Cylinder and Interference with a Plane Wall. Exp. Fluids, 9, 197-204, 1990. 7. Taniguchi, S., Miyakoshi, K., and Dohda, S., Interference between Plane Wall and Two-Dimensional Rectangular Cylinder. Trans. JSME 49-447, 2522-2529, 1983. 8. Durao, D. F. G., Gouveia, P. S. T., and Pereira, J. C. F., Velocity Characteristics of the Flow around a Square Cross Section Cylinder Placed near a Channel Wall. Exp. Fluids, 11, 298-304, 1991. 9. Devarakonda, R., and Humphrey, J. A. C., Turbulent Flow in the near Wakes of Single and Tandem Prisms, in Proc. Turbulent Shear Flows 10, Pennsylvania State University, University Park, Pennsylvania, USA, 1995. 10. Lyn, D. A., and Rodi, W., The Flapping Shear Layer Formed by Flow Separation from the Forward Corner of a Square Cylinder. J. Fluid Mech. 267, 353-376, 1994. 11. Lyn, D. A., Einav, S., Rodi, W., and Park, J.-H., A Laser-Doppler Velocimetry Study of Ensemble-Averaged Characteristics of the Turbulent Near-Wake of a Square Cylinder. J. Fluid Mech. 304, 285-319, 1995. 12. Kolar, V., Lyn, D. A., and Rodi, W., Ensemble-Averaged Measurements in the Turbulent Near-Wake of Two Side-by-Side Square Cylinders, to be submitted, 1996. 13. Bosch, G., ExperimenteUe und theoretische Untersuchung der instation~iren Str6mung um zylindrische Strukturen, Ph.D. Thesis, Univ. Karlsruhe, Karlsruhe, Germany, 1995. 14. Bosch, G., and Rodi, W., Simulation of Vortex Shedding Past a Square Cylinder near a Wall. Int. J. Heat Fluid Flow, in press. 15. Kappler, M. Experimentelle Untersuchung der instation~irenturbulenten Wirbelabl6sung an einem quadratischen Zylinder in einer Scherstr6mung und in Wandn~ihe, Diploma Thesis, Univ. Karlsruhe, Karlsruhe, Germany, 1995.

Received March 27, 1996; revised June 14, 1996

You might also like